University of Southern California Dissertations and Theses

Understanding perceptual processes in metaphor comprehension: specificity and context-dependence

(USC Thesis Other)

Understanding perceptual processes in metaphor comprehension: specificity and context-dependence

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 1 Understanding Perceptual Processes in Metaphor Comprehension: Specificity and Context-Dependence by Vesna Eliza Gamez-Djokic A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Neuroscience December, 2016 BRAIN AND CREATIVITY INSTITUTE DEPARTMENT OF NEUROSCIENCE FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA 2 Professor Lisa Aziz-Zadeh Vesna Gamez-Djokic Understanding Perceptual Processes in Metaphor Comprehension: Specificity and Context-Dependence Abstract A growing number of behavioral and neuroscientific studies have provided evidence in support of theories that stipulate that conceptual representations (both concrete and abstract concepts) are grounded in perceptual systems of the brain (Barsalou, 1999, 2008; Gallese & Lakoff, 2005; Niedenthal, Barsalou, Winkielman, Krauth-Gruber, & Ric, 2005). These theories have challenged more traditional views that typically have described conceptual processing in terms of computations on amodal symbols (Fodor, 1975; Pylyshyn, 1984, 2007). However, on closer inspection this empirical work has raised questions regarding the exact functional role that embodied (and symbolic) processes play in semantic representations. This question is especially relevant to abstract ideas that do not have obvious links to direct experiential information (i.e., morality). In this way, studies of metaphor comprehension offer a unique opportunity to understand the role of sensorimotor and affective processes in language comprehension as they often involve an abstract idea (e.g., immorality) that is expressed in terms of a sensorimotor domain using concrete language (e.g., ‘that was a rotten thing to do’), yet it is debated whether this should necessarily involve a strict literal interpretation of events ( Bowdle & Gentner, 2005; Keysar & Bly, 1999; Keysar, Shen, Glucksberg, & Horton, 2000; Lakoff & Johnson, 1980; Lakoff, 2014). The studies presented here, thus, seek to investigate if sensorimotor and affective processes are engaged during metaphor comprehension, as well as, further characterize the extent to which sensorimotor and affective activations in metaphor processing are 1) dependent on contextual factors (i.e., task and individual differences) or instead are automatic; and 2) reflective of ‘low-level’ (e.g., action kinematics) and/or ‘higher-level’ (e.g., action goals/outcomes) representations in the brain that putatively underlie distinct levels of sensorimotor specificity (e.g., distinct levels of motor hierarchy) relevant to the their meaning. Study 1 provides initial evidence that processing of disgust and fear metaphor recruit affective and sensorimotor systems of the brain that process features relevant to that affective experience (i.e., perceptual, internal states, and behavioral features). Novel disgust metaphors drawing on physical disgust language (i.e., pathogen-based disgust) compared to non-affective metaphor (i.e., motion-related) engaged areas of primary/secondary gustatory cortex (i.e., anterior insula/frontal operculum and OFC) and subregions of the basal ganglia (i.e., pallidum). Fear language stimuli mainly differed from disgust language stimuli in that they recruited the posterior insula (but also dorsal anterior insula for metaphor) and parietal/sensorimotor areas to a greater degree. However, disgust language stimuli mainly showed increased activity in subcortical brain regions (i.e., amygdala, basal ganglia) compared to fear language stimuli, but not within the anterior insula/frontal operculum. Similar results were found in our cross-modal 3 multivariate analysis, revealing significant classification accuracies within the anterior insula/frontal operculum for disgust versus neutral stimuli, but not when distinguishing disgust versus fear stimuli. Thus, we did not find dedicated brain regions specific to either emotion (i.e., anterior insula for disgust). Study 2, provides neural evidence that the processing of familiar moral disgust metaphors (both during reading and judgment) engage emotion-related brain regions relevant to physical disgust processing (i.e., areas of primary/secondary gustatory cortex and basal ganglia) to a greater extent than their literal counterparts matched in arousal/valence, but in a context-dependent manner. First, activity within emotion-related brain regions implicated in disgust processing significantly covaried with political orientation during the reading-period for moral disgust metaphor when contrasted with their literal counterparts, suggestive of the fact that embodiment effects are sensitive to individual differences. Secondly, emotion-related brain regions implicated in disgust processing were mainly seen during judgment across all participants irrespective of political orientation, pointing to the fact that embodiment effects may be modulated by the depth of processing required by the task (i.e., when a response is required). Lastly, literal paraphrases showed increased activity in brain areas associated with more deliberate, top-down processes in moral decision-making in the brain (i.e., VMPFC and DLPFC) when compared to moral disgust metaphor (both during reading and judgment), adding support to the idea that moral disgust metaphor may lead to differential moral processing at the neural level. Lastly, Study 3 shows that two families of familiar hand-action metaphors that differ in their encoded force-exchange patterns (e.g., ‘pull-type’ vs. ‘push-type’ metaphors) could be distinguished in a multivariate analysis using patterns of activity across voxels mainly within sensorimotor brain areas. These sensorimotor brain areas partially overlapped anatomically with activations of sensorimotor areas recruited during the univariate analysis of a motor localizer task (i.e., pulling and pushing actions involving an object vs. rest). The results show the importance of distinct levels of motor hierarchy in action execution, revealing significant classification accuracies both within primary motor areas (i.e., coding low- level action kinematics) and putatively ‘higher-level motor areas’ the inferior parietal lobule (IPL) (i.e., coding action goals/outcomes), revealing an unexpected level of sensorimotor specificity. Moreover, the highest classification accuracies and overlap with the motor task occurred within secondary somatosensory areas, suggesting a special role for somatosensory simulation. Nevertheless, significant voxels within other brain regions in the multivariate analysis do not exclude a role for other (supramodal) brain regions in the encoding of force exchange patterns in metaphor. Taken together these studies point to the involvement of sensorimotor and affective systems of the brain in the processing of metaphoric source domains. They further suggest that 1) sensorimotor and affective simulations during metaphor comprehension are context-dependent (i.e., sensitive to individual differences and depth of processing required by the task) and 2) point to the need for multiple levels of sensorimotor specificity (i.e., abstraction) along with other (supramodal) brain regions to flexibly accommodate processes of inference during metaphor comprehension. 4 Table of Contents Title… ...................................................................................................................... 1 Abstract .................................................................................................................... 2 Introduction .............................................................................................................. 5 Neural underpinnings of metaphors implying ‘moral disgust’ ............................... 28 Metaphor in politics: bringing affect to the decision space? .................................. 57 Distinguishing metaphors that differ in their force exchange patterns ................... 89 Discussion ............................................................................................................. 111 5 Introduction The use of metaphor occurs quiet frequently in typical discourse with some estimates suggesting that language users take advantage of 50 metaphors per every 1,000 words (Cameron, 2008). Our experience with metaphor in language suggests that one function of metaphor is to help re-conceptualize an abstract idea into a potentially easier to understand concrete domain (e.g., “she had a rough day”). According to one such influential proposal, conceptual metaphor theory (CMT), abstract domains (target domain) are actually thought to be structured in part through links with concrete sources (source domain) that co-occur during development, thereby, grounding abstract thought in sensorimotor and affective experience (Lakoff, 2014; Lakoff & Johnson, 1980). In support of this view, accumulating behavioral evidence suggests that much of thought may be metaphorical in nature (Gibbs, 2006; Kovecses, 2008; Lakoff, 1996; Lakoff & Johnson, 1980). This theory is in line, more generally, with embodied cognition theories that have argued that conceptual representations (both concrete and abstract concepts) are grounded in our physical experiences and involve ‘embodied simulations’ (e.g., the (re)-activation of perceptual mechanisms in part used for action recognition, execution, or motor imagery) (Barsalou, 1999, 2003; Barsalou, Simmons, Barbey, & Wilson, 2003; Gallese & Lakoff, 2005; Glenberg & Kaschak, 2002). An important prediction that follows from conceptual metaphor theory, is that metaphor comprehension should depend on the same types of sensorimotor and affective processes in the brain that drive inferences in the source domain so that they can be partially ‘re-used’ to make complex inferences in the target domain (Gibbs, 2006; Lakoff, 2014; Lakoff & Johnson, 1980). That is, “understanding metaphorical expressions like ‘grasp a concept’ or ‘get over an emotion’ involve simulating what it must be like to engage in these specific activities…”(Gibbs, 2006). Others, however, have argued that sensorimotor simulation is not necessary for metaphor comprehension as concrete language in metaphor could rather be processed through a categorization process (Bowdle & Gentner, 2005; Keysar & Bly, 1999; Keysar et al., 2000), whereby, for example, ‘sharks’ in the metaphor ‘some lawyers are sharks’ is immediately processed as belonging to the ‘abstract superordinate category’ or lexicalized metaphoric category ‘ruthless predator’ (Glucksberg, 1999, 2002). These categorization views of metaphor comprehension are more compatible with ‘disembodied’ views that find that abstract conceptual processing mainly involves computations on amodal symbols divorced from sensorimotor and affective systems of the brain. According to such views, sensorimotor activation during metaphor comprehension would most likely be a result of either post-comprehension imagery or spreading activation from the literal processing of the word ‘shark’, but would be largely epiphenomenal bearing little on the metaphoric interpretation (Mahon & Caramazza, 2008). Neuroscientific studies on metaphor, thus, present a unique window into understanding the role of sensorimotor and affective processing in language comprehension and abstract concept processing, as they involve an abstract idea (target domain) that is construed in terms of a concrete domain (source domain), yet it is currently debated whether or not this necessarily involves a direct comparison of source domain with target domain that entails sensorimotor simulation (Bowdle & Gentner, 2005; Keysar & Bly, 1999; Keysar et al., 2000). 6 Although a number of neuroscientific studies have provided some support for the idea that words and literal sentences relating to concrete sources activate sensorimotor representations in the brain, for reviews see (Hauk & Tschentscher, 2013; Meteyard, Cuadrado, Bahrami, & Vigliocco, 2012), studies looking at figurative language have been fewer and have yielded more mixed results, for a review see (Carr, Kavanagh, & Bergen, 2013). These findings have naturally cast doubt on whether or not sensorimotor activations should always be automatically activated for action verbs irrespective of the linguistic context (Desai, Binder, Conant, Mano, & Seidenberg, 2011; Desai, Conant, Binder, Park, & Seidenberg, 2013), both when action verbs are embedded in sentences that describe physical actions (e.g., ‘The teacher was ambling toward the school’) and when action verbs are used to express abstract ideas via analogy in a metaphoric context (e.g., ‘The story was ambling towards its conclusion’) (Troyer, Curley, Miller, Saygin, & Bergen, 2014). This echoes more recent findings that have questioned whether sensorimotor and affective activations during literal language processing are automatic or rather sensitive to contextual factors, more generally, for reviews see (Willems & Casasanto, 2011a; Yang, 2014). Lastly, it has been proposed that sensorimotor and affective representations in metaphor comprehension may reflect ‘more general or schematized representations’ (Troyer et al., 2014). This parallels recent evidence that suggests that embodied processes even during literal language processing may not need to always rely on ‘full-blow simulations’ involving modality- preferential representations, but perhaps need only involve secondary/multimodal representations (Barsalou et al., 2003; Fernandino et al., 2016; Simmons & Barsalou, 2003). For example, current controversy surrounds whether or not motor-related features of action verb meanings (e.g., pluck, grasp, pick) depend on the automatic activation of motor programs at distinct levels of motor hierarchy, either primarily low-level motor representations that process action kinematics (i.e., primary motor areas) and/or higher-level action representations that process action goals/outcomes (i.e., inferior parietal lobule (IPL)), irrespective of task or context (e.g., literal or metaphoric context) (Kemmerer, 2015; Spunt, Kemmerer, & Adolphs, 2015). Therefore, the pertinent questions currently in the neuroscience of embodied semantics and metaphor comprehension involve not only asking whether or not sensorimotor and affective processes are engaged during metaphor comprehension, but going beyond, to also provide an initial characterization of the context-dependent nature of sensorimotor and affective processes and, relatedly, understanding the level of sensorimotor specificity required (e.g., low-level action kinematics vs. higher-level action goals/outcomes). This, it would seem, is a precursor to understanding how sensorimotor and affective processes can make functional contributions to metaphor comprehension and language processing, more generally. Neural evidence for modality-preferential representations in semantic processing The notion that ‘modality-preferential’ (but not necessarily unimodal) representations relating to action and perception in the brain contribute to semantic processing first came from findings of category-specific semantic impairments in patients with lesions encompassing mainly modality-preferential areas of the brain (Warrington & McCarthy, 1983; Warrington & McCarthy, 1987; Warrington & Shallice, 1984). Specifically, category-specific 7 semantic impairments were found for words relating to living things (i.e., animal and foods) and non-living things (i.e., manipulable objects), which led Warrington and colleagues to conjecture that ‘differential weighting’ of visual information for biological entities and ‘functional attributes’ or affordances for non-living objects reflecting actual physical experience was possibly at the root cause of these semantic impairments (i.e., differential weighting hypothesis). Further studies extended these findings showing that deficits in words relating to action (Arevalo, Baldo, & Dronkers, 2012; Bak, O'Donovan, Xuereb, Boniface, & Hodges, 2001; Damasio & Tranel, 1993; Dreyer et al., 2015; Kemmerer, Rudrauf, Manzel, & Tranel, 2012; Neininger & Pulvermuller, 2001, 2003), audition (Bonner & Grossman, 2012; Trumpp, Kliese, Hoenig, Haarmeier, & Kiefer, 2013), and visual (Gainotti, 2010; Pulvermuller et al., 2010) semantics were selectively impaired when the lesion site corresponded to brain areas associated with the relevant motor, auditory, and visual processes involved in primary experience, respectively, for a review see (Barsalou, 2008; Gainotti, 2015). For example, more recently, Dreyer et al., 2015, reported evidence of impairments with tool-related words on a lexical decision task in a patient with a focal lesion to the ‘dorsolateral central sensorimotor system’. Despite variability of lesion sites and differences in methodology across studies, for critical reviews see (Caramazza & Mahon, 2003; Mahon, Anzellotti, Schwarzbach, Zampini, & Caramazza, 2009; Mahon & Caramazza, 2008, 2009), evidence of category-specific semantic impairments mainly seem to support sensorimotor models of semantic knowledge, for a review see (Gainotti 2015), which posit that conceptual knowledge and organization depend in part on the same perceptual mechanisms that were used to acquire them during actual sensorimotor/affective experience (Barsalou, 1999, 2008; Gallese & Lakoff, 2005). Further causal support for sensorimotor models of semantic knowledge can be found in patients with direct physical motor impairments brought on as a result of disease (i.e., Parkinson’s and Motor Neuron Disease) that present with deficits in action verb processing (Bak & Chandran, 2012; Bak et al., 2001; Boulenger et al., 2008; Cotelli et al., 2007). For instance, patients with Parkinson disease show selective difficulty naming actions as opposed to nouns (Cotelli et al., 2007). On the other hand, Parkinson patients off their meds, resulting in inhibition of motor areas, show a decreased masked priming effect when performing a combined masked priming and lexical decision task, but only for action verbs and not concrete nouns (Boulenger et al., 2008), for critiques see (Caramazza & Mahon, 2003; Mahon & Caramazza, 2008). Critically, for those with Motor Neuron Disease the degree of motor deficit often directly relates to difficulties with action verb processing, although the diffuse nature of brain damage in diseased populations does not preclude a role for other non-motor brain areas, for a review see (Bak & Chandran, 2012). Category-specific impairments found in lesion/disease populations also tend to support the notion that these deficits are related primarily to conceptual impairments rather than differences processing nouns vs. verbs (i.e., grammatical differences), for a review see (Kemmerer et al., 2012). Direct stimulation (TMS, rTMS, tDCS) studies additionally provide causal evidence that sensorimotor processes are involved in actual physical experience are linked to semantic processing in language. Buccino et al., 2005 used transcranial magnetic stimulation (TMS) to stimulate motor areas relevant for hand or leg actions somatotopically while participants read 8 hand and leg related sentences (Buccino et al., 2005). Stimulation of the hand or leg area in motor cortex corresponded to modulation of recorded motor evoked potentials (MEPs) in that effector but only when the read sentence (hand or leg related) matched the stimulated effector. In a similar vein, Pulvermuller et al., 2005 found that weak TMS pulses to either the leg or hand areas of motor cortex correspondingly affected reaction times in a lexical decision task involving hand or leg related words, such that stimulation of the matching effector led to a decrease in reaction time for lexical decisions referring to that effector (Pulvermuller, Hauk, Nikulin, & Ilmoniemi, 2005). Schomers et al., 2014, more recently, showed that TMS applied to the primary motor cortex (either the lip or tongue area) selectively delayed responses when participants processed words (e.g., pool vs. tool) relying on that particular articulatory effector during a word-picture matching task, suggesting that articulatory motor areas play a causal role in semantic processing (Schomers, Kirilina, Weigand, Bajbouj, & Pulvermuller, 2015). Lastly, Liuzzi et al 2010 found that stimulation of motor cortex (precentral/premotor areas) could interfere with learning of novel action words, adding to idea that the representations of action- related words involves sensorimotor systems of the brain (Liuzzi et al., 2010). Importantly, a number of studies show activation of modality-preferential representations in the brain during the processing of words and literal expressions about that modality. In the motor domain, neuroscientific studies (EEG, fMRI) have found evidence that action words (Hauk, Johnsrude, & Pulvermuller, 2004; Pulvermuller et al., 2005) and literal action phrases and sentences (Aziz-Zadeh, Wilson, Rizzolatti, & Iacoboni, 2006; Tettamanti et al., 2005) engage areas of motor cortex, showing somatotopic activation along the premotor strip in some cases but not all (Postle, McMahon, Ashton, Meredith, & de Zubicaray, 2008). Hauk et al 2004 found that action-related verbs involving the arm and feet activated premotor cortex in a somatotopic fashion corresponding to patterns of activation seen when subjects performed actions using each effector in a separate localizer task. Similarly, in the visual-motion domain, words (Kable, Kan, Wilson, Thompson-Schill, & Chatterjee, 2005; Kemmerer, Castillo, Talavage, Patterson, & Wiley, 2008) and literal expressions (Desai et al., 2011; Saygin, McCullough, Alac, & Emmorey, 2010) implying direct motion activate areas involved in motion processing, but see (Bedny & Caramazza, 2011; Dravida, Saxe, & Bedny, 2013). For example, Saygin et al., 2010 found increased activation in visual motion area MT for phrases such as ‘The wild horse crossed the barren field’ compared to sentences with little implied motion such as ‘The black horse stood in the barren field’. Although most empirical studies focus on the sensorimotor domain, a few other studies have shown similar embodiment effects in other sensory domains: Aziz-Zadeh et al., 2008 found that activity in the fusiform face area (FFA) and parahippocampal place area (PPA) were modulated depending on whether the sentences related to faces or places, respectively (Aziz-Zadeh et al., 2008b). As for olfactory and gustatory domains, Gonzalez et al 2006 found that scent-related words (e.g., garlic, cinnamon) activated olfactory cortex, while Barros- Loscertales et al 2012 found activation of primary and secondary gustatory cortices for taste- related words such as salt (Barros-Loscertales et al., 2012; Gonzalez et al., 2006). In the auditory domain, Kiefer et al., 2008 found that words (e.g., bells) relating to sounds activated the auditory cortices (Kiefer, Sim, Herrnberger, Grothe, & Hoenig, 2008). Lastly, although the extent to which basic emotions map onto distinct neural networks is still disputed, Ponz et al 2013 found that intracranial EEG recordings in the left anterior insula implicated in disgust processing, are 9 sensitive to disgust words as opposed to neutral words (Calder, Keane, Manes, Antoun, & Young, 2000; Ponz et al., 2014), for other emotion words see (Citron, 2012). While there is a growing amount of evidence in favor of the idea that sensorimotor and affective processes make functional contributions to semantic processing, there have also been a few studies that call into question whether these contributions are necessary and/or sufficient. For example, a few additional studies show evidence for double-dissociations, mainly patients with motor-related impairments who do not show the corresponding action-related semantic impairments. For example, apraxic individuals that have difficulty pantomiming or imitating object use do not always show comparable deficits in processing the meaning of object-related words (Buxbaum, Johnson, & Bartlett, 2002; Buxbaum & Saffran, 2002; Buxbaum, Sirigu, Schwartz, & Klatzky, 2003; Buxbaum, Veramonti, & Schwartz, 2000; Negri et al., 2007; Rosci, Chiesa, Laiacona, & Capitani, 2003). Similarly, Garcea et al., 2013 showed evidence of a patient with a lesion in the left hemisphere who had pronounced difficulties performing actions but did not have similar deficits processing action-related concepts (Garcea, Dombovy, & Mahon, 2013). Lastly, the extent to which specific sensorimotor and affective processes involved in actual physical experience actually contribute to semantic processing has been questioned by Arevalo et al., 2012, who failed to find a somatotopic-type effect of lesions sites on the processing of action words relating to the hand, mouth, and foot (Arevalo et al., 2012). Although, they did find impairments for the processing of foot-related action-words across all the lesion sites investigated. Delineating the limits and extent of the role of sensorimotor and affective systems in semantic processing is particularly relevant to the discussion of abstract concepts, as they lack obvious links to sensorimotor experience. Therefore, abstract concepts have typically presented as a challenge to theories of embodied semantics. Some initial evidence suggests that abstract concepts may be grounded in more diffuse connections with sensorimotor and affective systems of the brain drawing on multimodal situated simulations and/or affective states to a greater degree than concrete concepts (Barsalou, 2003; Kousta, Vigliocco, Vinson, Andrews, & Del Campo, 2011; Moseley et al., 2015; Vigliocco et al., 2014). Importantly, a recent study by Dreyer et al., 2015, showed evidence of a patient with a focal lesion to the left supplementary motor cortex who exhibited difficulty processing abstract emotion nouns compared to control nouns (i.e., animal, tool, and food nouns). These recent studies have provided evidence that suggests, in accordance with theories of embodied semantics, that even abstract concepts may be grounded in sensorimotor and affective experience. The studies in this thesis contribute to our understanding of the limit and extent to which sensorimotor and affective processes can impact abstract semantic processing by investigating how metaphors that express abstract ideas by drawing on affect (i.e., physical inducers of disgust and fear), as well as, action (i.e., pulling vs. pushing type actions) also recruit the relevant sensorimotor and affective systems of the brain. According to conceptual metaphor theory, abstract concepts (target domain) mainly become grounded through co-occurrences with more concrete experiences (source domain) over the course of development (Lakoff & Johnson 1980; Lakoff, 2014). These cross-domain conceptual mappings (i.e., conceptual metaphors) delineate how concrete domains (source 10 domain) provide structure to and, thereby, contribute to the representation of abstract concepts. Specifically, inferential structural mechanisms in the source domain are believed to be re-used to make complex inferences in the target domain. These conceptual mappings are thought to be reflected in metaphorical expressions and recruited during metaphor comprehension. For example, when we construe immorality as impurity and hence disgust (e.g., ‘that was a rotten thing to do’) specific inferences associated with physical disgust experience (e.g., motivated withdrawal) are believed to be ‘re-used’ to make complex inferences in the moral domain (i.e., beliefs of moral ‘wrongness’). This thesis investigates the hypothesis that metaphors that reflect specific conceptual mappings involving the abstract domains of immorality, communication, and the act of cognizing recruit specific sensorimotor and affective processes relevant to their metaphoric source domains, thereby, furthering our understanding of how abstract concepts may be grounded in sensorimotor and affective experience. The following section (modified) is reproduced from: Gamez-Djokic, V., Molnar-Szakacs, I., Aziz-Zadeh, L., Embodied Simulation, Building meaning through shared neural circuitry, Conceptual and Interactive Embodiment, Foundations of Embodied Cognition Volume 2, Chapter 12, pg. 216- 245, Edited by Fischer, M.H. and Coello, Y., Routledge Taylor and Francis, A Psychology Press Book. Contextual modulation of sensorimotor and affective systems of the brain in language processing Sensorimotor activations during language processing appear to be a robust phenomenon. Somatotopic activation of motor areas by action verbs have been found in both shallow semantic tasks, such as lexical decision tasks involving classification of action verbs from non-words (De Grauwe, Willems, Rueschemeyer, Lemhofer, & Schriefers, 2014) and in deeper semantic tasks such as making semantic similarity judgments for action verbs (Kemmerer et al., 2008). Furthermore, a few studies show that activation of motor areas occurs automatically and within ~200 ms or less following presentation of action-related language stimuli (Dalla Volta, Fabbri- Destro, Gentilucci, & Avanzini, 2014; Hauk et al., 2004; Hauk & Pulvermuller, 2004; Klepp et al., 2014; Pulvermuller et al., 2005; Shtyrov, Butorina, Nikolaeva, & Stroganova, 2014). Given these findings, it would appear that sensorimotor representations are automatic and immediate when processing action-related words, even when participants are not necessarily paying close attention. However, a number of growing studies rather suggest that sensorimotor and affective activations during language processing may be sensitive to various modulatory influences (i.e., task, individual differences, linguistic and extralinguistic contexts). For example, a number of recent fMRI studies support the idea that sensorimotor activations during language processing may depend on task demands and/or depth of processing, for a review see (Yang, 2014). In line with this idea, Papeo et al., 2009 found activation in the hand motor area when participants were asked to reflect on relevant motor properties of hand action verbs, but not when they were asked to simply count the number of syllables of hand action verbs (Papeo, Vallesi, Isaja, & Rumiati, 2009). This finding suggests that motor activity 11 does not occur when participants are not paying close attention to the meaning of action verbs. In other words, the mere presence of an action word form is not sufficient to cause language related motor activity. In a different study Tomasino et al., 2007 found that the task (lexical decision task vs. explicit mental imagery task) modulated activation in primary motor cortex with increased activation for mental imagery (Tomasino, Werner, Weiss, & Fink, 2007). The authors suggest that motor activity during action word processing may occur at ‘different levels of abstraction’ depending on the task demands. Motor activity during a lexical-decision task might reflect higher-level action representations, as in the case of action observation, and be distinct from motor imagery. Consistent with this notion, a recent study looking at connectivity between bilateral premotor cortices and the supplementary motor areas using Granger causality analysis found distinct patterns of connectivity between a passive verb reading task, a motor imagery task, and a hand movement task (Yang & Shu, 2014). Specifically, Yang & Shu, 2014 found a left-lateralized connectivity pattern for passive verb reading, but a more extensive and complex bilateral connectivity pattern for the other tasks. These findings demonstrate that the sensorimotor cortices during action word processing can become differentially engaged depending on the depth of processing required by the task. A number of recent studies provide further support for the idea that motor behaviors/expertise and motor contexts can modulate how sensorimotor areas are engaged during action-related language processing, for a review see (Yang, 2014). For instance, van Dam, Rueschemeyer, & Bekkering, 2010 found that the bilateral inferior parietal lobule (IPL), a ‘higher-level motor area’ sensitive to motor planning and action goals (Fogassi et al., 2005b; Fogassi & Luppino, 2005a; Iacoboni et al., 2005; Iacoboni et al., 1999; Molnar-Szakacs, Kaplan, Greenfield, & Iacoboni, 2006), showed increased activation for verbs denoting specific motor programs (‘to hammer’, ‘to shoot’) compared to verbs denoting more general motor programs (‘to repair’, ‘to hunt’). Importantly, both specific and general verb types compared to abstract verbs (‘to appreciate’) showed greater activity in the IPL (van Dam, Rueschemeyer, & Bekkering, 2010). Thus, the degree of motor activity seen in processing motor-related features of action verb meanings depends on the motor specificity relevant to the action goal. Similarly, Rueschemeyer et al., 2010 showed that the degree to which action nouns can be manipulated or not (‘hammer’ vs. ‘clock’) can also modulate activity in motor areas (Rueschemeyer, van Rooij, Lindemann, Willems, & Bekkering, 2010), for similar findings see (Aravena et al., 2014; Aravena et al., 2012; Martin & Chao, 2001; Saccuman et al., 2006). Thus, motor activity related to action word processing appears to be modulated by relevant motor-related features that reflect the number of sensorimotor and affective associations during primary experience. Relatedly, the results also suggest that embodied simulations may operate at different levels of granularities or ‘abstraction’. That is, high-level schematic information such as object-agent interactions may lead to less motor engagement, while more detailed motor plans involving either specific kinematic information or visuomotor and affective feedback may lead to greater motor engagement (Svensson, Ziemke, & Lindblom, 2007). Lastly, certain action words might be linked to a wider set of ‘affordances’ indexing more specific motor simulations, while others index more simulations of background situations such as events and settings (Glenberg & Gallese, 2012; Simmons, Hamann, Harenski, Hu, & Barsalou, 2008). This may lead to a more widely spread activation pattern in the brain. 12 Motor activity seen during processing of action-related language also closely reflects individual differences in sensorimotor experiences such as in the case of an acquired motor expertise. For instance, Willems, Hagoort, & Casasanto, 2010 found that right handers showed stronger activation in left premotor regions for verbs like ‘writing’, while left handers showed this effect in the opposite hemisphere (Willems, Hagoort, & Casasanto, 2010). Willems, Labruna, D’Esposito, Ivry, & Casasanto, 2011 also showed using off line theta-burst TMS that stimulation of the left premotor cortex (PMC) compared to stimulation of the right PMC led to increased facilitation effects during a lexical decision task when participants read verbs denoting actions typically performed with the dominant hand (‘to throw’, ‘to write’) but not when they read verbs denoting non-manual actions (‘to earn’, ‘to wander’) (Willems, Labruna, D'Esposito, Ivry, & Casasanto, 2011b). Lyons et al., 2010 extended this finding by looking at specific populations with a specific motor expertise, mainly expert ice hockey players vs. novices. They found that hockey experts showed significantly greater activity in the left premotor cortex than novices when listening to sentences about hockey (‘The hockey player knocked down the net’), but this was not the case for sentences describing everyday actions (‘The individual opened the fridge’) (Lyons et al., 2010). Thus, the specificity and personal relevance of sensorimotor experiences can modulate how sensorimotor systems are engaged during action-related language processing, also see (Casasanto & Chrysikou, 2011) for further body specific effects on cognition. The situational context can also modulate activity of sensorimotor neural systems during language processing, even when the language does not contain action words, but the situation indirectly evokes action-related conceptual representations. Specifically, indirect speech requests (‘It is hot in here!’) in the context of a room with a closed window can be interpreted as an indirect request to open the window and thus engage motor programs indirectly (van Ackeren, Casasanto, Bekkering, Hagoort, & Rueschemeyer, 2012). Van Ackeren, et al., 2012 investigated this by showing participants a visual cue (a room with a window closed) while listening to the utterance (‘It is very hot here’), implying that that a motor action needs to take place to alleviate the situation (e.g. opening the window). In contrast, showing a different visual cue (e.g., a car parked near a desert) while listening to the same sentence does not evoke a motor interpretation. They found that statements processed as indirect requests that implied a motor action showed increased activation in motor areas also activated in a motor planning task, compared to various controls including the same utterance interpreted simply as a statement. No such differences in activation were found in motor areas when the same utterance was simply interpreted as a statement. The results support the notion that background situations can provide simulations beyond those indexed by the specific words (Simmons et al., 2008). In this case, a sentence with no action words in a specific situational context can nevertheless engage motor simulations. Importantly, simulations of background situations indexed by the specific situational context here might not be enough. Although not discussed in the study, an affective simulation might have provided an important inference in this case, as well. ‘It is very hot here’ could lead to an affect-based simulation leading to the notion that an overheated body is an unpleasant experience and should, therefore, motivate appropriate actions to cool the body. Nevertheless, processing 13 affect-based simulations and background situations might rely on additional top-down inputs. Indeed, indirect requests also led to greater activity in areas previously implicated in theory of mind networks (ToM), including the left temporoparietal junction (TPJ) and the medial prefrontal cortices (mPFC), in addition to the more emotion-related brain regions including the anterior cingulate cortices (ACC) and the bilateral insula in line with an affect-based simulation. Thus, in addition to affective and sensorimotor simulations indexed by the linguistic input and background situations, language processing of indirect speech requests may require additional processing resources in other multimodal or supramodal brain regions (Pulvermuller, 2013; van Ackeren et al., 2012). The immediate linguistic context can also modulate motor engagement during the processing of action-related phrases. For example, Moody and Gennari, 2010 showed that activity in the inferior frontal gyrus (IFG) and areas of premotor cortex, also involved in force exertion on objects using the hand, were sensitive to the implied force suggested by noun-verb pairings in action-related phrases (‘pushing the piano’ vs. ‘pushing the chair’) (Moody & Gennari, 2010). Furthermore, sentences containing the syntactic negation marker (‘Now I don’t push the button’) have been shown to modulate activity in motor systems of the brain compared to affirmative action-related sentences. Tettamanti et al., 2008 had subjects passively listen to either negated or affirmative hand and mouth action sentences (‘Now I push the button’ vs. ‘Now I don’t push the button’), as well as abstract sentences (‘Now I appreciate loyalty’ vs. ‘Now I don’t appreciate loyalty’) as controls (Tettamanti et al., 2008). They found an overall main effect of polarity vs. concreteness, such that negated forms in both action-related and abstract sentences lead to a deactivation of pallido-cortical areas. Furthermore, when looking at specific interaction effects between polarity and concreteness they found that negated action sentences showed an overall reduction in a fronto-temporo-parietal system along with decreased connection strengths across these areas as assessed by dynamic causal modeling. They suggested that this finding indicates that negation may render the negated mental motor representation temporarily inaccessible. Tomasino et al., 2010 extended this finding by looking at negated and affirmative hand imperatives (‘Do grasp’ vs. ‘Don’t write’) and compared them to negated and affirmative imperatives with pseudo verbs (‘Do gralp’ vs. Don’t gralp’) (Tomasino, Weiss, & Fink, 2010). They found that activity in primary motor and premotor regions (also active in a hand motion task) were differentially decreased for negated hand imperatives compared to affirmative imperatives, as well as compared to both affirmative and negated imperatives with pseudo verbs. Thus, the mere presence or absence of a syntactic marker in action-related sentences can differentially modulate networks involved in actual action execution. In summary, an increasing number of studies support the idea that sensorimotor and affective activations during language processing are highly dependent on various contextual factors. For example, individual differences, task effects, and linguistic/extralinguistic contexts have all been shown to modulate the engagement of embodied processes in language comprehension. These modulatory factors are critical as they delineate the extent and limits to which embodied processes can make functional contributions to meaning processes. For example, evidence showing that individual differences can modulate the engagement of sensorimotor and affective systems in language processing would be in line with experience- 14 dependent views of semantic representation and organization, which would provide further support for the embodiment hypothesis. Similarly, specific task-dependent activations can show how the involvement of sensorimotor and affective systems in semantic processing may primarily depend on depth of conceptual processing needed or other factors, helping to explain why some studies fail to find sensorimotor and affective activations during language processing. They also suggest that embodied processes might rather make systematic context-dependent functional contributions to meaning processes underscoring the types of flexible simulation mechanisms that might be needed. This thesis investigates the hypothesis that sensorimotor and affective activations in metaphor comprehension may themselves be sensitive to modulatory factors, such as individual differences and/or task effects. Neural evidence for sensorimotor and affective processes in figurative language While an increasing number of fMRI studies find activation in sensorimotor and affective brain systems during processing of words and literal language related to distinct sensory and motor modalities, the results for figurative language processing have been much more mixed. For instance, Aziz-Zadeh et al., 2006 and Raposo et al., 2009 showed somatotopic activation within primary motor/ or premotor areas for literal action sentences (e.g., ‘biting the peach’), but did not report a similar finding for idiomatic action-phrases (e.g., ‘biting off more than you can chew”) (Aziz-Zadeh et al., 2006; Raposo, Moss, Stamatakis, & Tyler, 2009). Cacciari et al., 2011 additionally showed that TMS applied to the leg motor area following reading of fictive motion (‘The road turns left’) and motion-related metaphors (‘The lady turns her thoughts away from sorrow’) led to an increase in MEPs in the relevant leg muscles (Cacciari et al., 2011). On closer inspection, Cacciari et al. 2011 also found that TMS pulses applied to the leg motor areas after participants read motions sentences (either literal, metaphoric, or idiomatic) led to increased motor evoked potential (MEPs) in the relevant leg muscles for literal and metaphoric sentences, but again not for idiomatic sentences. Nevertheless, a few studies suggest that figurative language processing, both for idiomatic and metaphoric sentences, reliably recruits the relevant modality-preferential brain regions. Boulenger et al., 2009 showed that that action-related literal and idiomatic sentences showed somatotopic activations within motor cortex (Boulenger, Hauk, & Pulvermuller, 2009). Furthermore, two recent studies in sensory domains found that even highly conventional metaphors related to touch (“she had a rough day”) and taste (“the break-up was bitter for him”) reliably activated the relevant modality-preferential representations, mainly somatosensory cortex and areas of primary/secondary gustatory cortex, respectively (Citron & Goldberg, 2014; Lacey, Stilla, & Sathian, 2012). In addition, the study by Citron & Goldberg, 2014 showed that not only did familiar taste metaphors matched in arousal and valence (“The break up was bitter for him”) activate the anterior insula/frontal operculum, OFC more than their literal counterparts (“The break up was bad for him”), they also activated other emotion brain regions including the amygdala and parahippocampal area. These findings suggest that even highly conventional metaphors activate sensorimotor representations and, additionally, further suggest grounding in affective experience, more generally. 15 Desai et al., 2011 extended these findings by showing evidence that the degree to which sensorimotor activation is seen for action-related sentences decreases with increasing abstraction from literal to metaphoric and from non-familiar to more familiar metaphors, with again little to no sensorimotor activation for idioms. They found that while literal sentences activated the precentral gyrus and the left anterior inferior parietal lobe (aIPL), a ‘secondary or higher-level motor area’, metaphoric sentences also recruited the aIPL, but activity within primary motor areas was sensitive to how familiar the metaphors had been rated (Desai et al., 2011). Furthermore, they found that action-related metaphors activated only aIPL compared to abstract sentences, while idiomatic action sentences activated primarily classic language areas (i.e., inferior frontal gyrus (IFG), pars triangularis and orbitalis) (Desai et al., 2013). Relatedly, Saygin et al., 2010 and others also similarly showed that literal motion sentences activated areas within functionally localized motion areas (MT+) to a greater extent than did metaphoric motion sentences, again showing evidence of a ‘graded pattern of activity’ (Saygin et al., 2010). In an attempt to explain these results, Desai, Binder, Conat, Mano, & Seidenberg (2011) proposed that figurative expressions might undergo a process of change in which initially such phrases are deeply linked to sensorimotor representations, but over time these links become less important for meaning due to processes of conventionalization and instead rely on multimodal or even amodal representations (Aziz-Zadeh & Damasio, 2008a; Cardillo, Watson, Schmidt, Kranjec, & Chatterjee, 2012; Desai et al., 2011), calling into question whether sensorimotor and affective processes are always engaged in familiar metaphor comprehension (Desai et al., 2011). An alternative interpretation suggests that decreasing engagement of sensorimotor systems from literal to metaphoric sentences may underlie the fact that sensorimotor representations in metaphor are just ‘less-specific’ or ‘generalized’ compared to literal sentences, which could account for the ‘graded’ pattern of involvement found for literal compared to metaphoric sentences above (Troyer et al., 2014). This proposal finds evidence in a recent behavioral study looking at the impact of biological motion perception of point-light walkers on processing motion verbs in either a literal or metaphoric context (Troyer et al., 2014). They found that point walker primes facilitated the processing of literal sentences with motion verbs that are semantically distant from the walking prime motion (leaping, catapulting) but not those that resembled the walking motion (ambling, walking). Importantly, the opposite effect was found for motion verbs in a metaphoric context with motion verbs that are semantically distant form the walking motion being processed more slowly than semantically close verbs. Such facilitation or interference effects (i.e., action compatibility effects) are believed to be due to shared representations between primary experience (action, space, motion) and linguistically represented aspects of motion. Thus, the evidence suggests that action-related metaphors may depend on more schematized or higher-level sensorimotor representations, possibly more attuned to action goals, rather than specific low-level kinematics of the action. This is in line with the finding from Desai et al., 2011, 2013 that metaphors engaged mainly the IPL, a higher-level motor area attuned to action goals (Rizzolatti, Cattaneo, Fabbri-Destro, & Rozzi, 2014). The above studies suggest that sensorimotor and affective processes are modulated by the metaphoric (or literal) interpretation of action-related words, with less sensorimotor engagement for more familiar metaphors compared to novel action-related metaphors or literal action-related 16 sentences. However, it is unclear if this diminished contribution of sensorimotor and affective systems of the brain for familiar metaphor comprehension is primarily due to processes of conventionalization and/or because more schematized sensorimotor and affective representations are needed for the processing of more abstract ideas in familiar metaphor. Therefore, this thesis aims to investigate the level of sensorimotor specificity (or motor abstraction) involved in action execution (i.e., low-level action kinematics and/or higher-level action goals/outcomes) relevant to the processing of familiar hand-action metaphors that differ in subtle ways, mainly in complex force-exchange patterns between objects and agents, a level of sensorimotor specificity not previously investigated for abstract event descriptions in the brain. Overall, the studies comprising this thesis introduced briefly below explore neural evidence for these hypotheses. 1) Do metaphors relating to affect and action entail source domain activity within the relevant sensorimotor and affective areas in the brain? Relatedly, to what extent are there shared patterns of neural activity during physical experience and the processing of metaphors that putatively draw on the same underlying sensorimotor/affective processes in order to make more complex inferences in the target domain? 2) Are sensorimotor and affective activations in metaphor comprehension modulated by contextual factors (individual differences, task effects)? 3) What level of sensorimotor specificity (e.g., low-level kinematic actions vs. higher-level action goals) is involved for processing of metaphoric source domains that differ in subtle ways (i.e., force dynamics)? 1. Neural underpinnings of metaphors implying ‘moral disgust’ Although there is increasing evidence that metaphor processing entails neural activation of sensorimotor and affective systems of the brain relevant to the source domain, it is unclear to what extent metaphors that imply ‘moral disgust’ would engage sensorimotor and affective processes relevant to physical disgust processing in support of embodied cognition theories and CMT. Previous evidence suggests that the anterior insula/frontal operculum (i.e., areas of primary/secondary cortex) and basal ganglia may play an important role in physical disgust processing (i.e., gustatory/olfactory-based disgust) (Bastiaansen, Thioux, & Keysers, 2009; Calder, 2003; Calder et al., 2007). Study 1 investigates whether the neural substrates associated with the processing of visual and/or literal descriptions of physical disgust inducers (e.g., pathogen-based disgust) are also engaged during comprehension of disgust metaphors. To test this, we use both a univariate and multivariate approach. We predicted that disgust language compared to non-affective (i.e., motion-related) and fear-related (i.e., physical forces causing bodily harm) language, both for literal and metaphoric contexts, should recruit the relevant emotion-related brain regions implicated in physical disgust processing. In contrast, fear-related literal and metaphoric language should recruit in part a separate set of sensorimotor and affective processes in the brain compared to non-affective and disgust metaphor. Furthermore, using a cross-modal (but also within-modality) ROI-based multivariate pattern analysis (MVPA) we should be able to show that a classifier trained to distinguish between disgust and neutral or fear 17 inducing visual stimuli should be able to successfully classify between disgust and non-affective/ fear metaphors at the level of the anterior insula/frontal operculum. Study 1 thus tries to address the following questions about the processing of metaphor, more generally: (1) Are sensorimotor and affective systems engaged during comprehension of metaphors, such as for morality (e.g., Immorality is Impurity/Disgust)? (2) Are there shared patterns of activation between specific sensorimotor or affective processes involved in actual physical experience (i.e., disgust-inducing images) and metaphors drawing on these source domains (e.g., physical disgust language)? This paper was co-authored with Srini Narayanan, Elisabeth Wehling, Ben Bergen, Josh Davis, Tong Sheng, and Lisa Aziz-Zadeh. 2. Metaphor in politics: bringing affect to the decision space? The conceptual mapping, ‘Immorality is Impurity/(physical disgust)’, has been thought to have particular relevance for moral discourse and judgment (Haidt & Graham, 2007; George Lakoff, 1996), although this has been recently debated (Landy & Goodwin, 2015; Pizarro, Inbar, & Helion, 2011). Along similar lines, it is unclear what advantage using moral disgust metaphors that construe immorality as some kind of physical impurity (i.e., physical disgust) might have compared to literal paraphrases in political discourse, if any. Study 2 investigates whether or not familiar disgust metaphors that express specific moral political attitudes can distinctly modulate both more automatic emotion-related brain areas (also relevant to physical disgust processing) and brain areas implicated in more deliberate, top-down processes in moral decision-making when compared to literal paraphrases, matched for relevant psycholinguistic variables (i.e., arousal/valence, semantic similarity, etc.). In order to test this hypothesis, we recruited both conservative and liberal participants to read political statements and indicate their degree of agreement during a response period. We analyzed the reading-period and response-period separately to test for task-dependent effects. We further analyzed whether embodiment effects in moral disgust metaphor would reflect individual differences (i.e., political orientation effects). We hypothesized that moral disgust metaphors would distinctly modulate emotion-related brain regions, as well as, brain areas implicated in decision-making compared to literal paraphrases (across both the reading and judgment portions of the task), indicative of differential moral decision-making in the brain. Moreover, we predicted that activity within emotion-related brain regions during moral disgust metaphor processing (but also during the viewing of disgusting images) should be particularly sensitive to political orientation, as purity concerns have been found to be a strong psychological/physiological trait mainly amongst conservatives (Haidt & Graham, 2007; Smith et al., 2011). This paper was co-authored with Elisabeth Wehling and Lisa Aziz-Zadeh. 3. Distinguishing metaphors that differ in their force dynamics 18 It has been proposed that event descriptions in language, whether concrete or abstract, necessitate a 'naive physics' or knowledge about the nature of physical force exchange between objects and agents involved in each action. However, it is currently unknown whether metaphors that construe the processing and sharing of information as the manipulation and transfer of objects and, imply force either away or towards the agonist (i.e., the agent resisting the motion), engage sensorimotor processes in the brain. Previous studies suggest that processing action- related literal phrases reflect specific motor-features relevant to the meaning of action words (i.e., degree of implied force) (Kemmerer, 2015; Moody & Gennari, 2010), however, it is currently unknown whether hand-actions metaphors that differ in their encoded force dynamics should similarly recruit the same level of sensorimotor specificity and primarily reflect either low-level action kinematics (i.e., primary motor areas) and/or higher-level motor representations that reflect the underlying action goal/outcome (IPL/IFG (BA 44)) (Desai et al., 2013; Krasovsky, Gilron, Yeshurun, & Mukamel, 2014; Moody & Gennari, 2010). Thus, in study 3 we investigated whether there are unique patterns of activity across voxels in the brain that can distinguish between two metaphors that differ in their force dynamics and relate to cognizing and communicating: Expressing Ideas is a Pushing Force and Cognizing Ideas is a Pulling Force. To test this hypothesis, we used a whole-brain spherical searchlight multivariate analysis to perform a within-modality classification and trained a classifier to distinguish between these two metaphor types. Moreover, we functionally localized sensorimotor brain areas involved in both physical ‘pulling’ and ‘pushing’ actions on an object during a motor task to see if voxels with significant classification accuracies in the multivariate analysis partially overlapped anatomically with brain areas activated for the univariate analysis of the motor task. We hypothesized that putatively low-level (i.e., directed force) and higher-level motor representations (i.e., purposeful object transfer) that can distinguish the ‘pulling’ vs. ‘pushing’ physical actions, should partly also be able to distinguish between (‘Expressing Ideas is a Pushing Force) and (Cognizing Ideas is a Pulling Force), as both levels of motor representations are hypothesized to be relevant for making more complex inferences in the target domain. Furthermore, we predicted that these brain regions should overlap with sensorimotor areas involved in actual action execution. This paper was co-authored with Elisabeth Wehling and Lisa Aziz-Zadeh. 19 References Aravena, P., Courson, M., Frak, V., Cheylus, A., Paulignan, Y., Deprez, V., & Nazir, T. A. (2014). Action relevance in linguistic context drives word-induced motor activity. Front Hum Neurosci, 8, 163. doi:10.3389/fnhum.2014.00163 Aravena, P., Delevoye-Turrell, Y., Deprez, V., Cheylus, A., Paulignan, Y., Frak, V., & Nazir, T. (2012). Grip force reveals the context sensitivity of language-induced motor activity during "action words" processing: evidence from sentential negation. PLoS One, 7(12), e50287. doi:10.1371/journal.pone.0050287 Arevalo, A. L., Baldo, J. V., & Dronkers, N. F. (2012). What do brain lesions tell us about theories of embodied semantics and the human mirror neuron system? Cortex, 48(2), 242-254. doi:10.1016/j.cortex.2010.06.001 Aziz-Zadeh, L., & Damasio, A. (2008a). Embodied semantics for actions: findings from functional brain imaging. J Physiol Paris, 102(1-3), 35-39. doi:10.1016/j.jphysparis.2008.03.012 Aziz-Zadeh, L., Fiebach, C. J., Naranayan, S., Feldman, J., Dodge, E., & Ivry, R. B. (2008b). Modulation of the FFA and PPA by language related to faces and places. Soc Neurosci, 3(3-4), 229-238. doi:10.1080/17470910701414604 Aziz-Zadeh, L., Wilson, S. M., Rizzolatti, G., & Iacoboni, M. (2006). Congruent embodied representations for visually presented actions and linguistic phrases describing actions. Curr Biol, 16(18), 1818-1823. doi:10.1016/j.cub.2006.07.060 Bak, T. H., & Chandran, S. (2012). What wires together dies together: verbs, actions and neurodegeneration in motor neuron disease. Cortex, 48(7), 936-944. doi:10.1016/j.cortex.2011.07.008 Bak, T. H., O'Donovan, D. G., Xuereb, J. H., Boniface, S., & Hodges, J. R. (2001). Selective impairment of verb processing associated with pathological changes in Brodmann areas 44 and 45 in the motor neurone disease-dementia-aphasia syndrome. Brain, 124(Pt 1), 103-120. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11133791 Barros-Loscertales, A., Gonzalez, J., Pulvermuller, F., Ventura-Campos, N., Bustamante, J. C., Costumero, V., . . . Avila, C. (2012). Reading salt activates gustatory brain regions: fMRI evidence for semantic grounding in a novel sensory modality. Cereb Cortex, 22(11), 2554-2563. doi:10.1093/cercor/bhr324 Barsalou, L. W. (1999). Perceptual symbol systems. Behav Brain Sci, 22(4), 577-609; discussion 610-560. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11301525 Barsalou, L. W. (2003). Abstraction in perceptual symbol systems. Philos Trans R Soc Lond B Biol Sci, 358(1435), 1177-1187. doi:10.1098/rstb.2003.1319 Barsalou, L. W. (2008). Grounded cognition. Annu Rev Psychol, 59, 617-645. doi:10.1146/annurev.psych.59.103006.093639 Barsalou, L. W., Kyle Simmons, W., Barbey, A. K., & Wilson, C. D. (2003). Grounding conceptual knowledge in modality-specific systems. Trends Cogn Sci, 7(2), 84-91. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12584027 Bastiaansen, J. A., Thioux, M., & Keysers, C. (2009). Evidence for mirror systems in emotions. Philos Trans R Soc Lond B Biol Sci, 364(1528), 2391-2404. doi:10.1098/rstb.2009.0058 Bedny, M., & Caramazza, A. (2011). Perception, action, and word meanings in the human brain: the case from action verbs. Ann N Y Acad Sci, 1224, 81-95. doi:10.1111/j.1749- 6632.2011.06013.x 20 Bonner, M. F., & Grossman, M. (2012). Gray Matter Density of Auditory Association Cortex Relates to Knowledge of Sound Concepts in Primary Progressive Aphasia. Journal of Neuroscience, 32(23), 7986-7991. doi:10.1523/Jneurosci.6241-11.2012 Boulenger, V., Hauk, O., & Pulvermuller, F. (2009). Grasping ideas with the motor system: semantic somatotopy in idiom comprehension. Cereb Cortex, 19(8), 1905-1914. doi:10.1093/cercor/bhn217 Boulenger, V., Mechtouff, L., Thobois, S., Broussolle, E., Jeannerod, M., & Nazir, T. A. (2008). Word processing in Parkinson's disease is impaired for action verbs but not for concrete nouns. Neuropsychologia, 46(2), 743-756. doi:10.1016/j.neuropsychologia.2007.10.007 Bowdle, B. F., & Gentner, D. (2005). The career of metaphor. Psychol Rev, 112(1), 193-216. doi:10.1037/0033-295X.112.1.193 Buccino, G., Riggio, L., Melli, G., Binkofski, F., Gallese, V., & Rizzolatti, G. (2005). Listening to action-related sentences modulates the activity of the motor system: a combined TMS and behavioral study. Brain Res Cogn Brain Res, 24(3), 355-363. doi:10.1016/j.cogbrainres.2005.02.020 Buxbaum, L. J., Johnson, S. H., & Bartlett, M. (2002). Role of the left inferior parietal lobe in planning object-oriented actions: Evidence from ideomotor apraxia. J Cogn Neurosci, 64- 64. Retrieved from <Go to ISI>://WOS:000174072000225 Buxbaum, L. J., & Saffran, E. M. (2002). Knowledge of object manipulation and object function: dissociations in apraxic and nonapraxic subjects. Brain Lang, 82(2), 179-199. doi:Pii S0093-934x(02)00014-7 Doi 10.1016/S0093-934x(02)00014-7 Buxbaum, L. J., Sirigu, A., Schwartz, M. F., & Klatzky, R. (2003). Cognitive representations of hand posture in ideomotor apraxia. Neuropsychologia, 41(8), 1091-1113. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12667544 Buxbaum, L. J., Veramonti, T., & Schwartz, M. F. (2000). Function and manipulation tool knowledge in apraxia: Knowing 'what for' but not 'how'. Neurocase, 6(2), 83-97. doi:Doi 10.1080/13554790008402763 Cacciari, C., Bolognini, N., Senna, I., Pellicciari, M. C., Miniussi, C., & Papagno, C. (2011). Literal, fictive and metaphorical motion sentences preserve the motion component of the verb: a TMS study. Brain Lang, 119(3), 149-157. doi:10.1016/j.bandl.2011.05.004 Calder, A. J. (2003). Disgust discussed. Ann Neurol, 53(4), 427-428. doi:10.1002/ana.10565 Calder, A. J., Beaver, J. D., Davis, M. H., van Ditzhuijzen, J., Keane, J., & Lawrence, A. D. (2007). Disgust sensitivity predicts the insula and pallidal response to pictures of disgusting foods. Eur J Neurosci, 25(11), 3422-3428. doi:10.1111/j.1460- 9568.2007.05604.x Calder, A. J., Keane, J., Manes, F., Antoun, N., & Young, A. W. (2000). Impaired recognition and experience of disgust following brain injury. Nat Neurosci, 3(11), 1077-1078. doi:10.1038/80586 Cameron, L. (2008). Metaphor and Talk. Cambridge Handbook of Metaphor and Thought, 197- 211. Retrieved from <Go to ISI>://WOS:000304675500012 Caramazza, A., & Mahon, B. Z. (2003). The organization of conceptual knowledge: the evidence from category-specific semantic deficits. Trends Cogn Sci, 7(8), 354-361. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12907231 21 Cardillo, E. R., Watson, C. E., Schmidt, G. L., Kranjec, A., & Chatterjee, A. (2012). From novel to familiar: tuning the brain for metaphors. Neuroimage, 59(4), 3212-3221. doi:10.1016/j.neuroimage.2011.11.079 Carr, E. W., Kavanagh, L., & Bergen, B. K. (2013). How embodied is metaphor? Current Directions in Psychological Science, CDPS-12-0113. Casasanto, D., & Chrysikou, E. G. (2011). When left is "right". Motor fluency shapes abstract concepts. Psychol Sci, 22(4), 419-422. doi:10.1177/0956797611401755 Citron, F. M. (2012). Neural correlates of written emotion word processing: a review of recent electrophysiological and hemodynamic neuroimaging studies. Brain Lang, 122(3), 211- 226. doi:10.1016/j.bandl.2011.12.007 Citron, F. M., & Goldberg, A. E. (2014). Metaphorical sentences are more emotionally engaging than their literal counterparts. J Cogn Neurosci, 26(11), 2585-2595. doi:10.1162/jocn_a_00654 Cotelli, M., Borroni, B., Manenti, R., Zanetti, M., Arevalo, A., Cappa, S. F., & Padovani, A. (2007). Action and object naming in Parkinson's disease without dementia. Eur J Neurol, 14(6), 632-637. doi:10.1111/j.1468-1331.2007.01797.x Dalla Volta, R., Fabbri-Destro, M., Gentilucci, M., & Avanzini, P. (2014). Spatiotemporal dynamics during processing of abstract and concrete verbs: an ERP study. Neuropsychologia, 61, 163-174. doi:10.1016/j.neuropsychologia.2014.06.019 Damasio, A. R., & Tranel, D. (1993). Nouns and verbs are retrieved with differently distributed neural systems. Proc Natl Acad Sci U S A, 90(11), 4957-4960. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8506341 De Grauwe, S., Willems, R. M., Rueschemeyer, S. A., Lemhofer, K., & Schriefers, H. (2014). Embodied language in first- and second-language speakers: neural correlates of processing motor verbs. Neuropsychologia, 56, 334-349. doi:10.1016/j.neuropsychologia.2014.02.003 Desai, R. H., Binder, J. R., Conant, L. L., Mano, Q. R., & Seidenberg, M. S. (2011). The neural career of sensory-motor metaphors. J Cogn Neurosci, 23(9), 2376-2386. doi:10.1162/jocn.2010.21596 Desai, R. H., Conant, L. L., Binder, J. R., Park, H., & Seidenberg, M. S. (2013). A piece of the action: modulation of sensory-motor regions by action idioms and metaphors. Neuroimage, 83, 862-869. doi:10.1016/j.neuroimage.2013.07.044 Dravida, S., Saxe, R., & Bedny, M. (2013). People can understand descriptions of motion without activating visual motion brain regions. Front Psychol, 4, 537. doi:10.3389/fpsyg.2013.00537 Dreyer, F. R., Frey, D., Arana, S., von Saldern, S., Picht, T., Vajkoczy, P., & Pulvermuller, F. (2015). Is the Motor System Necessary for Processing Action and Abstract Emotion Words? Evidence from Focal Brain Lesions. Front Psychol, 6, 1661. doi:10.3389/fpsyg.2015.01661 Fernandino, L., Binder, J. R., Desai, R. H., Pendl, S. L., Humphries, C. J., Gross, W. L., . . . Seidenberg, M. S. (2016). Concept Representation Reflects Multimodal Abstraction: A Framework for Embodied Semantics. Cereb Cortex, 26(5), 2018-2034. doi:10.1093/cercor/bhv020 Fodor, J. A. (1975). The Language of Thought. New York, NY: Thomas Y. Crowell. 22 Fogassi, L., Ferrari, P. F., Gesierich, B., Rozzi, S., Chersi, F., & Rizzolatti, G. (2005b). Parietal lobe: from action organization to intention understanding. Science, 308(5722), 662-667. doi:10.1126/science.1106138 Fogassi, L., & Luppino, G. (2005a). Motor functions of the parietal lobe. Curr Opin Neurobiol, 15(6), 626-631. doi:10.1016/j.conb.2005.10.015 Gainotti, G. (2010). The influence of anatomical locus of lesion and of gender-related familiarity factors in category-specific semantic disorders for animals, fruits and vegetables: A review of single-case studies. Cortex, 46(9), 1072-1087. doi:10.1016/j.cortex.2010.04.002 Gainotti, G. (2015). Inborn and experience-dependent models of categorical brain organization. A position paper. Front Hum Neurosci, 9, 2. doi:10.3389/fnhum.2015.00002 Gallese, V., & Lakoff, G. (2005). The Brain's concepts: the role of the Sensory-motor system in conceptual knowledge. Cogn Neuropsychol, 22(3), 455-479. doi:10.1080/02643290442000310 Garcea, F. E., Dombovy, M., & Mahon, B. Z. (2013). Preserved tool knowledge in the context of impaired action knowledge: implications for models of semantic memory. Front Hum Neurosci, 7, 120. doi:10.3389/fnhum.2013.00120 Gibbs, R. W. (2006). Metaphor Interpretation as Embodied Simulation. Mind & Language, 21(3), 434-458. doi:10.1111/j.1468-0017.2006.00285.x Glenberg, A. M., & Gallese, V. (2012). Action-based language: a theory of language acquisition, comprehension, and production. Cortex, 48(7), 905-922. doi:10.1016/j.cortex.2011.04.010 Glenberg, A. M., & Kaschak, M. P. (2002). Grounding language in action. Psychon Bull Rev, 9(3), 558-565. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12412897 Glucksberg, S. (1999). When love is not a journey: What metaphors mean. Journal of Pragmatics, 31, 1541-1558. Glucksberg, S. (2002). Understanding figurative language: From metaphors to idioms. New York: Oxford University Press. Gonzalez, J., Barros-Loscertales, A., Pulvermuller, F., Meseguer, V., Sanjuan, A., Belloch, V., & Avila, C. (2006). Reading cinnamon activates olfactory brain regions. Neuroimage, 32(2), 906-912. doi:10.1016/j.neuroimage.2006.03.037 Haidt, J., & Graham, J. (2007). When Morality Opposes Justice: Conservatives Have Moral Intuitions that Liberals may not Recognize. Social Justice Research. Hauk, O., Johnsrude, I., & Pulvermuller, F. (2004). Somatotopic representation of action words in human motor and premotor cortex. Neuron, 41(2), 301-307. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14741110 Hauk, O., & Pulvermuller, F. (2004). Neurophysiological distinction of action words in the fronto-central cortex. Hum Brain Mapp, 21(3), 191-201. doi:10.1002/hbm.10157 Hauk, O., & Tschentscher, N. (2013). The Body of Evidence: What Can Neuroscience Tell Us about Embodied Semantics? Front Psychol, 4, 50. doi:10.3389/fpsyg.2013.00050 Iacoboni, M., Molnar-Szakacs, I., Gallese, V., Buccino, G., Mazziotta, J. C., & Rizzolatti, G. (2005). Grasping the intentions of others with one's own mirror neuron system. PLoS Biol, 3(3), e79. doi:10.1371/journal.pbio.0030079 Iacoboni, M., Woods, R. P., Brass, M., Bekkering, H., Mazziotta, J. C., & Rizzolatti, G. (1999). Cortical mechanisms of human imitation. Science, 286(5449), 2526-2528. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10617472 23 Kable, J. W., Kan, I. P., Wilson, A., Thompson-Schill, S. L., & Chatterjee, A. (2005). Conceptual representations of action in the lateral temporal cortex. J Cogn Neurosci, 17(12), 1855-1870. doi:10.1162/089892905775008625 Kemmerer, D. (2015). Are the motor features of verb meanings represented in the precentral motor cortices? Yes, but within the context of a flexible, multilevel architecture for conceptual knowledge. Psychon Bull Rev, 22(4), 1068-1075. doi:10.3758/s13423-014- 0784-1 Kemmerer, D., Castillo, J. G., Talavage, T., Patterson, S., & Wiley, C. (2008). Neuroanatomical distribution of five semantic components of verbs: evidence from fMRI. Brain Lang, 107(1), 16-43. doi:10.1016/j.bandl.2007.09.003 Kemmerer, D., Rudrauf, D., Manzel, K., & Tranel, D. (2012). Behavioral patterns and lesion sites associated with impaired processing of lexical and conceptual knowledge of actions. Cortex, 48(7), 826-848. doi:10.1016/j.cortex.2010.11.001 Keysar, B., & Bly, B. M. (1999). Swimming against the current: Do idioms reflect conceptual structure? Journal of Pragmatics, 31(12), 1559-1578. doi:Doi 10.1016/S0378- 2166(99)00004-1 Keysar, B., Shen, Y. I., Glucksberg, S., & Horton, W. S. (2000). Conventional language: How metaphorical is it? Journal of Memory and Language, 43(4), 576-593. doi:10.1006/jmla.2000.2711 Kiefer, M., Sim, E. J., Herrnberger, B., Grothe, J., & Hoenig, K. (2008). The sound of concepts: four markers for a link between auditory and conceptual brain systems. J Neurosci, 28(47), 12224-12230. doi:10.1523/JNEUROSCI.3579-08.2008 Klepp, A., Weissler, H., Niccolai, V., Terhalle, A., Geisler, H., Schnitzler, A., & Biermann- Ruben, K. (2014). Neuromagnetic hand and foot motor sources recruited during action verb processing. Brain Lang, 128(1), 41-52. doi:10.1016/j.bandl.2013.12.001 Kousta, S. T., Vigliocco, G., Vinson, D. P., Andrews, M., & Del Campo, E. (2011). The representation of abstract words: why emotion matters. J Exp Psychol Gen, 140(1), 14- 34. doi:10.1037/a0021446 Kovecses, Z. (2008). Metaphor and Emotion. Cambridge Handbook of Metaphor and Thought, 380-396. Retrieved from <Go to ISI>://WOS:000304675500022 Krasovsky, A., Gilron, R., Yeshurun, Y., & Mukamel, R. (2014). Differentiating intended sensory outcome from underlying motor actions in the human brain. J Neurosci, 34(46), 15446-15454. doi:10.1523/JNEUROSCI.5435-13.2014 Lacey, S., Stilla, R., & Sathian, K. (2012). Metaphorically feeling: comprehending textural metaphors activates somatosensory cortex. Brain Lang, 120(3), 416-421. doi:10.1016/j.bandl.2011.12.016 Lakoff, G. (1996). Moral politics : what conservatives know that liberals don't. Chicago: University of Chicago Press. Lakoff, G. (2014). Mapping the brain's metaphor circuitry: metaphorical thought in everyday reason. Front Hum Neurosci, 8. doi:ARTN 958 10.3389/fnhum.2014.00958 Lakoff, G., & Johnson, M. (1980a). The metaphorical structure of the human conceptual system. Cognitive Science, 4(2), 195-208. doi:10.1016/S0364-0213(80)80017-6 Lakoff, G., & Johnson, M. (1980b). Metaphors we live by. Chicago: University of Chicago Press. 24 Landy, J. F., & Goodwin, G. P. (2015). Does Incidental Disgust Amplify Moral Judgment? A Meta-Analytic Review of Experimental Evidence. Perspect Psychol Sci, 10(4), 518-536. doi:10.1177/1745691615583128 Liuzzi, G., Freundlieb, N., Ridder, V., Hoppe, J., Heise, K., Zimerman, M., . . . Hummel, F. C. (2010). The involvement of the left motor cortex in learning of a novel action word lexicon. Curr Biol, 20(19), 1745-1751. doi:10.1016/j.cub.2010.08.034 Lyons, I. M., Mattarella-Micke, A., Cieslak, M., Nusbaum, H. C., Small, S. L., & Beilock, S. L. (2010). The role of personal experience in the neural processing of action-related language. Brain Lang, 112(3), 214-222. doi:10.1016/j.bandl.2009.05.006 Mahon, B. Z., Anzellotti, S., Schwarzbach, J., Zampini, M., & Caramazza, A. (2009). Category- specific organization in the human brain does not require visual experience. Neuron, 63(3), 397-405. doi:10.1016/j.neuron.2009.07.012 Mahon, B. Z., & Caramazza, A. (2008). A critical look at the embodied cognition hypothesis and a new proposal for grounding conceptual content. J Physiol Paris, 102(1-3), 59-70. doi:10.1016/j.jphysparis.2008.03.004 Mahon, B. Z., & Caramazza, A. (2009). Concepts and categories: a cognitive neuropsychological perspective. Annu Rev Psychol, 60, 27-51. doi:10.1146/annurev.psych.60.110707.163532 Martin, A., & Chao, L. L. (2001). Semantic memory and the brain: structure and processes. Curr Opin Neurobiol, 11(2), 194-201. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11301239 Meteyard, L., Cuadrado, S. R., Bahrami, B., & Vigliocco, G. (2012). Coming of age: a review of embodiment and the neuroscience of semantics. Cortex, 48(7), 788-804. doi:10.1016/j.cortex.2010.11.002 Molnar-Szakacs, I., Kaplan, J., Greenfield, P. M., & Iacoboni, M. (2006). Observing complex action sequences: The role of the fronto-parietal mirror neuron system. Neuroimage, 33(3), 923-935. doi:10.1016/j.neuroimage.2006.07.035 Moody, C. L., & Gennari, S. P. (2010). Effects of implied physical effort in sensory-motor and pre-frontal cortex during language comprehension. Neuroimage, 49(1), 782-793. doi:10.1016/j.neuroimage.2009.07.065 Moseley, R. L., Shtyrov, Y., Mohr, B., Lombardo, M. V., Baron-Cohen, S., & Pulvermuller, F. (2015). Lost for emotion words: what motor and limbic brain activity reveals about autism and semantic theory. Neuroimage, 104, 413-422. doi:10.1016/j.neuroimage.2014.09.046 Negri, G. A. L., Rumiati, R. I., Zadini, A., Ukmar, M., Mahon, B. Z., & Caramazza, A. (2007). What is the role of motor simulation in action and object recognition? Evidence from apraxia. Cogn Neuropsychol, 24(8), 795-816. doi:10.1080/02643290701707412 Neininger, B., & Pulvermuller, F. (2001). The right hemisphere's role in action word processing: a double case study. Neurocase, 7(4), 303-316. doi:DOI 10.1093/neucas/7.4.303 Neininger, B., & Pulvermuller, F. (2003). Word-category specific deficits after lesions in the right hemisphere. Neuropsychologia, 41(1), 53-70. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12427565 Niedenthal, P. M., Barsalou, L. W., Winkielman, P., Krauth-Gruber, S., & Ric, F. (2005). Embodiment in attitudes, social perception, and emotion. Pers Soc Psychol Rev, 9(3), 184-211. doi:10.1207/s15327957pspr0903_1 25 Papeo, L., Vallesi, A., Isaja, A., & Rumiati, R. I. (2009). Effects of TMS on different stages of motor and non-motor verb processing in the primary motor cortex. PLoS One, 4(2), e4508. doi:10.1371/journal.pone.0004508 Pizarro, D., Inbar, Y., & Helion, C. (2011). On Disgust and Moral Judgment. Emotion Review, 3(3), 267-268. doi:10.1177/1754073911402394 Ponz, A., Montant, M., Liegeois-Chauvel, C., Silva, C., Braun, M., Jacobs, A. M., & Ziegler, J. C. (2014). Emotion processing in words: a test of the neural re-use hypothesis using surface and intracranial EEG. Soc Cogn Affect Neurosci, 9(5), 619-627. doi:10.1093/scan/nst034 Postle, N., McMahon, K. L., Ashton, R., Meredith, M., & de Zubicaray, G. I. (2008). Action word meaning representations in cytoarchitectonically defined primary and premotor cortices. Neuroimage, 43(3), 634-644. doi:10.1016/j.neuroimage.2008.08.006 Pulvermuller, F. (2013). How neurons make meaning: brain mechanisms for embodied and abstract-symbolic semantics. Trends Cogn Sci, 17(9), 458-470. doi:10.1016/j.tics.2013.06.004 Pulvermuller, F., Cooper-Pye, E., Dine, C., Hauk, O., Nestor, P. J., & Patterson, K. (2010). The word processing deficit in semantic dementia: all categories are equal, but some categories are more equal than others. J Cogn Neurosci, 22(9), 2027-2041. doi:10.1162/jocn.2009.21339 Pulvermuller, F., Hauk, O., Nikulin, V. V., & Ilmoniemi, R. J. (2005). Functional links between motor and language systems. Eur J Neurosci, 21(3), 793-797. doi:10.1111/j.1460- 9568.2005.03900.x Pylyshyn, Z. W. (1984). Computation and cognition: Towards a foundation for cognitive science. . Boston, MA: MIT Press. Pylyshyn, Z. W. (2007). Things and Places: How the Mind Connects with the World. Cambridge, MA: MIT Press. Raposo, A., Moss, H. E., Stamatakis, E. A., & Tyler, L. K. (2009). Modulation of motor and premotor cortices by actions, action words and action sentences. Neuropsychologia, 47(2), 388-396. doi:10.1016/j.neuropsychologia.2008.09.017 Rizzolatti, G., Cattaneo, L., Fabbri-Destro, M., & Rozzi, S. (2014). Cortical mechanisms underlying the organization of goal-directed actions and mirror neuron-based action understanding. Physiol Rev, 94(2), 655-706. doi:10.1152/physrev.00009.2013 Rosci, C., Chiesa, V., Laiacona, M., & Capitani, E. (2003). Apraxia is not associated to a disproportionate naming impairment for manipulable objects. Brain and Cognition, 53(2), 412-415. doi:10.1016/S0278-2626(03)00156-8 Rueschemeyer, S. A., van Rooij, D., Lindemann, O., Willems, R. M., & Bekkering, H. (2010). The function of words: distinct neural correlates for words denoting differently manipulable objects. J Cogn Neurosci, 22(8), 1844-1851. doi:10.1162/jocn.2009.21310 Saccuman, M. C., Cappa, S. F., Bates, E. A., Arevalo, A., Della Rosa, P., Danna, M., & Perani, D. (2006). The impact of semantic reference on word class: an fMRI study of action and object naming. Neuroimage, 32(4), 1865-1878. doi:10.1016/j.neuroimage.2006.04.179 Saygin, A. P., McCullough, S., Alac, M., & Emmorey, K. (2010). Modulation of BOLD response in motion-sensitive lateral temporal cortex by real and fictive motion sentences. J Cogn Neurosci, 22(11), 2480-2490. doi:10.1162/jocn.2009.21388 26 Schomers, M. R., Kirilina, E., Weigand, A., Bajbouj, M., & Pulvermuller, F. (2015). Causal Influence of Articulatory Motor Cortex on Comprehending Single Spoken Words: TMS Evidence. Cereb Cortex, 25(10), 3894-3902. doi:10.1093/cercor/bhu274 Shtyrov, Y., Butorina, A., Nikolaeva, A., & Stroganova, T. (2014). Automatic ultrarapid activation and inhibition of cortical motor systems in spoken word comprehension. Proc Natl Acad Sci U S A, 111(18), E1918-1923. doi:10.1073/pnas.1323158111 Simmons, W. K., & Barsalou, L. W. (2003). The similarity-in-topography principle: reconciling theories of conceptual deficits. Cogn Neuropsychol, 20(3), 451-486. doi:10.1080/02643290342000032 Simmons, W. K., Hamann, S. B., Harenski, C. L., Hu, X. P., & Barsalou, L. W. (2008). fMRI evidence for word association and situated simulation in conceptual processing. J Physiol Paris, 102(1-3), 106-119. doi:10.1016/j.jphysparis.2008.03.014 Spunt, R. P., Kemmerer, D., & Adolphs, R. (2015). The neural basis of conceptualizing the same action at different levels of abstraction. Soc Cogn Affect Neurosci. doi:10.1093/scan/nsv084 Svensson, H., Ziemke, T., & Lindblom, J. (2007). Making Sense of Embodied Cognition: Simulation Theories of Shared Neural Mechanisms for Sensorimotor and Cognitive Processes (pp. 241). Tettamanti, M., Buccino, G., Saccuman, M. C., Gallese, V., Danna, M., Scifo, P., . . . Perani, D. (2005). Listening to action-related sentences activates fronto-parietal motor circuits. J Cogn Neurosci, 17(2), 273-281. doi:10.1162/0898929053124965 Tettamanti, M., Manenti, R., Della Rosa, P. A., Falini, A., Perani, D., Cappa, S. F., & Moro, A. (2008). Negation in the brain: modulating action representations. Neuroimage, 43(2), 358-367. doi:10.1016/j.neuroimage.2008.08.004 Tomasino, B., Weiss, P. H., & Fink, G. R. (2010). To move or not to move: imperatives modulate action-related verb processing in the motor system. Neuroscience, 169(1), 246- 258. doi:10.1016/j.neuroscience.2010.04.039 Tomasino, B., Werner, C. J., Weiss, P. H., & Fink, G. R. (2007). Stimulus properties matter more than perspective: an fMRI study of mental imagery and silent reading of action phrases. Neuroimage, 36 Suppl 2, T128-141. doi:10.1016/j.neuroimage.2007.03.035 Troyer, M., Curley, L. B., Miller, L. E., Saygin, A. P., & Bergen, B. K. (2014). Action verbs are processed differently in metaphorical and literal sentences depending on the semantic match of visual primes. Front Hum Neurosci, 8. doi:ARTN 982 10.3389/fnhum.2014.00982 Trumpp, N. M., Kliese, D., Hoenig, K., Haarmeier, T., & Kiefer, M. (2013). Losing the sound of concepts: damage to auditory association cortex impairs the processing of sound-related concepts. Cortex, 49(2), 474-486. doi:10.1016/j.cortex.2012.02.002 van Ackeren, M. J., Casasanto, D., Bekkering, H., Hagoort, P., & Rueschemeyer, S. A. (2012). Pragmatics in action: indirect requests engage theory of mind areas and the cortical motor network. J Cogn Neurosci, 24(11), 2237-2247. doi:10.1162/jocn_a_00274 van Dam, W. O., Rueschemeyer, S. A., & Bekkering, H. (2010). How specifically are action verbs represented in the neural motor system: an fMRI study. Neuroimage, 53(4), 1318- 1325. doi:10.1016/j.neuroimage.2010.06.071 Vigliocco, G., Kousta, S. T., Della Rosa, P. A., Vinson, D. P., Tettamanti, M., Devlin, J. T., & Cappa, S. F. (2014). The neural representation of abstract words: the role of emotion. Cereb Cortex, 24(7), 1767-1777. doi:10.1093/cercor/bht025 27 Warrington, E. K., & McCarthy, R. (1983). Category specific access dysphasia. Brain, 106 ( Pt 4), 859-878. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/6652466 Warrington, E. K., & McCarthy, R. A. (1987). Categories of knowledge. Further fractionations and an attempted integration. Brain, 110 ( Pt 5), 1273-1296. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/3676701 Warrington, E. K., & Shallice, T. (1984). Category Specific Semantic Impairments. Brain, 107(Sep), 829-854. doi:DOI 10.1093/brain/107.3.829 Willems, R. M., & Casasanto, D. (2011a). Flexibility in embodied language understanding. Front Psychol, 2, 116. doi:10.3389/fpsyg.2011.00116 Willems, R. M., Hagoort, P., & Casasanto, D. (2010). Body-specific representations of action verbs: neural evidence from right- and left-handers. Psychol Sci, 21(1), 67-74. doi:10.1177/0956797609354072 Willems, R. M., Labruna, L., D'Esposito, M., Ivry, R., & Casasanto, D. (2011b). A functional role for the motor system in language understanding: evidence from theta-burst transcranial magnetic stimulation. Psychol Sci, 22(7), 849-854. doi:10.1177/0956797611412387 Yang, J. (2014). Influences of motor contexts on the semantic processing of action-related language. Cognitive, Affective, & Behavioral Neuroscience, 14(3), 912-922. doi:10.3758/s13415-014-0258-y Yang, J., & Shu, H. (2014). The Causal Interactions Between Bilateral M1 and SMA During Verb Comprehension, Motor Imagery and Hand Motion. Arch Neurosci, 1(3), e18185. doi:10.5812/archneurosci.18185 28 Neural Underpinnings of Metaphors Implying ‘Moral Disgust’ Vesna Gamez-Djokic 1,2 , Srini Narayanan 4 , Elisabeth Wehling 4 , Tong Sheng 6 , Benjamin Bergen 5 , Josh Davis 5 , Lisa Aziz-Zadeh 1,2,3* 1 Brain and Creativity Institute, University of Southern California, 2 Neuroscience Graduate Program, 3 Division of Occupational Science and Occupational Therapy, University of Southern California, 4 International Computer Science Institute, 5 Department of Cognitive Science, University of California, San Diego, 6 Palo Alto VA Medical Center Author contributions: V.G.D. contributed to design of study, creation of the stimuli, conducted the study, conducted analysis and data interpretation, and wrote the manuscript. S.N. conceived of the idea along with LAZ. E.W. contributed to manuscript write-up. T.S. was involved in study design, conducting the study, contributed to data analysis, B.B. and J.D. designed/created and piloted the linguistic stimuli, and L.A.Z oversaw all aspects of the study. All authors were involved with editing the manuscript. ABSTRACT Good and bad are frequently understood via a metaphor construing immoral deeds as impure or disgusting deeds (e.g., “A rotten thing to do.”). However, how the neural systems involved in the representation of affective states contribute to the representation of disgust metaphor is not well understood. In this study we specifically investigated whether the neural substrates associated with the processing of visual and/or literal descriptions of physical disgust (gustatory/olfactory-based disgust) (i.e., primary/secondary gustatory cortex and basal ganglia) were also engaged during the comprehension of metaphors implying ‘moral disgust’. Participants read literal (e.g., “The slabs of old deer meat are rotten…”) and metaphoric (e.g., “Congressman Smith’s statements about marriage are rotten…”) sentences that contained language relating to disgust (e.g., pathogen-based disgust), non-affective (i.e., motion-related), as well as fear (i.e., physical forces causing bodily harm) language descriptors. Comprehension of literal disgust sentences compared to non-affective literal sentences, showed increased activity bilaterally in brain regions relevant to physical disgust processing including areas of primary/secondary gustatory cortices (i.e., anterior insula/frontal operculum, OFC) and the basal ganglia. In parallel to this, comprehension of disgust metaphors compared to non-affective metaphors was also associated with increased activity in the ventral anterior insula, OFC, and pallidum but in the left hemisphere. Fear language stimuli mainly differed from disgust language stimuli in that they recruited the posterior insula (but also dorsal anterior insula for metaphor) and showed increased activity in parietal/sensorimotor areas. In contrast, disgust language stimuli mainly showed increased activity in subcortical brain regions (i.e., amygdala, basal ganglia) compared to fear language stimuli, but not within the anterior insula/frontal operculum. Similarly, the cross-modal multivariate analysis showed significant classification accuracies within the left anterior insula/frontal operculum for disgust versus neutral stimuli, but not for disgust versus fear stimuli. Our findings suggest that processing of disgust and fear metaphors recruit affective and sensorimotor systems of the brain that process features relevant to the corresponding affective experience (i.e., perceptual, internal states, and behavioral features), but not dedicated neural circuitry only specific to that emotion (e.g., anterior insula for disgust). Keywords: affective metaphor, embodiment, morality, fMRI, disgust, insula 29 INTRODUCTION A number of behavioral studies support a relationship between moral and physical disgust (e.g., rotten food, etc.), showing that disgust can impact an individuals’ reasoning and judgment of moral transgressions even if these feelings remain outside awareness or are unrelated to the current judgment (Eskine, Kacinik, & Prinz, 2011; Inbar, Pizarro, Knobe, & Bloom, 2009; Lee & Schwarz, 2011; Pizarro, Inbar, & Helion, 2011; Schnall, Haidt, Clore, & Jordan, 2008; Terrizzi, Shook, & Ventis, 2010; Wheatley & Haidt, 2005), for a critique see (Landy & Goodwin 2015). For instance, Schnall et al. (2008) found that experientially inducing disgust, via being in a disgusting room or exposed to a bad smell, led to harsher moral judgments. Relatedly, the concept of ‘moral disgust’ is relatively common in language (e.g., “A rotten crime”, “A disgusting thing to do”), and draws on physical disgust language (i.e., pathogen-based disgust). This metaphoric construal also has been found to similarly play a significant role in individuals’ reasoning about morality in politics, governance, and public political discourse (Lakoff, 1996). However, it remains to be seen to what extent comprehension of metaphors that imply ‘moral disgust’ and draw on physical disgust language actually invoke shared neural substrates with physical disgust experience (e.g., gustatory/olfactory-based disgust). According to Conceptual Metaphor Theory (CMT) (Lakoff & Johnson, 1980) the human conceptualization of moral transgressions as physically disgusting acts relies on an embodied, conceptual metaphor relating the abstract concept of immorality (target domain) to the concrete, physical experience of impurity and, in turn, disgust (source domain). This articulates with the theory of simulation semantics (Gallese & Lakoff, 2005; Barsalou et al., 2003; Barsalou 2009; Pulvermuller et al., 2005; Gallese 2003; Glenberg & Gallese 2011; Rizzolatti & Arbib 1998; Barsalou 1999), which hypothesizes that understanding is based on implicit ‘imaginative activation’ of embodied circuits in a simulation mode. In the context of simulation semantics in the brain, CMT predicts that metaphoric language is understood via the implicit simulation of sensorimotor, perceptual, and affective representations of source domain concepts in the brain (Gallese & Lakoff, 2005). This is in accordance with general theories of embodiment (also, ‘grounded cognition’ or ‘embodied semantics’), which posit that conceptual representations, abstract knowledge and reasoning are grounded in neural systems dedicated to direct world experience relating to motor actions, perceptions, or affect (Barsalou, 2008, 2009; Barsalou, Kyle Simmons, Barbey, & Wilson, 2003; Gallese & Lakoff, 2005; Lakoff & Johnson, 1980; Niedenthal, Barsalou, Winkielman, Krauth-Gruber, & Ric, 2005). In this view, the processing of metaphoric (or literal) disgust language should in part engage a set of shared or overlapping neural substrates with the processing of physical disgust experience. Previous research on embodied semantics has primarily focused on determining whether motor-related brain regions are engaged during processing of literal language about actions and has not directly focused on grounding in affective processes. These studies largely support the notion that motor-related brain regions are activated when processing action-related literal language (Aziz-Zadeh, Wilson, Rizzolatti, & Iacoboni, 2006; Hauk, Johnsrude, & Pulvermuller, 2004; Hauk & Pulvermuller, 2004; Pulvermuller, Hauk, Nikulin, & Ilmoniemi, 2005; Tettamanti et al., 2005), but see (Postle, McMahon, Ashton, Meredith, & de Zubicaray, 2008; Raposo, Moss, Stamatakis, & Tyler, 2009)). Outside the motor system, behavioral and neuroimaging studies indicate that conceptual processing engages modality preferential sensorimotor regions, when processing stimuli about food (Simmons et al., 2005), animals/tools (Simmons & Barsalou 30 2005), color (Simmons et al., 2007), and sounds (Kiefer et al., 2008). For literal sentences, Saygin, McCullough, Alac, & Emmorey (2010) showed that in the visual domain, there was increased activity in visual motion area MT for motion related language (e.g., ‘The wild horse crossed the barren field’) as compared to non-motion related language (‘The black horse stood in the barren field’) (Saygin et al., 2010). In another study, it was found that activity in the fusiform face area (FFA) and parahippocampal place area (PPA) were modulated by sentences related to faces (‘George Bush has wrinkles around his eyes’) or places (‘The Taj Majal faces a long thin reflecting pool’), respectively (Aziz-Zadeh et al., 2008). Lastly, studies exploring the gustatory and olfactory systems also indicate that scent-related words (e.g., garlic, cinnamon) activate olfactory cortex (Gonzalez et al., 2006) and that taste-related words (e.g., salt) activate primary and secondary gustatory cortices (Barros-Loscertales et al., 2012). Thus, a multitude of studies across different sensory modalities show activation of modality-preferential sensorimotor activations in the brain during the processing of words and literal expressions referring to that modality providing initial support for the theory of embodied semantics. While support for embodied semantics has been found for literal language, the results for metaphoric language processing have been mixed, with some asserting the degree of familiarity and conventionality of metaphors as an important factor (Aziz-Zadeh & Damasio, 2008a; Aziz- Zadeh et al., 2006; Desai, Binder, Conant, Mano, & Seidenberg, 2011; Desai, Conant, Binder, Park, & Seidenberg, 2013; Hauk & Tschentscher, 2013; Raposo et al., 2009; Willems & Casasanto, 2011a). In the current study, we focus on grounding in affective experience associated with physical disgust inducers (i.e., pathogen-based disgust) and ask whether novel disgust metaphors share an underlying neural substrate with the processing of physical disgust. Importantly, metaphors that construe morality in terms of impurity might gain their persuasiveness through partial activation of affective and sensorimotor representations pertaining to the basic emotion of disgust. Lending support for a role of affective experience in metaphor processing, Citron and Goldberg (2014) found that taste metaphors (“The break up was bitter for him”) activated areas of primary and secondary gustatory cortex, as well as the amygdala and parahippocampal area, when compared to their literal counterparts (“The break up was bad for him”) matched in valence and arousal (Citron & Goldberg 2011). Interestingly, while they similarly found activation within gustatory areas when they presented the taste words in isolation, they did not find activation within the amygdala and parahippocampal area. This finding suggests that taste metaphors might be ‘implicitly more emotionally engaging’ than their literal counterparts, matched for arousal and valence. Thus, grounding in affective brain systems might give these metaphors their ‘rhetorical advantage’ over equally accessible and familiar literal paraphrases (Citron & Goldberg, 2014). Indeed, Kousta et al. (2011), have proposed that abstract concepts (e.g. agony, joy), more generally, may rely to a greater degree on grounding in affective experience when compared to concrete concepts that might be grounded primarily in our sensorimotor experiences (Kousta, Vigliocco, Vinson, Andrews, & Del Campo, 2011; Vigliocco et al., 2014). For these reasons, it is important to understand how disgust metaphor modulates emotion-related brain regions relevant to physical disgust processing, as this could provide insight into how these metaphors could come to exert an impact on moral judgment in political discourse. Current studies support the idea that the anterior insula and adjacent frontal operculum (AIFO) play an important role in experiencing and observing the emotion of disgust (Bastiaansen, Thioux, & Keysers, 2009; Calder et al., 2007; Calder, Keane, Manes, Antoun, & 31 Young, 2000; Deen, Pitskel, & Pelphrey, 2011; Jabbi, Bastiaansen, & Keysers, 2008; Jabbi & Keysers, 2008; Wicker et al., 2003), also, more generally, areas of primary/secondary gustatory cortex (i.e., AIFO and OFC) for gustatory/olfactory-based disgust (Deen et al., 2011). In one study, Jabbi et al. (2008) found that the AIFO was recruited when participants experienced disgust via ingestion of a bitter substance, when they observed others experiencing disgust, and when they read image-driven texts that elicited disgust. The activation of the AIFO has also been implicated in the processing of distinct affective states, including experiencing pain and observing other people’s pain, as well as experiencing gustatory pleasure and observing pleased facial expressions in others (Jabbi & Keysers, 2008; Singer et al., 2004; Singer et al., 2006). In sum, there is evidence that emotion simulation may involve the same processes that underlie actual experiencing of the accordant emotion and that the AIFO may play an important role in disgust simulation as a center for interoceptive processing (i.e., perception of viscera/internal bodily changes) (Craig, 2009), particularly relevant to characterizing physical disgust experience. Importantly, the AIFO is an area of the brain with connections to numerous other emotion processing regions (e.g., OFC, ACC, basal ganglia, temporal lobe and amygdala) placing it in a unique position to associate external stimuli with internal bodily states that can, in turn, motivate appropriate changes in behavior, especially as it relates to higher cognitive function (Bastiaansen et al., 2009; Damasio, Damasio, & Tranel, 2013; Jabbi & Keysers, 2008). While previous functional imaging studies implicate the AIFO in disgust induced via odor, tastes, visual stimuli, and literal script-driven imagery (Adolphs, Tranel, & Damasio, 2003; Bastiaansen et al., 2009; Calder et al., 2000; Heining et al., 2003; Jabbi et al., 2008; Jabbi, Swart, & Keysers, 2007; Royet, Plailly, Delon-Martin, Kareken, & Segebarth, 2003; Small et al., 2003; Small & Prescott, 2005; Stark et al., 2007; Wicker et al., 2003; Zald, Lee, Fluegel, & Pardo, 1998), a number of studies have also implicated subregions of the basal ganglia (Calder et al., 2007; Jabbi et al., 2008; Jabbi et al., 2007; Phillips et al., 1998; Phillips et al., 1997; Sprengelmeyer, Rausch, Eysel, & Przuntek, 1998; van der Gaag, Minderaa, & Keysers, 2007; von dem Hagen et al., 2009; Wicker et al., 2003). It remains, however, to be seen if the processing of disgust metaphors would similarly preferentially activate the AIFO, as well as areas of the basal ganglia. For example, previous studies indicate that the left anterior insula is modulated by disgust words (e.g., “vomit”) as early as 200 ms in a combined surface/intracortical EEG study involving epileptic patients (Ponz et al., 2014). Therefore, it is conceivable that metaphors that draw on similar disgust language may also preferentially engage the anterior insula. Nevertheless, it is possible that the literal or metaphoric sentence context (e.g., “The sandwich was rotten” vs. “The congressman was rotten”) might impact processing of the same disgust language, since sentence context has been found to modulate sensorimotor recruitment in language comprehension in other studies (Citron & Goldberg 2014; Aziz-Zadeh et al., 2006; Raposo et al., 2009). To explore these questions, we looked at how processing sentences about literally disgusting scenes (e.g., “The slabs of deer meat are rotten. They saturate the air with noxious odor…”) and metaphoric sentences that imply moral disgust (e.g., “Congressman Smith’s statements about marriage are rotten. They saturate the air with noxious odor…”) engaged brain regions involved in affective systems of the brain relevant to physical disgust processing. Specifically, we were interested in neural regions previously implicated in disgust experience and recognition, mainly areas of primary/secondary gustatory cortex also involved in interoceptive/visceromotor processing (i.e., anterior insula/frontal operculum, OFC) and the 32 basal ganglia also previously implicated in motivated behavior (Haber, 2003; Taschibana & Hikosaka 2012). In order to test this, we compared activation in these emotion-related brain regions during the comprehension of literal disgust-related language vs. literal non-affective language and literal fear-related language. More importantly, we also compared the recruitment of these same regions during the comprehension of disgust metaphors vs. non-affective metaphors and fear-related metaphors. Additionally, in our multivariate analysis (within and cross-modality classification) we tested whether there are unique patterns of activity across voxels in the anterior insula and adjacent frontal operculum that allow us to distinguish disgust metaphors from non-affective or fear-related metaphors. This classification was based on activity patterns in the AIFO that distinguished disgust and non-affective or fear-related stimuli using pictures, literal sentences, or metaphoric sentences. We hypothesized that emotion-related brain regions previously implicated in physical disgust experience/recognition, specifically the anterior insula and adjacent frontal operculum, should reveal activity patterns specific to literal and metaphoric disgust language as compared to non-affective and fear-related language stimuli. METHODS Participants: Sixteen right-handed native English speakers (age range 20-30, 8 females, 8 males) provided informed consent and were paid for participating in this study. Three participants’ data was discarded (one was not a native speaker, two did not fully complete the task). This left 7 females and 6 males, and a total of 13 subjects. All subjects had normal hearing and vision, and no history of neurological or psychiatric illness. Stimuli: 1. Pictures: Forty-five images were selected per emotion category (neutral, fear, and disgust) for a total of 135 images. The images came from the International Affective Picture Site (IAPS) (http://csea.phhp.ufl.edu/Media.html#topmedia), the Geneva International Affective Picture Site (http://www.affective-sciences.org/researchmaterial), and the Internet. In a separate norming study 15 participants rated the pictures as disgust-inducing, fear-inducing, or other emotion on a 5-point Likert scale (e.g., 1-Not at all disgusted to 5- Extremely disgusted). A one- way ANOVA comparing average emotion ratings (fear, disgust, or other emotion) for individual fear pictures showed that fear pictures significantly induced fear (M = 2.77, SD = 0.68) more than any other emotion rating (F(2, 132) = 23.02, p <0.0001) and confirmed by independent t- tests (p<0.05). Similarly, disgust pictures significantly induced disgust (M = 2.86, SD = 1.10) more than any other emotion rating (F(2, 132) = 78.94, p <0.0001) and confirmed by independent t-tests (p<0.05). Lastly, our neutral pictures scored low on ratings for other emotion (M = 1.31, SD = 0.24) and did not elicit disgust or fear scoring the lowest average rating of 1 for both emotion categories. Participants also rated the pictures for arousal on a 5-point Likert scale (i.e., 1-Not at all arousing to 5-Strongly arousing) and valence on a 7-point Likert scale (1- Strongly unpleasant, 4-neither unpleasant nor pleasant, 7-Strongly pleasant). When comparing average ratings for our emotional pictures we found that disgust pictures (M = 2.87, SD = 0.82) and fear pictures (M = 2.66, SD = 0.66) did not significantly differ in arousal (t(44) = 1.03, p = 0.31), nor did disgust pictures (M = 1.88, SD = 0.56 ) and fear pictures (M = 2.04, SD = 0.70) significantly differ in valence (t(44) = -0.94, p = 0.36). The neutral images consisted of inanimate objects and scored low in arousal (M = 1.39, SD = 0.54) and were rated as slightly pleasant (M = 4.25, SD = 0.40) (4-neither unpleasant nor pleasant). 33 2. Language: Forty novel metaphorical paragraphs and 40 novel literal paragraphs, each consisting of three sentences, were used for each emotion category (neutral, fear, and disgust). Subjects viewed twenty metaphorical paragraphs and twenty literal paragraphs in total. The disgust language used consisted of any kind of physical disgust expressed vial language (i.e., vomit, feces, and rotten food, etc.). Importantly, our disgust metaphors did not reflect specific moral attitudes or political views. Examples are as follows: Disgust Literal: “The slabs of deer meat are rotten. They saturate the air with noxious odor. Anything exposed to them can become contaminated”. Disgust Metaphor: “Congressman Smith’s statements about marriage are rotten. They saturate the air with noxious odor. Anything exposed to them can become contaminated”). Fear stimuli included any kind of potential for physical harm expressed via language (i.e., falling, explosives, and snake bites, etc.). Examples are as follows: Fear Literal: “The unmanned drones are casting a shadow over the community. Their darkness is rapidly expanding. They are closing in on innocent adults and children”. Fear Metaphor: “Congressman Wilson’s Persuasive Measures Act is casting a shadow over the community. The darkness is rapidly expanding. It is closing in on innocent adults and children.” The non-affective language consisted of language mainly about motion (i.e., traveling slow or fast, moving up or down, etc.). Examples are as follows: Non-affective Literal: “The man in the truck has a long way to go before arriving to the job site. He has been heading north on the same road for over an hour already. Some days, traffic moves slower than others”. Non-affective Metaphor: “Congressman Thompson’s Battlefield Commemoration Act has a long way to go before it is approved. It has been on the same journey towards acceptance for over a year already. Some acts travel slower than others”. Every metaphorical paragraph had a literal counterpart, such that the last two sentences of the metaphorical and literal paragraphs matched as much as possible in that they consisted of the same valenced language or source domain language. While the last two sentences in a pair matched as much as possible, they were – dependent on the immediately preceding sentence – interpreted either literally or metaphorically. Stimuli were counterbalanced such that each participant only saw one paragraph from a matching pair (either literal or metaphoric). A total of 60 stimuli per category were normed using Amazon Mechanical Turk (MTurk). Each sentence was rated by 22 to 27 participants. The paragraphs were rated based on five criteria (i.e., “This is disgusting”, “This is frightening”, “This is pleasant”, “This is some other emotion”, “This is understandable”) each on a five point Likert scale (1-strongly disagree to 5- strongly agree). Ratings for each emotion category for literal and metaphoric stimuli were averaged across matching pairs. Next, the top 40 pairs that had the highest rating for one given emotion category and the lowest for all other emotion categories were selected. For example, disgust was calculated using: [disgust rating- mean rating of (fear, pleasant, other)]. As Table 1 shows, the selected affective and non-affective paragraph pairs demonstrated high specificity (all F-values > 251.16, all p-values < 0.05). Metaphoric stimuli did not significantly differ in average word length (M = 32.3, SD = 7.40) when compared to our literal stimuli (M = 31.53, SD = 9.23), (t(119) = 0.72, p=0.48). However, a two-way analysis of variance (ANOVA) with two levels (Literal vs. Metaphor) and three factors (Non-Affective, Disgust, Fear) showed significant differences in word length across emotion category (F(2, 234) = 14.56, p<0.05), no significant effect when comparing literal vs. metaphoric stimuli (F(1, 234) = 0.58, p=0.45), and no interaction term (F(2, 234) = 0.51, p=0.60). Differences in length between factors were mainly driven by significant differences in 34 word length between non-affective and fear-related stimuli across both literal and metaphoric stimuli, as revealed by independent t-tests (p<0.05). Metaphorical paragraphs (M = 15.05, SD = 2.2 sec) took significantly (t(119)=2.32, p = 0.013) longer to process than our literal paragraphs (M = 12.2, SD = 1.1 sec). Due to the fact that literal and metaphoric stimuli took different times to process we did not directly compare metaphoric and literal stimuli in our univariate analysis. All affective language stimuli were rated for arousal and valence in a separate post-hoc piloting task consisting of 17 participants (8 males and 9 females). Participants rated affective language stimuli for arousal on a 5-point Likert scale (i.e., 1-Not at all arousing to 5-Strongly arousing) and valence on a 7-point Likert scale (1-Strongly unpleasant to 7-Strongly pleasant) (Table 2). A two-way ANOVA with two levels (Literal vs. Metaphor) and two factors (Disgust, Fear) showed no significant difference in arousal between disgust and fear stimuli (F(1,156) = 0.08, p = 0.78), however there was a main effect of figurativeness (F(1,156) = 21.85, p = 0.00001) and a significant interaction effect (F(1,156) = 8.92, p = 0.0038). Post-hoc independent t-tests revealed that literal fear-related language was rated as significantly higher in arousal than fear metaphors (t(39)= 5.35, p = 0.00004). A two-way ANOVA with two levels (Literal vs. Metaphor) and two factors (Disgust, Fear) showed a significant difference in valence between disgust and fear stimuli (F(1,156) = 22.82, p < 0.0001) and a significant effect of figurativeness (F(1,156) = 46.7, p <0.0001), but no interaction term (F(1, 156) = 0.89, p = 0.35). Specifically, literal fear-related language was significantly lower in valence than fear metaphors (t(39) = - 4.89, p = 0.0001) and literal disgust language was significantly lower in valence than disgust metaphors (t(39) = -4.85, p = 0.0001). Non-affective language was rated as low in arousal for both metaphors (M = 1.94, SD = 0.27) and literal paragraphs (M = 1.95, SD = 0.31), as well as only slightly pleasant for metaphors (M = 4.12, SD = 0.30) and literal paragraphs (M = 4.38, SD = 0.5) with a score 4 of as being 4-neither pleasant nor unpleasant. Lastly, to exclude the possibility that differences in similarity across the first, second, and third sentences impacted our results we conducted a Latent Semantic Analysis (LSA) using the Sentence Comparison Interface with the general reading up to first year college (300 factors) topic space to compare similarity between successive sentences across all our stimuli (http://lsa.colorado.edu/). Two, one-way ANOVAs were carried out, one on the literal stimuli and one on the metaphorical stimuli. Each ANOVA had three factors (i.e., disgust, fear, non- affective) and two levels of sentence similarity (i.e., from sentence one to two, from sentence two to three). The first ANOVA showed no significant differences in sentence similarity (i.e., LSA cos value) across the first and second sentence, and, subsequently, second and third sentences across all literal stimuli types [emotion category (F(2,38) = 0.54, p=0.59), sentence similarity (F(1,19) = 0.002, p = 0.96), and interaction emotion category x sentence similarity (F(2,38) = 1.66, p=0.20)]. The second ANOVA showed a significant effect of emotion category (F(2,38) = 3.55, p = 0.04), but no effect of sentence similarity (F(1,19) = 3.22, p=0.09), or interaction effect emotion x sentence similarity (F(2,38) = 0.06, p=0.94). A post-hoc 2-tailed t- test showed that this effect was mainly due to the fact that the non-affective metaphor had a significantly larger LSA cos value (i.e., the sentences were more similar) than the fear metaphors (t(19) = -2.47, p=0.02). We note that while the inclusion of fear stimuli was used originally as an emotional control (i.e., to make sure activations found for disgust stimuli reflected disgust per se and not merely general differences in valence and arousal), as fear stimuli were not well matched with either disgust or non-affective stimuli across a number of dimensions above (valence, average word-length, and LSA cos value) our primary focus is on disgust metaphor. 35 Experimental Paradigm: Inside an fMRI scanner, participants were asked to silently read sentences for comprehension and at the end of each functional run to answer simple yes-no questions such as, “Was the last sentence you read political or not political in nature?”. Picture stimuli were viewed passively as a localizer task for affective brain regions. All stimuli were projected onto a screen in the scanner using Matlab Psychophysics toolbox (Psychtoolbox 3, www.psychtoolbox.org). Participants were presented with the linguistic stimuli first (comprehension task) which were divided into five runs each lasting 8.4 minutes. In order to counterbalance metaphoric and literal pairs, such that no participant ever saw both the metaphor and its literal pair, we only showed 20 metaphorical paragraphs of the first 20 pairs and 20 literal paragraphs from the other 20 pairs. Participants viewed a total of 3 metaphorical paragraphs and 3 literal paragraphs for each category (neutral, fear, and disgust) per run. Metaphor and literal paragraphs were presented in pseudorandom presentation (metaphors 17 secs, literal 13 secs, based on behavioral testing). This alternated with rest periods (13 secs) after each paragraph presentation. During the rest period, a fixation cross appeared during which participants fixated the cross and rested. After completing these five runs, participants were allowed to close their eyes and take a rest for approximately seven minutes during which we ran a structural scan (MPRAGE). Following this, participants were presented with the visual stimuli (picture task), which were divided in three runs each lasting 6.2 minutes. Participants viewed a total of 5 blocks per affect category (neutral, disgust, fear) per run (each block consisted of three non-repeated pictures presented for 4 sec with an inter-stimulus interval of 0.5 sec) for a total of 13 seconds in randomized presentation alternating with rest periods. Participants were instructed to view the images. MRI Data Acquisition: Functional MRI images were acquired with a Siemens MAGNETOM Trio 3T System with a 32-channel head matrix coil in the Dornsife Cognitive Neuroscience Imaging Center at the University of Southern California. A high-resolution anatomical scan was acquired for each subject: Structural T1-weighted magnetization-prepared rapid gradient echo (MPRAGE), TR=1950 ms, TE=2.26 ms, flip angle, x, FOV 256 x 256 mm squared, 1 mm resolution, 170 coronal slices. Whole-brain functional images were obtained with a T2* weighted single-shot gradient-recalled echo-planar imaging, echo-planar sequence (EPI) using blood-oxygenation-level-dependent contrast. Each functional image comprised of 37 contiguous axial slices (3.5 mm thick), acquired in interleaved mode, and with a repetition time (TR) of 2000 ms (echo time (TE) of 30 ms, flip angle (FA) 90 degrees, 64x64 matrix). Each participant underwent eight functional scanning sessions, with 252 volumes acquired for session of the comprehension task, and 186 volumes for each session of the picture task. Data Analysis: Univariate Analysis: All preprocessing and statistical analysis were carried out using FSL (Oxford Centre for Functional Magnetic Resonance Imaging of the Brain (FMRIB’s ) Software Library, http://www.fmrib.ox.ac.uk/fsl/index.html). Motion correction was performed using FMRIB’s Linear Image Registration Tool (MCFLIRT). Volumes were spatially smoothed using a 5 mm full-width at half-maximum Gaussian filter, prewhitened using FMRIB's Improved Linear Model (FILM), high-pass filtered (240 seconds), and registered into standard MNI space using the participant’s individual skull-stripped high-resolution T1 anatomical images using FMRIB’s Linear Image Registration Tool (FLIRT). The general linear model (GLM) was used 36 on individual voxels’ time series with FSL. The design matrix consisted of a synthetic hemodynamic response function (double gamma function) with its temporal derivative convolved with the input stimuli waveform in block design. In the main language task, six separate regressors were included in the model: 1) non- affective literal (NL); 2) disgust literal (DL); 3) fear literal (FL); 4) non-affective metaphor (NM); 5) disgust metaphor (DM); 6) fear metaphor (FM). In the functional localizer picture task design three separate regressors were included in the model: 1) neutral image (N); 2) disgust image (D); 3) fear image (F). The rest periods in between blocks served as the baseline contrast. Six motion parameters were also included in the design matrix as regressors to account for motion artifacts. The following additional contrasts were defined for each participant and used in the group analysis: For the language task (DL>NL, FL>NL, DM>NM, FM>NM). For the picture task (D>N, F>N, D>F). A mixed-effects analysis was carried out at the second level using FMRIB’s Local Analysis of Mixed Effects (FLAME 1). All statistical images were thresholded using Gaussian random field-based cluster analysis with an uncorrected z-score threshold of z>2.3 (p<0.01) and FWE corrected for multiple comparisons at a cluster extent threshold of p<0.05. This resulted in statistical maps of voxels significantly activated for the contrasts of interest as specified above. All statistical significance levels were conducted in a full-brain analysis. Additionally, small volume corrections were carried out within a priori defined regions of interest (ROIs). Small volume corrections in a priori defined ROIs were FWE corrected for multiple comparisons at the voxel level, voxel-corrected at a threshold of p<0.05. Region of interest analysis: We defined the following a priori ROIs involved in disgust processing including areas of primary/secondary gustatory cortex, mainly anterior areas of the insula (dorsal insula, ventral insula, but also posterior insula subregions for comparison), and adjacent frontal operculum, orbital frontal cortex (OFC), as well the various subregions of the basal ganglia (caudate, putamen, and pallidum) (Bastiaansen et al., 2009; A. J. Calder et al., 2007; Deen et al., 2011; Jabbi et al., 2008). Two additional ROIs involved in general emotional processing, including the bilateral amygdala and thalamus were also defined for comparison. The Harvard-Oxford Probabilistic Atlas was used to define the amygdala, frontal operculum, and OFC thresholded at P>0.65. The anterior insula and adjacent frontal operculum combined AIFO mask were created in two steps as follows. The anterior insula was first defined based on the location of the insular central sulcus (Afif & Mertens, 2010; Deen et al., 2011). Then hand- drawn ROIs of the anterior insula were created individually based on this anatomical distinction on each subject’s high-resolution T1 structural anatomical brain images and then transformed into the low-resolution functional space. Finally, we transformed the anatomically defined frontal operculum into the functional space and added it to the individually hand drawn anterior insula masks to create the combined AIFO mask. The entire insula was further subdivided into three insular subregions based on functionally identified clusters using cluster analysis (Deen et al., 2010). The clusters mean coordinates in MNI152 space were used to create three 5 mm spherical ROIs (e.g., the left ventral anterior insula (MNI coordinates, -33, 13, -7), left dorsal anterior insula (MNI coordinates, -38, 6, 2), and left posterior insula (MNI Coordinates, -38, -6, 5)). For the MVPA portion of the study we used the hand-drawn anatomically defined ROIs of the AIFO that averaged ~350 voxels. Multivariate Analysis: MVPA was carried out using the PyMVPA 0.6 software package (http://www.pymvpa.org/; [88]) and a linear support vector machine from LibSVM (http://www.csie.ntu.edu.tw/~cjlin/libsvm). Preprocessing consisted of concatenating all three 37 picture runs to form the picture data set, while all five language runs were concatenated to form the language data set for each subject. Each data set was then motion-corrected to the middle slice using FSL. Following this, each data set was linearly detrended and normalized to Z-scores using PyMVPA. Feature selection included restricting our features to voxels within both the left AIFO (LAIFO) and right AIFO (RAIFO), our main regions of interests. Additionally, for cross- modal classification analysis only we used a one-way ANOVA statistic to select the 30% most informative voxels within either the LAIFO or RAIFO masks using the training set only (picture data set) as this threshold best discriminated among the picture categories (Figure 1). Given our block design (picture blocks were 13 sec with a TR of 2 sec) we first took into account the hemodynamic lag (4 sec) following stimulus onset and selected the subsequent six volumes as individual samples for input to the classifier. This resulted in 90 samples per condition across all three picture functional runs combined. Similarly, for the literal blocks we also took six volumes per block (each block was 13 sec and TR was 2 sec) as input to the classifier, accounting for the hemodynamic lag. This resulted in 90 samples per condition across all five language functional runs. Lastly, for the metaphor samples we also took six volumes per block in order to match the number of samples across both the picture and literal runs. Similarly, six volumes of the metaphoric blocks (17 sec) after accounting for a hemodynamic lag (4 secs). The last six volumes were chosen based on the fact that the disgust content of the metaphorical paragraphs occurred in the last two sentences. Within-Modality MVPA Classification: Classification (training and testing) within the picture categories (N, D, and F) was carried out through a ‘a leave-one-run-out’ cross validation approach. For each subject all possible two-way classifications were obtained including “N vs. D”, “N vs. F”, and “D vs. F”. Similarly, for classification within the literal conditions (NL, DL, and FL) we carried out a ‘a leave-one-run-out’ cross validation approach, as well. For each subject all pairwise classifications were tested: “NL vs. DL”, “NL vs. FL”, and “DL vs. FL”. Lastly, for classification within the metaphoric conditions (NM, DM, FM) we also carried out a ‘a leave-on-run-out’ cross validation. For each subject all pairwise classifications were tested: “NM vs. DM”, “NM vs. FM”, and “DM vs. FM”. For statistical analysis a ‘a leave-one-run-out’ cross-validation approach was employed to train and test the classifier algorithm: In each cross- validation step, the classifier was trained on all but one fMRI run (train on 60 samples, test on 30 samples). This procedure was repeated, each time using a different run as test data set. In each cross-validation step, classifier performance was calculated as the classification accuracy (i.e. the number of correct guesses divided by the number of test trials). The overall performance for each classification task was estimated as the mean classification accuracy across all cross-validation steps. Cross-Modality MVPA Classification: We performed a cross-modal classification analysis where we trained the classifier on the picture task data and tested the performance of this classifier using the metaphor task data. For each subject all pairwise classifications were tested: “N vs. D”, “N vs. F”, and “D vs.F”. Next, we trained the classifier using the picture task data and tested the performance of this classifier using the literal task data. For each subject all pairwise classifications were tested: “N vs. D”, “N vs. F”, and “D vs.F”. MVPA Statistical Analysis: A permutation test was used to calculate significance by building a null-distribution. Only the training set was permuted by pairing each fMRI volume or sample with a random label, thus abolishing any information contained in the training set during training of the classifier. The permuted training set was tested on the un-permuted test data set 38 10,000 times to obtain 10,000 accuracy scores per subject. The null-distribution was created by randomly selecting one accuracy score from each subject’s permuted accuracy scores in order to obtain a group average, repeating this step 10,000 times in order to obtain a group-level null distribution. Significance of each correctly labeled accuracy scores was then calculated using the group-level null-distribution. RESULTS fMRI Univariate Results Literal Language: A whole-brain analysis for the contrast ‘disgust literal vs. non-affective literal’ revealed several large significant clusters of activity including a cluster within the left inferior frontal gyrus (IFG), pars opercularis (BA 44) that extended into areas of the IFG, pars triangularis (BA 45) and left frontal pole (BA 10). A significant cluster was also found within the left inferior temporal gyrus that extended into the left temporal occipital fusiform gyrus. Lastly, three additional emotion-related clusters were found. The first significant cluster within the left thalamus that included areas of basal ganglia (bilateral caudate) and the right amygdala with the second significant cluster within the left amygdala extending into areas of the basal ganglia (left pallidum). Lastly, a significant cluster within the left temporal pole extended into areas associated with primary/secondary gustatory cortex including the left insula and orbital frontal cortex (OFC) (Table 3). In line with previous findings that suggest a role for the AIFO in disgust processing and gustatory/olfactory processing, a small volume correction (SVC, p<0.05 FWE) showed significant clusters of activity in the bilateral amygdala, bilateral thalamus, bilateral caudate, left pallidum, left insula (ventral/dorsal and posterior insula), and bilateral OFC (Table 4). For the contrast ‘fear literal vs. non-affective literal’ we saw significant clusters of activation at the whole brain level within three clusters. The first significant cluster was within the left frontal pole but extended into areas of the left inferior frontal gyrus, pars opercularis (BA 44) and pars triangularis (BA 45), left middle frontal gyrus, and left precentral gyrus. Furthermore, we also found a significant cluster within the left posterior supramarginal gyrus that additionally encompassed areas of the left anterior supramarginal gyrus, left superior parietal lobule, and left postcentral gyrus. Lastly, a significant visual cluster within the left middle temporal gyrus extending into the left inferior lateral occipital cortex (LOC) was also found for this same contrast (Table 3). A small voxel correction (SVC, p<0.05 FWE) additionally showed that the left OFC (BA 11) showed significant clusters of activity for this contrast, as well (Table 4). Lastly, the contrast ‘disgust literal vs. fear literal’ was performed to determine to what extent the responses were specific to disgust. A small volume correction (SVC, p<0.05 FWE) showed significantly more activity within the left thalamus, bilateral caudate, bilateral amygdala, the left putamen, and the left pallidum for this contrast (Table 3). On the other hand, the contrast ‘fear literal vs. disgust literal’ showed at the whole brain level a large significant cluster of activity within the left superior parietal lobule that encompassed a number of sensorimotor brain areas including the bilateral precentral and postcentral gyri, but also the right superior parietal lobule. Additionally, a significant cluster of activity was found within the left frontal pole, as well as, two additional clusters within parietal areas. Mainly, a significant cluster was found within the left anterior supramarginal gyrus, as well as, another parietal cluster within the right angular 39 gyrus. Lastly, a small volume correction (SVC, p<0.05 FWE) further showed that the contrast ‘fear literal vs. disgust literal’ recruited the bilateral posterior insula. Metaphoric Language: The contrast ‘disgust metaphor vs. non-affective metaphor’ showed significant activity mainly within a priori defined ROIs (Table 4) within the left amygdala, left pallidum, left ventral anterior insula, and in the more ventral region of left OFC (SVC, p<0.05 FWE). Similarly, for the contrast ‘fear metaphors vs. non-affective metaphors’ a small volume correction (SVC, p<0.05 FWE) showed significant activity within left ventral anterior insula, left dorsal anterior insula, and left posterior insula. Lastly, the contrast ‘disgust metaphors vs. fear metaphors’ revealed significant increased activity within the left amygdala using a small volume correction (SVC, p<0.05 FWE). At the whole brain-level the contrast ‘fear metaphors vs. disgust metaphors’ revealed a large significant cluster within the right lateral occipital cortex that included areas of the right angular gyrus, right superior parietal lobule, right posterior supramarginal gyrus, and right precentral and postcentral gyri. An additional significant cluster was found within the right lingual gyrus that encompassed areas of the right precuneous. Lastly, a significant cluster of activity was found within the right middle frontal gyrus extending into the right frontal pole, as well as, a significant cluster within the right paracingulate gyrus. Lastly, using a small volume correction within a priori defined ROIS showed activity within the left dorsal anterior insula and posterior insula for this same contrast (SVC, p<0.05 FWE). fMRI Multivariate Results Within-Modality: In the LAIFO, a classifier trained on the picture emotion categories and tested on its ability to distinguish them was able to classify significantly above chance level across all three, two-way classifications: D vs. N (CR=0.69, p<0.05), D vs. F (CR=0.64, p<0.05), and F vs. N (CR=0.60, p<0.05). Two within-modality (language only) classifications were subsequently performed. First we trained a classifier to distinguish among the three emotion categories using literal language samples, and then we subsequently tested its performance on how well it could distinguish between the same emotion categories using metaphor samples. All two-way classifications were performed with average accuracy scores plotted as red vertical lines in Figure 2 (b). The classifier could significantly distinguish between D vs. N and D vs. F, but not F vs. N when compared to chance level. Secondly, we wanted to see how well we could distinguish among the three emotion categories based on activity patterns in the LAIFO using only the metaphor samples. Here we obtained significant average classification accuracies across all two-way classifications, again plotted as red vertical lines in Figure 2 (c): D vs. N (CR=0.56, p=0.0001), D vs. F (CR=0.56, p=0.0001), and F vs. N (CR=0.52, p=0.0473). Cross-Modality: In the cross-modal analysis, a classifier was trained on the task-related functional hemodynamic responses in the LAIFO to distinguish between pictures of different affective emotion categories. It was then tested on its ability to distinguish emotion categories across metaphor conditions. The two-way classification D vs. N showed that the classifier was able to significantly distinguish between neutral and disgust metaphoric stimuli above chance level with an average accuracy score of CR=0.56, p=0.0001 (indicated in Figure 2 (a) by the vertical red line) solely based on its ability to successfully distinguish between disgust and neutral pictures. The other two-way classifications did not achieve significant classification from chance level (e.g., D vs. F: CR=0.52, p=0.22; F vs. N: CR=0.51, p=0.05). Only the LAIFO, but 40 not the RAIFO achieved significant classification accuracies across both within-modality and cross-modality classifications. DISCUSSION In this study, we set out to test whether metaphors that imply ‘moral disgust’ engage affective and sensorimotor representations previously implicated in physical disgust processing, mainly areas of primary and secondary gustatory cortices (i.e., anterior insula/frontal operculum and OFC) (Adolphs et al., 2003; Bastiaansen et al., 2009; Calder et al., 2000; Heining et al., 2003; Jabbi et al., 2008; Jabbi et al., 2007; Royet et al., 2003; Small et al., 2003; Small & Prescott, 2005; Stark et al., 2007; Wicker et al., 2003; Zald et al., 1998), as well as, subregions of the basal ganglia previously implicated in disgust processing (Calder et al., 2007; Jabbi et al., 2008; Jabbi et al., 2007; Phillips et al., 1998; Phillips et al., 1997; Sprengelmeyer, Rausch, Eysel, & Przuntek, 1998; van der Gaag, Minderaa, & Keysers, 2007; von dem Hagen et al., 2009; Wicker et al., 2003). The univariate results highlight the importance of a left-lateralized set of brain regions relevant to the processing of physical disgust including the ventral anterior insula, OFC, and pallidum during the processing of disgust metaphor as compared to non-affective metaphor, with similar but more bilateral findings for literal disgust sentences compared to non- affective literal sentences. Differences in connectivity and function suggest that the ventral anterior subregion of the insula may be an initial stage for the processing of affective olfactory and gustatory stimuli or ‘a site for olfactory-gustatory convergence’ (Deen et al., 2011), while the dorsal region is rather involved in the later deployment of top-down attentional and control mechanisms that can subsequently impact behavior (Deen et al., 2011). Furthermore, the posterior insula is rather implicated in the affective dimension of somatosensory stimuli (Deen et al., 2011). Therefore, the fact that disgust metaphors activated the ventral anterior insula is in line with the idea that disgust language in a metaphorical context recruits modality-preferential affective gustatory/olfactory representations relating to the processing of pathogen disgust. Taken together these findings are in agreement with previous studies that have shown that words semantically related to taste (i.e., salt) recruited areas of primary/secondary gustatory cortex including the anterior insula and adjacent frontal operculum, as well as the orbital frontal cortex, a higher- order gustatory center (Barros-Loscertales et al., 2012). The univariate analysis also revealed greater activation in the basal ganglia for the processing of disgust language either in a literal or metaphoric context compared to non-affective language, in agreement with previous literature indicating a role for the basal ganglia in disgust processing (Calder et al., 2007; Jabbi et al., 2008; Jabbi et al., 2007; Phillips et al., 1998; Phillips et al., 1997; Sprengelmeyer et al., 1998; van der Gaag et al., 2007; von dem Hagen et al., 2009; Wicker et al., 2003). In particular, the left ventral pallidum was preferentially activated for processing disgust-related metaphors as compared to non-affective metaphors. In non-human primates, data indicate that the basal ganglia integrate motor and non-motor (including emotional) signals in the service of the regulation of goal-directed actions and may be involved in the generation of motivated withdrawal behaviors (Haber, 2003; Tachibana & Hikosaka, 2012), mainly the left ventral pallidum has been implicated in the generation of motivation signals (Mogenson, Jones, & Yim, 1980; Tachibana & Hikosaka, 2012). It is possible that, therefore, that activation of the left pallidum for disgust metaphors reflects the processing of 41 signals leading to motivated withdrawal behaviors relevant to the emotion of disgust. This would be in line with human studies showing that activity within the left pallidum (but also anterior insula) in response to disgusting foods, but not appetizing or bland foods, correlates with an individual’s disgust propensity and sensitivity (Calder et al., 2007; Stark et al., 2007), as well as, evidence showing that patients with neurological disease impacting the basal ganglia have deficits in disgust recognition and experience (Hayes, Stevenson, & Coltheart, 2007; Hennenlotter et al., 2004; Kipps, Duggins, McCusker, & Calder, 2007; Sprengelmeyer et al., 1998; Thieben et al., 2002; Wang, Hoosain, Yang, Meng, & Wang, 2003). Nevertheless, the specificity of the basal ganglia in disgust processing has been questioned with some arguing that deficits in disgust processing at least in patients with Huntington’s disease could also reflect a more generalized deficit in the processing of negative emotions (Milders, Crawford, Lamb, & Simpson, 2003; Sprengelmeyer et al., 1996). It is quite possible that the left pallidum activation in response to disgust metaphor compared to non- affective metaphor in our study, along with anterior insula activation, primarily reflects the processing of language stimuli with negative valence. This could also explain the left amygdala activation that was seen during the processing of disgust language stimuli compared to non- affective and fear language stimuli, especially as our disgust stimuli were rated as more negatively valenced compared to our fear stimuli. In fact, a number of studies on emotion word processing, more generally, show the involvement of a network of emotion-related brain regions, including the orbitofrontal and frontopolar cortex, anterior cingulate gyrus, but also the insula, and basal ganglia to be preferentially active during the processing of emotion words compared to non-emotion words or low arousal/valence words, for reviews see, (Moseley et al., 2011; 2015). Kousta et al., 2011 similarly showed that the inferior occipital gyrus, inferior temporal gyrus, middle temporal gyri, angular gyrus, but also the putamen and insula in the left hemisphere were all modulated non-linearly (U-shaped design) by valence across both concrete and abstract words. Thus, while activations of subregions of the anterior insula and basal ganglia, for disgust metaphor might reflect aspects of physical disgust processing, the present results do not rule out the possibility that these emotion-related activations merely reflect general differences in negative valence of our stimuli (Lindquist et al., 2012). The notion that the activations we found in the anterior insula/frontal operculum for disgust metaphor compared to non-affective metaphor merely reflects differences in valence of our stimuli, is in line with the fact that disgust metaphors couldn’t be distinguished from fear metaphors within this region. Disgust language stimuli mainly showed increased activity in subcortical brain regions (i.e., amygdala, basal ganglia) compared to fear language stimuli (for both literal and metaphoric language), but not within the anterior insula/frontal operculum. Next, while our cross-modal MVPA analysis showed that a classifier could significantly distinguish between disgust metaphor and non-affective metaphor based on unique patterns of activity across voxels within the LAIFO that successfully distinguished pathogen disgust pictures from neutral pictures, this unique pattern of activity across voxels in the LAIFO was not necessarily specific to the emotion of disgust. When we trained the binary classifier to distinguish between disgust and fear pictures using patterns of voxel activity in the LAIFO and, subsequently, tested its ability to distinguish between disgust and fear metaphors, the average classification accuracy was not significantly above the chance level. Interestingly, a within modality-classification analysis, in which the classifier was trained to distinguish between distinct emotion metaphors (non-affective, disgust, and fear) and 42 subsequently tested on its ability to distinguish between a separate test set of metaphors, showed that all possible two-way classifications (Disgust Metaphor vs. Non-Affective Metaphor, Disgust Metaphor vs. Fear Metaphor, and Fear Metaphor vs. Non-Affective Metaphor) were significantly above chance classification accuracies (Figure 2). Similarly, when the classifier was trained to distinguish between distinct emotion categories by first training on the literal samples and subsequently testing its performance on the metaphor samples, it successfully distinguished between disgust and fear metaphors. These results suggest that there might be a unique signal within the LAIFO that can distinguish among disgust, fear, and non-affective language. We note this regions key role in emotional simulation including interoceptive processing (Chapman & Anderson, 2012; Damasio et al., 2013; Lindquist, Wager, Kober, Bliss-Moreau, & Barrett, 2012). Indeed, the anterior insula/frontal operculum has been implicated, more generally, in interoceptive processing and simulation of emotional states (i.e., experiencing/recognizing others pain and/or gustatory pleasure) (Jabbi & Keysers 2008; Singer et al., 2004, 2006). However, given that the disgust language stimuli were rated as more negatively valenced than fear stimuli (matching in arousal) does not permit us to rule out the possibility that the classifier was picking up more general differences in valence of our stimuli. Future studies will need to replicate these results while controlling for valence. Importantly, our fear language stimuli differed from our disgust language stimuli in that they recruited the posterior insula (but also dorsal anterior insula for metaphor), as well as, parietal areas (i.e., supramarginal gyrus) to a greater degree. Additionally, significant clusters of activity within parietal areas extended into sensorimotor areas (i.e., precentral/postcentral gyrus). Activations within the posterior insula, implicated in the processing of valenced somatosensory stimuli (Deen et al., 2011), and somatosensory areas (i.e., postcentral gyrus but also somatosensory-related areas, supramarginal gyrus) most likely reflects the fact that our fear metaphors involved external physical forces that could cause harm to the body, thereby, inducing fear (i.e., falling off a ledge, etc.), which may have relied more heavily on somatosensory processes associated with tactile sensations relating to pain. Importantly, these activations do not appear to reflect differences in negative valence, as fear stimuli were actually rated as slightly less negatively valenced than our disgust stimuli overall, yet recruited these areas to a greater degree. Rather they suggest that one way in which the processing of our novel disgust and fear metaphors differed is that in fear, negative affect is a result of or is associated with external physical forces that are perceived to cause somatosensory pain, in contrast to the case of physical disgust which involves negative affect as a result of perception of how pathogen-based disgust inducers impact the viscera. This result rather points to the possibility that language users make use of embodied simulations relevant to emotion experience while processing affective metaphor. Taken together, our univariate and multivariate analysis suggest that disgust and fear metaphors may engage affective and sensorimotor systems of the brain relevant to that affective experience (i.e., perceptual, internal states, and behavioral features), but not necessarily dedicated neural circuitry only specific to either emotion. Future research is clearly needed to understand the various contributions of specific components of affective experience in affective metaphor comprehension. Still, if the cognitive and linguistic construal of immoral deeds as a matter of physical disgust involves affective and sensorimotor processes, then this may suggest a mechanism through which metaphoric language can associate, for example, general motivational tendencies (or negative affect) with novel complex social-moral ideas that can only be expressed 43 via language, eventually influencing behavior. More generally, our findings contribute to the discussion of whether or not abstract and non-abstract reasoning and language processing may be grounded in neural systems that relate to motor actions, perceptions, or affect (Barsalou, 2008, 2009; Barsalou et al., 2003; Gallese & Lakoff, 2005; Lakoff & Johnson, 1980; Niedenthal et al., 2005; Niedenthal, Winkielman, Mondillon, & Vermeulen, 2009). Acknowledgements: We would like to thank Jonas Kaplan and Tina Vuong for assisting with this study. This study was supported by the Intelligence Advanced Research Projects Activity (IARPA) via Department of Defense US Army Research Laboratory contract number W911NF- 12-C-0022. The U.S. Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright annotation thereon. Disclaimer: The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of IARPA, DoD/ARL, or the U.S. Government. References Adolphs, R., Tranel, D., & Damasio, A. R. (2003). Dissociable neural systems for recognizing emotions. Brain Cogn, 52(1), 61-69. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12812805 Aziz-Zadeh, L., & Damasio, A. (2008a). Embodied semantics for actions: findings from functional brain imaging. J Physiol Paris, 102(1-3), 35-39. doi:10.1016/j.jphysparis.2008.03.012 Aziz-Zadeh, L., Wilson, S. M., Rizzolatti, G., & Iacoboni, M. (2006). Congruent embodied representations for visually presented actions and linguistic phrases describing actions. Curr Biol, 16(18), 1818-1823. doi:10.1016/j.cub.2006.07.060 Barsalou, L. W. (2008). Grounded cognition. Annu Rev Psychol, 59, 617-645. doi:10.1146/annurev.psych.59.103006.093639 Barsalou, L. W. (2009). Simulation, situated conceptualization, and prediction. Philos Trans R Soc Lond B Biol Sci, 364(1521), 1281-1289. doi:10.1098/rstb.2008.0319 Barsalou, L. W., Kyle Simmons, W., Barbey, A. K., & Wilson, C. D. (2003). Grounding conceptual knowledge in modality-specific systems. Trends Cogn Sci, 7(2), 84-91. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12584027 Bastiaansen, J. A., Thioux, M., & Keysers, C. (2009). Evidence for mirror systems in emotions. Philos Trans R Soc Lond B Biol Sci, 364(1528), 2391-2404. doi:10.1098/rstb.2009.0058 Calder, A. J. (2003). Disgust discussed. Ann Neurol, 53(4), 427-428. doi:10.1002/ana.10565 Calder, A. J., Beaver, J. D., Davis, M. H., van Ditzhuijzen, J., Keane, J., & Lawrence, A. D. (2007). Disgust sensitivity predicts the insula and pallidal response to pictures of disgusting foods. Eur J Neurosci, 25(11), 3422-3428. doi:10.1111/j.1460- 9568.2007.05604.x Calder, A. J., Keane, J., Manes, F., Antoun, N., & Young, A. W. (2000). Impaired recognition and experience of disgust following brain injury. Nat Neurosci, 3(11), 1077-1078. doi:10.1038/80586 Calder, A. J., Lawrence, A. D., & Young, A. W. (2001). Neuropsychology of fear and loathing. Nat Rev Neurosci, 2(5), 352-363. doi:10.1038/35072584 44 Chapman, H. A., & Anderson, A. K. (2012). Understanding disgust. Ann N Y Acad Sci, 1251, 62- 76. doi:10.1111/j.1749-6632.2011.06369.x Citron, F. M., & Goldberg, A. E. (2014). Metaphorical sentences are more emotionally engaging than their literal counterparts. J Cogn Neurosci, 26(11), 2585-2595. doi:10.1162/jocn_a_00654 Craig, A. D. (2009). How do you feel--now? The anterior insula and human awareness. Nat Rev Neurosci, 10(1), 59-70. doi:10.1038/nrn2555 Damasio, A., Damasio, H., & Tranel, D. (2013). Persistence of feelings and sentience after bilateral damage of the insula. Cereb Cortex, 23(4), 833-846. doi:10.1093/cercor/bhs077 Deen, B., Pitskel, N. B., & Pelphrey, K. A. (2011). Three systems of insular functional connectivity identified with cluster analysis. Cereb Cortex, 21(7), 1498-1506. doi:10.1093/cercor/bhq186 Desai, R. H., Binder, J. R., Conant, L. L., Mano, Q. R., & Seidenberg, M. S. (2011). The neural career of sensory-motor metaphors. J Cogn Neurosci, 23(9), 2376-2386. doi:10.1162/jocn.2010.21596 Desai, R. H., Conant, L. L., Binder, J. R., Park, H., & Seidenberg, M. S. (2013). A piece of the action: modulation of sensory-motor regions by action idioms and metaphors. Neuroimage, 83, 862-869. doi:10.1016/j.neuroimage.2013.07.044 Eskine, K. J., Kacinik, N. A., & Prinz, J. J. (2011). A bad taste in the mouth: gustatory disgust influences moral judgment. Psychol Sci, 22(3), 295-299. doi:10.1177/0956797611398497 Gallese, V., & Lakoff, G. (2005). The Brain's concepts: the role of the Sensory-motor system in conceptual knowledge. Cogn Neuropsychol, 22(3), 455-479. doi:10.1080/02643290442000310 Haber, S. N. (2003). The primate basal ganglia: parallel and integrative networks. J Chem Neuroanat, 26(4), 317-330. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14729134 Hauk, O., Johnsrude, I., & Pulvermuller, F. (2004). Somatotopic representation of action words in human motor and premotor cortex. Neuron, 41(2), 301-307. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14741110 Hauk, O., & Pulvermuller, F. (2004). Neurophysiological distinction of action words in the fronto-central cortex. Hum Brain Mapp, 21(3), 191-201. doi:10.1002/hbm.10157 Hauk, O., & Tschentscher, N. (2013). The Body of Evidence: What Can Neuroscience Tell Us about Embodied Semantics? Front Psychol, 4, 50. doi:10.3389/fpsyg.2013.00050 Hayes, C. J., Stevenson, R. J., & Coltheart, M. (2007). Disgust and Huntington's disease. Neuropsychologia, 45(6), 1135-1151. doi:10.1016/j.neuropsychologia.2006.10.015 Heining, M., Young, A. W., Ioannou, G., Andrew, C. M., Brammer, M. J., Gray, J. A., & Phillips, M. L. (2003). Disgusting smells activate human anterior insula and ventral striatum. Ann N Y Acad Sci, 1000, 380-384. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14766651 Hennenlotter, A., Schroeder, U., Erhard, P., Haslinger, B., Stahl, R., Weindl, A., . . . Ceballos- Baumann, A. O. (2004). Neural correlates associated with impaired disgust processing in pre-symptomatic Huntington's disease. Brain, 127(Pt 6), 1446-1453. doi:10.1093/brain/awh165 Inbar, Y., Pizarro, D. A., Knobe, J., & Bloom, P. (2009). Disgust sensitivity predicts intuitive disapproval of gays. Emotion, 9(3), 435-439. doi:10.1037/a0015960 45 Jabbi, M., Bastiaansen, J., & Keysers, C. (2008). A common anterior insula representation of disgust observation, experience and imagination shows divergent functional connectivity pathways. PloS one, 3(8), e2939. Retrieved from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citati on&list_uids=18698355 Jabbi, M., & Keysers, C. (2008). Inferior frontal gyrus activity triggers anterior insula response to emotional facial expressions. Emotion, 8(6), 775-780. doi:10.1037/a0014194 Jabbi, M., Swart, M., & Keysers, C. (2007). Empathy for positive and negative emotions in the gustatory cortex. Neuroimage, 34(4), 1744-1753. doi:10.1016/j.neuroimage.2006.10.032 Kipps, C. M., Duggins, A. J., McCusker, E. A., & Calder, A. J. (2007). Disgust and happiness recognition correlate with anteroventral insula and amygdala volume respectively in preclinical Huntington's disease. J Cogn Neurosci, 19(7), 1206-1217. doi:10.1162/jocn.2007.19.7.1206 Kousta, S. T., Vigliocco, G., Vinson, D. P., Andrews, M., & Del Campo, E. (2011). The representation of abstract words: why emotion matters. J Exp Psychol Gen, 140(1), 14- 34. doi:10.1037/a0021446 Lakoff, G. (1996). Moral politics : what conservatives know that liberals don't. Chicago: University of Chicago Press. Lakoff, G., & Johnson, M. (1980). The metaphorical structure of the human conceptual system. Cognitive Science, 4(2), 195-208. doi:10.1016/S0364-0213(80)80017-6 Lee, S. W. S., & Schwarz, N. (2011). Wiping the Slate Clean: Psychological Consequences of Physical Cleansing. Current Directions in Psychological Science, 20(5), 307-311. doi:10.1177/0963721411422694 Lindquist, K. A., Wager, T. D., Kober, H., Bliss-Moreau, E., & Barrett, L. F. (2012). The brain basis of emotion: a meta-analytic review. Behav Brain Sci, 35(3), 121-143. doi:10.1017/S0140525X11000446 Milders, M., Crawford, J. R., Lamb, A., & Simpson, S. A. (2003). Differential deficits in expression recognition in gene-carriers and patients with Huntington's disease. Neuropsychologia, 41(11), 1484-1492. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12849766 Mogenson, G. J., Jones, D. L., & Yim, C. Y. (1980). From motivation to action: functional interface between the limbic system and the motor system. Prog Neurobiol, 14(2-3), 69- 97. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/6999537 Niedenthal, P. M., Barsalou, L. W., Winkielman, P., Krauth-Gruber, S., & Ric, F. (2005). Embodiment in attitudes, social perception, and emotion. Pers Soc Psychol Rev, 9(3), 184-211. doi:10.1207/s15327957pspr0903_1 Niedenthal, P. M., Winkielman, P., Mondillon, L., & Vermeulen, N. (2009). Embodiment of emotion concepts. J Pers Soc Psychol, 96(6), 1120-1136. doi:10.1037/a0015574 Phillips, M. L., Young, A. W., Scott, S. K., Calder, A. J., Andrew, C., Giampietro, V., . . . Gray, J. A. (1998). Neural responses to facial and vocal expressions of fear and disgust. Proc Biol Sci, 265(1408), 1809-1817. doi:10.1098/rspb.1998.0506 Phillips, M. L., Young, A. W., Senior, C., Brammer, M., Andrew, C., Calder, A. J., . . . David, A. S. (1997). A specific neural substrate for perceiving facial expressions of disgust. Nature, 389(6650), 495-498. doi:10.1038/39051 Pizarro, D., Inbar, Y., & Helion, C. (2011). On Disgust and Moral Judgment. Emotion Review, 3(3), 267-268. doi:10.1177/1754073911402394 46 Ponz, A., Montant, M., Liegeois-Chauvel, C., Silva, C., Braun, M., Jacobs, A. M., & Ziegler, J. C. (2014). Emotion processing in words: a test of the neural re-use hypothesis using surface and intracranial EEG. Soc Cogn Affect Neurosci, 9(5), 619-627. doi:10.1093/scan/nst034 Postle, N., McMahon, K. L., Ashton, R., Meredith, M., & de Zubicaray, G. I. (2008). Action word meaning representations in cytoarchitectonically defined primary and premotor cortices. Neuroimage, 43(3), 634-644. doi:10.1016/j.neuroimage.2008.08.006 Pulvermuller, F., Hauk, O., Nikulin, V. V., & Ilmoniemi, R. J. (2005). Functional links between motor and language systems. Eur J Neurosci, 21(3), 793-797. doi:10.1111/j.1460- 9568.2005.03900.x Raposo, A., Moss, H. E., Stamatakis, E. A., & Tyler, L. K. (2009). Modulation of motor and premotor cortices by actions, action words and action sentences. Neuropsychologia, 47(2), 388-396. doi:10.1016/j.neuropsychologia.2008.09.017 Royet, J. P., Plailly, J., Delon-Martin, C., Kareken, D. A., & Segebarth, C. (2003). fMRI of emotional responses to odors: influence of hedonic valence and judgment, handedness, and gender. Neuroimage, 20(2), 713-728. doi:10.1016/S1053-8119(03)00388-4 Schnall, S., Haidt, J., Clore, G. L., & Jordan, A. H. (2008). Disgust as embodied moral judgment. Pers Soc Psychol Bull, 34(8), 1096-1109. doi:10.1177/0146167208317771 Singer, T., Seymour, B., O'Doherty, J., Kaube, H., Dolan, R. J., & Frith, C. D. (2004). Empathy for pain involves the affective but not sensory components of pain. Science, 303(5661), 1157-1162. doi:10.1126/science.1093535 Singer, T., Seymour, B., O'Doherty, J. P., Stephan, K. E., Dolan, R. J., & Frith, C. D. (2006). Empathic neural responses are modulated by the perceived fairness of others. Nature, 439(7075), 466-469. doi:10.1038/nature04271 Small, D. M., Gregory, M. D., Mak, Y. E., Gitelman, D., Mesulam, M. M., & Parrish, T. (2003). Dissociation of neural representation of intensity and affective valuation in human gustation. Neuron, 39(4), 701-711. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12925283 Small, D. M., & Prescott, J. (2005). Odor/taste integration and the perception of flavor. Exp Brain Res, 166(3-4), 345-357. doi:10.1007/s00221-005-2376-9 Sprengelmeyer, R., Rausch, M., Eysel, U. T., & Przuntek, H. (1998). Neural structures associated with recognition of facial expressions of basic emotions. Proc Biol Sci, 265(1409), 1927-1931. doi:10.1098/rspb.1998.0522 Sprengelmeyer, R., Young, A. W., Calder, A. J., Karnat, A., Lange, H., Homberg, V., . . . Rowland, D. (1996). Loss of disgust. Perception of faces and emotions in Huntington's disease. Brain, 119 ( Pt 5), 1647-1665. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8931587 Stark, R., Zimmermann, M., Kagerer, S., Schienle, A., Walter, B., Weygandt, M., & Vaitl, D. (2007). Hemodynamic brain correlates of disgust and fear ratings. Neuroimage, 37(2), 663-673. doi:10.1016/j.neuroimage.2007.05.005 Tachibana, Y., & Hikosaka, O. (2012). The primate ventral pallidum encodes expected reward value and regulates motor action. Neuron, 76(4), 826-837. doi:10.1016/j.neuron.2012.09.030 Terrizzi, J. A., Shook, N. J., & Ventis, W. L. (2010). Disgust: A predictor of social conservatism and prejudicial attitudes toward homosexuals. Personality and Individual Differences, 49(6), 587-592. doi:10.1016/j.paid.2010.05.024 47 Tettamanti, M., Buccino, G., Saccuman, M. C., Gallese, V., Danna, M., Scifo, P., . . . Perani, D. (2005). Listening to action-related sentences activates fronto-parietal motor circuits. J Cogn Neurosci, 17(2), 273-281. doi:10.1162/0898929053124965 Thieben, M. J., Duggins, A. J., Good, C. D., Gomes, L., Mahant, N., Richards, F., . . . Frackowiak, R. S. (2002). The distribution of structural neuropathology in pre-clinical Huntington's disease. Brain, 125(Pt 8), 1815-1828. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12135972 van der Gaag, C., Minderaa, R. B., & Keysers, C. (2007). Facial expressions: what the mirror neuron system can and cannot tell us. Soc Neurosci, 2(3-4), 179-222. doi:10.1080/17470910701376878 Vigliocco, G., Kousta, S. T., Della Rosa, P. A., Vinson, D. P., Tettamanti, M., Devlin, J. T., & Cappa, S. F. (2014). The neural representation of abstract words: the role of emotion. Cereb Cortex, 24(7), 1767-1777. doi:10.1093/cercor/bht025 von dem Hagen, E. A., Beaver, J. D., Ewbank, M. P., Keane, J., Passamonti, L., Lawrence, A. D., & Calder, A. J. (2009). Leaving a bad taste in your mouth but not in my insula. Soc Cogn Affect Neurosci, 4(4), 379-386. doi:10.1093/scan/nsp018 Wang, K., Hoosain, R., Yang, R. M., Meng, Y., & Wang, C. Q. (2003). Impairment of recognition of disgust in Chinese with Huntington's or Wilson's disease. Neuropsychologia, 41(5), 527-537. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12559147 Wheatley, T., & Haidt, J. (2005). Hypnotic disgust makes moral judgments more severe. Psychol Sci, 16(10), 780-784. doi:10.1111/j.1467-9280.2005.01614.x Wicker, B., Keysers, C., Plailly, J., Royet, J. P., Gallese, V., & Rizzolatti, G. (2003). Both of us disgusted in My insula: the common neural basis of seeing and feeling disgust. Neuron, 40(3), 655-664. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14642287 Willems, R. M., & Casasanto, D. (2011a). Flexibility in embodied language understanding. Front Psychol, 2, 116. doi:10.3389/fpsyg.2011.00116 Zald, D. H., Lee, J. T., Fluegel, K. W., & Pardo, J. V. (1998). Aversive gustatory stimulation activates limbic circuits in humans. Brain, 121 ( Pt 6), 1143-1154. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9648549 48 TABLES AND FIGURES Table 1. Stimulus specificity ratings for Disgust, Fear, and Pictures Stimulus Ratings Disgust Pictures Fear Pictures Neutral Pictures Disgusting 3.16 ± 1.1 1.36 ± 0.9 1 ± 0 Frightening 1.2 ± 0.6 3.02 ± 0.8 1 ± 0 Other 1.17 ± 0.6 1.48 ± 0.2 1.03 ± 0.2 Arousal 4.04 ± 0.2 3.92 ± 0.8 1.7 ± 0.4 Valence 1.47 ± 0.2 1.51 ± 0.9 4.5 ± 0.5 Data are presented as mean ± SD for each emotion picture category. Each picture was rated on five rating scales (disgusting, frightening, other emotion, arousal, valence on a 5-point Likert scale). Table 2. Stimulus ratings for Disgust, Fear, and Neutral Paragraphs Stimulus Ratings Disgust Lit/Met Fear Lit/Met Non-affective Lit/Met Disgusting 4.45 ± 0.2 / 3.98 ± 0.4 2.62 ± 0.6 / 2.46 ± 0.4 1.78 ± 0.2 / 1.63 ± 0.2 Frightening 3.36 ± 0.6 / 3.17 ± 0.4 4.33 ± 0.2 / 3.67 ± 0.3 1.81 ± 0.2 /1.62 ± 0.2 Pleasant 1.39 ± 0.2 / 1.57 ± 0.2 1.48 ± 0.2 / 1.71 ± 0.2 2.51 ± 0.4 / 2.86 ± 0.5 Other 2.73 ± 0.3 / 2.92 ± 0.2 2.96 ± 0.2 / 2.95 ± 0.2 3.2 ± 0.2 / 3.1 ± 0.2 Arousal 3.23 ± 0.4 / 3.1 ± 0.3 3.48 ± 0.3 / 2.90 ± 0.3 1.95 ± 0.3 / 1.94 ± 0.3 Valence 1.61 ± 0.4 / 2.08 ± 0.3 1.92 ± 0.4 / 2.54 ± 0.4 4.38 ± 0.5 / 4.16 ± 0.3 Data are presented as mean ± SD of the averaged rating for each paragraph pair. Each paragraph pair was rated on six rating scales (disgusting, frightening, pleasant, other emotion, arousal, and valence on a 5-pont Likert scale). 49 Table 3. Whole brain activation patterns for main contrasts of interest Anatomic Region X Y Z Cluster size T Literal Task Disgust Lit > Non-affective Lit L IFG, pars triangularis -50 36 10 940 7.74 L IFG, pars opercularis -56 12 8 4.15 L Frontal pole -48 40 12 6.99 L Thalamus -10 -28 -2 584 5.12 R Thalamus 6 -18 0 3.74 L Caudate -10 4 12 3.60 R Caudate 10 4 12 4.36 R Amygdala 20 -2 -16 6.40 L Inferior temporal gyrus -44 -54 -12 256 5.70 L Temporal occipital fusiform -40 -48 -22 4.22 L Temporal pole -38 4 -18 246 5.34 L Insula -38 -12 -4 4.60 L OFC -32 18 -22 5.01 L Amygdala -22 -4 -14 245 6.98 L Pallidum -22 -6 -6 3.90 Fear Lit > Non-affective Lit L Frontal pole -42 42 2 1835 5.39 L IFG, pars triangularis -50 34 12 4.50 L IFG, pars opercularis -56 12 12 3.90 50 L Middle frontal gyrus -44 26 28 4.38 L Precentral gyrus -52 6 30 3.00 L Supramarginal gyrus, posterior -54 -46 42 1289 5.14 L Supramarginal gyrus, anterior -62 -34 32 4.84 L Superior parietal lobule -36 -52 64 3.17 L Postcentral gyrus -60 -22 40 2.96 L Middle temporal gyrus -54 -56 0 552 4.84 L Lateral Occipital Cortex, inferior -54 -70 10 4.65 Fear Lit > Disgust Lit L Superior parietal lobule -28 -52 66 5991 6.56 R Superior parietal lobule 26 -48 68 5.54 L Postcentral gyrus -22 -34 64 3.73 R Postcentral gyrus 22 -36 66 4.22 L Precentral gyrus -32 -20 72 4.48 R Precentral gyrus 28 -10 68 4.71 L Frontal pole -34 54 2 472 5.33 L Supramarginal gyrus, anterior -42 -28 34 302 4.64 R Angular gyrus 56 -54 34 289 4.66 Metaphor Task Fear Met > Disgust Met R Lateral occipital cortex, superior 48 -66 38 1918 4.81 R Angular gyrus 48 -54 44 3.96 R Superior parietal lobule 40 -44 62 4.33 R Supramarginal gyrus, posterior 50 -44 52 2.93 R Postcentral gyrus 52 -22 56 3.8 51 R Precentral gyrus 56 10 30 3.74 R Lingual gyrus 8 -62 6 1370 4.37 R Precuneous 20 -56 10 3.51 L Occipital pole -14 -94 26 4.16 R Middle frontal gyrus 32 26 34 1355 5.12 R Frontal pole 36 56 12 4.24 R Paracingulate gyrus 0 40 28 684 4.49 Activation threshold was set at p<0.05, FWE corrected for multiple comparisons at the cluster level for the whole brain. MNI (Montreal Neurological Institute) coordinates refer to peak voxel coordinates. Table 4. Small Volume Correction Analysis Anatomic Region X Y Z T Literal Task Disgust Lit > Non-affective Lit L OFC -32 18 -22 5.01 R OFC 32 22 -20 3.40 L Ventral anterior insula -34 8 -6 2.78 L Dorsal anterior insula -32 6 6 3.10 L Posterior insula -36 -8 -2 3.13 L Thalamus -4 -26 2 5.01 R Thalamus 2 -10 4 4.61 L Caudate -8 4 10 3.73 R Caudate 8 6 6 4.72 L Pallidum -22 -6 -6 3.9 L Amygdala -22 -4 -14 6.98 R Amygdala 20 -2 -16 6.40 52 Fear Lit > Non-affective Lit L OFC -24 22 -18 3.84 Disgust Lit > Fear Lit L Thalamus -6 -8 4 3.62 L Caudate -12 2 12 2.73 R Caudate 10 10 4 2.65 L Amygdala -20 -4 -14 3.62 R Amygdala 18 -4 -14 3.37 L Putamen -28 -18 0 2.6 L Pallidum -22 -14 2 2.47 Fear Lit > Disgust Lit L Posterior insula -38 -16 12 2.6 R Posterior insula 38 -18 18 2.9 Metaphor Task Disgust Met > Non-affective Met L OFC -38 26 -14 3.78 L Ventral anterior insula -30 14 -14 2.81 L Pallidum -22 -2 -4 2.93 L Amygdala -24 -4 -16 3.90 Fear Met > Non-affective Met 53 L Ventral anterior insula -28 16 -10 2.8 L Dorsal anterior insula -40 8 8 2.78 L Posterior insula -36 -12 4 2.71 Disgust Met > Fear Met L Amygdala -22 -6 -12 3.22 Fear Met > Disgust Met L Dorsal anterior insula -38 10 4 3.01 L Posterior insula -40 -8 4 3.05 A small volume correction was used within a priori defined ROIs was conducted and corrected for FWE [bilateral amygdala, anterior insula (ventral, dorsal), posterior insula, pallidum, putamen, caudate, thalamus, and orbital frontal cortex] (SVC, voxel-corrected p<0.05). MNI (Montreal Neurological Institute) coordinates refer to peak voxel coordinates. 54 Figure 1. A priori defined ROIs. (A) Anatomical masks of functionally distinct regions of the left insula, mainly the dorsal insula, ventral insula, and posterior insula; (B) Anatomical masks of different regions of left basal ganglia including caudate, putamen, and pallidum; (C) Anatomical mask of the orbital frontal cortex; (D) Anatomical mask of the post-hoc ROI the left amygdala. The corresponding areas on the right hemisphere were also included as ROIs (not shown). Figure 2. Classification Rate vs. Number of Voxels in left AIFO. In the cross-modal classification analysis a one-way ANOVA statistic was used to select the 30% most informative voxels within the left AIFO mask, using the training set only (picture data set) as this threshold discriminated best among the picture categories. 55 Figure 3. Disgust metaphors > Non-affective metaphors. Clusters of activation for the contrast ‘disgust metaphor vs. non-affective metaphor’ (SVC p<0.05, FWE). After small volume correction significant activation was found in the following ROIs: (A) left ventral anterior insula (red) and adjacent ventral region of the left OFC (yellow) (MNI coordinates, -33, 13, -7); (B) The left ventral pallidum (light blue) and left amygdala (dark blue) is shown (MNI coordinates, - 24, -8, -4). 56 Figure 4. Within and Cross-Modality MVPA Analysis. (A) Samples from picture task used to train classifier to distinguish between emotion categories. Performance of the classifier tested using metaphor samples; (B) Literal samples from the language task used to train the classifier to distinguish between emotion categories. Performance of the classifier tested using metaphor samples; (C) A ‘leave-one-run-out’ cross-validation scheme used. Trained on metaphor samples from first four runs to distinguish between emotion categories. Test performance of classifier using metaphor samples in the last run. Run all combinations and average CRs. CR = Correct Classification Rate. P- value calculated based on null distribution from permuting training set only (n=10,000) to test accuracy of the classifier above chance level. 57 Metaphor in Politics: Bringing Affect to the Decision Space? Vesna Gamez-Djokic 1,2 , Elisabeth Wehling 4 , Lisa Aziz-Zadeh 1,2,3 1 Brain and Creativity Institute, University of Southern California, 2 Neuroscience Graduate Program, 3 Division of Occupational Science and Occupational Therapy, University of Southern California, 4 International Computer Science Institute, University of California, Berkeley Author Contributions: V.G.D conceived of the study, designed the study with E.W., contributed to stimuli, piloted stimuli, conducted the study, conducted the analysis and data interpretation, and wrote the manuscript. E.W. designed the study with V.G.D., designed and created linguistic stimuli, and edited manuscript. L.A.Z. oversaw all aspects of the study and edited manuscript. ABSTRACT Moral evaluations often construe (im)morality in terms of (im)purity/disgust (e.g., “A rotten thing to do.”). However, it is unclear what specific advantage there might be to using moral disgust metaphors when compared to literal counterparts (e.g., “A bad/immoral thing to do”). According to Conceptual Metaphor Theory, the concept of immorality is intimately tied with our understanding of impurity (hence physical disgust) as co-occurrences of these two concepts during development establish cross-domain mappings. One prediction is that comprehending moral disgust metaphors should automatically recruit emotion-related processes relevant to physical disgust experience and this may have consequences for moral decision- making in the brain. In the current study we investigated how the processing of familiar disgust metaphors that express moral political attitudes could distinctly modulate both more automatic emotion-related brain areas also relevant to physical disgust processing (i.e., areas of gustatory cortex) and brain areas implicated in more deliberate, top-down processes in decision-making (i.e., ventral medial prefrontal cortex (VMPFC) and dorsolateral prefrontal cortex (DLPFC)) when compared to literal paraphrases. Conservative and liberal participants read each statement presented during a reading-period followed by a response-period during which they indicated their degree of agreement with the statement. Our results indicated that moral disgust metaphors (both during reading and response periods) recruited emotion-related brain regions relevant to physical disgust processing when compared to their literal counterparts (matched for arousal and valence), but in a context-dependent fashion. Activity within emotion-related brain regions implicated in disgust processing significantly covaried with political orientation during the reading period for moral disgust metaphor when compared to literal counterparts with increased activity for those scoring high on political conservatism. Emotion-related brain regions relevant to disgust processing were also found during the response part of the task across all participants. Critically, literal moral reading and judgment showed increased activity in regions in the VMPFC and DLPFC when compared to moral disgust metaphors, providing additional evidence of a differential impact of moral disgust metaphors on moral processing in the brain. Taken together these findings suggest that moral disgust metaphors engage more automatic emotional processes relevant to physical disgust processing to a greater extent than their literal counterparts, which instead engage brain areas associated with more top-down processes in moral decision-making. Keywords: Metaphor, Morality, Insula, VMPFC, Conservatives, Liberals 58 INTRODUCTION Language used to describe physical disgust is routinely also used to talk about the concept of ‘moral disgust’ (e.g., “A rotten crime”, “A disgusting thing to do”, etc.). However, it is not entirely clear whether the construal of moral transgressions as disgusting acts is merely a linguistic coincidence or if it reflects something deeper about the abstract concept of ‘moral disgust’ and how we conceptualize/reason about right and wrong, more generally (J. Borg, Lieberman, & Kiehl, 2008; Pizarro, Inbar, & Helion, 2011; Wehling, 2015). The notion that emotions play an important role in our understanding of morality is clearly articulated in Conceptual Metaphor Theory (CMT) (Lakoff & Johnson, 1980), which argues that the abstract concept of immorality (target domain) is structured in part through cross-domain mappings with the concrete domain of physical impurity and, hence physical disgust (source domain), due to co- occurrences of these two concepts during experience. Similarly, the concept of morality, more generally, is intimately tied with our understanding of wellbeing. In line with CMT, embodiment theories, have, more generally, argued that our conceptual knowledge both for concrete and abstract concepts is grounded in our physical experiences (Barsalou, 1999, 2003, 2008; Barsalou, Kyle Simmons, Barbey, & Wilson, 2003; Gallese & Lakoff, 2005; Lakoff & Johnson, 1980; Niedenthal, Barsalou, Winkielman, Krauth-Gruber, & Ric, 2005). Two relevant predictions follow from this theory: First, CMT would predict that the processing of moral disgust metaphor should recruit modality-preferential processes relevant to physical disgust processing (i.e., gustatory/olfactory based disgust), such that appropriate inferences can be drawn from the source to the target domain (e.g., mainly motivated withdrawal) (Gallese & Lakoff, 2005). This is, more generally, in line with theories of embodied simulation semantics that have argued that understanding language about action, affect, and perception draws on sensorimotor and affective simulations involving the brain’s modal regions (Barsalou, 2003; Gallese & Lakoff, 2005; Pulvermuller, 1999, 2001; Pulvermuller, Hauk, Nikulin, & Ilmoniemi, 2005). In support of this, a growing number of neuroscientific studies have shown that the processing of literal language about perceptible scenes, actions, and emotion engage brain systems that draw on the same perceptual mechanisms in the brain involved in initial sensorimotor and affective experience, for reviews see (Hauk & Tschentscher, 2013; Meteyard, Cuadrado, Bahrami, & Vigliocco, 2012; Willems & Casasanto, 2011a), although not always (for critical reviews see Arbib, Gasser, & Barres, 2014; Mahon & Caramazza, 2008). While growing evidence suggests that the processing of literal language pertaining to action, perception, and affect recruits modality-preferential representations in the brain, studies looking at metaphors drawing on action, perceptible scenes, and affect, in contrast offer much more mixed findings (for reviews see Carr, Kavanagh, & Bergen, 2013; Gamez-Djokic et al., 2015). On the basis of this evidence it has been suggested that embodiment effects in metaphor processing may actually depend on the degree of conventionality with novel metaphors showing the most sensorimotor involvement, conventional metaphors intermediate involvement, and frozen or idiomatic expressions showing little to no involvement (Desai, Binder, Conant, Mano, & Seidenberg, 2011; Desai, Conant, Binder, Park, & Seidenberg, 2013). According to this view, 59 comprehending increasingly familiar metaphors should depend mainly on categorization-like processes, whereby, the source domain would be categorized as belonging to an abstract superordinate category relevant to the target domain, instead of a process involving comparison of source and target domain that would entail sensorimotor simulation (Gentner & Bowdle 2005; see also Keysar & Bly, 1999; Keysar et al., 2000). Nevertheless, a number of recent studies do show evidence of robust activation of domain-specific representations in sensorimotor and affective brain systems even for highly conventional metaphors when compared to equally familiar literal paraphrases. For example, Lacey et al., 2012 showed that familiar texture-based metaphors (‘she had a rough day’) activated somatosensory areas implicated in texture discrimination compared to literal counterparts (‘she had a bad day’) (Lacey, Stilla, & Sathian, 2012), while Citron and Goldberg 2014 also showed that taste-related metaphors (‘the break up was bitter for him’) activated parts of primary and secondary gustatory cortex when compared to literal counterparts (‘the break-up was bad for him’). In the current study we thus predicted that familiar moral disgust metaphors should, nevertheless, engage emotion-related processes relevant to physical disgust processing (i.e., gustatory cortex and basal ganglia) when compared to literal counterparts as predicted by CMT. It is well known that the primary function of the emotion of disgust is to motivate withdrawal and rejection of harmful entities, forming part of a larger pathogen avoidance system (Calder, 2003; Krusemark & Li, 2011). It has been suggested that this system may have evolved to deal with more abstract threats (i.e., moral transgressions) (Chapman et al., 2011). Previous studies show the involvement of primary and secondary gustatory cortex (i.e., anterior insula/frontal operculum and OFC), (Bastiaansen et al., 2009; Calder, 2003; Calder et al., 2007; Deen et al., 2011; Jabbi et al., 2008; Murphy, Nimmo-Smith, & Lawrence, 2003; Phan, Wager, Taylor, & Liberzon, 2002), as well as, subregions of the basal ganglia during the processing of disgust-inducing stimuli (Calder et al., 2007; Hayes, Stevenson, & Coltheart, 2007; Hennenlotter et al., 2004; Kipps, Duggins, McCusker, & Calder, 2007; Sprengelmeyer, Rausch, Eysel, & Przuntek, 1998; Sprengelmeyer et al., 1996; Thieben et al., 2002; Wang, Hoosain, Yang, Meng, & Wang, 2003). Specifically, Jabbi et al., 2008 have shown the involvement of the anterior insula and adjacent frontal operculum during the ingestion of bitter substances, observation of others ingesting the same bitter substances, and reading of disgust image-driven scripts. Moreover, electrical stimulation of the anterior subregion of the insula in epileptic patients has also been shown to generate feelings of nausea and bad smells/taste (Penfield, 1959; Penfield & Faulk, 1955). Relatedly, Calder et al., 2007 showed that activity mainly within the ventral anterior insula and pallidal subregion of the basal ganglia correlated with individual differences in sensitivity to experience disgust when viewing food-related stimuli inducing physical disgust versus visual stimuli relating to food that elicited pleasant or neutral gustatory/olfactory experiences (Calder et al., 2007). Thus, we hypothesized that moral disgust metaphor compared to their literal counterparts would involve the ventral anterior insula and pallidal subregion of the basal ganglia due to their putative involvement in the recognition and experience of disgust (i.e., emotion simulation), especially the former in gustatory/olfactory based disgust (Deen et al., 2011) (Calder et al., 2007). 60 Secondly, CMT would also predict that moral disgust metaphors may have consequences for how individuals’ reason or make evaluations in moral politics (Lakoff, 1996). We reasoned that metaphors that draw on physical disgust language and recruit the relevant affective processes may in the right contexts be able to impact the decision-space (Lee & Schwarz, 2012). This affective grounding could give moral disgust metaphors a ‘persuasive advantage’ when compared to, for example, using semantically comparable literal sentences (Citron & Goldberg, 2014; for a review see Sopory & Dillard, 2006). Support for this hypothesis comes from recent research in moral psychology that suggests that physical disgust can directly impact moral judgment (Haidt 2001), however this is currently debated (Landy & Goodwin 2015; Schnall et al., 2015). A number of behavioral studies suggest that experiencing physical disgust can impact moral judgment (Eskine, Kacinik, & Prinz, 2011; Inbar, Pizarro, Knobe, & Bloom, 2009; Lee & Schwarz, 2011; Pizarro, Inbar, & Helion, 2011; Schnall, Haidt, Clore, & Jordan, 2008; Terrizzi, Shook, & Ventis, 2010; Wheatley & Haidt, 2005). For example, Schnall et al. 2008 found that inducing disgust irrelevant to the task at hand by either exposing participants to a smelly trash can, a dirty vs. clean testing environment, or through recall of disgusting experiences led to harsher moral judgments. Furthermore, Eskine et al. 2011 showed that drinking a bitter substance irrelevant to a moral judgment task influenced moral judgment and provoked increased feelings of ‘moral disgust’ when compared to drinking sweet or neutral tasting substances. Lastly, Wheatley & Haidt 2005 used hypnosis to associate a neutral word with feelings of disgust through ‘posthypnotic suggestion’, which led to harsher moral judgments for moral vignettes containing the neutral word as opposed to those that did not. Thus, a number of studies provide evidence of an important link between the experience of physical disgust and moral decision- making. Nevertheless, the impact of physical disgust on moral judgment is the subject of debate and appears to depend on 1) the type of disgust inducer (i.e., gustatory and olfactory but not other types of disgust) 2) the fact that participants lack awareness of the underlying cause of the disgust experience 3) an individual’s propensity and sensitivity to experience disgust (see Landy & Goodwin 2015; Schnall et al., 2015). For example, the impact of physical disgust on moral judgment is particularly strong in conservative moral judgment, especially for judgments that relate to sexual practices (e.g., homosexuality) (Graham, Haidt, & Nosek, 2009; Haidt & Graham, 2007). Recent research suggests that the impact of physical disgust on conservative moral judgments may be due to increased sensitivity to experience disgust. A number of studies show that disgust sensitivity correlates specifically with political conservatism, particularly interpersonal disgust, in the US and abroad (Brenner & Inbar, 2014; Inbar, Pizarro, Iyer, & Haidt, 2012; Inbar, Pizarro, Knobe, & Bloom, 2009). Moreover, Smith et al., 2011 found that physiological responses to disgust (SCR, skin conductance change) increased significantly with self-reported political conservatism. Specifically, SCR was mainly correlated with conservative political attitudes related to gay marriage (opposing gay-marriage) and pre-marital sex. Thus, it is possible that the degree to which moral disgust metaphors activate neural processes associated with physical disgust will similarly depend on individual differences, with increased effects for those who hold more conservative views. However, it is unclear whether this will result in an impact on the degree of agreement with political statements for moral disgust metaphor as compared to their literal counterparts. 61 In general, cognitive neuroscience research on morality has identified the importance of an impulsive emotional system that interacts with a more deliberate and explicit top-down system when making moral decisions (i.e., dual-process theories of moral judgment) (Greene, Nystrom, Engell, Darley, & Cohen, 2004; Moll, De Oliveira-Souza, & Zahn, 2008; Young & Dungan, 2012) (Forbes & Grafman 2010). Green et al., 2004 showed evidence that moral dilemmas (i.e., high-conflict, emotionally salient dilemmas vs. low-conflict, emotionally salient dilemmas), involving a prepotent emotional response that had to be suppressed in favor of the utilitarian option, recruited areas implicated in cognitive control/abstract reasoning (DLPFC) and conflict monitoring (dACC). This, along with evidence that manipulating cognitive load during moral judgments can interfere with utilitarian judgments, but not more emotional moral judgments, suggests two systems one emotional and the other cognitive that compete in a winner-take-all scenario during moral decision-making in the brain (Greene et al., 2008; for a different perspective see Hutcherson et al., 2015). These findings reflect ongoing research in decision-making, more generally (Bechara, 2005; Gupta, Koscik, Bechara, & Tranel, 2011). In one influential view, the amygdala and striatum belong to an ‘impulsive system’, while the VMPFC and DLFPC have rather been associated with a ‘reflective system’. The ‘reflective system’ control signals from the impulsive system and integrates them with other information for optimal decision-making in line with long- term goals (Bechara, 2005; Gupta et al., 2011). Specifically, the VMPFC has been primarily associated with the computation of value of stimuli or choices in decision-making (Bartra, McGuire, & Kable, 2013; Bechara, 2005; Clithero & Rangel, 2014; Gupta et al., 2011). Similarly, Shenhav & Greene 2010 showed that activity within the VMPFC correlates with moral value (i.e., number of lives saved). Although our moral political statements are far from the classical moral dilemmas in these studies, we reasoned that moral disgust metaphors should nevertheless distinctly modulate these interacting systems, as moral disgust metaphors should draw more heavily on automatic, emotion-related brain regions perhaps at the cost of less engagement of more deliberative, ‘explicit’ systems when making decisions about moral political statements (i.e., degree of agreement). Therefore, in this study we were interested in investigating whether or not relatively familiar disgust metaphors that express specific moral political attitudes (e.g., “Using taxpayer money to cover healthcare costs for the uninsured is rotten.”) could distinctly modulate both more automatic emotion-related brain areas also relevant to physical disgust processing (i.e., amygdala, basal ganglia, and areas of gustatory/olfactory cortex) and brain areas implicated in more deliberate, top-down processes in moral decision-making (i.e., VMPFC and DLPFC) when compared to literal paraphrases (e.g., “Using taxpayer money to cover healthcare costs for the uninsured is wrong.”). We recruited conservative and liberal participants to read moral statements presented during a reading-period followed by a response-period during which they indicated their degree of agreement. Moral political statements consisted of moral disgust metaphors and their literal counterparts matched for relevant psycholinguistic variables, which reflected either conservative or liberal views relating to socio-economic issues. We analyzed both the reading period and the response period after the reading of each statement in the scanner during which subjects indicated the degree to which they agreed with each statement on a 7-point 62 Likert scale (1-strongly agree à 7-strongly disagree). Lastly, in order to investigate how our liberal and conservative participants processed physical disgust directly we conducted a separate localizer task where we presented participants with physical disgusting inducing-images and neutral images. We defined a number of a priori emotion-related regions of interest (ROI) relevant to physical disgust experience, mainly the primary and secondary gustatory cortex (i.e., bilateral ventral anterior insula, adjacent frontal operculum, and orbital frontal cortex (OFC)) (Bastiaansen, Thioux, & Keysers, 2009; Calder, 2003; Calder et al., 2007; Deen, Pitskel, & Pelphrey, 2011; Jabbi, Bastiaansen, & Keysers, 2008), as well as, the pallidal subregion of the basal ganglia implicated in physical disgust processing (Calder et al., 2007; Hayes, Stevenson, & Coltheart, 2007; Hennenlotter et al., 2004; Kipps, Duggins, McCusker, & Calder, 2007; Sprengelmeyer, Rausch, Eysel, & Przuntek, 1998; Sprengelmeyer et al., 1996; Thieben et al., 2002; Wang, Hoosain, Yang, Meng, & Wang, 2003). We also investigated brain regions previously implicated in more automatic emotion-related processes, including the amygdala (and brain regions relevant to physical disgust processing), as well as, more deliberate, top-down processes implicated in cognitive neuroscience studies of moral decision-making in the brain, mainly the VMPFC and bilateral DLPFC involved in encoding value in decision-making and cognitive control/abstract reasoning, respectively (Hutcherson, Montaser-Kouhsari, Woodward, & Rangel, 2015; Greene, Nystrom, Engell, Darley, & Cohen, 2004; Moll, De Oliveira-Souza, & Zahn, 2008; Young & Dungan, 2012), as well as, in decision-making, more generally (Bechara, 2005; Gupta, Koscik, Bechara, & Tranel, 2011). We predicted that familiar disgust metaphors that express specific moral political attitudes (e.g., “Using taxpayer money to cover healthcare costs for the uninsured is rotten.”) would modulate both emotion-related brain areas relevant to physical disgust processing (i.e., areas of gustatory cortex and basal ganglia) and brain areas implicated in more deliberate, top- down processes in moral decision-making (i.e., VMPFC and DLPFC) distinctly when compared to literal paraphrases (e.g., “Using taxpayer money to cover healthcare costs for the uninsured is wrong.”), matched for relevant psycholinguistic variables (semantic similarity, familiarity, and arousal/valence). Moreover, we expected that emotion-related activations for moral disgust metaphors should have consequences for decision-making (e.g., possibly make statements seem more ‘wrong’) compared to literal counterparts. Lastly, we also anticipated that participants’ political orientation might impact the processing of moral disgust metaphors vs. literal counterparts, as well as, the viewing of disgust images vs. neutral images, as (im)purity concerns have been found to be a strong psychological/physiological trait amongst conservatives (Smith et al., 2011). METHODS Participants: Nineteen right-handed, native English speakers (age range 18-35, twelve females and seven males) provided informed consent and were paid for participating in this study. All subjects had normal hearing and vision, and no history of neurological illness. We recruited participants who self-identified as either liberal or conservative and held strong political beliefs. 63 Each subject was subsequently asked to complete a political questionnaire prior to the scanning session that assessed political orientation and political attitudes concerning social and economic issues. Picture Stimuli: 90 images were selected per emotion category (neutral, disgust) for a total of 180 images. The images came from the International Affective Picture Site (IAPS) (http://csea.phhp.ufl.edu/Media.html#topmedia), the Geneva International Affective Picture Site (http://www.affective-sciences.org/researchmaterial), and the Internet. A group of independent participants (subjects 10, 5 male, 5 female) viewed three consecutively presented disgust images (2 secs duration, inter-trial interval of .25 secs) and were then asked to rate their experience on that three-picture set indicating whether they experienced the emotion of disgust or another emotion using a 5-point Likert scale (i.e. 1- Did not experience disgust to 5- Strongly experienced disgust, 1- Did not experience another emotion to 5- Strongly experienced another emotion, (e.g., fear, anger, happiness, sadness)). Disgust image ratings were averaged across all participants and significantly (t(29) = -13.92, p<0.001) induced disgust (M = 2.88, SD = 0.63) and not another emotion (M = 1.20, SD = 0.18). The neutral images consisted of inanimate objects (e.g., cups, forks, empty apartments, etc.) and did not induce disgust (M = 1, SD = 0) and scored low for inducing other emotions (M = 1.28, SD = 0.22). Language Stimuli: Eighty familiar political disgust metaphors and eighty literal counterparts were created and matched in length (metaphors were M = 11.48, SD = 1.53 words in length, literal paraphrases were M = 11.49, SD = 1.46 words in length), (t(79) = -0.15, p = 0.89, n.s.) . Additionally, twenty apolitical literal sentences were created (literal sentences were M = 11, SD = 1.5 words in length) (e.g., “For traffic signs to be effective their visibility must be maintained.”). These sentences were included as an additional control, as well as, to ensure that participants were paying attention, as these sentences had obvious answers (i.e., either agree or disagree). However, due to timing limitations this control had fewer trials compared to metaphor and literal sentences and thus limit interpretation. Half of the political disgust metaphors and their literal counterparts reflected liberal views, while the other half reflected conservative views, surrounding socio-economic issues pertinent to society today. Every metaphorical sentence (e.g., “Using taxpayer money to cover healthcare costs for the uninsured is rotten.”) had a literal counterpart (e.g., “Using taxpayer money to cover healthcare costs for the uninsured is unethical.”), such that the last word of the sentence led to either a literal or metaphorical interpretation, however no subject saw the same sentence twice, as we created two counterbalanced sets. In the political disgust metaphors we used 40 words relating to physical disgust (e.g., rotten, dirty, and foul). In the literal paraphrases we used 40 words that provided a semantically similar interpretation given the context (e.g, unethical, wrong, and terrible). Each word (adjective) was used twice to create two counterbalanced sets for a total of eighty metaphors and eighty literal paraphrases, with each participant only seeing one set each or a total of 40 metaphors and 40 literal paraphrases. Metaphoric and literal language stimuli were normed for familiarity and valence/arousal by an independent set of participants (subjects 12, 5 males and 7 females). Half of all stimuli presented were conservative political statements, while the other half were liberal political statements. 64 To best assess arousal (e.g., 1-not at all arousing, 7-very arousing) and valence (e.g., 1-extremely negatively, 7-not at all negative) of the language stimuli, without interference from reactions to specific political beliefs, we compared ratings of physical disgust adjectives within sentences that only implied ‘moral disgust’ (e.g., “Congressman Smith’s views on marriage are rotten.”) with those that contained our semantically similar adjectives (e.g., “Congressman Smith’s statements on marriage are unethical.”). The results showed no significant differences in arousal (Metaphor M = 2.64, SD = 1.0, Literal M = 2.41, SD = 0.66, t(39) = 1.31, p = 0.19 n.s.) or valence (Metaphor M = 2.92, SD = 1.2, Literal M = 3.00, SD = 0.91, t(39) = -0.38, p=0.71 n.s.) between average ratings for each of our 40 disgust adjectives placed in sentences that implied ‘moral disgust’ and our 40 semantically similar adjectives placed in sentences that implied ‘moral disapproval’. No significant differences in familiarity were found (e.g., 1-not at all familiar, 7-extremely familiar) between metaphors and their literal counterparts (e.g., Metaphor M = 4.04, SD = 0.97, Literal M = 4.19, SD = 0.98, t(39) = 0.70, p=0.49 n.s.). Lastly, a separate post-hoc test (subjects 12, (males = 6 and females = 6)) showed that our metaphoric stimuli and their literal counterparts achieved a high degree of semantic similarity (1-not at all similar, 5- very similar, 7-equivalent in meaning) on average (M = 5.2, SD = 0.38). Experimental Paradigm: Participants first completed the language task. Inside the fMRI scanner participants read political statements (either metaphors, literal paraphrases, literal neutral sentences) for comprehension. Each sentence was presented on the screen for 10 secs. Subsequently, the text disappeared and a 7-point Likert scale appeared on the screen and participants indicated the degree to which they agreed with the statement using two hand held button-boxes in the scanner. The response period was of variable time that depended on when participants entered their response. Participants were instructed prior to the fMRI scan how to enter responses. The right-hand button box consisted of two buttons that could move the cursor either to the left or to the right on the Likert scale using the right thumb, while the left-hand hand button box simply had a single button that could be pressed to submit the response with the left thumb. Following the response period, a 11 secs rest period followed during which participants simply viewed a fixation cross on a gray screen. Participants completed 4 language runs each lasting (M = 9.88 mins, SD = 0.21). In each run participants viewed 10 political metaphorical, 10 political literal paraphrases, as well as 5 apolitical literal sentences, in a pseudorandom presentation alternating with rest periods. Each participant thus viewed 40 metaphorical paragraphs, 40 literal paraphrases, as well as 20 apolitical literal sentences in total. Stimuli were counterbalanced such that no participant saw a metaphor and its literal counterpart in the same scanning session. There were fewer apolitical sentences as due to timing limitations we couldn’t include more. As the apolitical sentences had straightforward answers, they were also used to assess whether participants were paying attention to the task and inputting appropriate responses (e.g., “Road signs should be properly maintained to ensure commuter safety.”). After completing these 4 language runs, participants were allowed to close their eyes and take a rest for 7 mins during which a structural image scan was taken. Following this, participants completed three runs of the picture task 5.5 mins. For each picture run participants viewed 10 picture blocks (each picture block consisted of three consecutively presented images with duration 2 secs and inter-trial interval of .25 secs). Each block lasted 6.5 seconds for each emotion category (neutral, disgust) in randomized presentation alternating with 10 second rest periods. All the stimuli were presented on a computer screen using Matlab Psychophysics toolbox (Psychtoolbox 3, www.psychtoolbox.org)). 65 MRI Data Acquisition: Functional MRI images were acquired with a Siemens MAGNETOM Trio 3T System with a 32-channel head matrix coil in the Dornsife Cognitive Neuroscience Imaging Center at the University of Southern California. A high-resolution anatomical scan was acquired for each subject: Structural T1-weighted magnetization-prepared rapid gradient echo (MPRAGE), TR=1950 ms, TE=2.26 ms, flip angle=9 degrees, 256 x 224 mm matrix, 1 mm resolution, 176 coronal slices. Whole brain functional images were obtained with a T2* weighted single-shot gradient-recalled echoplanar imaging, echo-planar sequence (EPI) using blood- oxygenation-level-dependent contrast. Each functional image comprised of 37 contiguous axial slices (3.5 mm thick), acquired in interleaved mode, and with a repetition time (TR) of 2000 ms (echo time (TE) of 30 ms, flip angle (FA) 90 degrees, 64x64 mm matrix). Preprocessing: All preprocessing and statistical analysis were carried out using FSL (Oxford Centre for Functional Magnetic Resonance Imaging of the Brain (FMRIB’s ) Software Library, http://www.fmrib.ox.ac.uk/fsl/index.html), motion correction using Motion Correction using FMRIB’s Linear Image Registration Tool (MCFLIRT), spatial smoothing using a 9 mm full- width at half-maximum Gaussian filter, prewhitening using FILM, high pass temporal filtering (90 seconds), and registration into standard MNI space using the participant’s individual skull- stripped high-resolution T1 anatomical images was carried out using FMRIB’s Linear Image Registration Tool (FLIRT). The general linear model (GLM) was used on each individual voxel’s time series with FSL. The design matrix consisted of a synthetic hemodynamic response function (double gamma function) with its temporal derivative convolved with the input stimuli waveform in an event-related design for the language task and, similarly, convolved with the input stimuli waveform in a block design in the picture task. GLM Statistical Analysis: In the language task, six separate repressors were included in the model: 1) metaphor 2) metaphor response 3) literal 4) literal response 5) apolitical literal 6) apolitical literal response. In the functional localizer picture task design two separate regressors were included in the model: 1) neutral image 2) disgust image. The rest periods in between served as the baseline contrast. Six motion parameters were also included in the design matrix as regressors to account for motion artifacts in the signal. The main contrasts of interest included 1) metaphor vs. literal 2) metaphor response vs. literal response 3) literal vs. metaphor 4) literal response vs. metaphor response, as well as, for the additional control 5) metaphor vs. apolitical 6) metaphor response vs. apolitical response 7) literal vs. apolitical 8) literal response vs. apolitical response. In the group analysis we also included a covariate consisting of the demeaned scores obtained in the political questionnaire that assessed overall political orientation on a 7-point Liker scale (e.g., 1-strongly liberal and 7-strongly conservative). In the picture task the main contrast of interest was disgust vs. neutral. The same covariate using the demeaned scores on our political orientation (liberal or conservative) questionnaire was also used in the picture task at the group level. Parametric Analysis: In our parametric analysis all regressors were modulated by response ratings that were modified from a 7-point Likert scale (1 Strongly Agree à 7 Strongly Disagree) as follows: Responses of either Strongly Disagree and Strongly Agree were given a maximum value of 4 with progressively less extreme choices linearly decreasing and Neutral (neither agree or disagree) response was given a value of 1. This reflected activations that were modulated by the strength of either 66 agreement or disagreement. We modeled both the parametrically modulated effect and the main effect in the GLM, and the parametric repressors were orthogonalized with the respective main effect repressor. Statistics: In all cases a mixed-effects analysis was carried out at the second level using FSL FMRIB’s Local Analysis of Mixed Effects (FLAME 1). All statistical images were thresholded using Gaussian random field-based cluster analysis with an uncorrected z-score threshold of z>2.3 and FWE corrected for multiple comparisons at the cluster level with cluster extent threshold of p<0.05. This resulted in statistical maps of voxels significantly activated for the contrasts of interests as defined above. Additional statistical significance was conducted within a priori defined regions of interest (ROIs) using a small volume correction and FWE corrected for multiple comparisons at the voxel level (SVC, voxel-corrected, p<0.05 FWE). Region of Interest Analysis (ROIs): We defined the following a priori ROIs previously implicated in disgust processing, mainly areas of primary/secondary gustatory and olfactory cortex including the ventral anterior insula, frontal operculum, and orbital frontal cortex (OFC), as well as the pallidal subregion of the basal ganglia (Bastiaansen et al., 2009; A.J. Calder et al., 2007; Deen et al., 2011; Jabbi et al., 2008; Touroutoglou A. et al., 2012). Additionally, we defined the following a priori ROIs previously implicated in moral decision-making in the brain including the amygdala, but also the VMPFC and DLPFC (Greene et al., 2004; Moll et al., 2008; Young & Dungan, 2012). The Harvard- Oxford Probabilistic Atlas in FSL was used to define bilaterally the amygdala, pallidum, frontal operculum, and OFC thresholded at either P>0.65. The ventral anterior insular subregion was defined based on functionally identified clusters using cluster analysis (Deen et al., 2010). The clusters mean coordinates in MNI152 space were used to create 7 mm spherical ROIs of the bilateral ventral anterior insula (MNI coordinates, ± 33, 13,-17)). For comparison we also identified bilaterally a functional ROI within the ventral anterior insula by creating a 7mm spherical ROI at the peak voxel for the contrast Disgust Pictures vs. Neutral Pictures, left ventral anterior insula (MNI coordinates, ± 38, 9, -14), results not shown. To define the peak MNI coordinates for our 7 mm spherical ROIs of the bilateral DLPFC (MNI coordinates, ± 30, 36, 42) we looked at two studies implicating these areas in cognitive control and executive function (Ide J.S. et al., 2013; Hayashi T. et al., 2013, respectively). Lastly, we defined the peak MNI coordinates for our 7 mm spherical ROI of the VMPFC (MNI coordinates, 0, 48, -6) based on two studies implicating this area in the generation of value signals (Kirk U. et al., 2011; Harvey A.H. et al., 2010). RESULTS Behavioral Results: Of the 19 subjects we recruited, 9 self-identified as liberal (M = 2.67, SD = 0.5), 1 self-identified as moderate (4), and 9 self-identified as conservative (M = 6.33, SD = 0.8) on a 7-point Likert scale (1-Strongly Liberal à 4-Neither Liberal or Conservative à 7-Strongly Conservative). We did not find significant differences when comparing 7-point Likert response ratings (1-Strongly Agree à 4-Neither agree nor disagree à 7-Strongly Disagree) for metaphor (M = 4.40, SD = 1.86) vs. literal paraphrases (M = 4.33, SD = 1.90), t(759) = 0.70, p=0.48, n.s.). A 2-way analysis of variance ANOVA with levels figurativeness (metaphor vs. literal) vs. 67 political orientation (liberal or conservative) additionally showed no statistically significant main effect of political orientation (F (1, 1436) = 0.31, p = 0.58, n.s.) or figurativeness (F(1, 1436) = 0.23, p = 0.63, n.s.) on response ratings and no interaction effect (F(1, 1436) = 0.11, p = 0.74, n.s.). Average response times for metaphor (M = 3.36, SD = 3.23) vs. literal paraphrases (M = 3.55, SD = 2.72) also did not differ significantly (t(719) = -1.31, p=0.19, n.s.). A 2-way analysis of variance ANOVA with levels political orientation (liberal or conservative) vs. figurativeness (metaphor vs. literal) additionally showed no statistically significant main effect of figurativeness (F(1, 1436) = 1.73, p = 0.19, n.s.) or political orientation (F(1, 1436) = 0.0006, p = 0.98, n.s.) on response times and no interaction effect (F(1, 1436) = 0.47, p = 0.49, n.s.). fMRI Language Task: At the whole-brain level we did not see significant differences for the contrast ‘metaphor vs. literal’ during the reading period nor for the contrast ‘metaphor response vs. literal response’ during the response period. However, a small volume correction analysis within a priori defined region of interest (ROI) did show significant activity in the ‘metaphor vs. literal’ contrast in the left amygdala (SVC, voxel-corrected p<0.05, FWE). Additionally, for the contrast ‘metaphor response vs. literal response’ a small volume correction showed significant activity within the left frontal orbital cortex (OFC), left frontal operculum extending into the adjacent anterior insula and left pallidum (SVC, voxel-corrected p<0.05, FWE). At a lower uncorrected z-score threshold z=1.65 and a corrected cluster extent threshold of p<0.05 the insula was bilaterally involved for this same contrast. The reverse contrasts ‘literal vs. metaphor’ and ‘literal response vs. metaphor response’ did not show significant activation differences at the whole-brain level. Nevertheless, a small volume correction within a priori defined ROIs revealed that the VMPFC and the DLPFC bilaterally were significantly more active in the ‘literal vs. metaphor’ condition (SVC, voxel-corrected p<0.05, FWE). Similarly, in the contrast ‘literal response vs. metaphor response’ the VMPFC and right DLPFC were significantly activated (SVC, voxel-corrected p<0.05, FWE). Lastly, at the whole brain level we did find that metaphor vs. apolitical and literal vs. apolitical contrasts engaged visual areas, mainly the lateral occipital cortex and occipital pole to a greater degree, but we did not see any differences during the other contrasts metaphor response vs. apolitical response and literal response vs. apolitical response (see supplementary section Table 1). We do not discuss these results, as the apolitical control had fewer trials compared to metaphor and literal sentences and thus limit our interpretations. At the whole brain level using political orientation as a covariate of interest we found for the contrast ‘metaphor vs. literal’ that activity within the bilateral thalamus and left caudate significantly correlated with political orientation scores in the contrast ‘metaphor vs. literal’ with increased activity for conservatives. Importantly, a small volume correction analysis within a priori defined ROIs further showed that the right pallidum, left OFC, and left ventral anterior insula covaried with political orientation scores for this same contrast with increased activity for conservatives (SVC, voxel-corrected p<0.05, FWE). Moreover, activity in the left anterior insula significantly covaried with political orientation mainly in the metaphor vs. baseline condition (see Figure 1) and not literal vs. baseline indicating that the metaphor condition was driving the effect in the contrast ‘metaphor vs. literal’. Brain regions that covaried with political orientation were only present in the contrast ‘metaphor vs. literal’, but not for any of the other contrasts 68 (e.g., ‘metaphor response vs. literal response’, ‘literal vs. metaphor’, or ‘literal response vs. metaphor response’). Our parametric analysis showed at the whole brain level that activity within the left frontal medial cortex, left frontal pole, and left OFC for the contrast ‘literal vs. metaphor’ was modulated overall by how strongly participants felt about a particular statement (either agreeing or disagreeing). Using a small volume correction analysis we, additionally, showed that in the contrast ‘metaphor response vs. literal response’ activity in the left hemisphere within the amygdala and ventral anterior insula was modulated by how strongly participants felt, regardless of whether they indicated that they agreed or disagreed with the statements presented (SVC, voxel-corrected p<0.05, FWE). Using political orientation as a covariate interest we were also able to see for the same contrast, ‘metaphor response vs. literal response’, that modulation of activity within the left ventral anterior insula by strength of response felt (either agreeing or disagreeing) was driven by political orientation with increased effect for those with higher self-identified conservative rating scores (SVC, voxel-corrected p<0.05, FWE). Lastly, we also found that modulation of activity within the right amygdala, VMPFC, and left DLPFC for the contrast ‘metaphor vs. literal’ by strength of response felt (either agree or disagreeing) was driven by political orientation with increased effect for conservatives (SVC, voxel-corrected p<0.05, FWE). fMRI Picture Localizer Task: At the whole brain level the contrast disgust pictures vs. neutral pictures showed activation within visual brain regions (left inferior lateral occipital cortex and right temporal occipital fusiform cortex), sensorimotor brain regions (right postcentral gyrus and precentral gyrus), parietal areas (left anterior supramarginal gyrus and bilateral superior parietal lobule), and emotion- related brain regions (brain stem, bilateral amygdala, left pallidum, bilateral anterior insula, and anterior cingulate gyrus (ACC)). A small volume correction analysis within a priori defined brain regions confirmed that the bilateral pallidum, bilateral amygdala, bilateral ventral anterior insula, bilateral frontal operculum, and bilateral OFC were significantly activated to a greater degree for the processing of disgust images when compared to neutral images (SVC, voxel-corrected p<0.05, FWE). Lastly, we found that individuals with higher conservative rating scores showed more activity within the bilateral amygdala, right ventral anterior insula, right OFC, and VMPFC (SVC, voxel-corrected p<0.05, FWE). DISCUSSION While the abstract concept of immorality is often expressed via analogy to the concrete domain of impurity/physical disgust (i.e., ‘that was a rotten things to do’) it is not clear whether this reflects merely a linguistic coincidence or if it underlies something deeper about how we process information in the moral domain. According to Conceptual Metaphor Theory, the concept of (im)morality is believed to be intimately tied with the concept of physical impurity (hence physical disgust), as co-occurrences between these two disparate domains over the course of development establish cross-domain mappings, a conceptual mapping reflected in language use (Lakoff & Johnson, 1980). In this study we sought to test two predictions that follow from this theory 1) comprehension of moral disgust metaphors should engage emotion-related brain 69 areas relevant to physical disgust processing 2) moral disgust metaphors may have consequences for how individual’s make moral evaluations reflected in differential recruitment of brain areas implicated in moral decision-making. We were specifically interested to see how comprehension and judgment of moral disgust metaphors could distinctly modulate more automatic, emotion- related brain regions also implicated in disgust processing (i.e., amygdala, basal ganglia, and gustatory/olfactory cortices), as well as, brain areas implicated in more deliberative, top-down processes (i.e., VMPFC, DLPFC) in moral decision-making in the brain when compared to semantically comparable literal sentences. We also sought to see to what extent activity in emotion-related brain regions relevant to physical disgust processing, both during the processing of moral disgust metaphor and during the viewing of physical disgust-inducing images, covaried with political orientation. Moral disgust metaphors vs. literal paraphrases The results showed neural differences between the processing of moral disgust metaphor and their literal counterparts. Specifically, moral disgust metaphors when compared to their literal counterparts showed greater engagement of the left amygdala during the reading period (SVC, p<0.05, FWE). Additionally, a number of brain areas relevant to physical disgust processing including the left pallidum, left frontal operculum/adjacent anterior insula, and left OFC showed increased activity for moral disgust metaphor compared to literal counterparts during the response period (SVC, p<0.05, FWE) (Bastiaansen et al., 2009) (Table 2). Given the fact that our stimuli were matched for arousal/valence and no discernable differences were found in the degree of agreement with moral disgust metaphors when compared to their literal counterparts, these activations may reflect affective processes relevant to physical disgust processing rather than differences in general negative (or positive) affect. The left frontal operculum/anterior insula and OFC together make up part of primary/secondary gustatory cortices suggesting that the judgment of moral disgust metaphors elicited modality-preferential activations. It is important to highlight, however, that growing evidence suggests that the anterior insula and adjacent frontal operculum may be, also more generally, involved in interoceptive processing (Craig, 2009), being activate during the experience of pain or pleasure and, similarly during observation of other’s pain or pleasure (Jabbi & Keysers, 2008; Jabbi, Swart, & Keysers, 2007; Singer et al., 2004; Singer et al., 2006). That is, the anterior insula may not be specific to physical disgust processing, but rather may be involved in emotion simulation, more generally. We point out that this brain region’s involvement in olfactory/gustatory based processes, close connection to visceromotor systems of the brain, and involvement in interoceptive processing are, however, qualities particularly relevant to characterizing physical disgust experience (Lindquist, Wager, Kober, Bliss-Moreau, & Barrett, 2012). As mentioned above, in addition to the involvement of anterior insula/adjacent frontal operculum we also found that the left pallidum was more active for moral disgust metaphor compared to literal counterparts during the response period. A number of studies have highlighted a role for the basal ganglia in the processing of disgusting stimuli (Bastiaansen et al., 2009, Calder et al., 2000; Calder et al., 2007; Phillips et al., 1998; Phillips et al., 1997; Sprengelmeyer et al., 1998; van der Gaag et al., 2007; von dem Hagen et al., 2009). Specifically, 70 previous studies have shown that subregions of the basal ganglia, mainly within the left pallidum, appear to be modulated by individual’s propensity and sensitivity to experience disgust during the viewing of disgusting foods, as opposed to pictures of appetizing or bland foods (Calder et al., 2007). Importantly, however, recent evidence also points to a more general difficulty with the processing of aversive stimuli in those with damage to the basal ganglia (Milders, Crawford, Lamb, & Simpson, 2003; Sprengelmeyer et al., 1996). Work on non-human primates suggest that the basal ganglia may in part function mainly to regulate motivated behaviors, receiving motor, affective, and top-down signals to guide behavior in a goal-directed manner (Haber, 2003; Tachibana & Hikosaka, 2012). In particular the left ventral pallidum may be involved in the generation of motivational signals (Mogenson, Jones, & Yim, 1980; Tachibana & Hikosaka, 2012). Taken together these findings suggest that judgment of moral disgust metaphors involves not only activation of modality-preferential areas (i.e., primary/secondary gustatory cortices), but also, processes associated with motivational withdrawal and rejection behavior (e.g., left pallidum), both relevant to the processing of physical disgust (Calder, 2003; Calder et al., 2007; Chapman & Anderson, 2012). It is important to note that activation of areas implicated in physical disgust processing across both liberals and conservatives occurred during the response or judgment phase and not during the initial comprehension phase of moral disgust metaphors. This suggests that perhaps sensorimotor and affective simulations are not necessary for comprehension of moral disgust metaphors, but rather may mainly reflect post-comprehension imagery (or post-conceptual processes) (Mahon & Caramazza 2008). Alternatively, the fact that activation of brain regions implicated in disgust processing occurred primarily during judgment, could also suggest that the level of depth of processing required by the task dictates the degree to which sensorimotor and affective activations are seen during language processing. That is, disgust simulation for moral disgust metaphors mainly occurs during the response phase because participants are required to process the metaphors in greater depth during this period to arrive at a response. This is in agreement with previous studies in the neuroscience of embodied semantics that have suggested that sensorimotor and affective simulation may depend on the specific task with greater depth of processing involving sensorimotor and affective simulations (for reviews see, Gamez-Djokic et al., 2015; Willems & Casasanto, 2011a; Yang, 2013; Jie Yang, 2014). Indeed, while some studies suggest that somatotopic activation of motor areas during the reading of hand-action verbs occurs automatically irrespective of context (De Grauwe, Willems, Rueschemeyer, Lemhofer, & Schriefers, 2014; Kemmerer, Castillo, Talavage, Patterson, & Wiley, 2008; Kemmerer & Tranel, 2008), others have challenged this idea. For example, Papeo et al., 2009 found that the level of depth of processing required by the task (semantic task vs. syllable counting task) impacted whether or not TMS-induced motor-evoked potentials from the right hand muscles (i.e., a measure of M1 activity) were seen or not for hand-action verbs during a ‘post-conceptual processing period’ (within 500 ms) (Papeo, Vallesi, Isaja, & Rumiati, 2009). The authors conclude that sensorimotor engagement occurs only at a later post-conceptual phase where it can be ‘modulated in a top-down manner by the specific demand of the task’ (Papeo, Vallesi, Isaja, & Rumiati, 2009). Alternatively, it is also possible that we did not see emotion- related brain regions relevant to physical disgust processing for moral disgust metaphor 71 comprehension due to the fact more familiar metaphors do not require sensorimotor/affective simulation. This is in line the hypothesis that at least for conventionalized metaphors, sensorimotor and affective simulation might not be necessary as concrete language in metaphor is rather immediately processed as an ‘abstract superordinate category’ relevant to the metaphoric target domain or lexicalized metaphoric category (e.g., rotten in ‘that was a rotten things to do’ would be immediately be perceived as belonging to the superordinate category ‘bad/immoral things’) (Lacey et al., 2012; Glucksberg et al., 2011; Bowdle & Gentner, 2005). However, although we did not see activation of the left anterior insula/adjacent frontal operculum and basal ganglia during comprehension of moral disgust metaphors when compared to literal paraphrases, we did, importantly see activation of the left amygdala (SVC, p<0.05, FWE) (Table 2). This suggests that comprehension of moral disgust metaphors may involve greater emotional processing than literal paraphrases, possibly at a subconscious or implicit level. Citron and Goldberg 2014, similarly, also found amygdala activation when comparing taste metaphors (“The break up was bitter for him”) with their literal counterparts (“The break up was bad for him”), controlled for arousal and valence. Although Citron & Goldberg, 2014 also saw activation of primary and secondary gustatory cortex including the frontal operculum, anterior insula, and OFC for both taste metaphors and the presentation of the taste words in isolation, they only found amygdala activation for taste metaphors. Based on this initial finding the authors proposed that familiar taste metaphors may be ‘implicitly more emotionally engaging’ than their literal counterparts. They further speculate that this provides evidence of additional affective grounding separate from modality-preferential activations within gustatory cortices that processes specific sensory features and/or arousal/valence associated with taste words. Although we did not present our disgust words in isolation, it is possible that we might have observed a similar finding and future studies should investigate this possibility. If so, these findings, along with the findings of Citron and Goldberg, 2014 would be in agreement with the work of Kousta et al., 2011 and Vigliocco et al., 2014 who have proposed that abstract language may depend mainly on grounding in affective experience, while concrete concepts may rather depend on grounding in sensorimotor experiences. Modulation by political orientation. Perhaps our most significant findings is that we found that activation of brain regions specifically implicated in disgust processing (e.g., anterior insula/adjacent frontal operculum and basal ganglia) during the comprehension of moral disgust metaphors were significantly modulated by political orientation. Specifically, we found for the contrast ‘metaphor vs. literal’ that the following brain regions implicated in disgust processing, including the right pallidum, the left ventral anterior insula, and left OFC correlated with political orientation with increased activity for those with higher conservative rating scores (SVC, p<0.05, FWE) (see Figure 1, Table 2). Importantly, this finding suggests that embodiment effects during metaphor comprehension may reflect individual differences in affective experience, in line within findings that conservatives have been found to show increased physiological responses to aversive/disgusting stimuli (Smith et al., 2011). The fact that the right pallidum, in addition to the left ventral anterior insula, both areas implicated in physical disgust processing (Bastiaansen 72 et al., 2009; Calder et al., 2007; Jabbi et al., 2008; Mogenson et al., 1980; Tachibana & Hikosaka, 2012), correlated with political orientation may suggest an increased tendency of ‘disgust simulation’ during comprehension of moral disgust metaphors for conservatives (SVC, p<0.05, FWE) (Table 2). Literal paraphrases vs. moral disgust metaphors When looking at the processing of literal paraphrases when compared to moral disgust metaphors, the reverse contrasts both during reading and response periods, we found increased activity mainly within the VMPFC and DLPFC. Specifically, comprehension of literal paraphrases versus moral disgust metaphor showed greater activation within the bilateral DLPFC and VMPFC (SVC, p<0.05, FWE) (Table 2). Similarly, judgments made on literal paraphrases compared to moral disgust metaphors recruited to a greater extent the VMPFC and right DLPFC (SVC, p<0.05, FWE) (Table 2). The VMPFC has been primarily associated with the computation of value of stimuli or choices in decision-making (Bartra, McGuire, & Kable, 2013; Bechara, 2005; Clithero & Rangel, 2014; Gupta et al., 2011) and recent evidence suggests it may play a similar role in moral judgment (Hutcherson et al., 2015; Shenhav & Greene, 2010, 2014). While the amygdala and striatum belong to an ‘impulsive system’, the VMPFC and DLFPC have rather been associated with a ‘reflective system’ that controls signals from the impulsive system and integrates them with other information for optimal decision-making in line with long-term goals (Bechara, 2005; Gupta et al., 2011). Moreover, in this view the VMPFC is ‘necessary for reactivating previously acquired information regarding the value of stimuli or events’ (Bechara, 2005; Gupta et al., 2011). Although speculative at best, it is possible that literal paraphrases relied to a greater degree on the ‘reflective system’ rather than the ‘impulsive system’ when compared to moral disgust metaphors because literal paraphrases involved greater ‘reactivation’ of previous affective states relevant to prior political beliefs compared to moral disgust metaphors. In contrast, judgments made for moral disgust metaphors may have relied more heavily on automatic affective information from disgust source domain language. Future studies need to examine this possibility further. Nevertheless, taken together our findings suggest that the processing of moral disgust metaphors recruit emotion-related brain regions implicated in disgust processing to a greater degree than literal paraphrases, which relied more heavily on more top-down processes implicated in a ‘reflective system’ in decision-making, associated with cognitive control (i.e., DLPFC) (Gupta et al., 2011), but also ‘overall value judgment’ in moral decision-making (i.e., VMPFC) (Hutcherson et al., 2015). Modulation by the strength of agreement In line with the above findings, we also found at the whole-brain level that the left frontal medial cortex, the left frontal pole, and left frontal orbital cortex were parametrically modulated by the strength of the response (either agree or disagree) for the contrast literal vs. metaphor. Using a priori defined ROIs we found that this included the VMPFC (SVC, p<0.05, FWE) (Table 2). Thus, again, we see that even for literal paraphrases participants felt strongly about (either agreeing or disagreeing), literal paraphrases recruited brain areas mainly from the ‘reflective system’. Interestingly, modulation by the strength of response within the VMPFC, but also the amygdala and left DLPFC, correlated significantly with political orientation in the 73 contrast ‘metaphor vs. literal’ with increased effect in conservatives (SVC, p<0.05, FWE) (Table 2). It is possible that for political statements conservatives felt particularly strong about there was a greater need to consider both automatic emotional processes related to the disgust source domain language, as well as, strong prior political beliefs, during the processing of moral disgust metaphor. However, future studies need to examine this possibility further. Next, in agreement with our previous results showing increased activity in brain areas implicated in disgust processing for processing of moral disgust metaphors compared to literal paraphrases, we, additionally, found that the amygdala and ventral anterior insula in the left hemisphere were also parametrically modulated by the strength of agreement with the presented moral sentence for the contrast ‘metaphor response vs. literal response’. Notably, parametric modulation by strength of response was found in the left ventral anterior insula to be correlated with political orientation. This is in line with our previous findings that show that the ventral anterior insula involved in the processing of olfactory and gustatory based disgust to be sensitive to political orientation during the processing of moral disgust metaphors versus literal counterparts. Taken together, results from our parametric modulation analysis are in agreement with our previous conclusions that moral disgust metaphors when compared to literal paraphrases recruit more emotional processes than literal paraphrases, while literal paraphrases mainly rely on brain regions involved in more top-down processes. Picture localizer task Lastly, we found increased activation within a range of emotion-related ROIs (i.e., bilateral amygdala, bilateral pallidum, bilateral ventral anterior insula, bilateral frontal operculum, and bilateral OFC) (SVC, p<0.05, FWE) (Table 1, 2) during physical disgust experience in our picture task for the contrast ‘disgust pictures vs. neutral pictures’. Activity within the bilateral amygdala, right ventral anterior insula, right OFC, and VMPC for the contrast disgust pictures versus neutral pictures also correlated with political orientation with increased activity for those with higher conservative rating scores (SVC, p<0.05, FWE) (Table 2). These findings confirm the involvement of our a priori emotion-related brain regions in physical disgust experience. More importantly, they also show evidence of increased emotion- related processing for disgusting stimuli in conservatives. This suggests that our findings for moral disgust metaphor may be linked to the fact that conservatives showed increased sensitivity to aversive/disgust stimuli in line with previous findings (Smith et al., 2011; Dodd et al., 2012; Hibbing, Smith, & Alford, 2014; Oxley et al., 2008). Behavioral data Behaviorally our results showed that moral disgust metaphor did not influence judgment when compared to their literal counterparts, in contrast to previous behavioral studies showing that disgust can impact moral judgment, particularly in conservatives (Eskine, Kacinik, & Prinz, 2011; Inbar et al., 2012; Inbar et al., 2009; Schnall et al., 2008; Terrizzi, Shook, & Ventis, 2010; Wheatley & Haidt, 2005). It is possible that as liberal and conservative participants held strong a priori political beliefs, as indicated by their responses on our political questionnaire, moral disgust metaphors may not have been able to influence judgment in a significant way. 74 Alternatively, it is possible that moral disgust metaphors only exert a ‘persuasive effect’ when judgments surround political issues relating to bodily/sexual purity, and not the socio-economic issues covered in our study. Previous behavioral studies indicate that physical disgust impacts moral judgment mainly on issues relating to bodily/sexual purity, such as gay marriage and abortion (Graham et al., 2009; Haidt & Graham, 2007; Smith et al., 2011). Future studies will need to delineate when moral disgust metaphor can actually impact behavior directly, also possibly by recruiting participants who do not hold strong a priori political beliefs. Conclusion We found notable brain differences in the way moral disgust metaphors were processed compared to their literal paraphrases. One of our main findings being that moral disgust metaphors showed increased activity within emotion-related brain regions, including areas implicated in disgust processing, compared to literal paraphrases across both reading and response periods. Most notably, activity in a number of these emotion-related brain regions during the comprehension of moral disgust metaphors vs. literal paraphrases correlated with political orientation ratings with increased activity for those with higher conservative ratings. This latter finding suggests that embodiment effects are sensitive to one’s previous affective experiences. Our findings are in line with embodied cognition views that propose that our conceptual system is grounded in part in sensorimotor and affective processes relating to real world experience with motor actions, perceptions, or affect (Lakoff & Johnson 1980; Gallese & Lakoff 2005; Barsalou et al., 2003; Barsalou 2008, 2009; Niedenthal et al., 2005; Kousta et al., 2011; Vigliocco et al., 2014), but point to a need to further consider how sensorimotor and affective processes making context-dependent functional contributions. In summary, our findings indicate that moral disgust metaphors may shape the cognitive decision-space by engaging affective processes to a greater extent than semantically comparable literal sentences, which appear to rely more on top-down processes such as the VMPFC implicated in ‘overall value judgment’ and the DLPFC, involved in cognitive control. Nevertheless, the fact that moral disgust metaphors may engage emotional processes to a greater degree than literal counterparts, may not always be enough to impact the final outcome of the moral decision-making process, which depends on the integration of (affective) information from multiple sources (e.g., prior political beliefs, disgust propensity, etc.). Future studies will need to discern exactly at what threshold this affective information will be enough to sway moral judgment directly. 75 References Arbib, M. A., Gasser, B., & Barres, V. (2014). Language is handy but is it embodied? Neuropsychologia, 55, 57-70. doi:10.1016/j.neuropsychologia.2013.11.004 Aziz-Zadeh, L., & Damasio, A. (2008a). Embodied semantics for actions: findings from functional brain imaging. J Physiol Paris, 102(1-3), 35-39. doi:10.1016/j.jphysparis.2008.03.012 Barsalou, L. W. (1999). Perceptual symbol systems. Behav Brain Sci, 22(4), 577-609; discussion 610-560. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11301525 Barsalou, L. W. (2003). Abstraction in perceptual symbol systems. Philos Trans R Soc Lond B Biol Sci, 358(1435), 1177-1187. doi:10.1098/rstb.2003.1319 Barsalou, L. W. (2008). Grounded cognition. Annu Rev Psychol, 59, 617-645. doi:10.1146/annurev.psych.59.103006.093639 Barsalou, L. W. (2009). Simulation, situated conceptualization, and prediction. Philos Trans R Soc Lond B Biol Sci, 364(1521), 1281-1289. doi:10.1098/rstb.2008.0319 Barsalou, L. W., Kyle Simmons, W., Barbey, A. K., & Wilson, C. D. (2003). Grounding conceptual knowledge in modality-specific systems. Trends Cogn Sci, 7(2), 84-91. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12584027 Bartra, O., McGuire, J. T., & Kable, J. W. (2013). The valuation system: a coordinate-based meta-analysis of BOLD fMRI experiments examining neural correlates of subjective value. Neuroimage, 76, 412-427. doi:10.1016/j.neuroimage.2013.02.063 Bastiaansen, J. A., Thioux, M., & Keysers, C. (2009). Evidence for mirror systems in emotions. Philos Trans R Soc Lond B Biol Sci, 364(1528), 2391-2404. doi:10.1098/rstb.2009.0058 Bechara, A. (2005). Decision making, impulse control and loss of willpower to resist drugs: a neurocognitive perspective. Nat Neurosci, 8(11), 1458-1463. doi:10.1038/nn1584 Borg, C., de Jong, P. J., Renken, R. J., & Georgiadis, J. R. (2013). Disgust trait modulates frontal-posterior coupling as a function of disgust domain. Soc Cogn Affect Neurosci, 8(3), 351-358. doi:10.1093/scan/nss006 Borg, C., Georgiadis, J. R., Renken, R. J., Spoelstra, S. K., Weijmar Schultz, W., & de Jong, P. J. (2014). Brain processing of visual stimuli representing sexual penetration versus core and animal-reminder disgust in women with lifelong vaginismus. PLoS One, 9(1), e84882. doi:10.1371/journal.pone.0084882 Borg, J., Lieberman, D., & Kiehl, K. A. (2008). Infection, incest, and iniquity: investigating the neural correlates of disgust and morality. J Cogn Neurosci, 20(9), 1529-1546. doi:10.1162/jocn.2008.20109 Botvinick, M. M. (2007). Conflict monitoring and decision making: reconciling two perspectives on anterior cingulate function. Cognitive, Affective, & Behavioral Neuroscience, 7(4), 356-366. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/18189009 Botvinick, M. M., Braver, T. S., Barch, D. M., Carter, C. S., & Cohen, J. D. (2001). Conflict monitoring and cognitive control. Psychol Rev, 108(3), 624-652. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11488380 Bowdle, B. F., & Gentner, D. (2005). The career of metaphor. Psychol Rev, 112(1), 193-216. doi:10.1037/0033-295X.112.1.193 Brenner, C. J., & Inbar, Y. (2014). Disgust sensitivity predicts political ideology and policy attitudes in the Netherlands. European Journal of Social Psychology. doi:10.1002/ejsp.2072 Calder, A. J. (2003). Disgust discussed. Ann Neurol, 53(4), 427-428. doi:10.1002/ana.10565 76 Calder, A. J., Beaver, J. D., Davis, M. H., van Ditzhuijzen, J., Keane, J., & Lawrence, A. D. (2007). Disgust sensitivity predicts the insula and pallidal response to pictures of disgusting foods. Eur J Neurosci, 25(11), 3422-3428. doi:10.1111/j.1460- 9568.2007.05604.x Calder, A. J., Keane, J., Manes, F., Antoun, N., & Young, A. W. (2000). Impaired recognition and experience of disgust following brain injury. Nat Neurosci, 3(11), 1077-1078. doi:10.1038/80586 Carr, E. W., Kavanagh, L., & Bergen, B. K. (2013). How embodied is metaphor? Current Directions in Psychological Science, CDPS-12-0113. Caruana, F., Jezzini, A., Sbriscia-Fioretti, B., Rizzolatti, G., & Gallese, V. (2011). Emotional and social behaviors elicited by electrical stimulation of the insula in the macaque monkey. Curr Biol, 21(3), 195-199. doi:10.1016/j.cub.2010.12.042 Chapman, H. A., & Anderson, A. K. (2012). Understanding disgust. Ann N Y Acad Sci, 1251, 62- 76. doi:10.1111/j.1749-6632.2011.06369.x Citron, F. M., & Goldberg, A. E. (2014). Metaphorical sentences are more emotionally engaging than their literal counterparts. J Cogn Neurosci, 26(11), 2585-2595. doi:10.1162/jocn_a_00654 Clithero, J. A., & Rangel, A. (2014). Informatic parcellation of the network involved in the computation of subjective value. Soc Cogn Affect Neurosci, 9(9), 1289-1302. doi:10.1093/scan/nst106 Craig, A. D. (2009). How do you feel--now? The anterior insula and human awareness. Nat Rev Neurosci, 10(1), 59-70. doi:10.1038/nrn2555 De Grauwe, S., Willems, R. M., Rueschemeyer, S. A., Lemhofer, K., & Schriefers, H. (2014). Embodied language in first- and second-language speakers: neural correlates of processing motor verbs. Neuropsychologia, 56, 334-349. doi:10.1016/j.neuropsychologia.2014.02.003 Deen, B., Pitskel, N. B., & Pelphrey, K. A. (2011). Three systems of insular functional connectivity identified with cluster analysis. Cereb Cortex, 21(7), 1498-1506. doi:10.1093/cercor/bhq186 Desai, R. H., Binder, J. R., Conant, L. L., Mano, Q. R., & Seidenberg, M. S. (2011). The neural career of sensory-motor metaphors. J Cogn Neurosci, 23(9), 2376-2386. doi:10.1162/jocn.2010.21596 Desai, R. H., Conant, L. L., Binder, J. R., Park, H., & Seidenberg, M. S. (2013). A piece of the action: modulation of sensory-motor regions by action idioms and metaphors. Neuroimage, 83, 862-869. doi:10.1016/j.neuroimage.2013.07.044 Dodd, M. D., Balzer, A., Jacobs, C. M., Gruszczynski, M. W., Smith, K. B., & Hibbing, J. R. (2012). The political left rolls with the good and the political right confronts the bad: connecting physiology and cognition to preferences. Philos Trans R Soc Lond B Biol Sci, 367(1589), 640-649. doi:10.1098/rstb.2011.0268 Eskine, K. J., Kacinik, N. A., & Prinz, J. J. (2011). A bad taste in the mouth: gustatory disgust influences moral judgment. Psychol Sci, 22(3), 295-299. doi:10.1177/0956797611398497 Etkin, A., Egner, T., & Kalisch, R. (2011). Emotional processing in anterior cingulate and medial prefrontal cortex. Trends Cogn Sci, 15(2), 85-93. doi:10.1016/j.tics.2010.11.004 Feinberg, M., & Willer, R. (2013). The moral roots of environmental attitudes. Psychol Sci, 24(1), 56-62. doi:10.1177/0956797612449177 77 Feldman, J., & Narayanan, S. (2004). Embodied meaning in a neural theory of language. Brain Lang, 89(2), 385-392. doi:10.1016/S0093-934X(03)00355-9 Gallese, V., & Lakoff, G. (2005). The Brain's concepts: the role of the Sensory-motor system in conceptual knowledge. Cogn Neuropsychol, 22(3), 455-479. doi:10.1080/02643290442000310 Gamez-Djokic, V., Molnar-Szakacs, I., & Aziz-Zadeh, L. (2015). Psychology Press Book. Giora, R. (1999). On the priority of salient meanings: Studies of literal and figurative language. . Journal of Pragmatics, 31, 919-929. Glucksberg, S. (1998). Understanding metaphor. . Current Directions in Psychological Science, 7, 39-43. Graham, J., Haidt, J., & Nosek, B. A. (2009). Liberals and conservatives rely on different sets of moral foundations. J Pers Soc Psychol, 96(5), 1029-1046. doi:10.1037/a0015141 Greene, J. D., Nystrom, L. E., Engell, A. D., Darley, J. M., & Cohen, J. D. (2004). The neural bases of cognitive conflict and control in moral judgment. Neuron, 44(2), 389-400. doi:10.1016/j.neuron.2004.09.027 Gupta, R., Koscik, T. R., Bechara, A., & Tranel, D. (2011). The amygdala and decision-making. Neuropsychologia, 49(4), 760-766. doi:10.1016/j.neuropsychologia.2010.09.029 Haber, S. N. (2003). The primate basal ganglia: parallel and integrative networks. J Chem Neuroanat, 26(4), 317-330. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14729134 Haidt, J., & Graham, J. (2007). When Morality Opposes Justice: Conservatives Have Moral Intuitions that Liberals may not Recognize. Social Justice Research. Hauk, O., & Tschentscher, N. (2013). The Body of Evidence: What Can Neuroscience Tell Us about Embodied Semantics? Front Psychol, 4, 50. doi:10.3389/fpsyg.2013.00050 Hayes, C. J., Stevenson, R. J., & Coltheart, M. (2007). Disgust and Huntington's disease. Neuropsychologia, 45(6), 1135-1151. doi:10.1016/j.neuropsychologia.2006.10.015 Helzer, E. G., & Pizarro, D. A. (2011). Dirty liberals! Reminders of physical cleanliness influence moral and political attitudes. Psychol Sci, 22(4), 517-522. doi:10.1177/0956797611402514 Hennenlotter, A., Schroeder, U., Erhard, P., Haslinger, B., Stahl, R., Weindl, A., . . . Ceballos- Baumann, A. O. (2004). Neural correlates associated with impaired disgust processing in pre-symptomatic Huntington's disease. Brain, 127(Pt 6), 1446-1453. doi:10.1093/brain/awh165 Hibbing, J. R., Smith, K. B., & Alford, J. R. (2014). Differences in negativity bias underlie variations in political ideology. Behav Brain Sci, 37(3), 297-307. doi:10.1017/S0140525X13001192 Hutcherson, C. A., Montaser-Kouhsari, L., Woodward, J., & Rangel, A. (2015). Emotional and Utilitarian Appraisals of Moral Dilemmas Are Encoded in Separate Areas and Integrated in Ventromedial Prefrontal Cortex. J Neurosci, 35(36), 12593-12605. doi:10.1523/JNEUROSCI.3402-14.2015 Inbar, Y., Pizarro, D., Iyer, R., & Haidt, J. (2012). Disgust Sensitivity, Political Conservatism, and Voting. Social Psychological & Personality Science. Inbar, Y., Pizarro, D. A., Knobe, J., & Bloom, P. (2009). Disgust sensitivity predicts intuitive disapproval of gays. Emotion, 9(3), 435-439. doi:10.1037/a0015960 78 Jabbi, M., Bastiaansen, J., & Keysers, C. (2008). A common anterior insula representation of disgust observation, experience and imagination shows divergent functional connectivity pathways. PLoS One, 3(8), e2939. doi:10.1371/journal.pone.0002939 Jabbi, M., & Keysers, C. (2008). Inferior frontal gyrus activity triggers anterior insula response to emotional facial expressions. Emotion, 8(6), 775-780. doi:10.1037/a0014194 Jabbi, M., Swart, M., & Keysers, C. (2007). Empathy for positive and negative emotions in the gustatory cortex. Neuroimage, 34(4), 1744-1753. doi:10.1016/j.neuroimage.2006.10.032 Kemmerer, D., Castillo, J. G., Talavage, T., Patterson, S., & Wiley, C. (2008). Neuroanatomical distribution of five semantic components of verbs: evidence from fMRI. Brain Lang, 107(1), 16-43. doi:10.1016/j.bandl.2007.09.003 Kemmerer, D., & Tranel, D. (2008). Searching for the elusive neural substrates of body part terms: a neuropsychological study. Cogn Neuropsychol, 25(4), 601-629. doi:10.1080/02643290802247052 Kiefer, M., & Pulvermuller, F. (2012). Conceptual representations in mind and brain: theoretical developments, current evidence and future directions. Cortex, 48(7), 805-825. doi:10.1016/j.cortex.2011.04.006 Kipps, C. M., Duggins, A. J., McCusker, E. A., & Calder, A. J. (2007). Disgust and happiness recognition correlate with anteroventral insula and amygdala volume respectively in preclinical Huntington's disease. J Cogn Neurosci, 19(7), 1206-1217. doi:10.1162/jocn.2007.19.7.1206 Kousta, S. T., Vigliocco, G., Vinson, D. P., Andrews, M., & Del Campo, E. (2011). The representation of abstract words: why emotion matters. J Exp Psychol Gen, 140(1), 14- 34. doi:10.1037/a0021446 Krusemark, E. A., & Li, W. (2011). Do all threats work the same way? Divergent effects of fear and disgust on sensory perception and attention. J Neurosci, 31(9), 3429-3434. doi:10.1523/JNEUROSCI.4394-10.2011 Lacey, S., Stilla, R., & Sathian, K. (2012). Metaphorically feeling: comprehending textural metaphors activates somatosensory cortex. Brain Lang, 120(3), 416-421. doi:10.1016/j.bandl.2011.12.016 Lakoff, G. (1996). Moral politics : what conservatives know that liberals don't. Chicago: University of Chicago Press. Lakoff, G., & Johnson, M. (1980). The metaphorical structure of the human conceptual system. Cognitive Science, 4(2), 195-208. doi:10.1016/S0364-0213(80)80017-6 Lee, S. W. S., & Schwarz, N. (2012). Metaphor in Judgment and Decision Making (M. J. Landau, Robinson, M. D., & B. P. Meier (Eds.) Ed.). Washington, DC: American Pscyhological Association. Lindquist, K. A., Wager, T. D., Kober, H., Bliss-Moreau, E., & Barrett, L. F. (2012). The brain basis of emotion: a meta-analytic review. Behav Brain Sci, 35(3), 121-143. doi:10.1017/S0140525X11000446 Lyons, I. M., Mattarella-Micke, A., Cieslak, M., Nusbaum, H. C., Small, S. L., & Beilock, S. L. (2010). The role of personal experience in the neural processing of action-related language. Brain Lang, 112(3), 214-222. doi:10.1016/j.bandl.2009.05.006 Mahon, B. Z., & Caramazza, A. (2008). A critical look at the embodied cognition hypothesis and a new proposal for grounding conceptual content. J Physiol Paris, 102(1-3), 59-70. doi:10.1016/j.jphysparis.2008.03.004 79 Meteyard, L., Cuadrado, S. R., Bahrami, B., & Vigliocco, G. (2012). Coming of age: a review of embodiment and the neuroscience of semantics. Cortex, 48(7), 788-804. doi:10.1016/j.cortex.2010.11.002 Milders, M., Crawford, J. R., Lamb, A., & Simpson, S. A. (2003). Differential deficits in expression recognition in gene-carriers and patients with Huntington's disease. Neuropsychologia, 41(11), 1484-1492. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12849766 Mogenson, G. J., Jones, D. L., & Yim, C. Y. (1980). From motivation to action: functional interface between the limbic system and the motor system. Prog Neurobiol, 14(2-3), 69- 97. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/6999537 Moll, J., De Oliveira-Souza, R., & Zahn, R. (2008). The neural basis of moral cognition: sentiments, concepts, and values. Ann N Y Acad Sci, 1124, 161-180. doi:10.1196/annals.1440.005 Murphy, F. C., Nimmo-Smith, I., & Lawrence, A. D. (2003). Functional neuroanatomy of emotions: a meta-analysis. Cognitive, Affective, & Behavioral Neuroscience, 3(3), 207- 233. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14672157 Narayanan, S. (1997). Knowledge-based Action Representations for Metaphor and Aspect (KARMA). Doctoral Dissertation, University of California, Berkeley. Niedenthal, P. M., Barsalou, L. W., Winkielman, P., Krauth-Gruber, S., & Ric, F. (2005). Embodiment in attitudes, social perception, and emotion. Pers Soc Psychol Rev, 9(3), 184-211. doi:10.1207/s15327957pspr0903_1 Oxley, D. R., Smith, K. B., Alford, J. R., Hibbing, M. V., Miller, J. L., Scalora, M., . . . Hibbing, J. R. (2008). Political attitudes vary with physiological traits. Science, 321(5896), 1667- 1670. doi:10.1126/science.1157627 Papeo, L., Vallesi, A., Isaja, A., & Rumiati, R. I. (2009). Effects of TMS on different stages of motor and non-motor verb processing in the primary motor cortex. PLoS One, 4(2), e4508. doi:10.1371/journal.pone.0004508 Penfield, W. (1959). The interpretive cortex; the stream of consciousness in the human brain can be electrically reactivated. Science, 129(3365), 1719-1725. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/13668523 Penfield, W., & Faulk, M. E., Jr. (1955). The insula; further observations on its function. Brain, 78(4), 445-470. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/13293263 Phan, K. L., Wager, T., Taylor, S. F., & Liberzon, I. (2002). Functional neuroanatomy of emotion: a meta-analysis of emotion activation studies in PET and fMRI. Neuroimage, 16(2), 331-348. doi:10.1006/nimg.2002.1087 Pizarro, D., Inbar, Y., & Helion, C. (2011). On Disgust and Moral Judgment. Emotion Review, 3(3), 267-268. doi:10.1177/1754073911402394 Pulvermuller, F. (1999). Words in the brain's language. Behav Brain Sci, 22(2), 253-279; discussion 280-336. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11301524 Pulvermuller, F. (2001). Brain reflections of words and their meaning. Trends Cogn Sci, 5(12), 517-524. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11728909 Pulvermuller, F., Hauk, O., Nikulin, V. V., & Ilmoniemi, R. J. (2005). Functional links between motor and language systems. Eur J Neurosci, 21(3), 793-797. doi:10.1111/j.1460- 9568.2005.03900.x 80 Rushworth, M. F., Buckley, M. J., Behrens, T. E., Walton, M. E., & Bannerman, D. M. (2007). Functional organization of the medial frontal cortex. Curr Opin Neurobiol, 17(2), 220- 227. doi:10.1016/j.conb.2007.03.001 Schnall, S., Haidt, J., Clore, G. L., & Jordan, A. H. (2008). Disgust as embodied moral judgment. Pers Soc Psychol Bull, 34(8), 1096-1109. doi:10.1177/0146167208317771 Shenhav, A., & Greene, J. D. (2010). Moral judgments recruit domain-general valuation mechanisms to integrate representations of probability and magnitude. Neuron, 67(4), 667-677. doi:10.1016/j.neuron.2010.07.020 Shenhav, A., & Greene, J. D. (2014). Integrative moral judgment: dissociating the roles of the amygdala and ventromedial prefrontal cortex. J Neurosci, 34(13), 4741-4749. doi:10.1523/JNEUROSCI.3390-13.2014 Singer, T., Seymour, B., O'Doherty, J., Kaube, H., Dolan, R. J., & Frith, C. D. (2004). Empathy for pain involves the affective but not sensory components of pain. Science, 303(5661), 1157-1162. doi:10.1126/science.1093535 Singer, T., Seymour, B., O'Doherty, J. P., Stephan, K. E., Dolan, R. J., & Frith, C. D. (2006). Empathic neural responses are modulated by the perceived fairness of others. Nature, 439(7075), 466-469. doi:10.1038/nature04271 Smith, K. B., Oxley, D., Hibbing, M. V., Alford, J. R., & Hibbing, J. R. (2011). Disgust Sensitivtiy and the Neurophysiology of Left-Right Political Orientations. PLoS One, 6(10). doi:10.1371/journal.pone.0025552 Sopory, P., & Dillard, J. P. (2006). The Persuasive Effects of Metaphor: A Meta-Analysis. Human Communication Research, 28(3), 382-419. Sprengelmeyer, R., Rausch, M., Eysel, U. T., & Przuntek, H. (1998). Neural structures associated with recognition of facial expressions of basic emotions. Proc Biol Sci, 265(1409), 1927-1931. doi:10.1098/rspb.1998.0522 Sprengelmeyer, R., Young, A. W., Calder, A. J., Karnat, A., Lange, H., Homberg, V., . . . Rowland, D. (1996). Loss of disgust. Perception of faces and emotions in Huntington's disease. Brain, 119 ( Pt 5), 1647-1665. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8931587 Tachibana, Y., & Hikosaka, O. (2012). The primate ventral pallidum encodes expected reward value and regulates motor action. Neuron, 76(4), 826-837. doi:10.1016/j.neuron.2012.09.030 Terrizzi, J. A., Shook, N. J., & Ventis, W. L. (2010). Disgust: A predictor of social conservatism and prejudicial attitudes toward homosexuals. Personality and Individual Differences, 49(6), 587-592. doi:10.1016/j.paid.2010.05.024 Thieben, M. J., Duggins, A. J., Good, C. D., Gomes, L., Mahant, N., Richards, F., . . . Frackowiak, R. S. (2002). The distribution of structural neuropathology in pre-clinical Huntington's disease. Brain, 125(Pt 8), 1815-1828. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12135972 Vigliocco, G., Kousta, S. T., Della Rosa, P. A., Vinson, D. P., Tettamanti, M., Devlin, J. T., & Cappa, S. F. (2014). The neural representation of abstract words: the role of emotion. Cereb Cortex, 24(7), 1767-1777. doi:10.1093/cercor/bht025 Wang, K., Hoosain, R., Yang, R. M., Meng, Y., & Wang, C. Q. (2003). Impairment of recognition of disgust in Chinese with Huntington's or Wilson's disease. Neuropsychologia, 41(5), 527-537. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12559147 81 Wehling, E. (2015). Moral disgust at its best. The important role of low-level mappings and structural parellism in political disgust and disease metaphors. In F. In Ervas, & E. Gola (Eds.). (Ed.), Metaphors and Communication (pp. 45-56). Amsterdam, NL: John Benjamins. Wheatley, T., & Haidt, J. (2005). Hypnotic disgust makes moral judgments more severe. Psychol Sci, 16(10), 780-784. doi:10.1111/j.1467-9280.2005.01614.x Willems, R. M., & Casasanto, D. (2011a). Flexibility in embodied language understanding. Front Psychol, 2, 116. doi:10.3389/fpsyg.2011.00116 Willems, R. M., Hagoort, P., & Casasanto, D. (2010). Body-specific representations of action verbs: neural evidence from right- and left-handers. Psychol Sci, 21(1), 67-74. doi:10.1177/0956797609354072 Yang, J. (2013). Context effects on embodied representation of language concepts. Oxford, UK: Academic Press. Yang, J. (2014). Influences of motor contexts on the semantic processing of action-related language. Cognitive, Affective, & Behavioral Neuroscience, 14(3), 912-922. doi:10.3758/s13415-014-0258-y Yang, J. (2014). The role of the right hemisphere in metaphor comprehension: A meta-analysis of functional magnetic resonance imaging studies. Hum Brain Mapp, 35, 107-122. Young, L., & Dungan, J. (2012). Where in the brain is morality? Everywhere and maybe nowhere. Soc Neurosci, 7(1), 1-10. doi:10.1080/17470919.2011.569146 82 TABLES AND FIGURES Table 1. Whole Brain Results, cluster-corrected for multiple comparisons (FWE), coordinates in MNI Space Anatomic Region X Y Z Cluster size T Picture Task Disgust > Neutral L Lateral Occipital Cortex, inferior -42 -76 -4 3903 11.68 L Temporal Occipital Fusiform -42 -56 -16 10.15 L Lateral Occipital Cortex, superior -32 -84 8 8.80 L Amygdala -26 -2 -14 2550 9.34 Brain Stem 6 -28 -6 7.18 R Amygdala 24 0 -16 6.96 R Insula 38 10 -14 7.56 L Insula -38 8 -14 5.86 L Pallidum -16 -8 -6 4.57 R Temporal Occipital Fusiform Cortex 38 -58 -12 1713 8.66 R Lateral Occipital Cortex, inferior 48 -60 -4 7.73 L Supramarginal Gyrus, anterior -60 -24 28 734 11.33 R Postcentral Gyrus 66 -14 42 706 7.94 R Precentral Gyrus 48 6 26 323 7.17 L Cingulate Gyrus, anterior -4 2 34 312 5.30 R Superior Parietal Lobule 26 -54 52 280 5.45 L Superior Parietal Lobule -38 -42 48 242 5.82 Language Task Metaphor > Literal 83 *With Political Orientation as Covariate R Thalamus 14 -4 12 1120 4.43 L Thalamus -12 -12 12 3.34 L Caudate -14 8 12 3.47 Language Task- Parametric Analysis Modulated by strength of response (either agree/disagree) Literal > Metaphor L Frontal Medial Cortex 0 36 -16 973 4.24 L Frontal Pole -18 44 -12 3.67 L Frontal Orbital Cortex -22 32 -20 36 4.01 Table 2. Small Volume Correction Results, voxel-corrected p<0.05 (FWE), coordinates in MNI Space Anatomic Region X Y Z T Corrected P-value Picture Task Disgust > Neutral L Orbital Frontal Cortex -34 18 -16 3.74 0.007 R Orbital Frontal Cortex 30 22 -12 4.53 0.002 84 L Frontal Operculum -32 14 10 4.66 0.002 R Frontal Operculum 36 18 6 4.44 0.002 L Ventral Anterior Insula -34 10 -14 5.00 0.001 R Ventral Anterior Insula 38 12 -10 4.85 0.001 L Amygdala -24 -4 -14 8.70 0.000003 R Amygdala 24 0 -16 6.96 0.00004 L Pallidum -16 -8 -6 4.57 0.001 R Pallidum 20 -4 -6 4.40 0.002 Disgust > Neutral *With Political Orientation as Covariate VMPFC -2 54 0 2.30 0.028 R Orbital Frontal Cortex 20 10 -20 4.71 0.002 R Ventral Anterior Insula 36 8 -10 2.84 0.018 L Amygdala -26 0 -22 2.58 0.015 R Amygdala 24 -2 -28 3.45 0.007 Language Task Metaphor > Literal L Amygdala -22 0 -20 2.42 0.020 Metaphor Response > Literal Response L OFC -36 26 -4 2.93 0.022 85 L Frontal Operculum extending into AI -42 24 0 2.46 0.011 L Pallidum -24 -18 -2 2.41 0.026 Literal > Metaphor VMPFC -4 52 -4 3.66 0.003 L DLPFC -26 34 44 3.11 0.007 R DLPFC 32 34 46 3.01 0.008 Literal Response > Metaphor Response VMPFC -6 42 -8 2.40 0.035 R DLPFC 28 34 44 2.7 0.009 Metaphor > Literal *With Political Orientation as Covariate L Ventral anterior insula -40 8 -10 3.10 0.017 L OFC -38 22 -10 3.20 0.020 R Pallidum 16 -4 -6 2.48 0.013 Language Task- Parametric Analysis Modulated by strength of response (either agree/disagree) Metaphor > Literal *With Political Orientation as Covariate R Amygdala 20 -8 -14 3.61 0.010 VMPFC -4 46 -6 2.61 0.018 86 L DLPFC -28 32 42 3.15 0.011 Metaphor Response > Literal Response L Amygdala -18 -6 -12 2.84 0.0196 L Ventral Anterior Insula -32 14 -10 2.26 0.0105 Metaphor Response > Literal Response *With Political Orientation as Covariate L Ventral Anterior Insula -34 6 -10 2.88 0.0186 Literal > Metaphor VMPFC 2 50 -6 2.66 0.0124 L OFC -22 32 -20 3.67 0.0200 87 Figure 1. Percent Signal Change in Left Anterior Insula for Contrast Metaphor > Baseline. Shows individual mean percent signal change within the left anterior insula for each subject when looking at metaphor vs. baseline against self-reported scores on our political orientation questionnaire. 88 SUPPLEMENTARY SECTION Table 1. Whole Brain Results, cluster-corrected for multiple comparisons (FWE), coordinates in MNI Space Anatomic Region X Y Z Cluster size T Language Task Metaphor > Apolitical R Lateral Occipital Cortex, superior -18 -64 52 3438 5.21 L Lateral Occipital Cortex, superior -28 -86 18 4.08 L Occipital Pole -16 -104 4 3010 5.13 Literal > Apolitical R Precuneous Cortex 8 -64 50 3294 2.77 R Lateral Occipital Cortex, superior 16 -60 60 3.24 L Occipital Pole -22 -96 16 3.53 R Occipital Pole 18 -100 16 3.86 89 Distinguishing Metaphors that Differ in their Force Exchange Patterns Vesna Gamez-Djokic 1,2 , Elisabeth Wehling 4 , Lisa Aziz-Zadeh 1,2,3 1 Brain and Creativity Institute, University of Southern California, 2 Neuroscience Graduate Program, 3 Division of Occupational Science and Occupational Therapy, University of Southern California, 4 International Computer Science Institute, University of California, Berkeley Author Contributions: V.G.D. conceived of and designed study with E.W., contributed to design and creation of stimuli, piloted the stimuli; conducted the study; conducted analysis and data interpretation, and wrote the manuscript. E.W. conceived of and designed study with V.G.D.; designed and created stimuli; edited manuscript. L.A.Z. oversaw all aspects of the study and edited manuscript. ABSTRACT Theories of embodied cognition suggest that understanding language about action and perception involves sensorimotor simulation or the recruitment of sensorimotor and affective processes involved in initial physical experience. Previous evidence suggests that action verbs and literal sentences containing action verbs activate areas of primary and/or premotor cortex that encode specific motor features relevant to the meaning of action verbs. For example, sensorimotor brain areas have been shown to be sensitive to the degree of implied force in action-related phrases (e.g., ‘pushing the piano’ vs. ‘pushing the chair’). However, it is currently unknown whether action-related metaphors that draw on different patterns of force exchange (e.g., ‘She’s pushing the agenda’ vs. ‘She’s grasping the idea’) need rely on the same level of motor specificity to encode the meaning of more abstract events. Thus, in this study we tested the hypothesis that more abstract events in metaphor engage sensorimotor processes relevant to action execution involving physical force exchange between objects and agents (i.e., physical pushing and pulling on an object). We used an MVPA whole-brain searchlight analysis to look for patterns of activity across voxels in the brain that could successfully train a classifier to distinguish between two familiar metaphors that differ in their encoded force dynamics: Metaphors drawing on action-verbs that imply an away-from-self force toward an antagonist and relate to the act of communicating (e.g., “She’s pushing the agenda.”) and metaphors drawing on action-verbs that imply a force that would move the antagonist towards the self and relate to cognizing (e.g., “She’s grasping the idea.”). The results of our within-modality 4-voxel radius whole-brain searchlight MVPA analysis revealed that patterns of activity across voxels in sensorimotor areas but also other brain regions could successfully distinguish between these two types of metaphors. Importantly, significant voxels mainly within sensorimotor brain areas in our within-modality MVPA analysis partially overlapped anatomically with brain activations in a univariate analysis of a motor task involving pushing and pulling actions on an object. Taken together the findings indicate that the precentral and postcentral cortices (mainly premotor and secondary somatosensory cortex), as well as, the inferior parietal lobe (IPL), a ‘higher-level motor area’, may be needed to encode information relevant to physical force exchange between objects and agents in metaphor, revealing a degree of motor specificity that would not otherwise be expected for more abstract event descriptions. Keywords: Metaphor, Action Verbs, Motor Cortex, Force Dynamics 90 INTRODUCTION Grounded cognition theories suggest that conceptual processing involves sensorimotor and affective mechanisms involved in primary experience (Barsalou, 1999, 2008; Gallese & Lakoff, 2005; Niedenthal, Barsalou, Winkielman, Krauth-Gruber, & Ric, 2005). From this viewpoint, it is predicted that understanding language about action and perception should depend on sensorimotor simulations in the brain (Barsalou, 2003; Gallese & Lakoff, 2005; Pulvermuller, 1999, 2001; Pulvermuller, Hauk, Nikulin, & Ilmoniemi, 2005). Embodied semantics theories have challenged more classical views that describe conceptual processing as mainly computations on amodal symbols removed from sensorimotor systems of the brain (Fodor, 1975; Pylyshyn, 1984). More moderate embodiment views suggest that symbolic processes may be sufficient for superficial conceptual processing, however, they maintain that sensorimotor processes are necessary for deep conceptual processing (Barsalou, 2003, 2008; Simmons & Barsalou, 2003). In contrast, some disembodied views that allow a role for sensorimotor processes have argued that they play a much less substantial role merely ‘color[ing] conceptual processing’ (Mahon & Caramazza, 2008). Additionally, these disembodied views have argued that evidence of activation of sensorimotor systems of the brain during the processing of action- related language most likely reflects post-comprehension imagery or spreading activation form processing specific action-related words, but these activations bear little on meaning processes. Abstract concepts generally do not have obvious links to direct experience and they have traditionally posed more of a challenge for embodied semantics. In this way, neuroscientific studies on metaphor comprehension offer a unique insight into abstract concept processing as they involve an abstract idea that is construed or expressed in terms of a sensorimotor domain, yet need not necessarily involve a ‘strict’ literal interpretation of events (Bowdle & Gentner, 2005; Keysar & Bly, 1999; Keysar, Shen, Glucksberg, & Horton, 2000). According to Conceptual Metaphor Theory (CMT), abstract concepts are grounded in sensorimotor and affective experience through mappings with easier to understand concrete concepts due to co- occurrences that take place during development (G. Lakoff, 2014; George Lakoff & Johnson, 1980a). Linguistic metaphors are thought to reflect underlying conceptual metaphors. In this view metaphor comprehension involves a direct comparison of source and target domains with inferences derived from sensorimotor simulation in the source domain projected to the target domain for more complex inferencing (Narayanan, 1997). In contrast, it has been argued that metaphor comprehension, especially for conventionalized metaphors, need not rely on directional projections from the source to the target domain involving sensorimotor simulation, rather metaphors may be interpreted directly through a categorization process (Bowdle & Gentner, 2005; Keysar & Bly, 1999; Keysar et al., 2000). For example, in the metaphor ‘she’s grasping the concept’ the verb ‘grasp’ need not be interpreted literally as relating to motoric experiences, but rather is immediately processed as the abstract superordinate category (or lexicalized metaphoric category) ‘to understand’ (Glucksberg 2002). Although an increasing number of fMRI studies find that the processing of words and literal language related to action, perception, and affect recruit sensorimotor and affective systems in the brain, for reviews see (Hauk & Tschentscher, 2013; Willems & Casasanto, 91 2011a), the results for figurative language processing have been much more mixed (Carr, Kavanagh, & Bergen, 2013). A number of studies show that the processing of action verbs or literal sentences containing action verbs do reliably activate sensorimotor systems in the brain involved in actually executing/observing or imaging that action. For example, action verbs that imply action with a specific body part (i.e., ‘kick’, ‘lick’, ‘spit’) show somatotopic activation in primary and premotor cortices (Aziz-Zadeh, Wilson, Rizzolatti, & Iacoboni, 2006; Buccino et al., 2005; Hauk, Johnsrude, & Pulvermuller, 2004; Kemmerer, Castillo, Talavage, Patterson, & Wiley, 2008; Pulvermuller et al., 2005; Rueschemeyer, van Rooij, Lindemann, Willems, & Bekkering, 2010; Willems, Hagoort, & Casasanto, 2010). Furthermore, Kemmerer et al., 2008 showed that verbs having to do with either action, motion, contact, change of state, and tool use (e.g., cutting verbs [cut, slice, hack], hitting verbs [hit, poke, jab], etc.) in a semantic judgment task engaged sensorimotor areas of the brain in line with findings for action execution/recognition. Particularly relevant to the present study, Moody and Gennari, 2009 showed that the premotor cortex was sensitive to the amount of force required to perform a described action (‘pushing a piano’ vs. ‘pushing a chair’) (Moody & Gennari, 2010). These studies, especially the latter study, suggest that activation of sensorimotor areas of the brain reflect ‘specific motor features’ of action verbs relevant to their meaning (Kemmerer et al., 2008) and, thus, much less likely to reflect spontaneous imagery, as suggested by Mahon & Caramazza 2008. While these studies support the idea that the processing of action verbs and action-related sentences involve sensorimotor representations of the brain with a high degree of sensorimotor specificity, it is less clear, however, whether or not familiar hand-action metaphors that draw on different patterns of force exchange would similarly show such motor specificity. Studies looking at figurative language related to action/perception have shown mixed findings. For example, Aziz-Zadeh et al., 2006 and Raposo et al., 2009 showed somatotopic activation within primary motor/ or premotor areas for literal action sentences (e.g., ‘biting the peach’), but did not report a similar finding for idiomatic action phrases (e.g., ‘biting off more than you can chew”) (Aziz-Zadeh et al., 2006; Raposo, Moss, Stamatakis, & Tyler, 2009). Desai et al., 2011 further showed that the degree to which sensorimotor activation is seen for action- related sentences decreases with increasing abstraction from literal to metaphoric and from non- familiar to more familiar metaphors, with little to no sensorimotor activation for idioms (Desai, Binder, Conant, Mano, & Seidenberg, 2011). Importantly, they found that literal sentences activated the precentral gyrus and the left anterior inferior parietal lobe (aIPL), a ‘higher-level motor area’. However, metaphoric sentences also recruited the aIPL, but critically activity within primary motor areas was sensitive to how familiar the metaphors had been rated. Furthermore, they found that action-related metaphors activated only aIPL compared to abstract sentences, while idiomatic action sentences activated primarily classic language areas (BA 45 & 47: pars triangularis and orbitalis; (Desai, Conant, Binder, Park, & Seidenberg, 2013). In an attempt to explain these results Desai, Binder, Conat, Mano, & Seidenberg (2011) proposed that figurative expressions might undergo a process of change, the ‘neural career of metaphor theory’, in which initially such phrases are deeply linked to sensorimotor representations, but over time these links become less important for meaning due to processes of conventionalization and instead rely on multimodal or even amodal representations. It has been proposed that processes of categorization 92 are the primary means by which we comprehend conventional metaphor, as discussed above (Bowdle & Gentner, 2005). Alternatively, a recent behavioral study, looking at the impact of biological motion perception of point-light walkers on processing motion verbs in either a literal or metaphoric context, suggests that sensorimotor representations in metaphor might simply be just ‘less- specific’ or ‘generalized’ compared to literal sentences (Troyer et al., 2014). They found that point walker primes facilitated the processing of literal sentences with motion verbs that are semantically distant from the walking prime motion (leaping, catapulting) but not those that resembled the walking motion (ambling, walking). Importantly, the opposite effect was found for motion verbs in a metaphoric context with motion verbs that are semantically distant form the walking motion being processed more slowly than semantically close verbs. Such facilitation or interference effects (action compatibility effects) are believed to be due to shared representations between primary experience (action, space, motion) and linguistically represented aspects of motion. Thus, according to this view sensorimotor and affective processes are shared between language and primary experience, but metaphor draws on a distinct subset of possibly ‘less- specific’, perhaps even abstracted embodied representations compared to motion-related literal sentences. These findings are in line with the above studies that have shown that familiar hand- action metaphors compared to abstract sentences mainly activated the IPL, while primary/premotor motor areas were mainly modulated by familiarity with increased activity for less familiar metaphors (Desai et al., 2011; 2013). Similarly, a study by van Dam et al., 2010 showed that the IPL may be sensitive to the level of ‘motor abstraction’ with more activation in this region during the processing of superordinate vs. more abstract verbs. Thus, it is currently unclear whether familiar action-related metaphors should depend on sensorimotor brain areas that reflect specific motor features at the level of action kinematics (e.g., primary motor areas) and/or areas that have been implicated in coding the underlying action goal/outcome (e.g., IPL) and reflect perhaps more schematized representations (Desai et al., 2013; Krasovsky, Gilron, Yeshurun, & Mukamel, 2014; Moody & Gennari, 2010). In CMT the concrete domain is said to help structure the more abstract domain, mainly by providing shared inferential mechanisms (Lakoff, 2014; Lakoff & Johnson, 1980b). For example, it has been proposed that the abstract concepts of ‘communication’ and ‘ideas’ are grounded in sensorimotor processes. Specifically, based on findings from linguistic corpus data it has been suggested that ideas are conceptualized as objects (Ideas are Objects), mainly objects that can be manipulated (Manipulating Ideas is Manipulating Objects), and furthermore that the act of communication can be conceptualized as the act of sending and receiving objects (Communication is Sending/Receiving Idea-Objects) (Boot & Pecher, 2011; Lakoff, 2014; Lakoff & Johnson, 1980a, 1980b; Reddy, 1979). Importantly, behavioral evidence suggests that thinking about the abstract domain of communication and (cognizing/ideas) may actually involve sensorimotor simulation. Glenberg and Kaschak (2002) showed that when participants read sentences reflecting concrete transfer (“Andy delivered the pizza to you”) and abstract transfer (“Liz told you the story”) this interfered with a sensibility response that required movement in the opposing direction for both literal and metaphoric object transfer. This effect, otherwise known as the action compatibility effect (ACE), occurs when participants read sentences about 93 actions that share features with the preparation/performance of an unrelated response carried out with the body (Glenberg & Gallese, 2012; Glenberg & Kaschak, 2002). Critically, this effect is taken as evidence in favor of the idea that language comprehension engages sensorimotor simulations, as they may partially overlap with sensorimotor processes underlying physical actions leading to interference or priming effects to occur. In a similar vein to CMT, Talmy’s Theory of Force Dynamics predicts that event descriptions in language, whether concrete or abstract, are understood in part by drawing upon a 'naïve physics' or experiential knowledge about the nature of force exchange between objects and agents involved in each action (i.e., force dynamics) (Boot & Pecher, 2011; Pecher, Boot, & Van Dantzig, 2011; Talmy, 1988). For example, understanding events about physical forces (e.g., “The ball kept rolling along the green”) or psychological forces (e.g., “She’s civil to him”, “John can’t go out of the house”) are said to draw upon basic force dynamic schemas (Talmy, 1988). At the simplest level a force dynamic schema description involves two force-exerting objects that are said to interact such that one objects is perceived as the focal force entity, the agonist, that exerts a force on the antagonist (i.e., the object resisting the motion), which either overcomes the focal force entity or succumbs to it (Talmy, 1988). According to this view the semantics underlying force dynamic schemas are represented within sensorimotor systems of the brain. In support of Talmy’s theory of force dynamics, Madden and Pecher (2010) showed that forces whether physical (e.g., “The bulldozer pushed the pile of dirt across the lot”) or psychological (e.g., “Her friends persuaded the girl to come to the party”) exhibited an action-compatibility effect or faster sensibility responses when preceded by an animation that involved similar underlying force dynamics (e.g, a circle pushing a rectangle that eventually tumbles etc.) than one that did not (Pecher et al., 2011). Although such action-compatibility effects suggest shared substrates between language describing concrete or abstract transfer (and force dynamics) with actual physical transfer (and force exertion), little is known about the exact nature of sensorimotor processes that may be engaged (Pecher et al., 2011). Thus, while the above behavioral results are rather convincing, it is unclear whether, for example familiar cognizing and communication metaphors that construe the processing and sharing of information as the manipulation and transfer of objects (and imply force either away or towards the agonist) would actually engage specific sensorimotor processes in the brain relevant to the processing of force exchange between objects and agents. Thus, in this study we investigated the extent to which two metaphors that differ with regard to their encoded force dynamics, and relate to communication and cognizing (Expressing Ideas is a Pushing Force and Cognizing Ideas is a Pulling Force), could be distinguished in the brain using a whole-brain searchlight multivariate pattern analysis approach (MVPA). While univariate approaches rely on the average signal across nearby voxels to detect differences among conditions, multivariate analysis of fMRI data involves the detection of patterns of activation across voxels in the brain that can successfully classify between conditions of interest, offering additional sensitivity when detecting minor differences (Pereira, Mitchell, & Botvinick, 2009). In our study we used a searchlight-based MVPA approach to search for localized patterns of activity across a 4-voxel searchlight sphere in the whole-brain that could train a classifier to distinguish between our metaphoric conditions. We reasoned that a within-modality searchlight- 94 based MVPA approach was the best approach to distinguish between hand-action metaphors that differ in subtle ways, mainly in their encoded force dynamics. We also reasoned that this approach would allow us to ask whether primarily differences in action kinematics at ‘lower- levels’ of motor representation (e.g., primary motor areas) or brain areas implicated in processing action goals/outcomes irrespective of specific action kinematics (e.g., IPL) at higher-levels of motor representation contained information relevant to distinguishing metaphors that differ in their force exchange patterns. Lastly, we also included a localizer task involving physical ‘pulling’ and ‘pushing’ actions on an object to see the degree of anatomical overlap between significant voxels in our within-modality MVPA analysis in the language task and significant activations in our univariate analysis during actual action execution in the localizer task. We hypothesized that both low-level action kinematics involving the primary motor areas (e.g., directed force) and ‘higher-level motor representations’ (e.g., purposeful object transfer) involving the IPL, should both be able to distinguish between our two metaphor types, mainly (‘Expressing Ideas is a Pushing Force) and (Cognizing Ideas is a Pulling Force). Lastly, we expected that significant classification accuracies in our within-modality MVPA analysis in the language task should overlap anatomically with activations in our univariate analysis of the motor task during actual execution of ‘pulling’ and ‘pushing’ physical actions, mainly within these sensorimotor brain areas. METHODS Participants: 10 right-handed, native English speakers (age range 18-25, 4 females and 6 males) provided informed consent and were paid for participating in this study. All subjects had normal hearing and vision, and no history of neurological illness. All subjects were students at the University of Southern California (USC). Language Stimuli: Sixty familiar metaphors with the conceptual mapping (Expressing Ideas is a Pushing Force) containing action-verbs implying an away-from-self force toward an antagonist, ‘push metaphors’, and sixty familiar metaphors with the conceptual mapping (Cognizing Ideas is a Pulling Force) implying a force that moves the antagonist towards the self, ‘pull metaphors’ were created. Eighteen ‘pull type’ action verbs and twenty ‘push type’ action verbs were used each no more than four times to create all stimuli. All sentences were created in the third person singular, present tense, progressive in order to maintain similar sentence structure, which was always in the form (she’s/he’s x-ing [the sentence topic]). Push metaphors (M = 7.03, SD = 1.44) and pull metaphors (M = 7.10, SD = 1.49) did not significantly differ in length (t(60)=-0.24, p = 0.81, n.s.). Next, forty-five metaphors from each set were selected for our study such that ‘pull metaphors’ were matched with ‘push metaphors’ along a number of psycholinguistic variables (e.g., arousal/valence, meaningfulness, familiarity, imageability). These metaphors were normed by an independent set of participants (subjects 12, 4 males and 8 females). A meaningfulness judgment task (i.e., “makes sense?” or “does not make sense?”) indicated that all metaphors could easily be comprehended. Participants rated each sentence for familiarity on a 7-point Likert Scale (1-not at all familiar, 7-very familiar) and showed that on average metaphors were rated as being relatively familiar (M = 4.12, SD = 0.93). Additionally, ‘pull metaphors’ (M = 4.22, SD = 95 1.13) and ‘push metaphors’ (M = 4.00, SD = 0.68) did not differ significantly in familiarity (t(44) = -1.19, p = 0.24, n.s.). Participants also rated metaphors for motor imagery on a 7-point Likert scale (1-not at all associated with physical action, 7-very much associated with physical action). Overall metaphors were rated low on motor imagery, ‘pull metaphors’ (M = 1.34, SD = 0.33) and ‘push metaphors’ (M = 1.30, SD = 0.35) did not significantly differ in motor imagery (t(44) = - 0.46, p =0.65, n.s.). Lastly, participants were also asked to specify the force strength of each of the 18 ‘pull type’ and 18 ‘push type’ action-verbs, if any, on a 7-point Likert scale (1- no force, 7-very large force), as well as, the direction of the force implied whether it is 1) towards the body 2) away from the body 3) both or other direction. Results showed that 76% of the time ‘pull type’ action-verbs were associated with a force towards the person executing the action while 86% of the time ‘push type’ action-verbs were associated with a force away from the person executing the action. Additionally, ‘pull type’ action-verbs (M = 2.8, SD = 1.00) and ‘push type’ action- verbs (M = 3.00, SD = 0.85) did not significantly differ in terms of implied force strength (t(17) = -0.61, p = 0.55, n.s.). Experimental Paradigm: Participants first completed the language task. Inside the fMRI scanner participants were presented with a topic phrase (‘The physics lecture’) on the screen that then disappeared from the screen and was followed by presentation of a sentence about that topic (‘She’s grasping the lecture’) on the screen. We presented the topic phrase in order to encourage that the following sentence and verb would be interpreted metaphorically. Participants were simply instructed to read the topic sentence followed by a sentence about that topic for comprehension and pay attention to whether the sentence made sense or not. A number of catch trials were included for which the topic sentence and sentence did not make sense (e.g., topic: ‘The pepper’, sentence: ‘She punched the pepper’) and recall of these catch trials was done to ensure participants were paying attention. The language task was divided into 3 runs each lasting 8.5 minutes each during which each participant saw 15 metaphoric sentences and 15 literal paraphrases (also 2 catch trials) for a total of 45 metaphoric sentences and 45 literal paraphrases. The topic sentence was presented on the screen for 1.8 seconds followed by an inter-trial interval of 0.2 seconds and then the sentence was presented for 6 seconds followed by a rest period of 8 seconds in an event-related design in a pseudo-randomized presentation. During each rest period participants simply fixated a cross on a gray screen. All the stimuli were presented on a computer screen using Matlab Psychophysics toolbox (Psychtoolbox 3, www.psychtoolbox.org)). Lastly, the motor task was presented after the language task in order to avoid any priming of conceptual categories associated with ‘pushing’ and ‘pulling’. Additionally, for the motor task participants were instructed visually not verbally (i.e., the verbal labels ‘push’ and ‘pull’ were not used) prior to the scan, rather using an fMRI safe object subjects were shown how to properly perform both a ‘push’ vs. ‘pull’ type action paired with either a ‘red screen’ or ‘green screen’ as cue indicating which action to perform. Inside the fMRI scanner participants performed the ‘push’ action when they saw a red screen and the ‘pull’ action when they saw a green screen for a duration of 7 seconds. After performance of each action participants rested for 8 seconds during which they simply fixated a cross on a gray screen. Stimuli presentations were randomized and consisted of two runs each lasting 7.5 minutes each during which participants made 15 pull-type actions and 15 push-type actions per run, for a total of 30 pull-type actions and 30 push-type 96 actions. However, due to timing limitations seven participants only completed a single run of the motor task, therefore, while we included this in a univariate analysis we did not do a multivariate analysis with the motor task. MRI Data Acquisition: Functional MRI images were acquired with a Siemens MAGNETOM Trio 3T System with a 32-channel head matrix coil in the Dornsife Cognitive Neuroscience Imaging Center at the University of Southern California. A high-resolution anatomical scan was acquired for each subject: Structural T1-weighted magnetization-prepared rapid gradient echo (MPRAGE), TR=1950 ms, TE=2.26 ms, flip angle=9 degrees, 256 x 224 mm matrix, 1 mm resolution, 176 coronal slices. Whole brain functional images were obtained with a T2* weighted single-shot gradient-recalled echoplanar imaging, echo-planar sequence (EPI) using blood- oxygenation-level-dependent contrast. Each functional image comprised of 37 contiguous axial slices (3.5 mm thick), acquired in interleaved mode, and with a repetition time (TR) of 2000 ms (echo time (TE) of 30 ms, flip angle (FA) 90 degrees, 64x64 mm matrix). fMRI- Multivariate Pattern Analysis: Preprocessing consisted of concatenating two motor task runs to form the motor data set, while all four language runs were concatenated to form the language data set for each subject. Each data set was then motion-corrected to the middle slice using FSL, Oxford Centre for Functional Magnetic Resonance Imaging of the Brain (FMRIB’s) Software Library, http://www.fmrib.ox.ac.uk/fsl/index.html. Multivariate pattern analysis (MVPA) was carried out using the PyMVPA 0.6 software package (http://www.pymvpa.org/; (Hanke et al., 2009) and a linear support vector machine from LibSVM (http://www.csie.ntu.edu.tw/~cjlin/libsvm). Each data set was linearly detrended and normalized to Z-scores using PyMVPA in order to remove intensity differences across runs. We then used a searchlight-based analysis to calculate classification accuracy scores within spheres with a four- voxel radius (a cluster of 257 voxels) centered on each voxel of the brain per subject. Within each spherical searchlight we performed the binary classification, ‘push metaphor’ vs. ‘pull metaphor’, and obtained accuracy scores using the Gaussian Naïve Bayes classifier (Pereira et al., 2009) that were then assigned to each center voxel in order to obtain whole-brain accuracy maps. We labeled every TR in our event-related design as belonging to either the ‘push metaphor’ or ‘pull metaphor’ conditions such that each TR served as an individual sample for input to the classifier. This resulted in x samples for each condition per run that served as input to the classifier. To avoid biases of any particular run, we used a threefold cross-validation scheme where we trained the classifier using samples form one run and, subsequently, tested the classifier on the other two remaining runs to obtain a classification accuracy score. We repeated this three times for each run and then averaged all the results of the training/testing combinations, leaving whole-brain maps for each subject that contained average accuracy scores in each voxel. We then subtracted the chance-level of 0.5 from each accuracy score and further demeaned each score to minimize variance at the group level. We then combined these maps using a random- effects group analysis to identify regions of above-chance performance. Significance was obtained at voxel-wise uncorrected p<0.001 with FDR correction p<0.01. fMRI- Univariate Analysis: All preprocessing and statistical analysis were carried out using Statistical parametric mapping (SPM12) http://www.fil.ion.ucl.ac.uk/spm/software/spm12/) 97 software. Preprocessing of fMRI data included spatial realignment, coregistration, normalization and smoothing. To control for motion, all functional volumes were realigned to the mean volume. Images were spatially normalized to standard MNI space using the participant’s individual skull-stripped high-resolution T1 anatomical images with a voxel size of 3x3x3 mm. Volumes were spatially smoothed using a 9 mm full-width at half-maximum Gaussian filter, and high-pass filtered (128 seconds) to remove low-frequency drifts in the signal. The design matrix consisted of a synthetic hemodynamic response function (double gamma function) with its temporal derivative convolved with the input stimuli waveform in block design. In the main language task, three separate regressors were included in the model: 1) pull-type metaphors; 2) push-type metaphors 3) catch trials. In the localizer motor task design two separate regressors were included in the model: 1) pull-type action; 2) push-type action. The rest periods in between trials served as the baseline contrast. Six motion parameters were also included in the design matrix as regressors to account for motion artifacts. The following additional contrasts were defined for each participant and used in the group analysis: For the language task (pull-type metaphor vs. push-type metaphor). For the motor task (pull-type action vs. push-type action, all- actions vs. rest). A mixed-effects analysis was carried out at the second level in order to compute group statistics. All statistical images were thresholded at voxel-wise uncorrected p<0.001 with FDR correction p<0.01 to account for multiple comparisons across the whole brain with a minimum cluster size of 5. RESULTS fMRI- Multivariate Pattern Analysis: As Table 1 shows, a within-modality whole-brain MVPA searchlight analysis utilizing a four-voxel radius spherical searchlight revealed that voxels surviving significance testing were found within frontal regions, mainly the bilateral frontal medial cortex, left inferior frontal gyrus (BA 44), left frontal operculum, and right middle frontal gyrus at the group level analysis in the language task. Additionally, significant voxels were also found across the bilateral cortical sensorimotor systems. Within the motor cortex significant voxels were found in the bilateral precentral gyrus (BA 4, 6), left postcentral gyrus mainly within the parietal operculum. There were also significant voxels within the left inferior parietal lobule (IPL) (Figure 1). The peak t-value was localized to the right parietal operculum part of a large cluster of significant voxels (597 voxels) that extended into parts of the right postcentral gyrus, right precentral gyrus (BA 4, 6), and the right IPL (mainly the anterior supramarginal gyrus). A number of voxels within temporal/visual areas also included significant voxels mainly within the left middle temporal gyrus, bilateral inferior lateral occipital cortex, and bilateral occipital pole. A number of limbic regions also had voxels that passed significance testing at the group-level analysis including the left frontal orbital cortex (extending into the parahippocampal gyrus, temporal pole, and putamen), bilateral posterior cingulate gyrus, left anterior cingulate gyrus, and insula (dorsal/ventral anterior regions). Additionally, areas of the subcortical motor system, the left pallidum and the bilateral cerebellum crus 1 also contained voxels that passed significance testing. Given that 6 participants completed only one out of the three runs of the motor task we 98 did not perform a cross-modality MVPA classification with the motor task and language task due to insufficient sample size. fMRI- Univariate Analysis: As Table 2 shows, the motor localizer task when combining both pull-type and push-type actions (i.e., all actions) vs. rest showed significant activity within a number of sensorimotor cortical areas including the bilateral precentral gyrus (BA 4, 6), bilateral postcentral gyrus mainly within the parietal operculum cortex, bilateral IPL (supramarginal gyrus), and right supplementary motor region (SMA) was found. Significant activity was also found within a number of subcortical brain regions including the bilateral putamen, bilateral thalamus, and right cerebellum. We also found significant activity in visual and temporal areas including the left occipital pole, right middle temporal gyrus (temporoccipital part), and right lateral occipital cortex. We also found similar findings at the uncorrected level p<0.001 for pull- type actions vs. rest and push-type actions vs. rest, not shown. Activations in the univariate motor task overlapped anatomically with activations in our multivariate analysis, mainly within the bilateral occipital lobe, left putamen, right precentral gyrus (BA 6), left precentral gyrus (BA 4, 6), left postcentral gyrus (mainly within the left parietal operculum), and left IPL (supramarginal gyrus (BA 40)) (Figure 2). For comparison purposes we ran a conventional univariate analysis of the language task, as well, but did not find significant activations when contrasting pull-type metaphors and push-type metaphors. DISCUSSION In this study we sought to understand if sensorimotor brain areas involved in action execution could successfully distinguish between two hand-action metaphors that differed in their patterns of force exchange. We presented participants with 1) Metaphors drawing on action- verbs that imply an away-from-self force toward an antagonist and relate to the act of communicating (e.g., “She’s pushing the agenda.”) and 2) metaphors drawing on action-verbs that imply a force that would move the antagonist towards the self and relate to cognizing (e.g., “She’s grasping the idea.”). We used a whole-brain searchlight multivariate pattern analysis (MVPA) approach to search for patterns of activity within a 4-voxel searchlight sphere across the whole brain that could train a classifier to successfully distinguish between our two types of metaphors. A motor localizer task allowed us to detect anatomical overlap between significant activations in sensorimotor brain areas actually involved in action execution (i.e., pulling and pushing actions on an object) in the univariate analysis of the motor localizer task and voxels with significant classification accuracy in the MVPA analysis of the language task that could distinguish between our two types of metaphor. The results of our within-modality MVPA analysis showed that a number of sensorimotor brain regions including the bilateral precentral (BA 4, 6) and postcentral cortices (mainly secondary somatosensory cortex) contained significant voxels that could distinguish pull-type metaphors from push-type metaphors (Figure 1, Table 1). Additionally, the left inferior parietal cortex (IPL), a ‘higher-level motor area’, mainly the supramarginal gyrus, also contained patterns of activity that could significantly distinguish between the two metaphors. Critically, these brain regions partially overlapped with significant 99 activity found in our univariate analysis of the motor localizer task, mainly within the bilateral precentral gyrus (BA 6), left postcentral gyrus (including the parietal operculum), and left IPL (supramarginal gyrus (BA 40)) (Figure 2). These findings suggest that the processing of hand- action metaphors engage sensorimotor processes relevant to their specific encoded force dynamics in line with embodied cognition theories, revealing an unexpected degree of motor- specificity for the processing of hand-action metaphors. Furthermore, the findings highlight that motor areas associated with distinct levels of motor hierarchy, both low-level kinematic action representations (e.g., primary motor areas) and higher-level aspects of action that are rather thought to reflect the underlying action goal/intention (e.g., inferior parietal lobule IPL)) (Fogassi et al., 2005b; Fogassi & Luppino, 2005a; Iacoboni et al., 2005; Iacoboni et al., 1999; Rizzolatti, Cattaneo, Fabbri-Destro, & Rozzi, 2014) contained patterns of activity that could distinguish between the two types of metaphors in our study. It is possible that significant voxels within primary motor/premotor areas found in our study reflect differences in implied force direction (or force strength) between our two metaphor types. Recall that the study by Moody and Gennari 2010 showed that the left anterior IFG was sensitive to the implied physical effort in the sentence (‘pushing the chair’ vs. ‘pushing the piano’). In contrast, the IPL, which was active across all action sentences compared to abstract sentences, was not modulated by the implied force strength in this study. While some studies have primarily found direction selective areas within primary motor cortex (Eisenberg, Shmuelof, Vaadia, & Zohary, 2010), other studies have also found direction selectivity in premotor, but also parietal regions (intraparietal sulcus and parietal reach area) (Fabbri, Caramazza, & Lingnau, 2010). Therefore, significant voxels in the precentral gyrus (BA 4, 6), but also perhaps IPL, in our study could similarly reflect force strength in a particular direction. A number of studies have additionally shown, however, that the parietal cortex, mainly the inferior parietal lobule (IPL), is mainly sensitive to action goals/outcomes regardless of the specific motor movements required to accomplish a specific task (e.g., pulling vs. pushing a door to open it) (Desmurget & Sirigu, 2009; Haggard, 2008; Hamilton & Grafton, 2008; Krasovsky et al., 2014; Tunik, Rice, Hamilton, & Grafton, 2007; van Elk, 2014; Wurm & Lingnau, 2015). Given such findings, significant patterns of activity in the left IPL that successfully distinguished between our metaphor types could also reflect higher-level action goals associated with physical force exchange between objects and agents (i.e., an away-from-self force toward an antagonist implies that the agent’s force is encountering resistance from the object, which may overcome or succumb to the agent’s force). Overall the results point to at least two levels of motor hierarchy the kinematic and the goal/outcome levels are involved in processing of more abstract events in metaphor, in contrast to previous studies that mainly found activity within the IPL for familiar action metaphors (Desai et al., 2011; 2013) and studies that suggest that it is primarily more ‘high-level’ schematized motor representations that contribute to metaphor processing (Troyer et al., 2014). The current study also showed that activations in our univariate motor task analysis overlapped anatomically with significant voxels in our multivariate analysis in regions extending beyond ‘strictly’ motor areas of the brain, mainly within areas of the occipital cortex, left postcentral gyrus (secondary somatosensory cortex/parietal operculum), and the left putamen. 100 The involvement of these areas across both the motor task and the language task, despite the distinct analysis, underscore the involvement of occipital to parietal and frontal regions in the semantics of goal-directed hand actions involving objects (Rizzolatti et al., 2014). At a basic level grasping an object involves reaching for the object and pre-shaping the hand, both of which depend not only on motor input, but also visual input associated with the object properties and finally somatosensory input associated with the act of exerting force against the object when finally gripping it and holding it (van Polanen & Davare, 2015). In one view, low-level visual and motor kinematic features are first processed in lateral occipital regions and inferior frontal regions, respectively. This involves the classically defined two pathways form visual to premotor areas, mainly the dorsolateral and dorsomedial circuits that are believed to process distinct aspects of the visuomotor transformations involved in reach to grasp movements. These more low-level kinematics are integrated further up the motor hierarchy in the parietal cortex, which putatively processes higher-level action representations relating to action outcomes/goals and prior intentions (Hamilton & Grafton, 2008; Hamilton, Wolpert, Frith, & Grafton, 2006). Using a repetition suppression technique Hamilton and Grafton, 2006 have found evidence for this action hierarchy. They found that when participants viewed action videos performed with the hand that used different types of grasps (e.g., hand taking a wine-bottle with precision vs. whole-hand grip) the lateral occipital cortex (LOC), middle interaparietal sulcus, and inferior frontal gyrus (IFG) were sensitive to repeated grasps when compared to novel grasps (i.e., low-level kinematics). In contrast, when focusing on hand videos showing different goals (e.g., hand taking a wine-bottle vs. a dumb-bell), the authors found that the anterior intraparietal sulcus (aIPS), cerebellum, and basal ganglia showed adaptation to repeated goals compared to novel goals. The similar pattern of distributed activity from occipital, to parietal and frontal regions that could distinguish between our two types of metaphor is strongly suggestive of the fact that these areas are coding semantics of goal-directed hand actions involving objects similar to that described by studies of action observation. Interestingly, in the above study by Hamilton and Grafton 2006 the basal ganglia along with the IPL showed repetition suppression effects when the goals of hand actions were repeated compared to novel goals. It is possible that significant patterns of activity in the left putamen in our MVPA analysis suggest a similar role for the basal ganglia in coding hand-action goals. Although the motor function of the basal ganglia is still debated, studies have primarily highlighted the basal ganglia in the regulation of movement, such as in the control of precision grip force in grasping (Prodoehl, Corcos, & Vaillancourt, 2009; Wasson, Prodoehl, Coombes, Corcos, & Vaillancourt, 2010) and not necessarily in coding action goals/outcomes. However, the IPL, which is believed to code for action goals/outcomes (e.g., grasping to eat vs. grasping to place), has also been shown to be sensitive to specific movement parameters such as ‘how much the finger should be lifted rather than that the finger should be lifted’ (Iacoboni et al., 2001; Iacoboni et al., 1999; van Dam, Rueschemeyer, & Bekkering, 2010). Thus, it is possible that the involvement of the basal ganglia (and for that matter the IPL) in both the motor and language tasks rather reflects movement kinematics, mainly force direction that are pertinent to achieving a specific action goal/plan and may not actually contribute to coding more abstract differences associated with patterns of force exchange between objects and agents in our motor and language task, but this remains to be investigated. 101 The peak t-value in our study was associated mainly with patterns of activity across voxels in the right secondary somatosensory cortex/parietal operculum, which has previously been associated with sensorimotor prediction in motor control (Blakemore & Sirigu, 2003; Sukhwinder, 2013). A number of studies have shown that activity within the SII is diminished when action and tactile stimulation occur simultaneously when compared to the presentation of the same tactile stimulation alone without self-initiated action (Blakemore, Goodbody, & Wolpert, 1998; Blakemore, Wolpert, & Frith, 2000; Sukhwinder, 2013). Moreover, Sukhwinder et al., 2013 further showed that dampening of activity within the SII during action could be reduced by introducing a time delay between motor and tactile experience, suggesting that the SII’s activity may be modulated by action in accordance to proposed forward models in motor control. In such a view during action execution or during action rehearsal/imagery a prediction of the sensory consequences of the action are simulated and are used to either cancel somatosensory input, as in the above case, or adjust ongoing movement by calculating a prediction error (i.e., comparing predicted sensory outcome with actual sensory outcome). Consistent with the idea of SII in sensorimotor prediction other studies have found that the somatosensory cortex (SII) along with the insula are active prior to the initiation of a planned action (Jackson, Parkinson, Pears, & Nam, 2011; Parkinson et al., 2011). This result suggests that the ability of SII to discriminate between our two metaphors may be due to its putative role in sensorimotor prediction. That is, somatosensory simulations may represent the largest differences between ‘pulling’ and ‘pushing’ type force dynamic schemas relevant to distinguishing between our two metaphor types. Taken together, our findings suggest that sensorimotor simulations underlie action semantics in the processing of hand-action metaphors that differ in their force dynamics in line with predictions from embodied cognition theories, including Conceptual Metaphor Theory and the Theory of Force Dynamics. We note that a limitation to our current study is that we were not able to carry out a cross-modal classification due to the fact that not all participants completed enough runs of the motor task. However, this analysis would have presumably been a more sensitive test for investigating overlap between sensorimotor processes that can distinguish across both physical actions that differ in their force dynamics and those involved in comprehension of hand-action metaphors that also similarly differ in their force dynamics. We hope to address this in a future study. Another limitation is that although our metaphors were rated as relatively familiar it is possible that the task may have encouraged deeper conceptual processing. In our semantic judgment task participants were asked to assess a given topic, ‘The physics lecture’, followed by a sentence about that topic, ‘She’s grasping the physics lecture’, for sensibility (i.e., does the sentence make sense given the topic). Furthermore, participants were asked to recall catch trials that did not make sense, such as presentation of the topic, ‘The green pepper’, followed by ‘She’s punching the green pepper’ at the end of each run. This particular task may have led to specific strategies that induced motor imagery or relied on more specific sensorimotor simulations than would normally occur in natural language use. Nevertheless, this would still be in line with embodied cognition theories which stipulate that people make use of sensorimotor simulations when interpreting metaphorical expressions (Gibbs 2006). The strategies employed may have made participants process our familiar metaphors similar to novel metaphors, which have been shown to engage sensorimotor brain areas to a greater degree than familiar metaphors and 102 possibly employ different cognitive strategies (Desai et al., 2011; 2013; Bowdle & Gentner 2005). Given this, however, it is possible that the strategy used by participants in our task resulted in sensorimotor simulations mediated by top-down processes rather than automatic bottom-up processes relevant to comparisons between source and target domains in the metaphor. This would be in line with a study by Papeo et al, 2009 who showed that when TMS was applied to the posterior MTG activity previously seen in the precentral gyrus when contrasting action verbs versus non-action verbs disappeared suggesting that embodiment processes may be employed by specific top-down strategies used in comprehension that are context-dependent rather than automatic as suggested by some embodiment theories (Papeo, Vallesi, Isaja, & Rumiati, 2009), for reviews on context-dependent activations see (Willems & Casasanto, 2011a; Yang, 2014). The above interpretation is also consistent with the fact that our multivariate analysis showed the involvement of a number of additional brain regions not found in our motor localizer task. Specifically, we found that a number of frontal areas, mainly the left frontal operculum, left inferior frontal gyrus –pars opercularis (BA 44), right middle frontal gyrus, and bilateral frontal medial cortex showed significant voxels in our MVPA whole-brain searchlight (Table 1). Similarly, we found that activity across a number of temporal/visual areas could also successfully distinguish between our two metaphor types, as described above this may relate to visuomotor transformations, mainly the left middle temporal gyrus (temporoccipital part), bilateral inferior lateral occipital cortex, and bilateral occipital pole temporal. Lastly, a number of limbic regions including the left frontal orbital cortex (extending into the parahippocampal gyrus, temporal pole, and putamen), bilateral posterior cingulate gyrus, left anterior cingulate gyrus, and insula (dorsal/ventral anterior regions) were also found to contain patterns of activity across voxels that could distinguish between the two metaphor types. The recruitment of emotion-related regions could reflect the fact that our two metaphor types may have differed in emotional engagement, which needs to be assessed in future studies. While our within-modality MVPA findings underscore the importance of other brain regions besides sensorimotor brain regions in distinguishing between our two types of metaphors, we note that voxels with significant classification accuracy in the MVPA analysis of the language task overlapped anatomically mainly with activations in sensorimotor brain areas in the univariate analysis of the motor localizer task. Furthermore, the peak t-value and largest cluster of significant voxels in our within-modality MVPA classification analysis in the language task was found within sensorimotor brain areas (mainly secondary somatosensory cortex). Thus, our findings strongly suggest that patterns of activity across the fronto-parietal action network that are involved in actions on objects by agents can successfully distinguish between two metaphors that differ in their encoded force dynamics, but point to a special role for somatosensory simulation. Further study will need to discern whether these sensorimotor brain areas are necessary or not for the processing of force dynamic semantics (i.e., force exchange patterns involving objects and agents) in non-verbal, literal sentences, and metaphoric sentences and continue to characterize the possibly distinct contributions of different levels of the motor hierarchy in metaphor comprehension. 103 References Aziz-Zadeh, L., Wilson, S. M., Rizzolatti, G., & Iacoboni, M. (2006). Congruent embodied representations for visually presented actions and linguistic phrases describing actions. Curr Biol, 16(18), 1818-1823. doi:10.1016/j.cub.2006.07.060 Barsalou, L. W. (1999). Perceptual symbol systems. Behav Brain Sci, 22(4), 577-609; discussion 610-560. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11301525 Barsalou, L. W. (2003). Abstraction in perceptual symbol systems. Philos Trans R Soc Lond B Biol Sci, 358(1435), 1177-1187. doi:10.1098/rstb.2003.1319 Barsalou, L. W. (2008). Grounded cognition. Annu Rev Psychol, 59, 617-645. doi:10.1146/annurev.psych.59.103006.093639 Blakemore, S. J., Goodbody, S. J., & Wolpert, D. M. (1998). Predicting the consequences of our own actions: the role of sensorimotor context estimation. J Neurosci, 18(18), 7511-7518. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9736669 Blakemore, S. J., & Sirigu, A. (2003). Action prediction in the cerebellum and in the parietal lobe. Exp Brain Res, 153(2), 239-245. doi:10.1007/s00221-003-1597-z Blakemore, S. J., Wolpert, D., & Frith, C. (2000). Why can't you tickle yourself? Neuroreport, 11(11), R11-16. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10943682 Boot, I., & Pecher, D. (2011). Representation of Categories Metaphorical Use of the Container Schema. Exp Psychol, 58(2), 162-170. doi:10.1027/1618-3169/a000082 Bowdle, B. F., & Gentner, D. (2005). The career of metaphor. Psychol Rev, 112(1), 193-216. doi:10.1037/0033-295X.112.1.193 Buccino, G., Riggio, L., Melli, G., Binkofski, F., Gallese, V., & Rizzolatti, G. (2005). Listening to action-related sentences modulates the activity of the motor system: a combined TMS and behavioral study. Brain Res Cogn Brain Res, 24(3), 355-363. doi:10.1016/j.cogbrainres.2005.02.020 Carr, E. W., Kavanagh, L., & Bergen, B. K. (2013). How embodied is metaphor? Current Directions in Psychological Science, CDPS-12-0113. Desai, R. H., Binder, J. R., Conant, L. L., Mano, Q. R., & Seidenberg, M. S. (2011). The neural career of sensory-motor metaphors. J Cogn Neurosci, 23(9), 2376-2386. doi:10.1162/jocn.2010.21596 Desai, R. H., Conant, L. L., Binder, J. R., Park, H., & Seidenberg, M. S. (2013). A piece of the action: modulation of sensory-motor regions by action idioms and metaphors. Neuroimage, 83, 862-869. doi:10.1016/j.neuroimage.2013.07.044 Desmurget, M., & Sirigu, A. (2009). A parietal-premotor network for movement intention and motor awareness. Trends Cogn Sci, 13(10), 411-419. doi:10.1016/j.tics.2009.08.001 Eisenberg, M., Shmuelof, L., Vaadia, E., & Zohary, E. (2010). Functional organization of human motor cortex: directional selectivity for movement. J Neurosci, 30(26), 8897-8905. doi:10.1523/JNEUROSCI.0007-10.2010 Fabbri, S., Caramazza, A., & Lingnau, A. (2010). Tuning Curves for Movement Direction in the Human Visuomotor System. Journal of Neuroscience, 30(40), 13488-13498. doi:10.1523/Jneurosci.2571-10.2010 Fodor, J. A. (1975). The Language of Thought. New York, NY: Thomas Y. Crowell. Fogassi, L., Ferrari, P. F., Gesierich, B., Rozzi, S., Chersi, F., & Rizzolatti, G. (2005b). Parietal lobe: from action organization to intention understanding. Science, 308(5722), 662-667. doi:10.1126/science.1106138 104 Fogassi, L., & Luppino, G. (2005a). Motor functions of the parietal lobe. Curr Opin Neurobiol, 15(6), 626-631. doi:10.1016/j.conb.2005.10.015 Gallese, V., & Lakoff, G. (2005). The Brain's concepts: the role of the Sensory-motor system in conceptual knowledge. Cogn Neuropsychol, 22(3), 455-479. doi:10.1080/02643290442000310 Gibbs, R. W. (2006). Metaphor Interpretation as Embodied Simulation. Mind & Language, 21(3), 434-458. doi:10.1111/j.1468-0017.2006.00285.x Glenberg, A. M., & Gallese, V. (2012). Action-based language: a theory of language acquisition, comprehension, and production. Cortex, 48(7), 905-922. doi:10.1016/j.cortex.2011.04.010 Glenberg, A. M., & Kaschak, M. P. (2002). Grounding language in action. Psychon Bull Rev, 9(3), 558-565. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12412897 Haggard, P. (2008). Human volition: towards a neuroscience of will. Nat Rev Neurosci, 9(12), 934-946. doi:10.1038/nrn2497 Hamilton, A. F., & Grafton, S. T. (2008). Action outcomes are represented in human inferior frontoparietal cortex. Cereb Cortex, 18(5), 1160-1168. doi:10.1093/cercor/bhm150 Hamilton, A. F., Wolpert, D. M., Frith, U., & Grafton, S. T. (2006). Where does your own action influence your perception of another person's action in the brain? Neuroimage, 29(2), 524-535. doi:10.1016/j.neuroimage.2005.07.037 Hauk, O., Johnsrude, I., & Pulvermuller, F. (2004). Somatotopic representation of action words in human motor and premotor cortex. Neuron, 41(2), 301-307. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14741110 Hauk, O., & Tschentscher, N. (2013). The Body of Evidence: What Can Neuroscience Tell Us about Embodied Semantics? Front Psychol, 4, 50. doi:10.3389/fpsyg.2013.00050 Iacoboni, M., Koski, L. M., Brass, M., Bekkering, H., Woods, R. P., Dubeau, M. C., . . . Rizzolatti, G. (2001). Reafferent copies of imitated actions in the right superior temporal cortex. Proc Natl Acad Sci U S A, 98(24), 13995-13999. doi:DOI 10.1073/pnas.241474598 Iacoboni, M., Molnar-Szakacs, I., Gallese, V., Buccino, G., Mazziotta, J. C., & Rizzolatti, G. (2005). Grasping the intentions of others with one's own mirror neuron system. PLoS Biol, 3(3), e79. doi:10.1371/journal.pbio.0030079 Iacoboni, M., Woods, R. P., Brass, M., Bekkering, H., Mazziotta, J. C., & Rizzolatti, G. (1999). Cortical mechanisms of human imitation. Science, 286(5449), 2526-2528. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10617472 Jackson, S. R., Parkinson, A., Pears, S. L., & Nam, S. H. (2011). Effects of motor intention on the perception of somatosensory events: a behavioural and functional magnetic resonance imaging study. Q J Exp Psychol (Hove), 64(5), 839-854. doi:10.1080/17470218.2010.529580 Kemmerer, D., Castillo, J. G., Talavage, T., Patterson, S., & Wiley, C. (2008). Neuroanatomical distribution of five semantic components of verbs: evidence from fMRI. Brain Lang, 107(1), 16-43. doi:10.1016/j.bandl.2007.09.003 Keysar, B., & Bly, B. M. (1999). Swimming against the current: Do idioms reflect conceptual structure? Journal of Pragmatics, 31(12), 1559-1578. doi:Doi 10.1016/S0378- 2166(99)00004-1 105 Keysar, B., Shen, Y. I., Glucksberg, S., & Horton, W. S. (2000). Conventional language: How metaphorical is it? Journal of Memory and Language, 43(4), 576-593. doi:10.1006/jmla.2000.2711 Krasovsky, A., Gilron, R., Yeshurun, Y., & Mukamel, R. (2014). Differentiating intended sensory outcome from underlying motor actions in the human brain. J Neurosci, 34(46), 15446-15454. doi:10.1523/JNEUROSCI.5435-13.2014 Lakoff, G. (2014). Mapping the brain's metaphor circuitry: metaphorical thought in everyday reason. Front Hum Neurosci, 8. doi:ARTN 958 10.3389/fnhum.2014.00958 Lakoff, G., & Johnson, M. (1980a). The metaphorical structure of the human conceptual system. Cognitive Science, 4(2), 195-208. doi:10.1016/S0364-0213(80)80017-6 Lakoff, G., & Johnson, M. (1980b). Metaphors we live by. Chicago: University of Chicago Press. Mahon, B. Z., & Caramazza, A. (2008). A critical look at the embodied cognition hypothesis and a new proposal for grounding conceptual content. J Physiol Paris, 102(1-3), 59-70. doi:10.1016/j.jphysparis.2008.03.004 Moody, C. L., & Gennari, S. P. (2010). Effects of implied physical effort in sensory-motor and pre-frontal cortex during language comprehension. Neuroimage, 49(1), 782-793. doi:10.1016/j.neuroimage.2009.07.065 Niedenthal, P. M., Barsalou, L. W., Winkielman, P., Krauth-Gruber, S., & Ric, F. (2005). Embodiment in attitudes, social perception, and emotion. Pers Soc Psychol Rev, 9(3), 184-211. doi:10.1207/s15327957pspr0903_1 Papeo, L., Vallesi, A., Isaja, A., & Rumiati, R. I. (2009). Effects of TMS on different stages of motor and non-motor verb processing in the primary motor cortex. PLoS One, 4(2), e4508. doi:10.1371/journal.pone.0004508 Parkinson, A., Plukaard, S., Pears, S. L., Newport, R., Dijkerman, C., & Jackson, S. R. (2011). Modulation of somatosensory perception by motor intention. Cogn Neurosci, 2(1), 47-56. doi:10.1080/17588928.2010.525627 Pecher, D., Boot, I., & Van Dantzig, S. (2011). Abstract Concepts: Sensory-Motor Grounding, Metaphors, and Beyond. Psychology of Learning and Motivation: Advances in Research and Theory, Vol 54, 54, 217-248. doi:10.1016/B978-0-12-385527-5.00007-3 Pereira, F., Mitchell, T., & Botvinick, M. (2009). Machine learning classifiers and fMRI: a tutorial overview. Neuroimage, 45(1 Suppl), S199-209. doi:10.1016/j.neuroimage.2008.11.007 Prodoehl, J., Corcos, D. M., & Vaillancourt, D. E. (2009). Basal ganglia mechanisms underlying precision grip force control. Neurosci Biobehav Rev, 33(6), 900-908. doi:10.1016/j.neubiorev.2009.03.004 Pulvermuller, F. (1999). Words in the brain's language. Behav Brain Sci, 22(2), 253-279; discussion 280-336. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11301524 Pulvermuller, F. (2001). Brain reflections of words and their meaning. Trends Cogn Sci, 5(12), 517-524. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11728909 Pulvermuller, F., Hauk, O., Nikulin, V. V., & Ilmoniemi, R. J. (2005). Functional links between motor and language systems. Eur J Neurosci, 21(3), 793-797. doi:10.1111/j.1460- 9568.2005.03900.x Pylyshyn, Z. W. (1984). Computation and cognition: Towards a foundation for cognitive science. . Boston, MA: MIT Press. 106 Raposo, A., Moss, H. E., Stamatakis, E. A., & Tyler, L. K. (2009). Modulation of motor and premotor cortices by actions, action words and action sentences. Neuropsychologia, 47(2), 388-396. doi:10.1016/j.neuropsychologia.2008.09.017 Reddy, M. J. (1979). The conduit metaphor -- a case of frame conflict in our language about langauge. . Cambridge: Cambridge University Press. Rizzolatti, G., Cattaneo, L., Fabbri-Destro, M., & Rozzi, S. (2014). Cortical mechanisms underlying the organization of goal-directed actions and mirror neuron-based action understanding. Physiol Rev, 94(2), 655-706. doi:10.1152/physrev.00009.2013 Rueschemeyer, S. A., van Rooij, D., Lindemann, O., Willems, R. M., & Bekkering, H. (2010). The function of words: distinct neural correlates for words denoting differently manipulable objects. J Cogn Neurosci, 22(8), 1844-1851. doi:10.1162/jocn.2009.21310 Simmons, W. K., & Barsalou, L. W. (2003). The similarity-in-topography principle: reconciling theories of conceptual deficits. Cogn Neuropsychol, 20(3), 451-486. doi:10.1080/02643290342000032 Sukhwinder, S. (2013). Modulation of somatosensory processing by action. Neuroimage, 70(Apr 15), 356-362. Talmy, L. (1988). Force Dynamics in Language and Cognition. Cognitive Science, 12(1), 49- 100. doi:Doi 10.1016/0364-0213(88)90008-0 Tunik, E., Rice, N. J., Hamilton, A., & Grafton, S. T. (2007). Beyond grasping: representation of action in human anterior intraparietal sulcus. Neuroimage, 36 Suppl 2, T77-86. doi:10.1016/j.neuroimage.2007.03.026 van Dam, W. O., Rueschemeyer, S. A., & Bekkering, H. (2010). How specifically are action verbs represented in the neural motor system: an fMRI study. Neuroimage, 53(4), 1318- 1325. doi:10.1016/j.neuroimage.2010.06.071 van Elk, M. (2014). The left inferior parietal lobe represents stored hand-postures for object use and action prediction. Front Psychol, 5, 333. doi:10.3389/fpsyg.2014.00333 van Polanen, V., & Davare, M. (2015). Interactions between dorsal and ventral streams for controlling skilled grasp. Neuropsychologia, 79(Pt B), 186-191. doi:10.1016/j.neuropsychologia.2015.07.010 Wasson, P., Prodoehl, J., Coombes, S. A., Corcos, D. M., & Vaillancourt, D. E. (2010). Predicting grip force amplitude involves circuits in the anterior basal ganglia. Neuroimage, 49(4), 3230-3238. doi:10.1016/j.neuroimage.2009.11.047 Willems, R. M., & Casasanto, D. (2011a). Flexibility in embodied language understanding. Front Psychol, 2, 116. doi:10.3389/fpsyg.2011.00116 Willems, R. M., Hagoort, P., & Casasanto, D. (2010). Body-specific representations of action verbs: neural evidence from right- and left-handers. Psychol Sci, 21(1), 67-74. doi:10.1177/0956797609354072 Wurm, M. F., & Lingnau, A. (2015). Decoding actions at different levels of abstraction. J Neurosci, 35(20), 7727-7735. doi:10.1523/JNEUROSCI.0188-15.2015 Yang, J. (2014). Influences of motor contexts on the semantic processing of action-related language. Cognitive, Affective, & Behavioral Neuroscience, 14(3), 912-922. doi:10.3758/s13415-014-0258-y 107 TABLES AND FIGURES Table 1. Brain regions with significant voxels from MVPA searchlight analysis (4-voxel radius) Anatomic Region Hem. X Y Z Size (Voxels) T- value Pull Metaphors vs. Push Metaphors Frontal medial cortex R 10 38 -12 6 6.19 Frontal medial cortex L -6 34 -20 48 7.03 Anterior cingulate gyrus L -10 -12 38 97 10.30 Posterior cingulate gyrus L -2 -42 38 37 7.34 Frontal orbital cortex L -44 24 -8 11 6.59 Frontal orbital cortex L -18 10 -20 248 7.59 Parahippocampal gyrus L -18 2 -24 6.19 Temporal pole L -22 8 -28 8.49 Putamen L -18 14 -10 6.66 IFG, pars opercularis (BA 44) L -60 16 20 12 7.47 Middle frontal gyrus R 30 16 32 116 10.89 Frontal operculum cortex L -44 10 8 32 8.33 Parietal operculum cortex (SII) R 60 -16 14 597 19.00 Postcentral gyrus (BA2/ 3) R 58 -14 32 7.52 Precentral gyrus (BA 44/ 6) R 60 6 22 6.42 Premotor cortex (BA 6) L -28 -18 60 40 6.93 Precentral gyrus (BA 6/ 44) L -54 6 34 5 6.54 Precentral gyrus (BA 4/ 6) L -58 0 28 107 8.71 Parietal operculum cortex (SII) L -52 -16 20 45 7.57 Postcentral gyrus (BA 1/ 2/ 3) R 60 -12 32 135 10.78 108 IPL, anterior supramarginal gyrus L -62 -32 32 30 6.82 Middle temporal gyrus, temporocc. L -64 -50 -10 19 7.45 Inferior lateral occipital cortex R 48 -60 6 30 7.30 Inferior lateral occipital cortex L -54 -62 -6 33 7.38 Occipital pole R 22 -90 -2 52 6.69 Occipital pole L -6 -90 -2 204 8.60 Cerebellum crus 1 R 40 -76 -38 134 8.15 Cerebellum crus 1 L -26 -84 -28 48 10.10 Insula L -30 18 6 29 7.96 Insula/Frontal operculum cortex L -42 18 -4 7 6.19 Insula/Putamen L -32 10 0 8 6.23 Pallidum L -14 0 -6 8 7.65 The statistical maps were significant at voxel-wise uncorrected p<0.001 and FDR corrected p<0.01, minimum cluster size = 5 voxels. Peak t-values are shown in MNI coordinates. Table 2. Significant cluster-corrected brain regions in whole-brain univariate analysis Anatomic Region Hem. X Y Z Size (Voxels) T- value Pull & Push Actions > Rest Middle temporal gyrus, temporocci. R 51 -58 5 5 3.65 Occipital pole L -12 -97 2 1373 10.48 Lateral occipital cortex, superior R 18 -58 65 9 4.12 Superior parietal lobule R Precentral gyrus (BA 44/ 6) L -63 2 14 6 5.39 109 Parietal operculum cortex (SII) R 45 -28 23 126 5.36 Supramarginal gyrus R Precentral gyrus (BA 44/ 6) R 66 5 14 5 4.45 Parietal operculum cortex (SII) L -51 -25 23 84 5.50 Supramarginal gyrus L Precentral gyrus (BA 6/ 4) R 3 -16 56 788 7.16 SMA R Postcentral gyrus (BA 1/ 3) R 48 -13 56 429 7.07 Precentral gyrus (BA 4/ 6) Cerebellum V1 R 24 -49 -22 12 3.58 Putamen R 21 11 -1 5 3.56 Putamen L -18 8 2 27 3.84 Thalamus R 12 -19 5 5 4.06 Thalamus L -12 -19 2 3 3.95 The statistical maps were significant at voxel-wise uncorrected p<0.001 with FDR correction p<0.05 to account for multiple comparisons across the whole brain with a minimum cluster size of 5. 110 Figure 1. Significant Voxels in MVPA Searchlight Analysis that Distinguish ‘Pull-type’ vs. ‘Push-type’ Metaphors. Shows voxels that survived significance testing that could distinguish between ‘push-type’ and ‘pull-type’ metaphors as identified by a within-modality searchlight analysis (4-voxel radius) within areas of the left precentral and bilateral postcentral gyri at MNI coordinate (-52, -4, 60). The color bars indicate t-values of the group-random effects analysis. Figure 2. Anatomical Overlap in Univariate Analysis of Motor Localizer Task and Multivariate Analysis of Language task. Shows the significant activations of the univariate motor localize task (all actions vs. rest) overlaid on top of the significant voxels found in the multivariate searchlight analysis of the language task. The largest area of overlap being localized to the left postcentral gyrus at MNI coordinate (-52, -4, 60). 111 Discussion According to theories of embodiment, including Conceptual Metaphor Theory (CMT), understanding metaphor drawing on action, perception, and affect involves in part simulating what it would be like to engage in that specific activity and involves sensorimotor and affective systems in the brain that captured the initial experience (Gallese & Lakoff, 2005; Gibbs, 2006; Lakoff, 2014). Others, however, have argued that understanding metaphor rather involves a categorization process without, necessarily, a need for sensorimotor simulation of the source domain (Bowdle & Gentner, 2005; Glucksberg, 1990; Keysar & Bly, 1999; Keysar, Shen, Glucksberg, & Horton, 2000). The results presented here focus on investigating the role that sensorimotor and affective processes play in the processing of metaphors drawing on either action (i.e., pushing and pulling type hand-actions) or affect (i.e., gustatory/olfactory-based disgust). The results, briefly summarized below, contribute neural evidence in favor of the idea that metaphor comprehension involves activation of sensorimotor and affective systems of the brain. Moreover, they provide information about modulatory influences that impact activation of sensorimotor and affective systems of the brain in metaphor comprehension and provide anatomical detail about the level of sensorimotor and affective specificity that might be involved (e.g., low-level action kinematics vs. higher-level action goals), helping to elucidate the reach, functional consequences, and limits of embodied processes in metaphor comprehension. Study 1 provides initial evidence that comprehending metaphors that draw on physical disgust (i.e., gustatory/olfactory based disgust) and imply ‘moral disgust’ draw upon affective and sensorimotor systems of the brain relevant to physical disgust processing involving a distributed set of processes that captured the initial affective experience (i.e., perceptual features, internal states, and behaviors) and not necessarily in any dedicated neural circuitry only for the emotion of disgust (e.g., anterior insula for disgust). When comparing disgust metaphors versus non-affective metaphors we observed greater activation in primary and secondary gustatory cortices in the left hemisphere (i.e., anterior insula/frontal operculum and OFC), as well as, within the basal ganglia (i.e., pallidum), the latter previously implicated in the regulation of behavior/motivational processes and the former in interoceptive and evaluative processes. Fear language stimuli compared to disgust language stimuli recruited to a greater degree the posterior insula (also dorsal anterior insula for fear metaphor), as well as, parietal/sensorimotor areas (i.e., supramarginal gyrus/precentral gyrus). In contrast, disgust language stimuli compared to fear language stimuli showed increased activity in subcortical brain areas (i.e., amygdala and basal ganglia), but not within the anterior insula or other areas of primary/secondary gustatory cortices. Similarly, a cross-modal multivoxel-pattern analysis showed that when we trained a classifier to distinguish disgust versus non-affective stimuli (either using pictures or literal sentences) we could distinguish disgust metaphors from non-affective metaphors based on unique patterns of activity across voxels within the left anterior insula/adjacent frontal operculum, but not for the same analysis when trying to distinguish between disgust and fear stimuli. Taken together, the findings suggest that disgust metaphors that imply ‘moral disgust’ (or fear-related metaphors) may draw upon perceptual mechanisms involved in initial experience with physical disgust (and fear) inducers (i.e., sensory, internal, and behavioral). Future studies should consider the 112 distributed nature of the representation of affective experience to further investigate the components of affective experience most functionally relevant to metaphor comprehension. Study 2 builds on study 1 and addresses the following limitations. Specifically, in study 1 sensorimotor and affective activations during disgust metaphor comprehension could mainly reflect 1) processes associated with arousal and valence of disgust stimuli 2) the use of novel metaphor which may have been more likely to induce disgust-inducing imagery 3) spreading activation from the physical disgust words used, unrelated to the metaphoric meaning. Study 2 revealed that processing of familiar moral disgust metaphors when compared to literal paraphrases matched in arousal and valence (across both the reading and response periods), engaged areas of primary/secondary gustatory cortices (i.e., ventral anterior insula), but also the basal ganglia (i.e., pallidum), in line with the findings for novel disgust metaphor in study 1. Importantly, however, these activations were found to be context-dependent. Emotion-related brain regions implicated in disgust processing in moral disgust metaphor comprehension when compared to literal paraphrases (but also during the viewing of disgusting images) correlated with political orientation, with greater activity for conservatives. This finding suggests that the degree to which embodiment effects are seen for metaphor comprehension is sensitive to individual differences (i.e., political orientation) that have relevance to the metaphoric target domain. Next, across all participants increased activation of emotion-related brain regions implicated in disgust processing were seen during the moral judgment part of the task when compared to literal counterparts, suggesting that embodiment effects are also sensitive to the depth of processing required by the task (i.e., when a response is required). Moreover, literal paraphrases compared to moral disgust metaphors activated to a greater extent areas involved in top-down processing in moral decision-making (i.e., VMPFC, DLPFC), providing additional evidence of a differential impact of moral disgust metaphor on moral processing in the brain. Study 3 provides evidence that both low-level and higher-level motor representations may be needed to encode information relevant to physical force exchange between objects and agents in more abstract event descriptions in metaphor. A 4-voxel radius whole-brain searchlight MVPA analysis revealed that patterns of activity across the precentral and postcentral cortices (i.e., primary/premotor and secondary somatosensory cortex), as well as, within areas of the inferior parietal lobule (i.e., supramarginal gyrus) could significantly distinguish between the two types of metaphors. Significant voxels within sensorimotor systems of the brain in our MVPA analysis partially overlapped anatomically with brain activations in a univariate analysis of the motor task involving both pushing and pulling actions on an object versus rest, suggesting that they reflect processes associated with action execution. Furthermore, as the highest accuracy values and greatest degree of overlap with the motor task occurred within areas of secondary somatosensory cortex the results points to a special role for somatosensory simulation in the encoding of force dynamic semantics in hand-action metaphors. A number of other frontal and visual brain regions also contained patterns of voxels that could significantly distinguish between the two types of metaphors in our multivariate analysis. Taken together, a distributed pattern of occipital to parietal and frontal regions previously implicated in coding low-level action kinematics and higher-level goal-directed hand-actions involving objects appears to be involved in the representation of force dynamic semantics in metaphor in support of embodied cognition 113 theories and CMT, but does not exclude a role for other supramodal brain regions. Future studies will need to use cross-modal multivariate classification analysis for a more sensitive test of overlap between action semantics during action execution and during the comprehension of hand-action metaphors that differ in their force exchange patterns. At the broadest level, these studies provide evidence that the processing of metaphor about action and affect engages sensorimotor and affective systems of the brain. While previous neuroscientific studies on embodied semantics have shown evidence of sensorimotor and affective activations for literal language, studies on figurative language processing have, in contrast, been fewer and have shown mixed findings (see Introduction). Importantly, these studies provide an initial characterization of 1) the context-dependent nature of sensorimotor and affective engagement in metaphor comprehension and 2) the level of sensorimotor specificity involved (e.g., low-level action kinematics and/or higher-level action goals). Taken together these studies suggest that sensorimotor and affective processes may in part make functional contributions to metaphor comprehension through embodied simulations that are context- dependent and draw upon both low-level and higher-level sensorimotor representations, but do not exclude a role for other (i.e., heteromodal/supramodal) brain regions. Novel disgust metaphors compared to non-affective metaphors engaged emotion-related processes relevant to, but not specific to the emotion of disgust; Fear language stimuli could mainly be distinguished from disgust language stimuli in that they differentially recruited sensorimotor brain areas and subcortical areas (i.e., amygdala, basal ganglia), but not the anterior/frontal operculum. Moreover, processing of familiar moral disgust metaphors recruited emotion-related brain regions relevant to physical disgust processing to a greater degree than literal paraphrases, but in a context-dependent manner, showing sensitivity to individual differences (i.e., political orientation) and task effects. Lastly, metaphors relating to hand-actions that differed in their encoded force dynamics could be distinguished across distinct levels of motor hierarchy (both low-level and higher-level action representations) in the brain showing a level of sensorimotor specificity not previously investigated for metaphor processing, but also pointed to the involvement of other (supramodal) brain regions. In this discussion I mainly address neuroscientific theories that have made predictions about the nature of the relationship between sensorimotor and affective processes and linguistic representations in the brain, and then discuss their relevance to our findings of contextual- dependency and specificity of sensorimotor and affective representations in metaphor comprehension. I will primarily identify and discuss lines of future research. Following this, given that each paper presented here provides a standalone discussion of potential issues and ways to address them, I will end by very briefly discussing our findings in the context of current theories of metaphor comprehension. Theories of embodied cognition share in common the view that sensorimotor mechanisms involved in actual physical experience underlie semantic representations in the brain and have challenged the classical cognitive characterization of semantic processing as a series of computations on abstract or ‘amodal’ symbols divorced from basic sensorimotor and affective systems in the brain (e.g., feature lists, word-word co-occurrences, etc.) (Fodor, 1975; Z.W. 114 Pylyshyn, 1984, 2007). This view has been heavily criticized on philosophical grounds (i.e., the symbol grounding problem) (Harnad, 1990). Essentially, a circularity issue occurs if symbols that are abstract, amodal, and propositional are always defined in terms of other symbols without any direct links to the physical world (i.e., Chinese room experiment) (Searle, 1990). With growing behavioral and neuroscientific evidence suggesting some role for sensorimotor and affective systems in language processing (see Introduction), different theories from what have been termed ‘disembodied’, ‘weakly embodied’, and ‘strongly embodied’ views have considered the role of sensorimotor and affective systems in the brain during the processing of language relating to action, perception, and affect (Meteyard, Cuadrado, Bahrami, & Vigliocco, 2012). We focus specifically on how each view incorporates contextual dependency of sensorimotor and affective systems and, relatedly, what predictions they might make about the level of sensorimotor specificity involved (e.g., low-level action kinematics amd/or higher-level action goals). The ‘disembodied view’, otherwise known as grounding by interaction, suggests that semantic processing occurs in amodal areas (i.e., classic perisylvian language areas) that are separate from but can interact with sensorimotor systems (Mahon & Caramazza, 2008, 2009). Importantly, they have argued that sensorimotor representations are not necessary for semantic processing but mainly add ‘coloring’ to concepts: “Within the grounding by interaction framework, sensory and motor information colors conceptual processing, enriches it, and provides it with a relational context” (Mahon & Caramazza, 2008). This abstract or amodal semantic system is also to a degree in line, with ‘inborn views’ of categorical brain organization that stipulate the existence of innate brain mechanisms that exist primarily as a consequence of processes of natural selection (Caramazza & Hillis, 1990; Fodor, 1975; Patterson, 2000; Z. W. Pylyshyn, 1973), as opposed to mainly being shaped by experience and immediate interaction with the environment (for a review see Gainotti, 2015). According to proponents of this view, findings of sensorimotor and affective activations during language processing most likely reflect either post-comprehension imagery driven by top-down modulation from linguistic neural representations. Or, reflect spontaneous activity from words or literal language about action, perception, or emotion as a consequence of linkages between the language and sensorimotor/affective systems of the brain that do not contribute directly or in a major way to meaning processes (i.e., epiphenomenal activity) (Mahon & Caramazza, 2008). In support of this idea Papeo et al., 2015 replicated previous findings showing that TMS stimulation of the motor cortex during the processing of hand-action words compared to non-action words resulted in motor evoked potentials MEPs in the hand, but found that this effect was abolished if repetitive TMS (rTMS) was applied to the left posterior middle temporal gyrus (lpMTG), but not other areas or during sham stimulation (Papeo et al., 2015). They argued that rTMS of the lpMTG, which also resulted in disturbed semantic-verb processing, suggests that the lpMTG, which has been previously implicated in conceptual processing of verbs (Bedny, Caramazza, Grossman, Pascual-Leone, & Saxe, 2008; Kable, Kan, Wilson, Thompson-Schill, & Chatterjee, 2005), reflects activation of amodal conceptual representations that then drive activation of specific motor representations that occur post-comprehension. 115 Associationist learning theory, henceforth Strongly Embodied 1, have argued that activation of sensorimotor and affective systems during language processing contribute directly to meaning processes. Importantly, phonological and orthographic representations of words during development become associated with specific motor programs involved in action execution via Hebbian learning mechanisms and, thereby, are said to acquire ‘referential meaning’ (Pulvermuller, 2001; Pulvermuller, Hauk, Nikulin, & Ilmoniemi, 2005). That is, distributed cell assemblies or ‘functional units that reflect specific cortical distributions’ are acquired and shaped by experience by correlated neuronal activations. In this view the word form itself may automatically and immediately facilitate activation of motor programs associated with it even in cases when subjects are not paying close attention, such as during word recognition and, presumably, irrespective of the linguistic context (i.e., figurative vs. literal context). According to this view, abstract words would be similarly grounded. For example, abstract emotion words would acquire referential meaning mainly through associations with sensorimotor/affective areas of the brain that process, for example, emotional facial expressions and, subsequently, associated internal body states (Moseley, Carota, Hauk, Mohr, & Pulvermuller, 2012). In support of this view a number of studies support the idea of automatic, early activations of action-related verbs (see Introduction for early activations around ~200 ms), which argue against the idea that this activation reflects mainly post-comprehension imagery. While this view suggests that dominant features of a concept are coded as distributed neuronal cell assemblies that are relatively ‘stable and automatic’ (Pulvermuller et al., 2005; Yang, 2014), more recent accounts allow that top-down modulation mechanisms or higher-level cognitive influences involving multimodal or supramodal brain areas might allow for some contextual flexibility by highlighting non-dominant conceptual associations (Kiefer & Pulvermuller, 2012; Pulvermuller, 2013). Theories of embodied simulation semantics, Strongly Embodied 2, although in some ways compatible with the ‘associationist view’ rather find that action-related language draws on embodied simulations in line with Barsalou’s perceptual symbols theory (Aziz-Zadeh, Wilson, Rizzolatti, & Iacoboni, 2006; Buccino et al., 2005; Gallese & Lakoff, 2005; Gallese & Sinigaglia, 2011; Molnar-Szakacs, Kaplan, Greenfield, & Iacoboni, 2006; Tettamanti et al., 2005). In this view, the linguistically represented actions would be understood in part through motor simulation of the events described, such that brain regions involved in actually performing/observing/ or imaging that action could be re-used to interpret the events described (Barsalou, 1999, 2008; Barsalou, Kyle Simmons, Barbey, & Wilson, 2003; Gallese & Lakoff, 2005; Gallese & Sinigaglia, 2011; Glenberg & Gallese, 2012; Glenberg & Kaschak, 2002). Importantly, however, in this view a particular action word would cause motor activity ‘indirectly’ as it triggers a possible simulation of the events described linguistically through ‘paired controller (inverse model) and predictor (feedforward models) models’ inspired from motor control theory that are sensitive to contextual factors (Glenberg & Gallese, 2012). Such simulations provide a mechanism whereby linguistically represented events can be interpreted in lieu of contextually-appropriate higher-level goals similar to how the motor system is organized to also allow for contextually-appropriate actions in the service of a higher-level goal (Rizzolatti, Cattaneo, Fabbri-Destro, & Rozzi, 2014). In support of this view, neuroscientific studies on linguistic negation have found evidence that the presence of the syntactic negation marker, the 116 morpheme (‘not’), in hand and mouth action sentences show reduced activity in motor systems of the brain compared to the same sentences in the affirmative context suggesting that negation entails reduced access to the negated mental representation and/or inhibition of motor simulation/planning as alluded by simulation-based models of negation (Tettamanti et al., 2008; Tomasino, Weiss, & Fink, 2010). This argues rather for an indirect relationship between action words and motor systems of the brain through an intermediate simulation step flexibly mediated by the linguistic input in meaning processes (Tomasino et al., 2010). A computational account of how linguistically represented actions can be ‘indirectly’ flexibly linked to motor simulations or ‘X-schemas’ (i.e., dynamic schemas inspired by theories of motor control) has been proposed including to model metaphor comprehension (Narayanan, 1997). This computational account has further stipulated that in certain contexts detailed and specific motor programs might not be needed and higher-level schemas that reflect the overall event-structure and goal would be sufficient for relevant inferences to be drawn in language comprehension (i.e., for good enough comprehension) (Narayanan, 1997). Relatedly, Aziz-Zadeh & Damasio 2008 have also stipulated that the depth of involvement of sensorimotor systems in action-related language comprehension may be dictated by ‘higher-level goals’ that might be captured by a hierarchical system of conceptual representation involving progressive levels of convergence/divergence zones for multisensory integration (Aziz-Zadeh & Damasio, 2008a; Damasio, 1989; Meyer & Damasio, 2009). A related group of theories, generally termed ‘weakly embodied’, stipulate that modality- specific brain regions, multimodal brain regions or higher-level convergence/divergence zones, and statistical regularities in language use are all involved in different ‘mixtures’ depending on the context (linguistic or extralinguistic) (Barsalou et al., 2003; Binder & Desai, 2011; Kemmerer, 2015; Meteyard et al., 2012; Pulvermuller, 2013; Simmons & Barsalou, 2003). They also stress, similar to Aziz-Zadeh & Damasio 2008, the need to place the neuroscientific evidence in the context of complex neural network models of conceptual knowledge that support flexible, multi-level or hierarchical convergence and divergence zones for multisensory integration (Aziz-Zadeh & Damasio, 2008a; Damasio, 1989; Meyer & Damasio, 2009). For instance, Barsalou’s Language as Situated System (LASS) theory consists of a linguistic system that processes statistical regularities about word forms during language usage (i.e., word-word associations, phrases, syntactic structures, etc.) that provide very superficial conceptual processing that may be sufficient to accomplish certain strictly conceptual tasks (Simmons, Hamann, Harenski, Hu, & Barsalou, 2008). This linguistic system however interacts with a system for conceptual representations similar to Damasio’s, 1989 ‘convergence zone hypothesis’ that involves hierarchically organized convergence zones (CZs) that link various low-level sensory/motor features within a modality and across modalities that create regions of cross- modal convergence. Importantly, LASS differs from Damasio’s proposal in that concept retrieval not only results from ‘situated simulations’ within modality specific brain regions, but CZs at higher-levels may also be sufficient on their own for conceptual processing without the need for a mechanism of ‘time-locked retroactivation’ that involves re-activation of modality-specific activations (Barsalou et al., 2003; Fernandino et al., 2016; Simmons & Barsalou, 2003). In LASS convergence zones that link similar features may become localized to similar regions in space allowing for the emergence of properties and categories (i.e., similarity-in-topography principle). 117 This is in line with recent findings showing that the sensory-motor attribute ratings of words that differ in (color, shape, visual motion, sound, and manipulation) correlate with activity mainly within ‘higher-level’ (secondary sensory and multimodal) sensorimotor brain regions and not necessarily low-level primary cortical areas, suggesting that for certain forms of conceptual processing embodied multimodal abstraction may be sufficient (Fernandino et al., 2016). In summary, all of the different perspectives above focused on specifying the nature of the relationship between sensorimotor and affective systems and language processes in the brain. Current theories can be said to differ in terms of the extent, impact, and limits that sensorimotor and affective representations can have on semantic processing from ‘strongly embodied’ to ‘weakly embodiment’, and ‘disembodied’ views (Meteyard et al., 2012). Mainly, 1) The disembodied view suggests that sensorimotor/affective systems are not necessary but make post- comprehension contributions that can ‘color’ or enhance semantic processes by providing ‘relational context’ and/or make largely epiphenomenal contributions 2) The strongly embodied 1 position suggests that embodied processes are necessary and automatically activated (i.e., should be relatively stable) 3) The strongly embodied 2 view suggests that embodied processes are necessary but are ‘flexibly’ modulated by various contextual factors (i.e., linguistic representations ‘indirectly’ activate sensorimotor simulations) 4) Finally, the weakly embodied view suggests that embodied processes are not strictly necessary for superficial conceptual processing, but otherwise make context-dependent functional contributions involving embodied simulations (i.e., two representational systems one embodied the other symbolic). Furthermore, and this mainly refers to embodiment views, predictions have been made about the level of sensorimotor and affective specificity involved. It can be said that embodiment theories generally find the need for a hierarchical sensorimotor integration system for conceptual processing involving convergence/divergence zones as proposed by Damasio, 1989. They have further suggested that low-level, modality-specific representations may not always be needed, rather representations at higher-levels of sensorimotor convergence (multimodal/supramodal areas) could be sufficient for embodied simulation. The strongly embodied 2 view, suggests that the underlying ‘goal’ may control the level of sensorimotor/affective abstraction needed. Relatedly, the weakly embodied position suggests that this effect is mediated by depth of conceptual processing. In contrast, more recent discussions of the strongly embodied 1 position argue that both mainly modality-preferential and higher-level multimodal/supramodal areas should contribute to semantic processing in language comprehension, as higher-level multimodal/supramodal areas merely reflect patterns of associations that reinstate specific modality-specific representations, but may not necessarily contain perceptual representational content themselves (Moseley, Kiefer, & Pulvermuller, 2015). Can the theories presented above accommodate our current findings of context-dependent involvement of embodied processes (i.e., individual differences, task effects), as well as, evidence of both the involvement of low-level and higher-level sensorimotor representations (also multimoda/supramodal regions) during metaphor processing? Briefly, it is important to note that ultimately the question of whether or not embodied processes are ‘necessary and/or sufficient’ for language processing can only be answered by studies that explore direct causal 118 links, such as that offered by lesion/disease studies or by artificially induced disruptions (TMS) (see the Introduction for a review). That being said, fMRI studies show evidence for involvement of sensorimotor and affective systems, but also information about modulatory factors that may impact their involvement and task-dependent timing of activations. Thus, they still provide relevant data on the relationship between embodied processes and language comprehension. While the results of study 1 and 3 provide evidence of ‘automatic’ sensorimotor and affective engagement during metaphor processing, study 2 rather showed that sensorimotor and affective systems do not always appear to be automatically activated, but rather may make context-dependent functional contributions. Specifically, study 2 showed that emotion-related brain regions relevant to physical disgust processing were activated for moral disgust metaphors compared to literal paraphrases, but in a context-dependent manner. First, emotion-related activations during metaphor comprehension were sensitive to individual differences, correlating with political orientation (i.e., increased activity for conservatives). Secondly, these activations were found across all participants during the judgment portion of the task suggesting that embodiment effects are also sensitive to task/depth of processing (i.e., when a response is required). Importantly, in study 2 moral disgust metaphors and literal paraphrases were matched for arousal and valence, suggesting that embodiment effects are not merely a byproduct of individual sensitivities to negative affect in conservatives, but rather individual differences reflected in a link between the experience of disgust and political orientation. This rather suggests that these activations reflect functional contributions to meaning processes, possibly through co-occurrences that link physical disgust and the domain of morality during development as proposed by CMT. Taken together the results of study 2 point to the possibility that embodied processes are not automatically activated or always necessary, but rather make context-dependent functional contributions. This is most closely aligned with the position taken by the weakly embodied view, which suggests that embodied simulations make systematic context-dependent contributions to language processing, but are not always necessary. The positions taken by both strongly embodied 1 and 2 views are in line with our findings of modulatory influences due to individual differences in study 2 during the processing of moral disgust metaphors. However, they would predict that activation of embodied processes during moral disgust metaphor comprehension compared to literal paraphrases should be found across all participants, as opposed to only in those who scored high on political conservatism. We discuss reasons for why we might not have seen activation of emotion-related brain regions relevant to physical disgust processing during the reading period across all participants. One possibility is that the moral political literal paraphrases used also recruited emotion-related brain regions involved in physical disgust processing. According to CMT, the abstract concept of (im)morality is grounded through co-occurrences during experience with our notion of well- being (ill-being). As the moral-political statements involved opinions about immoral political stances, both metaphor and literal paraphrases might have evoked the same underlying conceptual metaphor, immorality is impurity and hence disgust (Lakoff, 2014; Lakoff & Johnson, 1980). This would have limited our ability to detect differences between moral disgust metaphor and literal paraphrases, except in conservatives where this effect might be the strongest for moral disgust metaphor. Future studies should compare familiar moral disgust metaphors 119 against literal sentences matched in arousal and valence, but not matched for meaning. The findings from study 1, which showed that novel disgust metaphors that imply moral disgust engaged emotion-related brain regions relevant to physical disgust processing when compared to non-affective metaphors, suggest that we may find similar results for familiar disgust metaphors when compared to literal sentences matched for arousal/valence but not for semantic similarity. The idea that abstract concepts such as ‘immorality’ rely on grounding in affective experience, more generally, finds support in a few recent studies. First, Kousta et al., 2011; Vigliocco et al., 2014 found that when concrete and abstract words were corrected for a number of psycholinguistic factors, including imageability (i.e., a factor generally associated with concreteness), abstract concepts activated the rostral ACC – ‘a region previously implicated in emotion processing’ – to a greater degree than concrete concepts (Kousta, Vigliocco, Vinson, Andrews, & Del Campo, 2011; Vigliocco et al., 2014). Critically, when matching for imageability, the abstract words were rated as higher in valence and arousal when compared to concrete words, with the rACC modulated by the hedonic valence of abstract words (degree of negative or positive valence), suggesting that abstract words may have greater affective associations than concrete words when controlling for all the relevant psycholinguistic factors. On the basis of such evidence Kousta et al., (2011) have proposed that abstract concepts (e.g. agony, joy), more generally, may rely to a greater degree on grounding in affective experience when compared to concrete concepts that might be grounded primarily in our sensorimotor experiences. In line with this idea, Citron and Goldberg 2014 further showed that taste metaphors (“The break up was bitter for him”) compared to their literal counterparts (“The break up was bad for him”), controlled for arousal and valence, activated the amygdala and parahippocampla area. Although Citron & Goldberg, 2014 also saw activation of primary and secondary gustatory cortex including the frontal operculum, anterior insula, and OFC for both taste metaphors and the presentation of the taste words in isolation, they only found amygdala/parahippocampal activation for taste metaphors. The authors suggest that familiar taste metaphors may be ‘implicitly more emotionally engaging’ than their literal counterparts. This might reflect additional grounding in affective experience for abstract concepts beyond grounding in modality preferential brain areas (i.e., gustatory/olfactory cortices) (Citron & Goldberg, 2014). A second and mutually compatible hypothesis concerns the possibility that familiar moral disgust metaphors in study 2 are not so much grounded in affective experience, but mainly sensorimotor representations associated with relevant emotional behavior (i.e., facial expressions, gesture, etc). It is possible that we may not have been able to detect engagement of embodied processes across all participants during reading of moral disgust metaphor, because we looked primarily at grounding in affective experience and not also sensorimotor grounding. Importantly, it has been proposed that abstract emotion words denoting internal states may not be easy to learn through classical referential means, as it is difficult to point a particular ‘thing’ and label it as ‘fear’ (Moseley et al., 2012). Rather it has been suggested that the meaning of abstract emotion words are learned over the course of experience through associations with various ‘behavioral contexts’ and ‘situations’ (Barsalou, 2003; Bennett & Hacker, 2006; Moseley et al., 2012; Wittgenstein, 1953). This view is supported by research in psychology that has characterized the many ways in which various internal states can be indexed by analyzing their 120 accompanying behavior (i.e., facial expressions, gestures, and approach/avoidance behaviors (Ekman, 1993, 2009; Ekman, Friesen, Osullivan, & Scherer, 1980), but also see (Barrett, 2006; Barrett, Lindquist, & Gendron, 2007; Lindquist, Barrett, Bliss-Moreau, & Russell, 2006) for how language shapes emotion. In this way, emotion words may acquire referential meaning through grounding in the relevant emotional behaviors and, thereby, eventually their accompanying internal states. In support of this idea, a number of neuroscientific studies have tried to show that the processing of abstract emotion words depends on motor systems of the brain in addition to limbic ones (Dreyer et al., 2015; Moseley et al., 2012; R. L. Moseley et al., 2015). For example, Moseley et al., 2012 showed that abstract emotion words (i.e., dread, spite) that were rated as low in arousal and valence compared to animal words, activated sensorimotor brain areas (i.e., left motor/premotor cortex, supplementary motor cortex, and supramarginal gyrus) in a similar ‘somatotopic manner’ as did action words referring to either the face or arms, suggesting grounding in emotional expression and gesture. Adding support to this idea, Dreyer et al., 2015 showed in a single patient study that a lesion to the left supplementary motor cortex resulted in impairments in abstract emotional noun processing compared to the recognition of nouns in different categories (i.e., food, animals, and tools) (see Introduction). Future studies should look at how moral disgust metaphor might be grounded in a more distributed set of affective but aslo sensorimotor processes. Lastly, the embodiment theories reviewed above suggest the need for a hierarchical sensorimotor integration system for conceptual processing involving convergence/divergence zones as proposed by Damasio et al., 1989. Relatedly, they find that in some cases full-blown simulations involving low-level, modality-preferential representations may not always be needed, rather representations at higher-levels of sensorimotor convergence could be sufficient for comprehension. The Strongly Embodied 2 view, suggests that the underlying ‘goal’ may control the level of sensorimotor/affective abstraction needed. This ‘goal’ might presumably be dictated by the linguistic or extralinguistic context, but also the task. Relatedly, the Weakly Embodied position, suggests that this effect is mediated by depth of conceptual processing, which presumably would be similarly sensitive to various contextual factors including linguistic or extralinguistic context and task. Specifying exactly what sensorimotor programs or simulations might suffice or be needed for comprehension has been a challenge for embodiment theories, more generally, as a number of potentially relevant simulations at varying levels of sensorimotor/affective specificity could be recruited (Narayanan, 1997; Pezzulo et al., 2012). An alternative, alluded to by recent descriptions of the Strongly Embodied 1 view (Moseley et al., 2015), suggests that heteromodal areas or regions of higher-level sensory/motor convergence contribute mainly to concept access or retrieval, but they are not sufficient in an of themselves for storing meaningful representations. Therefore, they would predict that as originally described by Damasio, 1989 that a system of time-locked retroactivation would require that representations in lower-levels of convergence-divergence including modality-preferential areas would need to be re-activated for semantic processing. While study 1 and 2 do not reveal much about the level of sensorimotor and affective specificity involved, Study 3 revealed that two action-related metaphors that differed in their encoded force dynamics could be distinguished at both low-level and higher-level motor 121 representations. This is in line, more generally, with the idea that higher-level motor representations may not be sufficient to encode the details of force dynamics and rather suggest that a full-blown simulation might be needed for proper (i.e., good enough) inferences to be drawn from source to target domain to comprehend hand-action metaphors. This is in contrast to a recent behavioral finding by Troyer et al., 2014 that found evidence that sensorimotor representations during action-related language processing appear to be somehow ‘less specific’ or more generalized in nature. Nevertheless, it is possible that this may have been influenced by the task used in study 3, which may have led to deeper conceptual processing (see Discussion for study 3). It is possible that a hierarchical system for sensory integration involving convergence/divergence zones may be one pre-requisite for functional context-dependent effects in language comprehension. Thus, it is possible that some of the modulatory effects that have been observed thus far reflect this neuroarchitectural framework. Future studies should, therefore, investigate whether or not different levels of sensorimotor and affective specificity are modulated by the underlying ‘goal’, whether it’s related to the communicative goal, action goal, or task goal will be an important consideration. Lastly, although our results do not rule out the possibility that sensorimotor and affective activations during metaphor comprehension mainly reflect post-comprehension imagery and/or epiphenomenal activity in line with the Disembodied view, two lines of evidence suggest otherwise. While activation of emotion-related brain regions implicated in physical disgust processing during moral disgust metaphor in study 2 during the judgment period could reflect post-comprehension imagery, the fact that we found that these brain regions correlated with political orientation during the comprehension period rather suggests that these activations may reflect functional contributions. Secondly, the fact that familiar hand-action metaphors that differed in subtle ways (i.e., force exchange patterns) could be distinguished at the level of sensorimotor systems in the brain, suggests a level of sensorimotor specificity that greatly undermines an epiphenomenal activity or post-comprehension imagery interpretation. Nevertheless, the findings in study 1 could reflect post-comprehension imagery, especially given that these were novel metaphors and, hence, may have naturally induced imagery. However, this would be compatible with embodied simulation views (especially embodied 2 view), which would suggest that the degree of imageability may actually underlie aspects related to functional contributions to meaning processes. Postscript: Overall, the studies presented in this thesis support the idea that sensorimotor and affective systems of the brain are engaged during metaphor comprehension and provide an initial characterization of 1) modulatory factors that impact the engagement of sensorimotor and affective systems in metaphor comprehension and 2) examined the level of sensorimotor specificity involved in specific inferences underlying conceptual mappings predicted by CMT. The results of these studies, while in line with the idea that metaphor processing entails activation of sensorimotor and affective systems relevant to the metaphoric source domain in accordance with CMT, do not necessarily rule out categorization processes in metaphor comprehension. Specifically, in study 2 we did not find evidence that emotion-related brain 122 regions relevant to disgust processing were engaged for moral disgust metaphors compared to literal paraphrases across all participants during the reading period, rather the effect was dependent on political orientation. Therefore, it is possible that categorization type processes were involved during familiar metaphor comprehension. That is, rather than an explicit comparison process whereby features between the source domain are mapped onto the target domain (i.e., or a process of finding relevant correspondences), it is possible that the source domain language was immediately categorized as belonging to an abstract superordinate category (e.g., rotten in ‘that was a rotten thing to do’ is immediately perceived as the lexicalized metaphoric category ‘bad things’). Recent research in metaphor comprehension highlights the importance of both processes, but has been divided as to what factors allow one process to dominate over the other. According to one proposal, ‘the neural career of metaphor’ reviewed in the Introduction, the engagement of embodied processes, and hence direct comparison processes, inversely depend on the degree of conventionality of the metaphor. In more conventionalized metaphors concrete language is processed as a lexically encoded category relevant to the topic that need not involve sensorimotor simulation. Indeed, Desai et al., 2011, 2013 found evidence of a graded pattern of sensorimotor engagement from novel to more familiar metaphors. Given this, future research should delineate how a hierarchical system of divergence/convergence zones for multisensory integration along with linguistic representations (e.g., statistical regularities such word-word co-occurrence) allow for this kind of functional context-dependent (or graded) contributions of embodied processes in metaphor comprehension. References Aziz-Zadeh, L., & Damasio, A. (2008a). Embodied semantics for actions: findings from functional brain imaging. J Physiol Paris, 102(1-3), 35-39. doi:10.1016/j.jphysparis.2008.03.012 Aziz-Zadeh, L., Wilson, S. M., Rizzolatti, G., & Iacoboni, M. (2006). Congruent embodied representations for visually presented actions and linguistic phrases describing actions. Curr Biol, 16(18), 1818-1823. doi:10.1016/j.cub.2006.07.060 Barrett, L. F. (2006). Are Emotions Natural Kinds? Perspect Psychol Sci, 1(1), 28-58. doi:10.1111/j.1745-6916.2006.00003.x Barrett, L. F., Lindquist, K. A., & Gendron, M. (2007). Language as context for the perception of emotion. Trends Cogn Sci, 11(8), 327-332. doi:10.1016/j.tics.2007.06.003 Barsalou, L. W. (1999). Perceptual symbol systems. Behav Brain Sci, 22(4), 577-609; discussion 610-560. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11301525 Barsalou, L. W. (2003). Abstraction in perceptual symbol systems. Philos Trans R Soc Lond B Biol Sci, 358(1435), 1177-1187. doi:10.1098/rstb.2003.1319 Barsalou, L. W. (2008). Grounded cognition. Annu Rev Psychol, 59, 617-645. doi:10.1146/annurev.psych.59.103006.093639 Barsalou, L. W., Kyle Simmons, W., Barbey, A. K., & Wilson, C. D. (2003). Grounding conceptual knowledge in modality-specific systems. Trends Cogn Sci, 7(2), 84-91. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12584027 123 Bedny, M., Caramazza, A., Grossman, E., Pascual-Leone, A., & Saxe, R. (2008). Concepts are more than percepts: the case of action verbs. J Neurosci, 28(44), 11347-11353. doi:10.1523/JNEUROSCI.3039-08.2008 Bennett, M. R., & Hacker, P. M. (2006). Language and cortical function: conceptual developments. Prog Neurobiol, 80(1), 20-52. doi:10.1016/j.pneurobio.2006.07.002 Binder, J. R., & Desai, R. H. (2011). The neurobiology of semantic memory. Trends Cogn Sci, 15(11), 527-536. doi:10.1016/j.tics.2011.10.001 Bowdle, B. F., & Gentner, D. (2005). The career of metaphor. Psychol Rev, 112(1), 193-216. doi:10.1037/0033-295X.112.1.193 Buccino, G., Riggio, L., Melli, G., Binkofski, F., Gallese, V., & Rizzolatti, G. (2005). Listening to action-related sentences modulates the activity of the motor system: a combined TMS and behavioral study. Brain Res Cogn Brain Res, 24(3), 355-363. doi:10.1016/j.cogbrainres.2005.02.020 Caramazza, A., & Hillis, A. E. (1990). Where do semantic errors come from? Cortex, 26(1), 95- 122. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/2354648 Citron, F. M., & Goldberg, A. E. (2014). Metaphorical sentences are more emotionally engaging than their literal counterparts. J Cogn Neurosci, 26(11), 2585-2595. doi:10.1162/jocn_a_00654 Damasio, A. R. (1989). Time-locked multiregional retroactivation: a systems-level proposal for the neural substrates of recall and recognition. Cognition, 33(1-2), 25-62. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/2691184 Dreyer, F. R., Frey, D., Arana, S., von Saldern, S., Picht, T., Vajkoczy, P., & Pulvermuller, F. (2015). Is the Motor System Necessary for Processing Action and Abstract Emotion Words? Evidence from Focal Brain Lesions. Front Psychol, 6, 1661. doi:10.3389/fpsyg.2015.01661 Ekman, P. (1993). Facial Expression and Emotion. American Psychologist, 48(4), 384-392. doi:Doi 10.1037//0003-066x.48.4.384 Ekman, P. (2009). Darwin's contributions to our understanding of emotional expressions. Philosophical Transactions of the Royal Society B-Biological Sciences, 364(1535), 3449- 3451. doi:10.1098/rstb.2009.0189 Ekman, P., Friesen, W. V., Osullivan, M., & Scherer, K. (1980). Relative Importance of Face, Body, and Speech in Judgments of Personality and Affect. J Pers Soc Psychol, 38(2), 270-277. doi:Doi 10.1037//0022-3514.38.2.270 Fernandino, L., Binder, J. R., Desai, R. H., Pendl, S. L., Humphries, C. J., Gross, W. L., . . . Seidenberg, M. S. (2016). Concept Representation Reflects Multimodal Abstraction: A Framework for Embodied Semantics. Cereb Cortex, 26(5), 2018-2034. doi:10.1093/cercor/bhv020 Fodor, J. A. (1975). The Language of Thought. New York, NY: Thomas Y. Crowell. Gallese, V., & Lakoff, G. (2005). The Brain's concepts: the role of the Sensory-motor system in conceptual knowledge. Cogn Neuropsychol, 22(3), 455-479. doi:10.1080/02643290442000310 Gallese, V., & Sinigaglia, C. (2011). What is so special about embodied simulation? Trends Cogn Sci, 15(11), 512-519. doi:10.1016/j.tics.2011.09.003 Gibbs, R. W. (2006). Metaphor Interpretation as Embodied Simulation. Mind & Language, 21(3), 434-458. doi:10.1111/j.1468-0017.2006.00285.x 124 Glenberg, A. M., & Gallese, V. (2012). Action-based language: a theory of language acquisition, comprehension, and production. Cortex, 48(7), 905-922. doi:10.1016/j.cortex.2011.04.010 Glenberg, A. M., & Kaschak, M. P. (2002). Grounding language in action. Psychon Bull Rev, 9(3), 558-565. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12412897 Glucksberg, S. (1990). Understanding metaphorical comparisons: Beyond similarity. Psychological Review, 21, 3-18. Harnad, S. (1990). The Symbol Grounding Problem. Physica D, 42(1-3), 335-346. doi:Doi 10.1016/0167-2789(90)90087-6 Kable, J. W., Kan, I. P., Wilson, A., Thompson-Schill, S. L., & Chatterjee, A. (2005). Conceptual representations of action in the lateral temporal cortex. J Cogn Neurosci, 17(12), 1855-1870. doi:10.1162/089892905775008625 Kemmerer, D. (2015). Are the motor features of verb meanings represented in the precentral motor cortices? Yes, but within the context of a flexible, multilevel architecture for conceptual knowledge. Psychon Bull Rev, 22(4), 1068-1075. doi:10.3758/s13423-014- 0784-1 Keysar, B., & Bly, B. M. (1999). Swimming against the current: Do idioms reflect conceptual structure? Journal of Pragmatics, 31(12), 1559-1578. doi:Doi 10.1016/S0378- 2166(99)00004-1 Keysar, B., Shen, Y. I., Glucksberg, S., & Horton, W. S. (2000). Conventional language: How metaphorical is it? Journal of Memory and Language, 43(4), 576-593. doi:10.1006/jmla.2000.2711 Kiefer, M., & Pulvermuller, F. (2012). Conceptual representations in mind and brain: theoretical developments, current evidence and future directions. Cortex, 48(7), 805-825. doi:10.1016/j.cortex.2011.04.006 Kousta, S. T., Vigliocco, G., Vinson, D. P., Andrews, M., & Del Campo, E. (2011). The representation of abstract words: why emotion matters. J Exp Psychol Gen, 140(1), 14- 34. doi:10.1037/a0021446 Lakoff, G. (2014). Mapping the brain's metaphor circuitry: metaphorical thought in everyday reason. Front Hum Neurosci, 8. doi:ARTN 958 10.3389/fnhum.2014.00958 Lakoff, G., & Johnson, M. (1980). The metaphorical structure of the human conceptual system. Cognitive Science, 4(2), 195-208. doi:10.1016/S0364-0213(80)80017-6 Lindquist, K. A., Barrett, L. F., Bliss-Moreau, E., & Russell, J. A. (2006). Language and the perception of emotion. Emotion, 6(1), 125-138. doi:10.1037/1528-3542.6.1.125 Mahon, B. Z., & Caramazza, A. (2008). A critical look at the embodied cognition hypothesis and a new proposal for grounding conceptual content. J Physiol Paris, 102(1-3), 59-70. doi:10.1016/j.jphysparis.2008.03.004 Mahon, B. Z., & Caramazza, A. (2009). Concepts and categories: a cognitive neuropsychological perspective. Annu Rev Psychol, 60, 27-51. doi:10.1146/annurev.psych.60.110707.163532 Meteyard, L., Cuadrado, S. R., Bahrami, B., & Vigliocco, G. (2012). Coming of age: a review of embodiment and the neuroscience of semantics. Cortex, 48(7), 788-804. doi:10.1016/j.cortex.2010.11.002 Meyer, K., & Damasio, A. (2009). Convergence and divergence in a neural architecture for recognition and memory. Trends Neurosci, 32(7), 376-382. doi:10.1016/j.tins.2009.04.002 125 Molnar-Szakacs, I., Kaplan, J., Greenfield, P. M., & Iacoboni, M. (2006). Observing complex action sequences: The role of the fronto-parietal mirror neuron system. Neuroimage, 33(3), 923-935. doi:10.1016/j.neuroimage.2006.07.035 Moseley, R., Carota, F., Hauk, O., Mohr, B., & Pulvermuller, F. (2012). A role for the motor system in binding abstract emotional meaning. Cereb Cortex, 22(7), 1634-1647. doi:10.1093/cercor/bhr238 Moseley, R., Kiefer, M., & Pulvermuller, F. (2015). Perceptual and Emotional Embodiment: Foundations of Embodied Cognition (Vol. 1): A psychology press book. Moseley, R. L., Shtyrov, Y., Mohr, B., Lombardo, M. V., Baron-Cohen, S., & Pulvermuller, F. (2015). Lost for emotion words: what motor and limbic brain activity reveals about autism and semantic theory. Neuroimage, 104, 413-422. doi:10.1016/j.neuroimage.2014.09.046 Narayanan, S. (1997). Knowledge-based Action Representations for Metaphor and Aspect (KARMA). Doctoral Dissertation, University of California, Berkeley. Papeo, L., Lingnau, A., Agosta, S., Pascual-Leone, A., Battelli, L., & Caramazza, A. (2015). The origin of word-related motor activity. Cereb Cortex, 25(6), 1668-1675. doi:10.1093/cercor/bht423 Patterson, K., Hodges, J.R. (2000). Semantic dementia: one window on the structure and organisation of semantic memory (2nd Edn. ed. Vol. 2). North-Holland: Elsevier. Pezzulo, G., Barsalou, L. W., Cangelosi, A., Fischer, M. H., McRae, K., & Spivey, M. J. (2012). Computational Grounded Cognition: a new alliance between grounded cognition and computational modeling. Front Psychol, 3, 612. doi:10.3389/fpsyg.2012.00612 Pulvermuller, F. (2001). Brain reflections of words and their meaning. Trends Cogn Sci, 5(12), 517-524. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11728909 Pulvermuller, F. (2013). How neurons make meaning: brain mechanisms for embodied and abstract-symbolic semantics. Trends Cogn Sci, 17(9), 458-470. doi:10.1016/j.tics.2013.06.004 Pulvermuller, F., Hauk, O., Nikulin, V. V., & Ilmoniemi, R. J. (2005). Functional links between motor and language systems. Eur J Neurosci, 21(3), 793-797. doi:10.1111/j.1460- 9568.2005.03900.x Pylyshyn, Z. W. (1973). The role of competence theories in cognitive psychology. J Psycholinguist Res, 2(1), 21-50. doi:10.1007/BF01067110 Pylyshyn, Z. W. (1984). Computation and cognition: Towards a foundation for cognitive science. . Boston, MA: MIT Press. Pylyshyn, Z. W. (2007). Things and Places: How the Mind Connects with the World. Cambridge, MA: MIT Press. Rizzolatti, G., Cattaneo, L., Fabbri-Destro, M., & Rozzi, S. (2014). Cortical mechanisms underlying the organization of goal-directed actions and mirror neuron-based action understanding. Physiol Rev, 94(2), 655-706. doi:10.1152/physrev.00009.2013 Searle, J. R. (1990). Is the Brains Mind a Computer-Program. Scientific American, 262(1), 26- 31. Retrieved from <Go to ISI>://WOS:A1990CE06200002 Simmons, W. K., & Barsalou, L. W. (2003). The similarity-in-topography principle: reconciling theories of conceptual deficits. Cogn Neuropsychol, 20(3), 451-486. doi:10.1080/02643290342000032 126 Simmons, W. K., Hamann, S. B., Harenski, C. L., Hu, X. P., & Barsalou, L. W. (2008). fMRI evidence for word association and situated simulation in conceptual processing. J Physiol Paris, 102(1-3), 106-119. doi:10.1016/j.jphysparis.2008.03.014 Tettamanti, M., Buccino, G., Saccuman, M. C., Gallese, V., Danna, M., Scifo, P., . . . Perani, D. (2005). Listening to action-related sentences activates fronto-parietal motor circuits. J Cogn Neurosci, 17(2), 273-281. doi:10.1162/0898929053124965 Tettamanti, M., Manenti, R., Della Rosa, P. A., Falini, A., Perani, D., Cappa, S. F., & Moro, A. (2008). Negation in the brain: modulating action representations. Neuroimage, 43(2), 358-367. doi:10.1016/j.neuroimage.2008.08.004 Tomasino, B., Weiss, P. H., & Fink, G. R. (2010). To move or not to move: imperatives modulate action-related verb processing in the motor system. Neuroscience, 169(1), 246- 258. doi:10.1016/j.neuroscience.2010.04.039 Vigliocco, G., Kousta, S. T., Della Rosa, P. A., Vinson, D. P., Tettamanti, M., Devlin, J. T., & Cappa, S. F. (2014). The neural representation of abstract words: the role of emotion. Cereb Cortex, 24(7), 1767-1777. doi:10.1093/cercor/bht025 Wittgenstein, L. (1953). Philosophical investigations. Oxford: Basil Blackwell. Yang, J. (2014). Influences of motor contexts on the semantic processing of action-related language. Cognitive, Affective, & Behavioral Neuroscience, 14(3), 912-922. doi:10.3758/s13415-014-0258-y

Abstract (if available)

Abstract A growing number of behavioral and neuroscientific studies have provided evidence in support of theories that stipulate that conceptual representations (both concrete and abstract concepts) are grounded in perceptual systems of the brain (Barsalou, 1999, 2008

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Core, social and moral disgust processing in youth with autism spectrum disorder (ASD)

PDF

Contribution of fundamental brain networks to individual differences in social cognition

PDF

Schema architecture for language-vision interactions: a computational cognitive neuroscience model of language use

PDF

Facilitating learning and conceptual change through an intervention on text graphic processing

PDF

The development of object recognition in the newborn brain

PDF

Engagement of the action observation network through functional magnetic resonance imaging with implications for stroke rehabilitation

PDF

Intrinsic functional connectivity of the default mode network predicts the purposefulness of youths’ intended adult lives

PDF

Behavioral and neural correlates of core-self and autobiographical-self processes

PDF

Homeostatic imbalance and monetary delay discounting: effects of feeding on RT, choice, and brain response

PDF

Individual differences in heart rate response and expressive behavior during social emotions: effects of resting cardiac vagal tone and culture, and relation to the default mode network

PDF

A neuropsychological exploration of low-SES adolescents’ life goals and their motives

PDF

Experience modulates neural activity during action understanding: exploring sensorimotor and social cognitive interactions

PDF

Biological and behavioral correlates of emotional flexibility and associations with exposure to family aggression

PDF

The brain's virtuous cycle: an investigation of gratitude and good human conduct

PDF

Selflessness takes time: altruistic (but not cooperative) prosocial behavior increases with decision time

PDF

Neuropsychological mechanisms predictive of adolescent addictive behaviors

PDF

Endocrinology of social behavior and reward

PDF

Effect of mindfulness training on attention, emotion-regulation and risk-taking in adolescence

PDF

Development of error monitoring in preschool to 12th-grade students and relations in late childhood and adolescence to social-affective processing and emotions about math

PDF

Social and motor skills in autism spectrum disorder & developmental coordination disorder: functional & structural neurobiology

Asset Metadata

Creator Gamez-Djokic, Vesna Eliza (author)

Core Title Understanding perceptual processes in metaphor comprehension: specificity and context-dependence

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Neuroscience

Publication Date 10/11/2016

Defense Date 07/22/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag cognitive neuroscience,conceptual metaphor theory,metaphor,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Immordino-Yang, Mary Helen (committee chair), Aziz-Zadeh, Lisa (committee member), Bechara, Antoine (committee member)

Creator Email gamezdjo@usc.edu,vgdjokic@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-313470

Unique identifier UC11214643

Identifier etd-GamezDjoki-4873.pdf (filename),usctheses-c40-313470 (legacy record id)

Legacy Identifier etd-GamezDjoki-4873.pdf

Dmrecord 313470

Document Type Dissertation

Format application/pdf (imt)

Rights Gamez-Djokic, Vesna Eliza

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA