Interhemispheric interaction in bilateral redundancy gain: Effects of physical similarity

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Request accessible transcript

Transcript (if available)

Content INTERHEMISPHERIC INTERACTION IN BILATERAL REDUNDANCY GAIN:
EFFECTS OF PHYSICAL SIMILARITY
Copyright 2001
by
Nancy Lee Marks
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PSYCHOLOGY)
May 2001
Nancy Lee Marks
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3027746
_ _ ®
UMI
UMI Microform 3027746
Copyright 2001 by Bell & Howell Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
Bell & Howell Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFO RNIA 9 00 07
This dissertation, w ritten by
N a n c y L e e M a r k s
under the direction o f Dissertation
Committee, and approved by all its members,
has b e e n presented to and a c c e p te d by The
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
Dean of Graduate Studies
DISSERTATION COMMITTEE
Chairperson
\
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ii
Acknowledgements
Portions of the research reported in this dissertation were supported
by a grant from the National Science Foundation to Joseph B. Hellige
(SBR-9507924).
I would like to thank members of the Hellige laboratory, especially
Barbara Cherry, Luis Lesmes, and Gretchen Scott, for their contributions to
discussions of this research over the years. I am grateful to Ryan Barnes,
Yezzennya Castro, and Li Zhang for their assistance in testing participants,
generating stimulus materials, and entering data.
I would like to acknowledge the members of my dissertation
committee, David Lavond, William McClure, Richard Thompson, and David
Walsh, whose support, and suggestions were most helpful. My committee
chair and advisor, Joseph B. Hellige, has earned my most profound
gratitude for his patience, enthusiasm, and insightful guidance of every
step of this research and its interpretation. It has been a rare privilege to
claim him as my mentor for the past six years.
Over the years many people have touched my life and helped me to
believe in myself. Among them I want to thank my family and my many
friends and mentors, including especially Loretta Dahmus, Ray Eastman,
Carol Hurzeler, Fred Kitterle, Mark Ludorf, and Lauren Scharff.
Finally, I want to express my appreciation of my dear husband,
Kenneth Marks, without whose support and encouragement this
dissertation would never have been written.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table of Contents
Section Page Number
List of Figures......................................................................................................... iv
List of Tables................................................................ vi
Abstract............................... viii
Introduction...............................................................................................................1
Experiment 1 ......................................................................................................... 13
Design and Method.......................................................................................16
Results and Discussion............................................................................. 21
Experiment 2 ..................................................................... 27
Design and Method.................................................. 27
Results and Discussion............................................................................. 29
Experiment 3 ......................................................................................................... 36
Design and Method.......................................................................................... 36
Results and Discussion..................................................................................38
Experiment 4 .........................................................................................................48
Design and Method...................................................................................... 49
Results and Discussion ........................................................................52
Experiment 5 ........................................ 57
Design and Method.......................................................................................58
Results and Discussion..............................................................................60
Experiment 6 ......................................................................................................... 63
Design and Method....................................................... 64
Results and Discussion..............................................................................67
General Discussion............................................................................................. 77
References............................................................................................................ 88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
iv
List of Figures
Figure Page Number
Figure 1. Samples of the stimuli and of the 8 trial types for Experiment 1 ..18
Figure 2. Upper panel: Percentage of errors for each of the four visual field
conditions in Experiment 1. Lower panel: Qualitative error (QE) scores
for each of the four visual field conditions in Experiment 1................... 22
Figure 3. Upper panel: Percentage of errors for three visual field conditions
in Experiment 1 by case of stimuli. Lower panel: Qualitative error (QE)
scores for three visual field conditions in Experiment 1 by case of
stimuli.................... 26
Figure 4. Samples of the stimuli and of the 8 trial types for Experiment 2..30
Figure 5. Upper panel: Percentage of errors for each of the four visual field
conditions in Experiment 2. Lower panel: Qualitative error (QE) scores
for each of the four visual field conditions in Experiment 2 ................... 32
Figure 6. Upper panel: Percentage of errors for three visual field conditions
in Experiment 2 by case/font of stimuli. Lower panel: Qualitative error
(QE) scores for three visual field conditions in Experiment 2 by
case/font of stimuli........................ 35
Figure 7. Upper panel: Percentage of errors for each of the four visual field
conditions in Experiment 3. Lower panel: Qualitative error (QE) scores
for each of the four visual field conditions in Experiment 3 ................... 40
Figure 8. Upper panel: Percentage of errors for three visual field conditions
in Experiment 3 by case/font of stimuli. Lower panel: Qualitative error
(QE) scores for three visual field conditions in Experiment 3 by
case/font of stimuli........................................................................................44
Figure 9. Percentage of first-letter errors (FE), last-letter errors (LE) and
other errors (OE) by case of stimuli, collapsed across visual fields in
Experiment 2 (upper panel) and Experiment 3 (lower panel)............... 46
Figure 10. Samples of stimuli and trial types for Experiment 4 ......... 51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V
Figure 11. Upper panel: Percentage of errors for each of the three visual
field conditions in Experiment 4. Lower panel: Qualitative error (QE)
scores for each of the three visual field conditions in Experiment 4....54
Figure 12. Samples of stimuli and trial types for Experiment 5.................... 59
Figure 13. Upper panel: Percentage of errors for each of the three visual
field conditions in Experiment 5. Lower panel: Qualitative error (QE)
scores for each of the three visual field conditions in Experiment 5....61
Figure 14. Samples of stimuli and trial types for Experiment 6........... ..66
Figure 15. Upper panel: Percentage of errors for each of the four visual field
conditions in Experiment 6. Lower panel: Qualitative error (QE) scores
for each of the four visual field conditions in Experiment 6 ................... 69
Figure 16. Percentage of errors for each of four bilateral trial types in
Experiment 6 ..................................... 70
Figure 17. Percentage of first-letter (FE), last-letter (LE) and other errors
(OE) for Bilateral Consistent and Bilateral Inconsistent trials.... 73
Figure 18. Upper panel: Percentage of errors for three visual field
conditions in Experiment 6 by stimulus format. Lower panel: Qualitative
error (QE) scores for three visual field conditions in Experiment 6 by
stimulus format................................ 75
Figure 19. QE scores for digit trigrams and dot trigrams, alone (dashed
lines: Experiments 4 and 5) and in combination (solid lines:
Experiment 6).................................. 76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
vi
List of Tables
Table Page Number
Table 1. Percentage of CVC Identification Errors for LVF, RVF and
Redundant Bilateral (BVF) Trials in Each of 13 Published Experiments..6
Table 2. Qualitative Error (QE) Scores for LVF, RVF and Redundant Bilateral
(BVF) Trials in each of 13 Published Experiments......................................7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
“Nothing will ever be attempted if all possible objections must be first
overcome."
Dr. Johnson, 1709-1784
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
viii
Abstract
To study interhemispheric interaction, recent divided visual field
experiments have included conditions in which the same stimulus is
presented in both visual half-fields and thus to both hemispheres
simultaneously. Bilateral redundancy gain is the enhancement of
performance which occurs when both hemispheres have access to the
visual information needed to produce a response as compared with
performance on the better of the two unilateral conditions. To test whether
physical identity of the redundant stimuli is essential to produce this gain,
the appearance of the two stimuli was varied on bilateral trials, either
through manipulating case and font of consonant-vowel-consonant (CVC)
non-word trigrams (Experiments 1 to 3) or through varying the format in
which number trigrams were presented (as digits or structured dot
patterns) (Experiment 6). The two copies of the trigram presented on
bilateral trials could be either physically identical (Bilateral Consistent
trials) or physically different (Bilateral Inconsistent trials). Substantial
bilateral redundancy gain (ranging from 12.3% to 14.2%) was realized in all
bilateral conditions. For Experiments 1 and 2, the accuracy results for
Bilateral Inconsistent trials were virtually indistinguishable from those for
Bilateral Consistent trials. In Experiment 3, there was even a small but
reliable advantage in favor of Bilateral Inconsistent trials. Experiment 6
represented an even stronger test of the hypothesis, because the two
copies of the target were not only different in appearance, but different in
format. Again, both Bilateral Consistent and Bilateral Inconsistent trials
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ix
produced substantial bilateral redundancy gain, but in this experiment,
there was a reliable difference in favor of Bilateral Consistent trials,
suggesting that physical identity of the stimuli may have made some
contribution to the enhanced performance on Bilateral Consistent trials.
Possible sources of this consistency effect are discussed, including neural
coactivation, which may be enhanced when the two stimuli are processed
in the same cortical area. Since bilateral redundancy gain is realized even
when the two stimuli presented on bilateral trials are not physically
identical, it must result at least partially at stages of information processing
beyond low-level visual sensory processes. To clarify what these
processes are will require further research.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1
Interhemispheric Interaction in Bilateral Redundancy Gain:
Effects of Physical Similarity
"Two hemispheres are better than one": This phrase forms part of the
title of two recent reports on investigations of interhemispheric interaction
(Lepore, Ptito, & Jasper, 1986; Reuter-Lorenz, Stanczak, & Miller, 1999).
Although each hemisphere has different abilities, propensities and biases,
normally the two hemispheres work together, and each contributes in one
way or another to almost everything we do (Banich, 1998a; Hellige, 1990;
Hellige, 1993; Liederman, 1998). Under normal conditions, our perceptions,
cognitions and actions are unified rather than fragmented. Such unity is
possible because the two hemispheres are highly interconnected through
both cortical and subcortical pathways. The corpus callosum appears to play
the dominant role in interhemispheric interactions, both because of its size
and because it provides direct connections between most areas of the
cerebral hemispheres. The two hemispheres share the results of their
processing in a dynamic interplay, sometimes led by one or the other. A
fuller understanding of the ways the two hemispheres work together to
coordinate and enhance performance is essential for our understanding of
the nature and dynamics of human information processing.
Much attention has been focused recently on the role of the corpus
callosum in the transfer and integration of information between the two
cerebral hemispheres (Banich, 1995; Banich, 1998a; Liederman, 1998). The
study of interhemispheric interaction builds on earlier work in exploring and
defining hemispheric asymmetries. Rather than focusing on the ways in
which the two cerebral hemispheres differ in their information processing
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
abilities and propensities, much current research is directed toward
delineating the ways in which the massively interconnected hemispheres
cooperate in producing unity of experience, thought, and action (Hellige,
Taylor, Lesmes, & Peterson, 1998).
Information is shared between the two hemispheres through a variety
of cortical and subcortical routes, but the corpus callosum appears to play
the dominant role in interhemispheric interactions (Marks, 2000). Its 200
million fibers provide direct connections between nearly all areas of the
cortex, permitting sensory, association, and motor signals to be shared
through a variety of anatomic and temporal channels. Sensory signals are
segregated by the topographic organization of the corpus callosum, so that
visual, auditory, and somaesthetic information crosses in different regions,
and higher order information is also organized in distinct anatomic
channels. The corpus callosum is also important in the temporal
sequencing of shared information. In a visual recognition task, for example,
the sharing of low level sensory information begins before higher order
information about object identity or semantic associations becomes
available, and that information begins to be shared before response
selection and programming are initiated. These stages of information
transfer and integration may take place in parallel and interact, leading to the
final output of a unitary behavior.
There are a number of experimental paradigms for studying the ways
in which the corpus callosum mediates interhemispheric transfer and
integration of visual information, including experiments in which stimuli are
projected briefly to one visual half-field or the other, thereby determining
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
which hemisphere receives the critical information directly and initiates
processing. Quantitative and qualitative aspects of performance are
compared for left-visual-field/right-hemisphere (LVF/RH) trials and right-
visual-field/left-hemisphere (RVF/LH trials).
In one paradigm, performance on trials when all relevant task
information is directed to a single visual field (called the within-hemisphere
condition) is compared with performance on trials when task information is
divided between the two visual fields (called the between-hemispheres
condition). In some situations, typically when the task is easy (e.g., matching
two letters presented in the same case), performance is better for the within-
hemisphere condition. In other situations, typically when the task is more
demanding (e.g., matching two letters presented in different cases),
performance is better for the between-hemispheres condition. Banich and
her colleagues (Banich, 1998b; Belger & Banich, 1998; Weissman & Banich,
2000; Weissman, Banich, & Puente, 2000) have suggested that as the
complexity of the task increases or the demand for selective attention is
greater, the benefits of spreading the processing load between the
hemispheres outweigh the costs of transferring information across the
corpus callosum and coordinating the work of the two hemispheres. Banich
attributes the enhancement of performance to the expansion of processing
capacity from dividing computations over additional neural space.
Liederman and her colleagues (e.g., Liederman & Meehan, 1986;
Liederman, Merola, & Martinez, 1985) have also compared inter-versus
intrahemispheric processing, but they have stressed the role of the corpus
callosum in shielding ongoing processing in one hemisphere from
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
potentially interfering processing in the other. One task they have used
involves identifying upright or inverted letters. They have found that when
each hemisphere can perform one of the two conflicting tasks, dividing the
inputs between the two hemispheres leads to enhancement of performance,
particularly when the number of stimuli is increased from two to four.
Another way of investigating the transfer and integration of visual
information between the hemispheres is to compare performance on trials
when all relevant task information is directed to a single visual half-field (the
unilateral condition) with performance on trials when the same information
is presented simultaneously to both visual half-fields, so that both
hemispheres receive the same information directly (the bilateral redundant
condition). This paradigm is especially useful when the two hemispheres
both have competence for a task, but performance differs in some qualitative
manner. By comparing performance on redundant bilateral trials with
performance on unilateral trials, we can examine how collaboration of the
two hemispheres influences both the level of performance that is achieved
and the processing strategy that emerges (for discussion of the redundant
bilateral paradigm, see Banich & Karol, 1992; Brown & Jeeves, 1993;
Hasbrooke & Chiarello, 1998; Marks & Hellige, 1999; Mohr, Pulvermuller,
Mittelstadt, & Rayman, 1996; Mohr, Pulvermuller, Rayman, & Zaidel, 1994;
Mohr, Pulvermuller, & Zaidel, 1994).
For example, Banich and Karol (1992) used bilateral redundant
presentation in a rhyme task to investigate whether bihemispheric
processing could be predicted from the pattern of performance on unilateral
trials. Participants were asked to decide whether probe words rhymed with
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5
an earlier presented central target word. On bilateral redundant trials,
sometimes the same word appeared twice (either in the same or different
case and font), sometimes two different words appeared but they led to the
same decision, and sometimes two different words appeared which led to
different decisions. The type of processing that emerged on bilateral trials
could not be predicted from the pattern of performance on unilateral trials,
suggesting that physical identity interacted with semantic identity in
bihemispheric processing.
Bilateral redundant presentation has also been used to investigate
hemispheric differences in the processing of consonant-vowel-consonant
(CVC) nonsense syllables (Cherry, Hellige, & McDowd, 1995; Eng & Hellige,
1994; Eviatar, Hellige, & Zaidel, 1997; Hellige & Cowin, 1996; Hellige &
Marks, in press; Hellige & Scott, 1997; Hellige, Taylor, & Eng, 1989; Hellige
etal., 1998; Kee, Cherry, Neale, McBride, & Segal, 1998; Luh & Levy, 1995;
Marks & Hellige, 1999; Taylor, 1998/1999). In the CVC task, participants are
required to identify nonsense syllables presented briefly to the LVF/RH, the
RVF/LH, or with the same syllable in each visual field (bilateral redundant
presentation). The primary results of these experiments have been quite
consistent, despite a number of procedural differences that are not relevant
for present purposes. Thus, these previous experiments, whose results are
summarized in Tables 1 and 2, provide a solid foundation on which to build
the present experiments.
As expected for a verbal task, the CVC identification error rate is
consistently lower on RVF/LH trials than on LVF/RH trials (see Table 1) (for
review, see Hellige, 1993). With two exceptions, there have been even fewer
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
Table 1
Percentage of CVC Identification Errors for LVF, RVF and Redundant
Bilateral (BVF) Trials in Each of 13 Published Experiments
Percentage of CVC Identification Errors
Experiment LVF RVF BVF Predicted
BVF °
1. Hellige et at., 1989, Exp. 1 3 61.3% 38.8% 38.6% 23.8%
2. Luh & Levy, 1995 a 62.5% 42.7% 45.1% 26.7%
3. Hellige et al., 1989, Exp. 2 61.3% 42.9% 38.8% 26.3%
4. Eng & Hellige, 1994 61.5% 47.4% 37.7% 29.2%
5. Hellige et al., 1994 b 61.0% 45.0% 40.0% 27.5%
6. Cherry et al., 1995 58.9% 45.2% 39.2% 26.6%
7. Hellige & Cowin, 1996, Exp. 2 61.4% 47.2% 38.7% 29.0%
8. Eviatar et al., 1997 55.4% 48.1% 40.0% 26.6%
9. Hellige et al., 1998 54.7% 43.4% 38.1% 23.7%
10. Kee et al., 1998 b 63.2% 49.8% 39.6% 31.5%
11. Taylor, 1998/1999 57.0% 45.8% 38.0% 26.1%
12. Marks & Hellige, 1999, Exp. 1 62.9% 42.9% 26.8% 26.8%
13. Marks & Hellige, 1999, Exp. 2 58.4% 39.8% 44.1% 48.0%
Means of 13 experiments 60.0% 44.5% 38.8% 28.6 %
8 b
Used a central fixation digit. Left-handed participants omitted.
Q
BVF error percentage predicted by statistical summation (see text for
explanation). For Marks & Hellige (1999), Exp. 2, see their Appendix A.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
Table 2
Qualitative Error (QE) Scores for LVF, RVF and Redundant Bilateral (BVF)
Trials in each of 13 Published Experiments
QE Scores
Experiment LVF RVF BVF Unilateral
Average °
1. Hellige et al., 1989, Exp. 1 a .53 .35 .45 .44
2. Luh & Levy, 1995 a .58 .25 .44 .42
3. Hellige et al., 1989, Exp. 2 .43 .12 .37 .28
4. Eng & Hellige, 1994 .52 .27 .45 .40
5. Hellige et al., 1994 b .54 .31 .44 .43
6. Cherry et al., 1995 .35 .21 .32 .28
7. Hellige & Cowin, 1996, Exp. 2 .27 .10 .21 .19
8. Eviatar et al., 1997 .36 .18 .23 .27
9. Hellige et al., 1998 .39 .12 .35 .26
10. Kee et al., 1998 b .50 .32 .37 .41
11. Taylor, 1998/1999 .41 .26 .39 .34
12. Marks & Hellige, 1999, Exp. 1 .46 .20 .35 .33
13. Marks & Hellige, 1999, Exp. 2 .46 .23 .39 .35
Means of 13 experiments .45 .22 .37 .34
3 b
Used a central fixation digit. Left-handed participants omitted.
C
Mean of the QE scores for LVF and RVF trials.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
errors on bilateral trials than on RVF/LH trials. The two exceptions come
from the only two experiments to combine CVC identification with a
secondary task that required observers to also identify a small digit
presented at the fixation point at the same time that a CVC stimulus was
presented to one or both visual half-fields.
The amount of bilateral redundancy gain found in the CVC
identification task is linked to attentional factors. When only a single target is
presented on unilateral trials, and nothing appears in the contralateral visual
field, the amount of gain in performance on redundant bilateral trials is quite
modest and much less than would be predicted by statistical summation.
Consider what level of bilateral performance would be expected if each copy
of the stimulus were processed as efficiently as it would have been had it
been presented alone. If this were the case, and if a correct identification of
either copy would lead to a correct response on bilateral redundant trials,
then the level of bilateral performance can be predicted from the level of
LVF/RH and RVF/LH performance using statistical summation. That is,
Pbvf = Plvf + P rvf - (P lvf)(P rvf), where Pbvf is the predicted probability of a
correct response on bilateral trials, P lvf is the probability of a correct
response on LVF/RH trials, and Prvf is the probability of a correct response
on RVF/LH trials.
One possible explanation for the generally low level of bilateral
redundancy gain found in most previous experiments is that on bilateral
redundant trials, neither stimulus can be processed in the same way and
with the same resources as it would have been had it been presented alone.
To test for this possibility, in Marks and Hellige (1999) Experiment 1 we
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
modified the unilateral trials to include an alphabetic noise stimulus in the
contralateral visual field, thus more closely approximating the division of
attention between the two identical stimuli on bilateral redundant trials. As
predicted, bilateral redundancy gain was substantially larger than what had
been reported for earlier experiments. In fact, the gain was virtually identical
to that predicted by statistical summation (see Table 1). The achievement of
this level of bilateral redundancy gain clearly depends on some degree of
cooperation between the two cerebral hemispheres. It leaves open the
question of the level or levels at which information is shared and integrated,
from low-level sensory or perceptual processes, to higher level processes,
to control of the final unitary response.
The performance enhancement that often results when redundant
copies of the stimulus are presented bilaterally could be explained in a
variety of ways (which are not mutually exclusive). One possibility is that the
two hemispheres operate independently, and whichever completes
processing first initiates a response (statistical summation or horse-race
model: Coney, 1985). Another possibility is that some sort of neural
summation occurs, often referred to as coactivation ( e.g., Allen, 1968;
Fournier & Eriksen, 1990; Miller, 1982; Schuez & Preissl, 1996). It is also
possible that some interaction occurs between independent processes and
some kind of neural summation (e.g., Mordkoff & Yantis, 1991).
Sometimes a horse-race model is invoked to explain or to test the
locus of processing enhancement (Coney, 1985). For example, Hasbrooke
and Chiarello (1998) used bilateral redundant presentation to investigate
hemispheric differences in lexical ambiguity resolution. Targets related to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
the dominant or subordinate meanings of ambiguous word primes were
presented either unilaterally or bilaterally. Priming was observed for
dominant meanings in both unilateral conditions, but for subordinate
meanings in only the RVF/LH condition. Bilateral performance showed
redundancy gain in both reaction times and accuracy and a priming pattern
similar to that of the RVF/LH condition. They used a statistical model of
independent processing in the two hemispheres to see if it would account
for the observed gain. Because modeled reaction times were always faster
than observed reaction times, Hasbrooke and Chiarello concluded that their
results provided evidence for interhemispheric cooperation, with some cost
associated with that cooperation.
For an example of an explanation in terms of neural summation,
consider the work of Mohr, Pulvermuller, and Zaidel and colleagues (Mohr et
al., 1996; Mohr, Pulvermuller, Rayman, & Zaidel, 1994; Mohr, Pulvermuller, &
Zaidel, 1994). In one experiment, these researchers presented function
words, content words, or pseudowords for lexical decision in either unilateral
or bilateral redundant conditions. They found a large right visual field
advantage (RVFA) for content words, but no visual field advantage for
function words. Bilateral redundant presentation improved performance for
both content and function words, but not for pseudowords. Mohr et al.
attributed the gain they observed for words (but not for nonwords) to
cooperation between the lexicons in the two hemispheres or to the existence
of bilaterally distributed word representations. They suggested that their
results were consistent with the view that the neuronal counterparts of words
are Hebbian cell assemblies consisting of strongly connected excitatory
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11
neurons of both hemispheres. Bilateral gain then results from spatial
summation of EPSPs in these cell assemblies.
The present experiments were designed to extend our investigation of
interhemispheric interaction by examining the information processing locus
of effects related to target redundancy and interhemispheric cooperation.
Because in past experiments using consonant-vowel-consonant (CVC)
nonsense syllables (e.g., Cherry et al., 1995; Eng & Hellige, 1994; Eviatar et
al., 1997; Hellige & Cowin, 1996; Hellige & Marks, in press; Hellige & Scott,
1997; Hellige et al., 1989; Hellige et al., 1998; Kee et al., 1998; Marks &
Hellige, 1999; Taylor, 1998/1999), the redundant stimuli were always
physically identical, it has not been possible to separate perceptual from
semantic or other more abstract aspects of interhemispheric interaction. For
example, consider Marks and Hellige (1999) Experiment 1, in which bilateral
redundant presentation of the target CVC resulted in much better
performance than unilateral presentation of the target CVC together with a
noise stimulus in the opposite visual field. It is unclear whether this bilateral
redundancy gain (measured by the enhancement in performance for the
bilateral redundant condition compared with the better of the unilateral
conditions) depended on the fact that the two CVC stimuli were physically
identical, on the fact that they consisted of the same letters and had the
same name, on the fact that they led to the same response, or on some
combination of these types of similarity. As shown by Banich and Karol
(1992; see also Banich & Shenker, 1994) in their studies of other aspects of
interhemispheric interaction, these possibilities could be differentiated by
introducing a case and font manipulation, such that redundant targets are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
sometimes in the same case and font as each other and sometimes in a
different case and font. To the extent that low level sensory or perceptual
effects account for a significant portion of the performance advantage on
redundant bilateral trials, then varying the case and font of the two targets
would reduce or eliminate bilateral gain. To the extent that more abstract
information processing accounts for a significant portion of the bilateral gain,
the gain would remain even when case and font are varied. Similar
arguments could be made about the qualitative nature of processing,
especially if the strategy that emerges on bilateral trials is determined by the
information processing locus at which the hemispheres interact for a
particular task.
In the present experiments, both the nature and the physical
characteristics of the stimuli were varied. The first two experiments
constituted a replication of Marks and Hellige (1999) Experiment 1, with the
addition of case and font manipulations. Bilateral redundant trials included a
consistent condition, in which both CVCs were presented in the same case
and font, and an inconsistent condition, in which the two CVCs differed in
case (Experiment 1) or case and font (Experiment 2). The purpose was to
determine whether physical identity of the stimuli presented is required to
produce a bilateral redundancy gain in accuracy, and secondarily to compare
qualitative performance with physically identical and non-identical stimuli.
Experiment 3 was a replication of Experiment 2, with separate adjustment of
stimulus exposure durations for uppercase and lowercase stimuli. This was
done in order to more nearly equate accuracy for the two types of stimuli,
because accuracy was found to be much greater for uppercase than for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
lowercase stimuli in Experiments 1 and 2. Experiment 4 was designed to
test whether the patterns of performance found for CVCs and other linguistic
materials would extend to numbers presented as Arabic digit trigrams.
Experiment 5 explored the processing of numbers presented as structured
dot pattern trigrams, and a final experiment compared performance on
physically identical number trigrams (both digit or both dot cluster trigrams)
with performance on non-identical number trigrams (a digit trigram
presented in one visual field with a dot cluster trigram representing the
same number in the other visual field). The logic of Experiment 6 was the
same as that for Experiments 1 to 3. The primary purpose was to determine
whether the pattern of performance found in Experiments 1 to 3 would extend
to the case where the two stimuli were different both in physical form and in
kind.
In this series of experiments, the presentation of number trigrams
presented either as digits or as structured dot clusters provided an
opportunity to compare performance on these non-linguistic stimuli with
performance on CVCs. The presentation of consistent and inconsistent
bilateral CVCs and number trigrams was designed to examine the effect of
physical similarity on quantitative and qualitative measures of performance.
EXPERIMENT 1
Experiment 1 was a replication of Experiment 1 from Marks and
Hellige (1999), with the addition of a physical identity manipulation. Bilateral
redundant trials included a consistent condition, in which both CVCs were
presented in the same case, and an inconsistent condition, in which the two
CVCs which led to the same response differed in case. The primary purpose
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
was to determine whether physical identity of the stimuli presented is
required to produce a bilateral redundancy gain in accuracy. If low level
sensory or perceptual effects account for a significant portion of the
performance advantage on redundant bilateral trials, then varying the case of
the two copies of the target should reduce or eliminate bilateral gain. If more
abstract information processing accounts for a significant portion of the
bilateral gain, the gain should remain even when case is varied.
A secondary purpose was to compare qualitative performance with
physically identical and non-identical stimuli. To examine the qualitative
nature of CVC processing, investigators have classified CVC identification
errors into three theoretically motivated types suggested first by Levy, Heller,
Banich and Burton (1983): first-letter errors (FEs), last-letter errors (LEs) and
other errors (OEs). An FE occurs when the first letter is missed but the last
letter is correct (e.g., CAG or CEG for DAG), with the correctness of the vowel
being irrelevant. An LE occurs when the last letter is missed but the first
letter is correct (e.g., DAC or DEC for DAG), with the correctness of the vowel
being irrelevant. All other types of errors are classified as OEs. The resulting
error pattern is quite different for the LVF/RH and RVF/LH conditions. The
difference is captured nicely by the following qualitative error or QE score
introduced by Levy et al. and computed separately for each visual field
condition:
QE = (LE - FE)/(FE + LE + OE)
Thus, the QE score provides an indication of the extent to which the number
of LEs is greater than the number of FEs when corrected for differences in
the overall error rate among the visual field conditions.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
As shown in Table 2, the QE scores for CVC recognition are uniformly
larger on LVF/RH trials than on RVF/LH trials. It has been suggested that the
relatively large difference between the number of FEs and the number of LEs
on LVF/RH trials reflects the fact that the right hemisphere lacks phonetic
processing ability and, as a consequence, treats the CVC as three individual
letters which are processed in a relatively slow sequential manner. The
lower QE scores on RVF/LH trials suggests that the left hemisphere
distributes attention more quickly or more evenly across the three letters,
perhaps because of its phonetic processing ability, superiority for
orthographic coding, or superiority for processing the individual elements of
a multi-element display (for discussion, see the papers whose results are
summarized in Tables 1 and 2).
It might be expected that when the hemispheres are predisposed
toward different ways of performing a task, the mode of processing
associated with superior performance on unilateral trials would also emerge
on redundant bilateral trials. With respect to CVC processing, this would
mean that the QE scores on bilateral trials would be identical to the QE
scores on RVF/LH trials. Despite the intuitive appeal of this possibility and
the fact that the error rate on bilateral trials is equal to or less than the error
rate on RVF/LH trials, the actual results have been quite different. As shown
in Table 2, the QE score on bilateral trials is uniformly greater than the QE
score on RVF/LH trials, with the difference being statistically significant in all
but one case. In some experiments the bilateral QE score is so large that it
does not differ significantly from the LVF/RH QE score, though in all cases
the bilateral QE score is numerically in between the two unilateral QE scores
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16
and is often only slightly larger than the average of the two unilateral QE
scores (the rightmost column of Table 2). Experiment 1 allowed us to
explore whether physical differences between two simultaneously presented
stimuli influence the qualitative nature of processing as indexed by the QE
score.
Design and Method
Participants
Twelve men and 16 women university students were recruited to
participate in Experiment 1. Participants received extra credit in psychology
courses. Only right-handed native speakers of English who had normal or
corrected-to-normal vision in both eyes by self report served as participants.
Participants ranged in age from 19 to 39 years (M =21.1, SD = 3.7).
Apparatus and stimulus materials
All stimuli were prepared and presented on a Macintosh Ilex computer
with a 13-inch AppleColor High Resolution RGB Monitor (M1297) using the
MacProbe software package from Aristometrics, Inc. (Castro Valley,
California). Stimuli consisted of 36 consonant-vowel-consonant (CVC) non
word trigrams formed from the consonants D, F, G, K, P, S, and T and the
vowels A, E, and 0. Each CVC appeared in both uppercase and lowercase
forms. In addition, a trigram consisting of three uppercase X s was used as
a noise stimulus. The noise stimulus was similar in appearance to the
target stimuli and was thus expected to capture attention in the same
manner as a redundant copy of the target in bilateral redundant trials, even
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17
though an X never appeared in a target stimulus and participants were told
this was the case. An additional 12 different CVCs were used for a block of
37 practice trials. The CVCs used in this experiment were the same as
those used in previous experiments ( e.g., Hellige et al., 1989; Marks &
Hellige, 1999). The trigrams were in Helvetica Black 24 pt. font, oriented
vertically, and displayed in white against a dark background on the computer
monitor. Each trigram subtended 0.5° of visual angle horizontally and 3.2 0 of
visual angle vertically when viewed at a distance of approximately 57 cm
from the screen. Participants used a chin rest as well as a forehead bar to
limit head movement and maintain viewing distance.
The stimuli were presented in two positions on the screen: LVF,
centered 3° left of fixation, and RVF, centered 3° right of fixation. In the LVF/RH
condition, the CVC (uppercase or lowercase) was presented to the left of
fixation and a trigram consisting of three X s (the noise stimulus) was
presented to the right; in the RVF/LH condition, the CVC (uppercase or
lowercase) was presented to the right of fixation and the noise stimulus was
presented to the left; in the bilateral conditions, the same CVC was
presented in both visual fields. In the Bilateral Consistent condition, both
CVCs were in the same case. In the Bilateral Inconsistent condition, one
CVC was in uppercase and the other CVC was in lowercase. In half of the
Bilateral Inconsistent trials, the uppercase CVC was presented in the LVF
and the lowercase CVC was presented in the RVF; the other half of the trials
were reversed. In all conditions, each side of the screen contained either a
CVC or a noise stimulus. Samples of the stimuli and the 8 possible trial
types are illustrated in Figure 1.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
18
f "\ r ....... \
D X X d
A + X X + a
G X X g
V ) v J
Unilateral LVF/ RH Unilateral RVF/LH
f I
f ^
D D d d
A + A
a + a
G G
g g
V J V J
Bilateral Consistent Bilateral Consist ent
Uppercase Lowercase
f I
r
D d d D
A + a a + A
G g g G
v J v J
Bilat eral Inconsistent Bilateral Inconsist ent
LVF Uppercase RVF Uppercase
Figure 1. Samples of the stimuli and of the 8 trial types for Experiment 1.
Fonts are depicted accurately, but the displays are not drawn to scale
(LVF/RH = left visual field/right hemisphere; RVF/LH = right visual field/left
hemisphere.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
19
A practice block of 37 trials was selected randomly from the 96
possible trials using the 12 practice CVCs. The practice block was followed
by six experimental blocks of 49 trials each. Each CVC served as a target
eight times, once in each of the trial types. The order of trials was
randomized for each participant, with the constraint that no individual CVC or
trial type could immediately follow itself. The first trial in each block served as
a filler trial and thus was not scored, leaving a total of 288 scored trials.
A pattern mask was displayed in each screen position following the
offset of the stimuli. The masks were composed of thin black and white
horizontal bars. Each mask subtended approximately 3° of visual angle
horizontally and 4° of visual angle vertically and was centered 3° to the right
or left of fixation.
Procedure
Participants filled out a questionnaire based on the Edinburgh
Handedness Inventory (Oldfield, 1971) and were judged to be right handed if
they wrote and drew with their right hands and did not report any pressure to
use their right hand rather than their left hand in childhood. Each participant
was tested individually in one session, during which he or she was seated
in a quiet, dimly lit testing room. Participants were told that on each trial a
CVC would be presented in the left visual field (with a string of three X ’s in
the right visual field); in the right visual field (with a string of three X ’s in the
left visual field); or in both visual fields simultaneously. They were instructed
to ignore the X ’s, because the CVC would never contain an X. The
experimenter emphasized that when there were two copies of the CVC, they
would always be the same. During the instructions, samples of the stimuli
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
were presented on the screen. After they completed the practice block,
participants were asked whether they had any questions about the task.
For each trial a tone was sounded and a small fixation cross
appeared in the center of the screen for 1000 ms, followed by the
presentation of the trigrams. Immediately at the offset of the trigrams, the
pattern mask appeared in both visual fields for 210 ms. Participants were
instructed to maintain fixation on the fixation cross for the duration of the trial
and to report the CVC by first pronouncing it and then spelling it, guessing
when they were unsure. The experimenter entered their responses in the
computer, and the next trial was immediately initiated. The importance of
maintaining fixation on the fixation cross for the duration of the trial was
emphasized.
To insure an error rate near 50% for qualitative analysis, exposure
duration of the stimuli was adjusted throughout the experiment, as follows:
on the first practice trial, the exposure duration was set at 180 ms.
Thereafter, each correct response shortened the duration of the next trial by
15 ms (a single screen refresh cycle for our 75 Hz monitor), and each
incorrect response lengthened the duration of the next trial by 15 ms. The
maximum exposure duration was set at 210 ms in order to avoid eye
movements. The effect of this method of adjustment is that mean
percentage of errors is constrained to be near 50%, but that performance is
allowed to vary across the three visual field conditions. Participants were
allowed short rest breaks between blocks.
On each trial, accuracy was measured for the trigram as a whole and
for each letter position within the trigram. Means for each trial type were
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 1
calculated by the MacProbe program. Dependent variables for statistical
analysis were mean percentage of errors and mean QE score (which
provides an indication of the extent to which the number of LEs is greater
than the number of FEs when corrected for differences in the overall error
rate among the visual field conditions).
Results and Discussion
The trial-by-trial adjustment of exposure duration resulted in
approximately equal durations for all four visual field conditions (LVF/RH
M = 72 ms; RVF/LH M = 73 ms; Bilateral Consistent M = 73 ms; Bilateral
Inconsistent M = 73 ms) and produced an overall CVC identification error
rate of 49.1% (SD = 2.4%).
Total Error Percentage
Total error percentages were submitted to an AN OVA with gender as
the between-subjects factor and visual field condition (LVF, RVF, Bilateral
Consistent, Bilateral Inconsistent) as the within-subjects factor. There was
no main effect of gender, F(1,26) = 2.37, MSE = 21.71,p = .14. However,
there was a significant interaction of gender with visual field, F(3,78) = 3.35,
MSE = 12.29, p < .05, which came about because the RVF/LH error rate was
lower for males (M = 48.3%, SD = 5.2%) than for females (M = 58.2%,
SD = 17.0%), p < .05, but male and female performance did not differ in the
other visual field conditions, all ps > .20 (all pairwise comparisons reported
are Newman-Keuls post hoes).
As shown in the upper panel of Figure 2, the error rates were not
equal for the four visual field conditions, F(3,78) = 27.85, MSE = 123.32,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
22
7 0%
a t
a >
5 6 0 % -
c
a t
o
a t
a
v. 5 0 % -
Z 4 0 % -
I-
3 0%
LVF/RH RVF/LH BVF-Con BVF-lncon
Visual Field/Hemisphere
0.40
0 . 3 0 -
o t
o
o
< n
L U
O
0 .20 -
0.10
LVF/RH RVF/LH BVF-Con BVF-lncon
Visual Field/Hemisphere
Figure 2. Upper panel: Percentage of errors for each of the four visual field
conditions in Experiment 1. Lower panel: Qualitative error (QE) scores for
each of the four visual field conditions in Experiment 1. Because visual field
was a within-subject variable, error bars in all figures show the standard
error of the difference scores computed by subtracting each participant’s
mean score from his or her score for each visual field and, thus, do not
include between-subjects variability (e.g. Biederman & Gerhardstein, 1993).
(LVF/RH = left visual field/right hemisphere; RVF/LH = right visual field/left
hemisphere; BVF-Con = Bilateral Consistent; BVF-lncon = Bilateral
Inconsistent.)
T
r ~ ---------------1 ------------------1 ------------------ r
i-----------------1 -----------------i-----— ------r
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23
p < .001. Subsequent pairwise comparisons showed that the error rate was
significantly smaller on RVF/LH trials (M = 53.9%, SD = 14.0%) than on
LVF/RH trials (M = 62.9%, SD = 11.4%) and significantly smaller on both
Bilateral Consistent trials (M = 39.4%, SD = 6.3%) and Bilateral Inconsistent
trials (M = 40.0%, SD = 7.7%) than on either type of unilateral trial, all
ps < .01. For LVF/RH, RVF/LH, and Bilateral Consistent trials, this pattern of
results is similar to that found for previous CVC experiments (see Table 1).
To test whether physical identity is important for bilateral redundancy
gain, the crucial comparison was of performance on bilateral trials in which
both stimuli were in the same case (Bilateral Consistent) with that on
bilateral trials in which the two stimuli were in a different case (Bilateral
Inconsistent). For both Bilateral Consistent and Bilateral Inconsistent trials,
total error percentage was near 40%. There was virtually no difference
between accuracy when the two copies of the stimulus were in the same
case and accuracy when the two copies of the stimulus were in a different
case, p = .85.
Although error rates for bilateral trials were much lower than error
rates for unilateral trials, they were still significantly higher than would be
predicted by the statistical summation formula, f(23) = 4.02, p < .001. The
statistical summation formula yields a predicted bilateral error rate of 32.8%,
compared to the mean bilateral error rate in Experiment 1 of 39.7%. This
result is in contrast to Marks and Hellige (1999) Experiment 1, which also
used XXX as a noise stimulus to equate the number of stimuli present for all
trial types. The large bilateral gain in that experiment was attributed to the
division of attention between the LVF and RVF stimuli. Note that in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
24
present experiment, the same noise stimulus (uppercase XXX) was used
on all trials, even ones on which the target was a lowercase CVC. It is
possible that this noise stimulus did not lead observers to divide attention
between the two stimuli on unilateral trials to the same extent as on bilateral
trials. The present experiment also differed from Marks and Hellige (1999)
Experiment 1 in that it included two conditions which had very different levels
of difficulty (see the discussion of case effects below). The result was that
although performance overall averaged near 50%, both mean uppercase
performance and mean lowercase performance were rather far from 50%,
with the consequence that some participants showed floor and ceiling
effects in some conditions. As a result, performance on some visual field
conditions was constrained, and this may have reduced bilateral gain for
some participants.
However, the bilateral redundancy gain of 12.3% (calculated as the
difference between the RVF/LH error rate and the mean bilateral error rate)
was still greater than on previous CVC experiments without a noise stimulus
(see Table 1). It is clear from the finding of a statistically reliable bilateral
redundancy gain, found both when the two stimuli were physically identical
and when the two stimuli differed in case, that physical identity is not
required for performance to be enhanced relative to unilateral presentations
in this experiment.
QE Scores
In order to examine hemispheric differences in the qualitative nature
of processing, QE scores were computed for each visual field condition in
the manner described earlier: QE = (LE - FE) / (FE + LE + OE). As shown in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25
the lower panel of Figure 2, the QE scores were not equal for the four visual
field conditions, F(3,78) = 6.33, MSE = .017, p < .001. Subsequent pairwise
comparisons showed that the QE score was significantly smaller on RVF/LH
trials (M = .19, SD = .18) than on LVF/RH trials (M = .33, SD = .26) or Bilateral
Consistent trials (M = .31, SD = .23) or Bilateral Inconsistent trials (M = .31,
SD = .26). For LVF/RH, RVF/LH, and Bilateral Consistent trials, this pattern of
QE scores is similar to some earlier findings for CVCs (see Table 2).
Case Effects
To examine the effects of the case manipulation, an additional AN OVA
was performed with visual field (LVF, RVF, BVF) and case (uppercase,
lowercase) as the within-subjects factors. The case effects were not of
theoretical interest here, but were included to check for any interaction with
visual field. (In this analysis, only Bilateral Consistent trials were included,
because case was mixed in the Bilateral Inconsistent trials).
For total error percentage, there was a main effect of visual field,
F{2,54) = 23.19, MSE = 339.93, p < .001 (descriptive statistics are reported
with the previous analysis). There was also a main effect of case,
F(1,27) = 139.17, MSE = 164.96, p < .001. Lowercase stimuli (M = 63.8%,
SD = 6.1%) led to substantially more errors than uppercase stimuli
(M = 40.4%, SD = 5.9%). There was also a significant interaction of case and
visual field, F(2,54) = 6.80, MSE = 50.58, p < .01. The upper panel of Figure 3
shows total error percentages by case. The interaction appears to have
arisen because the difference between performance on lowercase and
uppercase trials was smaller for LVF/RH trials than for RVF/LH or Bilateral
Consistent trials.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26
8 0 %
09
ffi
0.
m 4 0 % -
c o
* ■ ■ ■ =
o Uppercase
- - 0 - - Lowercase
2 0%
RVF LVF BVF-Con
Visual Field/Hemisphere
0.50
0 . 4 0 -
©
hm
o
W 0 . 3 0 -
u i
O
0 .2 0 -
0.10
LVF/RH RVF/LH BVF-Con
Visual Field/Hemisphere
Figure 3. Upper panel: Percentage of errors for three visual field conditions
in Experiment 1 by case of stimuli. Lower panel: Qualitative error (QE) scores
for three visual field conditions in Experiment 1 by case of stimuli.
(LVF/RH = left visual field/right hemisphere; RVF/LH = right visual field/left
hemisphere; BVF-Con = Bilateral Consistent.)
— Uppercase
- - 0 - - Lowercase
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
27
For QE scores, there was a main effect of visual field, F(2,54) = 10.41,
MSE = .038, p < .001 (descriptive statistics are reported with the previous
analysis). There was no main effect of case, F < 1, and no significant
interaction of visual field and case, F(2,54) = 1.91, MSE = .022, p = .16.
Uppercase QE scores (M = .30, SD = .23) were not significantly different
from lowercase QE scores {M = .26, SD = .21). The lower panel of Figure 3
shows QE scores by case.
The main finding of Experiment 1 is that physical identity of bilateral
redundant stimuli is not required to produce substantial bilateral redundancy
gain. There was no significant difference between accuracy on bilateral trials
when the two stimuli were physically identical and accuracy on bilateral trials
when the two stimuli were different in case. However, it remains possible
that the failure to find a difference between performance for consistent and
inconsistent bilateral trials may be partially attributable to the similarity
between the uppercase and lowercase stimuli, because both were in the
same font. In order to test this possibility, for Experiment 2, both case and
font were varied, in order to increase the physical difference between the two
stimuli on Bilateral Inconsistent trials.
EXPERIMENT 2
Design and Method
Participants
Ten men and 14 women university students were recruited to
participate in Experiment 2. They were from the same pool as Experiment 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
28
and selected in the same manner. Participants ranged in age from 18 to 22
years {M = 19.5, SD = 1.3).
Apparatus and stimulus materials
The apparatus and stimulus materials for Experiment 2 were the
same as those for Experiment 1, with the following exceptions: The trigrams
were oriented vertically and displayed in black against a white background
on the computer monitor. Each trigram subtended approximately 0.60 of
visual angle horizontally and 3.3 ° of visual angle vertically when viewed at a
distance of approximately 57 cm from the screen. Uppercase trigrams were
in 30 pt. Trebuchet MS font, and lowercase trigrams were in 30 pt. Rockwell
font. These fonts were chosen because the uppercase and lowercase
versions of the letters in the trigrams were quite different. Also, the
uppercase and lowercase trigrams were approximately the same size. All
trigrams were constructed in a computer graphics program (Canvas).
In the Bilateral Consistent condition, both CVCs were in the same
case and font. In the Bilateral Inconsistent condition, one CVC was in
uppercase Trebuchet MS and the other CVC was in lowercase Rockwell.
Samples of the stimuli and the 8 possible trial types are illustrated in
Figure 4.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
Procedure
The procedure for Experiment 2 was the same as for Experiment 1.
Results and Discussion
The trial-by-trial adjustment of exposure duration resulted in
approximately equal durations for all four visual field conditions (LVF/RH
M = 37 ms; RVF/LH M = 38 ms; Bilateral Consistent M = 38 ms; Bilateral
Inconsistent M - 38 ms) and produced an overall CVC identification error
rate of 43.7% (SD = 5.4%).
Total Error Percentage
Total error percentages were submitted to an AN OVA with gender as
the between-subjects factor and visual field condition (LVF, RVF, Bilateral
Consistent, Bilateral Inconsistent) as the within-subjects factor. There was a
main effect of gender, F(1,22) = 12.95, MSE = 76.71, p < .01. Males
(M = 39.9%, SD = 4.6%) committed fewer errors than females (M = 46.5%,
SD = 4.2%). However, there was no significant interaction of gender with
visual field, F(3,66) = 1.59, MSE = 67.92, p = .20.
As shown in the upper panel of Figure 5, the error rates were not
equal for the four visual field conditions, F(3,66) = 63.62, MSE = 67.92,
p < .001. Subsequent pairwise comparisons showed that the error rate was
significantly smaller on RVF/LH trials (M = 45.9%, SD = 9.2%) than on
LVF/RH trials (M = 61.9%, SD = 9.3%) and significantly smaller on both
Bilateral Consistent trials (M = 33.6%, SD = 9.1%) and Bilateral Inconsistent
trials (M = 33.6%, SD = 8.6%) than on either type of unilateral trial, ps < .001.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
c ^
r .......................... .... ............\
D X X d
A + X X + a
G X X g
V .... . ... .......... ..J V J
Uni lateral L V F / R H Uni late ra l R V F / L H
C \ ( : ^
D D d d
A + A a + a
G G
g g
V J V J
Bi 1 at eral Consi st ent
Bi 1 at eral Consi st ent
Uppercase
Lowercase
f 1
r ..... \
D d d D
A + a
<
+
G g
g g
V J L J
Bi I at eral Inconsistent Bi I at eral I neons i st ent
LVF Uppercase RVF Uppercase
Figure 4. Samples of the stimuli and of the 8 trial types for Experiment 2.
Fonts are depicted accurately, but the displays are not drawn to scale
(LVF/RH = left visual field/right hemisphere; RVF/LH = right visual field/left
hemisphere.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31
To test whether physical identity is important for bilateral redundancy
gain, the crucial comparison was of performance on bilateral trials in which
both stimuli were in the same case and font (Bilateral Consistent) with that
on bilateral trials in which the two stimuli were in a different case and font
(Bilateral Inconsistent). For both Bilateral Consistent and Bilateral
Inconsistent trials, total error percentage was 33.6%. There was
virtually no difference between accuracy when the two copies of the stimulus
were in the same case and font and accuracy when the two copies of the
stimulus were in a different case and font.
Although error rates for bilateral trials were much lower than error rates for
unilateral trials, they were again significantly higher than would be predicted
by the statistical summation formula, f(23) = 2.75, p < .05. The statistical
summation formula yields a predicted bilateral error rate of 32.8%,
compared to the mean bilateral error rate in Experiment 2 of 39.7%. Because
Experiment 2 was very similar to Experiment 1, the same reasons may
account for these findings as were suggested for Experiment 1. That is, the
use of an uppercase noise stimulus for all trials may not have led observers
to divide their attention between the two visual half-fields equally on
unilateral and bilateral trials, and the difference in difficulty between
uppercase and lowercase stimuli may have resulted in floor and ceiling
effects for some participants in some conditions.
Nonetheless, the bilateral redundancy gain of 16.7% (calculated as
the difference between the RVF/LH error rate and the mean bilateral error
rate) was greater than on previous CVC experiments without a noise
stimulus (see Table 1). It is clear from the finding of a statistically reliable
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
32
7 0 %
S3
o >
« 6 0 % -
c
0 )
o
X m
< D
a.
5 0 % -
U J
( 0
4 0% -
30%
LVF/RH RVF/LH BVF-Con BVF-lncon
Visual Field/Hemisphere
0.60
0 . 5 0 -
0 .4 0 -
0 .3 0 -
0 .2 0 -
1 0 -
0.00
I i I ' |
LVF/RH RVF/LH BVF-Con BVF-lncon
Visual Field/Hemisphere
Figure 5. Upper panel: Percentage of errors for each of the four visual field
conditions in Experiment 2. Lower panel: Qualitative error (QE) scores for
each of the four visual field conditions in Experiment 2. Because visual field
was a with in-subject variable, error bars in all figures show the standard
error of the difference scores computed by subtracting each participant’s
mean score from his or her score for each visual field and, thus, do not
include between-subjects variability (e.g. Biederman & Gerhardstein, 1993).
(LVF/RH = left visual field/right hemisphere; RVF/LH = right visual field/left
hemisphere; BVF-Con = Bilateral Consistent; BVF-lncon = Bilateral
Inconsistent.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33
bilateral redundancy gain, found both when the two stimuli were physically
identical and when the two stimuli differed in case and font, that physical
identity is not required for performance to be enhanced relative to unilateral
presentations in this experiment, just as in Experiment 1.
QE Scores
In order to examine hemispheric differences in the qualitative nature
of processing, QE scores were computed for each visual field condition in
the manner described earlier: QE = (LE - FE) / (FE + LE + OE). As shown in
the lower panel of Figure 5, the QE scores were not equal for the four visual
field conditions, F(3,66) = 27.76, MSE = .025, p < .001. Subsequent pairwise
comparisons showed that the QE score was significantly smaller on RVF/LH
trials (M = . 11, SD = .22) than on LVF/RH trials (M = .50, SD = .16) and
intermediate on both Bilateral Consistent trials (M = .23, SD = .24) and
Bilateral Inconsistent trials (M = .36, SD = .25). All four visual field conditions
differed significantly, ps < .05. For LVF/RH, RVF/LH, and Bilateral Consistent
trials, this pattern of QE scores is similar to earlier findings for CVCs (see
Table 2). The finding of significantly higher QE scores for Bilateral
Inconsistent trials than for Bilateral Consistent trials is in contrast with the
results for Experiment 1.
Case and Font Effects
To examine the effects of the case and font manipulation, an
additional AN OVA was performed with visual field (LVF, RVF, BVF) and
case/font (uppercase Trebuchet MS, lowercase Rockwell) as the within-
subjects factors. The case/font effects were not of theoretical interest here,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34
but were included to check for any interaction with visual field. (In this
analysis, only Bilateral Consistent trials were included, because case/font
was mixed in the Bilateral Inconsistent trials).
For total error percentage, there was a main effect of visual field,
F(2,46) = 58.89, MSE = 164.07, p < .001 (descriptive statistics are reported
with the previous analysis). There was also a main effect of case,
F(1,23) = 85.99, MSE = 181.71, p < .001. Lowercase stimuli (M = 57.5%,
SD = 7.4%) led to substantially more errors than uppercase stimuli
(M = 36.7%, SD = 8.1%). However, there was no significant interaction of
case and visual field, F < 1. The upper panel of Figure 6 shows total error
percentages by case.
For QE scores, there was a main effect of visual field, F(2,46) = 31.08,
MSE = .048, p < .001 (descriptive statistics are reported with the previous
analysis). There was also a main effect of case, F(1,23) = 30.76, MSE = .094,
p < .001. Uppercase QE scores (M = .47, SD = .18) were substantially higher
than lowercase QE scores (M = .18, SD = .21), but there was no significant
interaction of visual field and case, F(2,46) = 2.21, MSE = .045, p = .12.
Consideration of this case effect on QE scores will be postponed until the
discussion of Experiment 3. The lower panel of Figure 6 shows QE scores
by case.
The main finding of Experiments 1 and 2 is that physical identity of
bilateral redundant stimuli is not required to produce substantial bilateral
redundancy gain. There was no difference between accuracy on bilateral
trials when the two stimuli were physically identical and accuracy on bilateral
trials when the two stimuli were different in case or in case and font.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
35
8 0 % -
7 0 % -
a >
o>
•£ 6 0 % -
6)
U
£ 5 0 % -
C L
2 4 0 % -
u i
_ 3 0 % -
c o
o
H 2 0 % -
1 0 % -
LVF/RH RVF/LH BVF-Con
Visual Field/Hemisphere
0.80
0 . 6 0 -
o 0.40-
< 0
L U
O
0 .2 0 -
0 . 0 0 -
LVF/RH RVF/LH BVF-Con
Visual Field/Hemisphere
Figure 6. Upper panel: Percentage of errors for three visual field conditions
in Experiment 2 by case/font of stimuli. Lower panel: Qualitative error (QE)
scores for three visual field conditions in Experiment 2 by case/font of
stimuli. (LVF/RH = left visual field/right hemisphere; RVF/LH = right visual
field/left hemisphere; BVF-Con = Bilateral Consistent.)
Uppercase
O Lowercase
\
1
X
T
O
X
s
Uppercase
- o -
Lowercase
t---------------------r
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
36
However, it remains possible that the pattern of results found in Experiments
1 and 2 may be influenced by the very different error rates for uppercase
versus lowercase stimuli. Specifically, the failure to find a difference between
performance for consistent and inconsistent bilateral trials may be partially
attributable to the difference in difficulty for recognizing uppercase and
lowercase stimuli, which may result in unequal contributions of the two
stimuli (and thus hemispheres) to performance. In order to test this
possibility, for Experiment 3 stimulus exposure durations were adjusted
separately for each case, to reduce the difference in error rates between
uppercase and lowercase trials. An additional question was whether
bringing error rates for all conditions closer to 50% would also lead to
increased bilateral redundancy gain, more in line with that predicted by
statistical summation, thus replicating Marks and Hellige (1999)
Experiment 1.
EXPERIMENT 3
Design and Method
Participants
Twelve men and 12 women university students participated in
Experiment 3. They were recruited from the same pool as Experiment 1 and
selected in the same manner. Participants ranged in age from 18 to 22
years (M = 19.5, SD= 1.2).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
37
Procedure
Experiment 3 was a replication of Experiments 1 and 2, except for the
manner of adjusting stimulus exposure duration. Because Experiments 1
and 2 showed that uppercase stimuli were much easier to identify than
lowercase stimuli, the exposure duration was separately adjusted for
uppercase and lowercase stimuli as follows: on the first practice trial, the
exposure duration was set at 180 ms for stimuli in both cases. Thereafter,
each correct response shortened the duration of the next trial in the same
case by 15 ms, and each incorrect response lengthened the duration of the
next trial in the same case by 15 ms. The maximum exposure duration was
set at 210 ms in order to avoid eye movements.
On Bilateral Inconsistent trials, the exposure duration was separately
calculated for the stimulus in each visual field, using the current uppercase
exposure duration for the uppercase stimulus and the current lowercase
exposure duration for the lowercase stimulus. When the exposure duration
for the uppercase CVC had elapsed, a pattern mask appeared in its place,
while the lowercase CVC continued to be presented for the appropriate
remaining time. The lowercase CVC was then replaced by a pattern mask.
The time from the onset of the pattern mask replacing the uppercase CVC to
the offset of both pattern masks was 210 ms. Performance on Bilateral
Inconsistent trials did not affect the exposure duration of subsequent trials
for either case. The exposure duration for each case for the final practice trial
was used for the first experimental trial, and the adjustment procedure
continued throughout the experiment.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38
Results and Discussion
The trial-by-trial adjustment of exposure duration resulted in
approximately equal durations for LVF/RH trials (M = 46 ms), RVF/LH trials
(M = 48 ms), and Bilateral Consistent trials (M = 49 ms) and produced an
overall CVC identification error rate of 41.2% (SD = 2.5%). Stimulus exposure
durations for uppercase trigrams (M = 38 ms) were shorter than for
lowercase trigrams (M = 57 ms), and so on Bilateral Inconsistent trials the
mean exposure duration for the uppercase trigram was 38 ms and the mean
exposure duration for the lowercase trigram was 57 ms. The separate
adjustment of exposure duration for uppercase and lowercase stimuli
reduced, but did not eliminate, the difference in total error percentage
between uppercase stimuli (M = 44.1%, SD = 5.1%) and lowercase stimuli
(M = 48.0%, SD = 1.8%), t(23) = 3.63, p < .01. The inability to completely
equate performance for uppercase and lowercase stimuli is due to
constraints imposed by the refresh rate of our computer monitor. Because of
equipment limitations, we are unable to reduce stimulus duration below 15
ms (one screen refresh cycle). Some participants find the CVC task
sufficiently easy (particularly for uppercase stimuli) that they achieve error
rates well below 50% for some conditions even with exposure durations a
this minimum. Nonetheless, the remaining 3.9% difference between
uppercase and lowercase error rates was much smaller the 20.8%
difference found in Experiment 2.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39
Total Error Percentage
Total error percentages were submitted to an AN OVA with gender as
the between-subjects factor and visual field condition (LVF, RVF, Bilateral
Consistent, Bilateral Inconsistent) as the within-subjects factor. There was a
main effect of gender, F(1,22) = 5.47, MSE = 19.56, p < .01. In contrast to
Experiment 2, females (M = 40.1%, SD = 2.4%) committed fewer errors than
males (M = 42.2%, SD = 2.1%). Again, there were no significant interactions
involving gender, F < 1. Because the main effects and interactions involving
gender in Experiments 1 to 3 do not replicate from one experiment to the
next, and because they have not been found in previous CVC experiments, it
seems unlikely that they are reliable or meaningful.
As shown in the upper panel of Figure 7, the error rates were not
equal for the four visual field conditions, F(3,66) = 65.80, MSE = 79.52,
p < .001. Subsequent pairwise comparisons showed that the error rate was
significantly smaller on RVF/LH trials (M = 46.2%, SD = 9.0%) than on
LVF/RH trials (M = 59.6%, SD = 10.5%) and significantly smaller on both
Bilateral Consistent trials (M = 32.4%, SD = 4.9%) and Bilateral Inconsistent
trials (M = 26.6%, SD = 6.7%) than on either type of unilateral trial, ps < .001.
If physical identity is important for bilateral redundancy gain, then we would
expect performance on bilateral trials in which both stimuli were in the same
case and font (Bilateral Consistent) to be better than that on bilateral trials in
which the two stimuli were in a different case and font (Bilateral
Inconsistent). However, performance was actually significantly better for
inconsistent than for consistent trials, t(23) = 3.88, p < .001. Eta squared for
total error percentages was .40. The accuracy difference of 5.8% in favor of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
40
7 0 %
g , 6 0%
a
C
4)
O
a, 5 0 % -
a,
L.
o
im
L o
4 0 % -
U J
n
o
I-
3 0 % -
2 0 %
LVF/RH RVF/LH BVF-Con BVF-lneon
Visual Field/Hemisphere
0.6 0
0 . 4 0 -
o
u
CO
UJ
a
0 . 2 0 -
LVF/RH RVF/LH BVF-Con BVF-lncon
Visual Field/Hemisphere
Figure 7. Upper panel: Percentage of errors for each of the four visual field
conditions in Experiment 3. Lower panel: Qualitative error (QE) scores for
each of the four visual field conditions in Experiment 3. (LVF/RH = left visual
field/right hemisphere; RVF/LH = right visual field/left hemisphere; BVF-
Con = Bilateral Consistent; BVF-lncon = Bilateral Inconsistent.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
41
inconsistent trials is puzzling, but it clearly does not support the hypothesis
that physical identity is necessary or even helpful for bilateral redundancy
gain.
The mean bilateral error rate of 29.5% for Experiment 3 was not
significantly different from the error rate predicted by the statistical
summation formula of 27.0%, f(23) = 1.46, p = .16, yielding a bilateral
redundancy gain of 16.7% (calculated as the difference between the RVF/LH
error rate and the mean bilateral error rate). This level of bilateral redundancy
gain replicates that found by Marks and Hellige (1999), but is in contrast with
the results of the present Experiments 1 and 2. Perhaps the uppercase XXX
noise stimulus was more effective in the context of the additional differences
between uppercase and lowercase stimuli. Also, because stimulus
exposure durations were separately adjusted for uppercase and lowercase
stimuli, performance for all conditions was closer to 50%, as in Marks and
Hellige (1999), so there were fewer instances of floor and ceiling effects.
QE Scores
In order to examine hemispheric differences in the qualitative nature
of processing, QE scores were computed for each visual field condition in
the manner described earlier: QE = (LE - FE) / (FE + LE + OE). As shown in
the lower panel of Figure 7, the QE scores were not equal for the four visual
field conditions, F(3,66) = 12.06, MSE = .033, p < .001. Subsequent pairwise
comparisons showed that the QE score was significantly smaller on RVF/LH
trials (M = .18, SD = .24) than on LVF/RH trials (M = .49, SD = .22) and
intermediate on both Bilateral Consistent trials (M = .30, SD = .25) and
Bilateral Inconsistent trials (M = .33, SD = .36). All four visual field conditions
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
42
differed significantly, ps < .05, except that Bilateral Consistent and Bilateral
Inconsistent QE scores did not differ from each other. For LVF/RH, RVF/LH,
and Bilateral Consistent trials, this pattern of QE scores is consistent with
the results of earlier CVC experiments (see Table 2).
The finding of no significant difference between Bilateral Consistent
and Bilateral Inconsistent QE scores is similar to the results of Experiment
1. It does not replicate the results of Experiment 2, in which Bilateral
Inconsistent QE scores (M = .36) were significantly greater than Bilateral
Consistent QE scores (M = .23). In Experiment 3, there was only a small
(and statistically non-significant) difference, although it was in the same
direction. In Experiment 1, Bilateral Consistent and Bilateral Inconsistent QE
scores were virtually identical (M = .31). Because it does not seem likely that
there is a reliable difference, it may be premature to speculate on the
reasons for an effect which may not exist.
Case and Font Effects
To examine the effects of the case and font manipulation, an
additional ANOVA was performed with visual field (LVF, RVF, BVF) and
case/font (uppercase Trebuchet MS, lowercase Rockwell) as the within-
subjects factors. The case/font effects were not of theoretical interest here,
but were included to check for any interaction with visual fie!d(ln this
analysis, only Bilateral Consistent trials were included, because stimulus
format was mixed in the Bilateral Inconsistent trials).
For total error percentage, there was a main effect of visual field,
F(2,46) = 45.95, MSE = 193.21, p < .001 (descriptive statistics are reported
with the previous analysis). There was also a main effect of case,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
43
F( 1,23) = 13.17, MSE = 41.48, p < .01. However, there was no significant
interaction of visual field and case, F < 1. The upper panel of Figure 8 shows
total error percentages by case. As discussed above, the 3.9% difference
between the uppercase and lowercase error rates is much smaller than the
20.8% difference found in Experiment 2. Thus it appears unlikely that the
pattern of results found for Bilateral Consistent versus Bilateral Inconsistent
trials in Experiments 1 and 2 is a result of the difference in error rates
between uppercase and lowercase stimuli.
For QE scores, there was a main effect of visual field, F(2,46) = 23.61,
MSE = .047, p < .001 (descriptive statistics are reported with the previous
analysis). There was also a main effect of case, F(1,23) = 13.26, MSE = .085,
p < .01. The lower panel of Figure 8 shows QE scores by case. Uppercase
QE scores (M = .42, SD = .21) were substantially higher than lowercase QE
scores (M = .25, SD = .24), but there was no significant interaction of visual
field and case, F < 1. This case effect on QE scores is different from the
results for Experiment 1, where only case was manipulated and no
significant difference was found between uppercase QE scores and
lowercase QE scores. However, it replicates a similar finding for Experiment
2, where both QE scores {M = .42, SD = .21) were substantially higher than
lowercase QE scores (M = .25, SD = .24), but there was no significant
interaction of visual field and case, F < 1. This case effect on QE scores is
different from the results for Experiment 1, where only case was manipulated
and no significant difference was found between uppercase QE scores and
lowercase QE scores. However, it replicates a similar finding for Experiment
2, where both case and font were also manipulated. In order
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44
7 0 %
— Uppercase
a > 6 0%
U i
( 0
£
Q )
O
G )
C L
5 0%
I 4 0 %
UJ
o 3 0 %
- - o - - Lowercase
2 0 % ------- 1 ----------------1 ----------------r
LVF/RH RVF/LH BVF-Con
Visual Field/Hemisphere
Uppercase
0.60-
- - o - - Lowercase
1 ' 0.40
o
o
CO
UJ
O
0.00-
LVF/RH RVF/LH BVF-Con
Visual Field/Hemisphere
Figure 8. Upper panel: Percentage of errors for three visual field conditions
in Experiment 3 by case/font of stimuli. Lower panel: Qualitative error (QE)
scores for three visual field conditions in Experiment 3 by case/font of
stimuli. (LVF/RH = left visual field/right hemisphere; RVF/LH = right visual
field/left hemisphere; BVF-Con = Bilateral Consistent; BVF-lncon = Bilateral
Inconsistent.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
to explore a possible case effect on QE scores further, the percentage of
each type of error was calculated by case, collapsed across visual field, for
Experiments 2 and 3 (see Figure 9). Recall that the QE Score is a ratio which
depends on the rates of first-letter errors (FEs), last-letter errors (LEs) and
other errors (OEs). In both experiments, lowercase trigrams produced higher
rates of FEs and OEs than uppercase trigrams. However, there were not
proportionately more LEs for lowercase trigrams. The reasons for this are
unclear.
The main finding of Experiments 1 to 3 is that substantial bilateral
redundancy gain was realized even when the two copies of the CVC were not
physically identical. It should be noted, however, that although they are not
physically identical, different case and font representations of the same letter
may be physically similar. For example, uppercase and lowercase P, K, O,
and S are very similar, and this is true even though care was taken to choose
two fonts that were quite dissimilar. Also, even letters whose uppercase and
lowercase forms are less similar, like F and T, still have some visual
features in common. Finally, for all letters, there may be strong
interconnections between the visual representations in the brain of their
uppercase and lowercase forms. According to some theorists, abstract letter
identities, which representation in visual word recognition (e.g., Besner,
Coltheart, & Davelaar, 1984; Evett & Humphreys, 1981). They may develop
because different forms of the same letter occur in the same visual contexts
(Polk & Farah, 1997). For example, if the letter form a appears between c and
t in the word cat, then the form A will also occur in a visually similar context
(between C and Tin CAT). Statistical regularities in the occurrence of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
46
3 0 %
--------
O
UJ 2 0 % -
« * —
o
a >
05
(0
+ ■ >
c
o
s ™
.T
Uppercase
e u
0.
o- • Lowercase
o %
OE FE LE
Error Type
3 0%'
(A
O
u.
m 2 0 %
< 4 —
O
0 )
a*
n
c 10%
a >
a
0 3
a .
o %
FE
Uppercase
■ o - - Lowercase
LE ^0E
Error Type
Figure 9. Percentage of first-letter errors (FE), last-letter errors (LE) and other
errors (OE) by case of stimuli, collapsed across visual fields in Experiment 2
(upper panel) and Experiment 3 (lower panel).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
47
letters in similar contexts could lead to Hebbian learning in neural networks.
According to other theorists, it is also possible that participants engage in
fast visual generation of lowercase forms when they are presented with
uppercase letters (Boles, 1992).
Because it appears that uppercase and lowercase representations of
letters may be quite similar at early stages of visual processing, the finding
of bilateral redundancy gain for Bilateral Inconsistent trials in Experiments 1
to 3 might be the result of visual similarity, and therefore due to sharing of
low level perceptual information between the two hemispheres. To test for
bilateral redundancy gain when the sensory information available to the two
hemispheres is more distinctive, we wanted to find two kinds of trigram
stimuli which would lead to the same response, but which would share less
perceptual information than uppercase and lowercase CVCs. Some
previous experiments have investigated the processing of numbers
represented in various formats, including, dot patterns, dials, and bar graphs
(e.g., Boles, 1986; Hatta & Dimond, 1980). Arabic digits and dot patterns
were selected for the present series of experiments because they could
easily be arranged in trigrams representing the same three-digit numbers.
Before we could test whether bilateral redundancy gain would be found
when one hemisphere received a digit trigram and the other a dot trigram
representing the same number, it was important to determine whether the
presentation of two digit trigrams or two dot trigrams would lead to
substantial redundancy gain. Therefore, in Experiment 4, digit trigrams
served as the target stimuli; in Experiment 5, structured dot pattern trigrams
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
48
were used; and in a final experiment, both digit and dot pattern trigrams were
presented.
EXPERIMENT 4
Bilateral redundancy gain has been demonstrated using linguistic
material, including words (lexical decision: e.g., Hasbrooke & Chiarello,
1998; Mohr, Pulvermuller, & Zaidel, 1994; Zaidel & Rayman, 1994), nonword
trigrams (CVCs, CCCs) (e.g., Eng & Hellige, 1994; Hellige & Marks, in press;
Hellige et al., 1989; Luh & Wagner, 1997; Marks & Hellige, 1999), and
symbol trigrams (SSS) (Hellige & Marks, in press; Kim, 1996; Luh & Wagner,
1997). However, it had not been demonstrated for numbers represented as
digits. There is some evidence of an RVFA for the identification of digits
(Boles, 1986), especially when more than one digit is presented at a time
(e.g., Dimond & Beaumont, 1971; Hatta & Dimond, 1980). Experiment 4 was
designed to test whether the bilateral gain in performance found for CVCs
and other linguistic materials would extend to numbers. Because digits are
essentially verbal material (Boles, 1986), bilateral redundant presentation of
digits might be expected to produce a redundancy gain in accuracy.
It is also interesting to compare qualitative performance for digit
trigrams with the patterns of qualitative performance so consistently found
for CVC trigrams. If hemispheric strategies for the deployment of attention
depend on the nature of the stimulus material, then performance may be
somewhat different for digit trigrams than for CVCs (cf. Hellige & Marks, in
press; Kim, 1996; Luh & Wagner, 1997). The relatively lower QE score found
for the RVF/LH is usually attributed to the phonetic processing abilities of the
left hemisphere, which may permit it to treat the CVC in a more holistic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
49
manner (for discussion, see Eng & Hellige, 1994; Hellige et al., 1989; Levy et
al., 1983; Luh & Levy, 1995; Luh & Wagner, 1997). Note that a CVC syllable
has a pronunciation as unit that is different from the naming of individual
letters in sequence. This is not true, at least not to the same degree and in
the same manner, for three-digit numbers, so QE differences between the
two unilateral viewing conditions may be less for number trigrams.
Design and Method
Participants
Twelve men and 12 women served as participants in Experiment 4.
They were recruited from the same pool as Experiment 1 and selected in the
same manner. Participants ranged in age from 18 to 36 (M = 21.8, SD = 4.4).
Apparatus and stimulus materials
The same apparatus and software were used for Experiment 4 as for
Experiment 2. Stimuli for Experiment 4 consisted of 48 digit trigrams
constructed from the digits 1 to 6 and an additional 12 digit trigrams for the
practice trials. Each of the 6 digits appeared approximately equally often in
each of the three positions: first, second, and third. The digits were
presented in Geneva 31 pt. font, and each digit was centered in a box
measuring 1.3° of visual angle on a side. In addition, a trigram consisting of
three 8's was used as a noise stimulus. The digit stimuli were designed to
correspond to the dot pattern stimuli to be used in Experiment 5, which were
in turn designed to resemble dice. The digit and dot stimuli were made
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50
comparable so that they could be combined in Experiment 6. The same
fixation cross and mask were used as in Experiment 2.
The trigrams were oriented vertically, and displayed in black against a
white background on the computer monitor. Each trigram subtended
approximately 1.3 0 of visual angle horizontally and 4.6 0 of visual angle
vertically when viewed at a distance of approximately 57 cm from the screen.
The nearest edge of the trigram was 2.4° from the center of the screen.
Participants used a chin rest as well as a forehead bar to limit head
movement and maintain viewing distance. Samples of the stimuli and trial
types are illustrated in Figure 10.
A practice block of 36 trials was followed by three blocks of
experimental trials, with 49 trials each. The first trial in each block served as
a filler trial and thus was not scored. Across the set of 144 scored trials,
each of the 48 stimuli appeared once in the LVF, once in the RVF, and once
bilaterally (with the same stimulus in each visual field). The order of trials
was randomized for each participant, with the constraints that an individual
trigram was not repeated immediately and no more than four consecutive
trials were in any given visual field condition.
Procedure
Each participant was tested individually in one session, during which
he or she was seated in a quiet, dimly lit testing room. Participants were told
that on each trial a number would be presented in the left visual field (with a
string of three 8's in the right visual field); in the right visual field (with a string
of three 8's in the left visual field); or the same number would appear in both
visual fields simultaneously. They were instructed to ignore the 8's, because
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
51
1 8
2
+
8
4 8
V________________________ /
8 1
8
+
2
8 4
J
Unilateral LVF/RH Unilateral RVF/LH
1
+
B ilateral
Figure 10. Samples of stimuli and trial types for Experiment 4. Stimuli are
depicted accurately, but the displays are not drawn to scale. (LVF/RH = left
visual field/right hemisphere; RVF/LH = right visual field/left hemisphere.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
52
there would never be an 8 in the target stimulus. The experimenter
emphasized that when there were two copies of the number, they would
always be the same. Participants were instructed to maintain fixation on the
fixation cross for the duration of the trial and to report the target stimulus as a
series of three single digits, from top to bottom. During the instructions,
samples of the stimuli were presented on the screen. After the practice
block, participants were asked whether they had any questions about the
task.
For each trial a tone was sounded and the fixation cross appeared in
the center of the screen for 1000 ms. Then the stimuli appeared for the
appropriate exposure duration, followed immediately by a mask for 210 ms.
For the first practice trial, the exposure duration was set at 180 ms. The
exposure duration on subsequent trials was adjusted upwards by 15 ms if
the response was incorrect or downwards by 15 ms if it was correct, without
regard to the visual field condition. The minimum exposure duration was 15
ms (one screen refresh) and the maximum exposure duration was 210 ms.
The effect of this method of adjustment is that mean percentage of errors is
constrained to be near 50%, but that performance is allowed to vary across
the three visual field conditions.
Results and Discussion
The trial-by-trial adjustment of exposure duration resulted in
approximately equal durations for all three visual field conditions (LVF/RH
M = 54 ms, RVF/LH M = 55 ms, BVF M = 56 ms) and produced an overall
error rate of 46.5% (SD = 4.2%).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
Total Error Percentage
Total error percentages were submitted to an AN OVA with gender as
the between-subjects factor and visual field (LVF, RVF, BVF) as the within-
subjects factor. There was no significant effect of gender, F < 1, and no
significant interaction of gender and visual field, F < 1. As shown in the upper
panel of Figure 11, the error rates were not equal for the three visual field
conditions, F(2,44) = 75.16, MSE = 77.23, p < .001. Subsequent pairwise
comparisons showed that the error rate was significantly smaller on RVF/LH
trials (M = 45.8%, SD = 6.7%) than on LVF/RH trials {M = 62.4%, SD = 9.6%)
and significantly smaller on BVF trials (M = 31.3%, SD = 8.4%) than on either
type of unilateral trial, ps < .001.
The mean bilateral error rate of 31.3% for Experiment 4 was not
significantly different from the error rate predicted by the statistical
summation formula of 28.4%, t(23) = 1.53, p = .14. This yields a bilateral
redundancy gain of 14.4% (calculated as the difference between the RVF/LH
error rate and the BVF error rate), which replicates the level of gain found in
Marks and Hellige (1999) Experiment 1, with digit trigrams instead of CVCs.
This pattern of performance is remarkably similar to that found for the
corresponding visual field conditions in Experiment 3, using CVCs. The
RVFA found for linguistic material was replicated with numbers represented
as digits, as was the finding of substantial bilateral redundancy gain for trials
on which physically identical items were presented in both visual fields
(Bilateral Consistent trials in Experiment 3 and BVF trials in the current
experiment). The task of naming digit trigrams may invoke the same
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
54
7 0 %
® 6 0 % -
0 5
a
+ ■ *
c
0
H 5 0 % -
a
b.
£ 4 0 % -
U J
«
o 3 0 % -
2 0 %
0.50
0 . 4 5 -
| 0 . 4 0 -
o
a
t n
o 0 . 3 5 -
0 . 3 0 -
0.25
Figure 11. Upper panel: Percentage of errors for each of the three visual field
conditions in Experiment 4. Lower panel: Qualitative error (QE) scores for
each of the three visual field conditions in Experiment 4. (LVF/RH = left visual
field/right hemisphere; RVF/LH = right visual field/left hemisphere;
BVF = Bilateral.)
i
1 --------------- 1 ----------------r~
LVF/RH RVF/LH BVF
Visual Field/Hemisphere
1 1 --------------------------------------------1 —
LVF/RH RVF/LH BVF
Visual Field/Hemisphere
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
55
processes as the task of naming CVC trigrams, because digits are
essentially verbal material (Boles, 1986).
QE Scores
In order to examine hemispheric differences in the qualitative nature
of processing, QE scores were computed for each visual field condition in
the manner described earlier: QE = (LE - FE)./ (FE + LE + OE). As shown in
the lower panel of Figure 11, the QE scores were not equal for the three
visual field conditions, F(2,44) = 4.66, MSE = .026, p < .05. Subsequent
pairwise comparisons showed that the QE score was significantly smaller
on RVF/LH trials (M = .29, SD = .27) than on LVF/RH trials (M = .43, SD = .23),
p < .05. However, the QE score for BVF trials {M = .41, SD = .36) did not differ
significantly from that for LVF/RH trials. This pattern of QE scores is similar
to that reported for some earlier CVC experiments (see Table 2), insofar as
the QE score is lower for the RVF/LH trials than for the LVF/RH trials and
numerically intermediate for the BVF trials. However, it is different from the
pattern found for CVCs in that the RVF/LH QE score for Experiment 4 was
higher than that reported for the CVC studies summarized in Table 2 and the
present Experiments 1 to 3. A one-sample f-test comparing the mean
RVF/LH QE score for Experiment 4 with the means for the 16 previous
studies (M = .21, SD = .08) showed a statistically significant difference,
f(15) = -4.24, p < .01. Although the RVF/LH QE was higher than has been
found for CVCs, the LVF/RH QE score was not different from that found for
CVCs (M= .44, SD= .09), t< 1.
Also, as suggested above, digit trigrams show less difference
between LVF/RH and the RVF/LH QE scores than CVCs do. The difference
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
56
between unilateral QE scores {M = .13, SD = .20 for Experiment 4) is smaller
than that reported for the CVC studies (M = .23, SD = .08), f(15) = 5.31,
p < .001. The finding of a higher RVF/LH QE score may indicate a slower or
less holistic deployment of attention across the elements of the digit
trigrams than is the case for CVCs, possibly because of the absence of a
higher-order unitary representation of the digit trigram analogous to the
pronunciation of the CVC as a unit. Hemispheric strategies for the
deployment of attention are apparently sensitive to the nature of the stimuli
and task demands (cf. Hellige & Yamauchi, 1999). Nonetheless, it is notable
that RVF/LH QE scores are significantly lower than LVF/RH QE scores for
digit trigram identification, perhaps indicating that the left hemisphere
distributes attention more rapidly or holistically across the three elements of
the trigram than the right hemisphere does.
The main finding of Experiment 4 for the purposes of the present
series of experiments is that bilateral redundant presentation of digit
trigrams produces substantial bilateral gain, similar to that found with CVCs.
In order to extend the investigation of the locus of bilateral gain to a
completely different type of stimulus material, Experiment 5 was designed to
test performance on the recognition of numerosity from dot patterns. Another
purpose was to provide an alternative physical format requiring the
recognition of a three-digit number, in addition to the digit trigrams used in
Experiment 4. (Experiment 6 then combined the two number formats to
investigate whether or not physical identity is necessary for bilateral gain in
the recognition of numbers.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
57
EXPERIMENT 5
Some previous research has addressed the ability to detect
numerosity in dot displays, but usually the dots are arranged randomly
(Kimura, 1966). The participant may be required to report the number of dots
(Boles, 1986) or to manipulate the quantity in some calculation (Berger &
Landolt, 1990) or to make a comparison on the basis of numerosity
(Norman, Jeeves, Milne, & Ludwig, 1992). Klein and Mclnnes (1988)
presented numbers represented as digits, number words, or dot patterns
arranged like dice and required participants to make an odd/even judgment.
They found an interaction of stimulus type with visual field, with words
tending to show an RVFA while digits and dots tended to show an LVF/RH
advantage. Boles (1986) reported that there is generally an LVF/RH
superiority for dot enumeration. Trigrams consisting of dot patterns have not
been used as stimuli in past experiments, so it was unclear whether there
would be a visual field advantage in this task.
In order to make the task a recognition rather than a counting task, the
stimuli were constructed in the familiar pattern of six-sided dice. People
generally show an ability to apprehend the numbers represented on the
faces of several dice at a time, so it was expected that they would be able to
perform this task. There are apparently differences in the ways that people
process various representations of numerical quantities. For example,
digits, word names, random dot clusters, bar graphs, dials, and clock faces
have been used in various experiments (e.g., Berlucchi, Brizzolara, Marzi,
Rizzolatti, & Umilta, 1979; Boles, 1986; Cohen, 1975; Hatta, 1978). The more
visual representations of quantity usually lead to an LVF/RH advantage
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
58
(LVFA), in contrast to an RVFA for digits. The issue to be explored was
whether redundancy gain would be observed with trig rams of structured dot
clusters arranged in a familiar format. An additional aim was to compare the
qualitative pattern of performance found for dot clusters to those found for
digit trigrams and CVCs.
Design and Method
Participants
Ten men and 14 women served as participants in Experiment 5. They
were recruited from the same pool as Experiment 1 and selected in the
same manner. Participants ranged in age from 18 to 35 (M = 22.67,
SD = 4.3).
Apparatus and stimulus materials
The same apparatus and software were used for Experiment 5 as for
Experiment 2. Stimuli for Experiment 5 were the same as those for
Experiment 4, except that the digits 1 to 6 were replaced by dot patterns in
which each dot subtended 0.3°. Dot patterns were arranged in the standard
dice format, except for the 8’s used for the noise stimulus, which were
arranged in a domino pattern. Samples of stimuli and trial types are
illustrated in Figure 12. No mask was used for the dot pattern trigrams,
because pilot studies indicated that these stimuli were too difficult to identify
at brief exposure durations when masked.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59
9#9
• • •
•
•
+
* 9 *
9
• • •
• •
9 # 9
9%
• •
9 9
V________________________ J
Unilateral LVF/RH
9#9
9*9 9
9*9
v
9*9
+
9
9
• • • 9 9
mz®
9*9 9 9
V_________________; _______ /
Unilateral RVF/LH
B ilateral
Figure 12. Samples of stimuli and trial types for Experiment 5. Stimuli are
depicted accurately, but displays are not drawn to scale. (LVF/RH = left visual
field/right hemisphere; RVF/LH = right visual field/left hemisphere.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
60
Procedure
The procedure for Experiment 5 was the same as for Experiment 4.
Results and Discussion
The trial-by-trial adjustment of exposure duration resulted in
approximately equal durations for all three visual field conditions (LVF/RH
M = 104 ms, RVF/LH M = 103 ms, BVF M = 106 ms) and produced an overall
error rate of 46.8% (SD = 7.9%). The exposure durations were nearly twice
as long as those required to produce a comparable level of performance for
digit trigrams or CVCs, indicating that the recognition of dot patterns is a
more difficult task for participants. This may be partly because these stimuli
are less familiar.
Total Error Percentage
Total error percentages were submitted to an AN OVA with gender as
the between-subjects factor and visual field (LVF, RVF, BVF) as the within-
subjects factor. There was no significant effect of gender, F(1,22) = 1.48,
MSE = 183.00, p = .24, and no significant interaction of gender and visual
field, F(2,44) = 1.33, MSE = 57.52, p = .27. As shown in the upper panel of
Figure 13, the error rates were not equal for the three visual field conditions,
F(2,44) = 35.68, MSE = 57.52, p < .001. Subsequent pairwise comparisons
showed that the error rates on RVF/LH trials (M = 52.2%, SD = 9.7%) and
LVF/RH trials (M = 52.1%, SD = 10.2%) did not differ significantly, p = .97, and
both were significantly larger than the error rate on BVF trials (M = 36.1%,
SD = 10.2%), ps < .001. There was no visual field advantage for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
61
5 5 %
S’ 5 0%
( 0
c
o
a 4 5%
L U
_ 4 0 %
( 0
o
H
3 5 %
LVF/RH RVF/LH BVF
Visual Field/Hemisphere
0.55
0 . 5 0 -
w
£
o
o
c n
in
o
0 . 4 5 -
0.40
Figure 13. Upper panel: Percentage of errors for each of the three visual field
conditions in Experiment 5. Lower panel: Qualitative error (QE) scores for
each of the three visual field conditions in Experiment 5. (LVF/RH = left visual
field/right hemisphere; RVF/LH = right visual field/left hemisphere;
BVF = Bilateral.)
I
1 -------------- 1 -------------- r~
LVF/RH RVF/LH BVF
Visual Field/Hemisphere
x
t---------------------1-------------------- r
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
62
the recognition of dot pattern trigrams, in contrast to the RVFA found so
consistently for CVCs and found in Experiment 4 for digit trigrams. The
failure to find an LVFA (cf. Boles, 1986) may be due to the fact that
Experiment 4 was designed to be a recognition task rather than a dot
enumeration task.
Although error rates for bilateral trials were much lower than error
rates for unilateral trials, they were still significantly higher than would be
predicted by the statistical summation formula, f(23) = 5.60, p < .001. The
statistical summation formula yields a predicted bilateral error rate of 27.4%,
compared to the mean bilateral error rate in Experiment 5 of 36.1%. It is
possible that the noise stimulus, because it was visually very different from
the standard six-sided die faces, was not entirely effective in leading
observers to divide their attention between the two visual fields to a similar
extent for unilateral and bilateral trials. Nonetheless, the bilateral redundancy
gain of 16.0% (calculated as the difference between the RVF/LH error rate
and the BVF error rate) was substantial.
QE Scores
In order to examine hemispheric differences in the qualitative nature
of processing, QE scores were computed for each visual field condition in
the manner described earlier: QE = (LE - FE) / (FE + LE + OE). As shown in
the lower panel of Figure 13, the QE scores were not equal for the three
visual field conditions, F(2,44) = 7.05, MSE = .020, p < .05. Subsequent
pairwise comparisons showed that the QE score was significantly smaller
on LVF/RH trials (M = .44, SD = .21) than on RVF/LH trials (M = .50, SD = .22),
p < .05. However, the QE score for BVF trials (M = .49, SD = .27) did not differ
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
63
significantly from that for RVF/LH trials. This pattern of QE scores is quite
different from that found for digit trigrams, where the LVF/RH QE score was
higher than the RVF/LH QE score. The relatively lower LVF/RH QE score may
reflect the more spatial nature of the stimuli. It is possible that the right
hemisphere’s superiority for spatial tasks allows it to deploy attention more
holistically across the three elements of the trigram than the left hemisphere
is able to do, in contrast to the findings for CVCs and digit trigrams, but
similar to some previous findings for trigrams of familiar non-letter symbols
(Kim, 1996; Luh & Wagner, 1997; but cf. Hellige & Marks, in press). The QE
scores for the structured dot stimuli in Experiment 5 differ from those found
in CVC experiments in that the RVH/LH QE score is much higher (just as
was the case in Experiment 4 for digit trigrams). Again, this elevation may
reflect the absence of a higher-order unitary response for number trigrams,
compared to the phonetic pronunciation available for CVCs (Hellige &
Yamauchi, 1999).
The main finding of Experiment 5 for the purposes of the present
series of experiments is that bilateral redundant presentation of dot pattern
trigrams produces substantial bilateral gain, similar to that found with CVCs
and digit trigrams. It can therefore legitimately be investigated whether
bilateral gain will also be found when one hemisphere receives a digit
trigram and the other a dot pattern trigram representing the same three-digit
number.
EXPERIMENT 6
Experiment 6 was the same as Experiments 4 and 5, with the addition
of a physical identity manipulation. Three-digit numbers were represented
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
64
as either digit trigrams or as structured dot pattern trigrams. Bilateral
redundant trials included a consistent condition, in which both trigrams were
presented in the same format, and an inconsistent condition, in which the
two trigrams, although representing the same number, differed in format.
The logic of Experiment 6 is similar to that of Experiments 1 to 3, but there is
an additional element. In Experiments 1 to 3, the same CVC was
represented either in uppercase or in lowercase, so the two stimuli on
inconsistent trials were not physically identical. However, both were made up
of letters, so the decoding process was the same for both of them. In
Experiment 6, the two stimuli both represented a three-digit number, but the
two representations were somewhat more different than uppercase and
lowercase letters. The decoding process for structured dot patterns requires
the detection of numerosity, whereas the digit trigrams are interpreted as
orthographic symbols for the names of the numbers they represent. One
question to be addressed, then, is whether information in two different codes
can be combined to enhance performance. The coding difference and the
physical difference may also influence the strategy that emerges on Bilateral
Inconsistent trials. The pattern of performance observed for numerical
stimuli will be compared to that observed for linguistic stimuli in Experiments
1 to 3.
Design and Method
Participants
Ten men and 14 women served as participants in Experiment 6. They
were recruited from the same pool as Experiment 1 and selected in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
same manner. Participants ranged in age from 18 to 40 (M = 22.17,
SD = 4.7).
Apparatus and stimulus materials
The same apparatus and software were used as for the earlier
experiments. Stimulus materials for Experiment 6 consisted of the combined
materials for Experiments 4 and 5. In addition, Bilateral Inconsistent trials
were added representing the same numbers as digit trigrams in one visual
field and dot trigrams in the other visual field. Samples of stimuli and trial
types are illustrated in Figure 14.
Procedure
The procedure for Experiment 6 was the same as for Experiments 4
and 5, except that additional trials were added to insure a sufficient number
of trials of each type for analysis. Also, because the mean stimulus duration
required for dot trigrams in Experiment 5 was twice as long as that required
for digit trigrams in Experiment 4, stimulus exposure durations were
separately adjusted for digit trigrams and dot trigrams. The adjustment
procedure was similar to that used in Experiment 3 for uppercase and
lowercase stimuli. Digit trigrams were replaced by a mask at stimulus offset;
dot trigrams were not masked but were replaced by a blank field at offset
until the digit mask had been displayed for 210 ms.
A practice block of 36 trials was selected randomly from the 96
possible trials using the 12 practice number trigrams. The practice block
was followed by eight experimental blocks of 49 trials each. Each of the 48
number trigrams served as a target eight times, once in each of the trial
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
66
8
•
+
00
+
•
•
00
• • •
• • •
• •
• •
Unilateral LVF/RH Unilateral RVF/ LH
1 1
• •
2
+
2
•
•
+
•
4 4
• •
• •
• •
• •
Bilateral Consistent
Digit s
Bilat eral Consist ent
Dice
Bilat eral Inconsist ent
LVF Dig it s
1
• •
1
2
+
•
•
I •
«
+
2
4
• t
• 9
• •
• •
4
Bilat eral Inconsist ent
RVF Digits
Figure 14. Samples of stimuli and trial types for Experiment 6. Stimuli are
depicted accurately, but displays are not drawn to scale. (LVF/RH = left visual
field/right hemisphere; RVF/LH = right visual field/left hemisphere.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
67
types. The order of trials was randomized for each participant, with the
constraint that no individual target or trial type could immediately follow itself.
The first trial in each block served as a filler trial and thus was not scored,
leaving a total of 384 scored trials. Because there were so many trials,
participants were offered a longer break (up to five minutes) between the
fourth and fifth blocks of experimental trials.
Results and Discussion
The trial-by-trial adjustment of exposure duration resulted in
approximately equal durations for LVF/RH trials (M = 103 ms), RVF/LH trials
(M = 104 ms), and Bilateral Consistent trials (M = 102 ms) and produced an
overall error rate of 46.6% (SD = 6.2%). Stimulus exposure durations for digit
trigrams (M = 84 ms) were shorter than for dot trigrams (M = 122 ms), and so
on Bilateral Inconsistent trials the dot trigram was presented for 122 ms and
the digit trigram for 84 ms. The separate adjustment of exposure duration for
digit and dot trigrams resulted in nearly identical total error percentages for
digit trigrams (M = 48.8%, SD = 3.9%) and dot trigrams (M = 48.3%,
SD = 8.9%), f(23) = .36, p = .73.
Total Error Percentage
Total error percentages were submitted to an AN OVA with gender as
the between-subjects factor and visual field condition (LVF, RVF, Bilateral
Consistent, Bilateral Inconsistent) as the within-subjects factor. There was
no significant effect of gender, F(1,22) = 2.04, MSE = 139.88, p = .17 and no
significant interaction of gender and visual field, F < 1. As shown in the upper
panel of Figure 15, the error rates were not equal for the four visual field
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
68
conditions, F(3,66) = 26.38, MSE = 98.70, p < .001. Subsequent pairwise
comparisons showed that the error rate was significantly smaller on RVF/LH
trials (M = 51.52%, SD = 10.5%) than on LVF/RH trials (M = 58.98%,
SD = 13.5%) and significantly smaller on both Bilateral Consistent trials
(M = 35.20%, SD = 6.7%) and Bilateral Inconsistent trials (M = 40.28%,
SD = 10.3%) than on either type of unilateral trial, p < .001.
If physical identity is important for bilateral redundancy gain, then we
would expect performance on bilateral trials in which both stimuli were in the
same format (Bilateral Consistent) to be better than that on bilateral trials in
which the two stimuli were in a different format (Bilateral Inconsistent). In
fact, performance was better for consistent than for inconsistent trials, but
only marginally in the Newman-Keuls post hoc, p = .08. The accuracy
difference of 5.1% in favor of consistent trials may support the hypothesis
that physical identity can be helpful for bilateral redundancy gain. Because
this comparison is the crucial test of the hypothesis that physical identity is
necessary to produce bilateral redundancy gain, a separate f-test of the
Bilateral Consistent error percentages versus the Bilateral Inconsistent error
percentages was conducted, and it showed a statistically significant
difference, f(23) = -2.70, p = .01. Eta squared for the total error percentages
was .24.
To examine this consistency effect more closely, performance was
compared for four bilateral trial types (see Figure 16). Bilateral Consistent
trials were broken down into those in which both visual fields contained a
digit trigram and those in which both visual fields contained a dot pattern
trigram. Bilateral Inconsistent trials were broken down into those in which a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
69
7 0 %
6 0 % -
a
5 0 % -
U J
4 0 % -
o
(-
3 0%
LVF/RH RVF/LH BVF/Con BVF/lncon
Visual Field/Hemisphere
©
0 . 5 0 -
o
o
(/)
UJ
O
0 . 4 0 -
LVF/RH RVF/LH BVF/Con BVF/lncon
Visual Field/Hemisphere
Figure 15. Upper panel: Percentage of errors for each of the four visual field
conditions in Experiment 6. Lower panel: Qualitative error (QE) scores for
each of the four visual field conditions in Experiment 6. (LVF/RH = left visual
field/right hemisphere; RVF/LH = right visual field/left hemisphere; BVF-
Con = Bilateral Consistent; BVF-lncon = Bilateral Inconsistent.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
70
5 0 %
g , 45%
a *
o
S 4 0 % -
Q .
3 5 % -
< 8
o
3 0 %
T
2 5 % -■----------1 --------------------1 --------------------1 --------------------p
BVF-Dots BVF-Digits BVF-DN BVF-ND
Bilateral Configuration
Figure 16. Percentage of errors for each of four bilateral trial types in
Experiment 6. (BVF-Dots = Bilateral Consistent with dot pattern trigrams in
both visual fields; BVF-Digits = Bilateral Consistent with digit trigrams in both
visual fields; BVF-DN = Bilateral Inconsistent with dot pattern trigrams in the
left visual field and digit trigrams in the right visual field; BVF-ND = Bilateral
Inconsistent with digit trigrams in the left visual field and dot pattern trigrams
in the right visual field.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
71
digit trigram appeared in the LVF and a dot trigram appeared in the RVF, and
those in which a dot trigram appeared in the LVF and a digit trigram
appeared in the RVF. The error rate on Bilateral Consistent trials with digit
trigrams in both visual fields (M = .30, SD = .08) was the lowest and
accounted for all the excess bilateral redundancy gain for consistent trials,
all ps < .05. The error rate on Bilateral Consistent trials with dot trigrams in
both visual fields (M = .41, SD = .10) was not significantly different from the
error rate on either type of Bilateral Consistent trial, all ps > .10. Bilateral
Inconsistent trials with a dot pattern trigram on the left and a digit trigram on
the right (M = .37, SD = .10) had a lower error rate than Bilateral Inconsistent
trials with a digit trigram on the left and a dot pattern trigram on the right (M =
.44, SD= .14), p < .05.
Although error rates for bilateral trials were much lower than error
rates for unilateral trials, they were still significantly higher than would be
predicted by the statistical summation formula, f(23) = 4.07, p < .001. The
statistical summation formula yields a predicted bilateral error rate of 30.2%,
compared to a mean bilateral error rate in Experiment 6 of 37.7%. Perhaps
this is a result of the inclusion of dot trigrams, for which as suggested above,
the noise stimulus (three 8’s in a domino pattern) may not be entirely
effective in leading observers to divide their attention to a similar extent for
unilateral and bilateral trials. Nonetheless, the bilateral redundancy gain of
13.8% (calculated as the difference between the RVF/LH error rate and the
mean bilateral error rate) was substantial.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
72
QE Scores
In order to examine hemispheric differences in the qualitative nature
of processing, QE scores were computed for each visual field condition in
the manner described earlier: QE = (IE - FE) / (FE + LE + OE). As shown in
the lower panel of Figure 15, the QE scores were not equal for the four visual
field conditions, F(3,66) = 10.50, MSE = .015, p < .001. Subsequent pairwise
comparisons showed that the QE scores were significantly smaller on
LVF/RH trials (M = .42, SD = .20) and RVF/LH trials (M = .39, SD = .20) than
on either Bilateral Consistent trials (M = .49, SD = .21) or Bilateral
Inconsistent trials (M = .57, SD = .21). All four visual field conditions differed
significantly, ps < .05, except that LVF/RH and RVF/LH QE scores did not
differ from each other, p = .50. The higher QE score for Inconsistent
compared to Consistent Bilateral trials is similar to the findings for
Experiment 2, which as noted above were not replicated in Experiment 3. In
both Experiment 2 and Experiment 6, the elevated QE score for Bilateral
Inconsistent trials appears to arise because there were more last-letter
errors (LEs) (ps < .01) but the same level of first-letter errors (FEs) (ps > .20)
for Bilateral Inconsistent trials as compared with Bilateral Consistent trials
(see Figure 17).
Stimulus Format Effects
To examine the effects of the stimulus format manipulation, an
additional AN OVA was performed with visual field (LVF, RVF, BVF) and
stimulus format (digits trigrams, dot trigrams) as the within-subjects factors.
(In this analysis, only Bilateral Consistent trials were included, because
stimulus format was mixed in the Bilateral Inconsistent trials).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
73
3 0 %
0)
S b ,
O
£ 2 0 %
©
«
4 > >
£ 10%
a
k.
a s
o.
0 %
"•— Bilateral C on sisten t
- - O - - Bilateral In co n sisten t
T
FE
T
LE
i
OE
Error Type
40%
Bilateral C o n sisten t
- - O - - Bilateral In co n sisten t
3 0 % -
o
in
2 0 % -
as
o >
(8
S
o s
o
k .
0, 1 0 %-
o.
0 %
LE OE FE
Error Type
Figure 17. Percentage of first-letter (FE), last-letter (LE) and other errors (OE)
for Bilateral Consistent and Bilateral Inconsistent trials. Upper panel:
Experiment 2. Lower panel: Experiment 6.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
74
For total error percentage, there was a main effect of visual field,
F(2,46) = 29.92, MSE - 226.90, p < .001 (descriptive statistics are reported
with the previous analysis). There was no main effect of stimulus format,
F < 1. However, there was a significant interaction of stimulus format and
visual field, F(2,46) = 20.21, MSE = 72.72, p < .001. The upper panel of
Figure 18 shows total error percentages by stimulus format. Essentially, the
patterns of identification accuracy seen for digits in Experiment 4 and dots in
Experiment 5 were replicated when stimuli in the two formats were
intermixed. Digit trigrams continued to show an RVFA (LVF/RH M = 64.5%,
SD = 13.6%; RVF/LH M = 50.3%, SD = 9.6%), whereas dot trigrams showed
no visual field advantage (LVF/RH M = 53.2%, SD = 17.0%; RVF/LH
M = 52.3%, SD = 12.2%), and digit trigrams (BVF M = 30.2%, SD = 8.2%)
showed greater redundancy gain than dot trigrams (BVF M = 40.9%,
S D = 10.1%).
For QE scores, there was a main effect of visual field, F(2,46) = 3.30,
MSE = .034, p < .05 (descriptive statistics are reported with the previous
analysis). However, there was no main effect of stimulus format, F < 1, and
no interaction of visual field and stimulus format, F < 1. The lower panel of
Figure 18 shows QE scores by stimulus format. Most notably there was no
differences between LVF/RH {M = .44, SD = .29) and RVF/LH QE scores
(M = .40, SD = .28) for digit trigrams, p = .92. This is in contrast to the results
of Experiment 4, in which the RVF/LH QE score (M = .29) was significantly
lower than the LVF/RH QE score (M = .43) when only digits were presented.
Although it appears that the RVF/LH QE scores for dot pattern trigrams in
Experiment 6 (M = .41, SD = .21) are lower than they were in Experiment 4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
75
8 0 %
a, 7 0%
u >
m
**e
S 6 0 %
o
h m
©
a.
5 0%
s_
o
S n
£ 4 0 %
re
° 3 0%
2 0%
0.60
0 . 5 5 -
£ 0 . 5 0 -
o
o
C O
UJ
O 0 . 4 5 -
0 . 4 0 -
0.3 5
Figure 18. Upper panel: Percentage of errors for three visual field conditions
in Experiment 6 by stimulus format. Lower panel: Qualitative error (QE)
scores for three visual field conditions in Experiment 6 by stimulus format.
(LVF/RH = left visual field/right hemisphere; RVF/LH = right visual field/left
hemisphere; BVF-Con = Bilateral Consistent.)
Dot Trigrams
O — " Digit Trigrams
T
,o
/ I
t
T
i --------------------1 ------------------- r
LVF/RH RVF/LH BVF-Con
Visual Field/Hemisphere
_ - o - .
Dot Trigrams
Digit Trigrams
s
s
1------------------------ 1 -------------------------- J-
LVF/RH RVF/LH BVF-Con
Visual Field/Hemisphere
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
£
o
o
u >
s
0.60
0.50
0.40
0.30-
T
□-----
1
0.2 0 ----------1 ---------------------1 -------------------- r
LVF/RH RVF/LH BVF-Con
Visual Field/Hemisphere
- - O— Exp. 4 Digit Trigrams
- - - Exp. 5 Dot Trigrams
— • ----- Exp. 6 Digit Trigrams
— ■— Exp. 6 Dot Trigrams
Figure 19. QE scores for digit trigrams and dot trigrams, alone (dashed
lines: Experiments 4 and 5) and in combination (solid lines: Experiment 6).
(LVF/RH = left visual field/right hemisphere; RVF/LH = right visual field/left
hemisphere; BVF-Con = Bilateral Consistent.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
77
(M = .50), there is no significant difference, p - .87. Figure 19 shows the QE
score by visual field for digit trigrams and dot trigrams in Experiments 4 to 6.
The solid lines (which reproduce the lower panel of Figure 18) represent the
results of Experiment 6, in which digit trigrams and dot pattern trigrams were
intermixed. The dashed lines represent results when each format was
presented in isolation. It is apparent that the LVF/RH QE score did not
change for digit trigrams in Experiment 6 compared with Experiment 4, but
the RVF/QE score was much higher for digit trigrams when they were
presented along with dot trigrams. This may be because the left hemisphere
is able to adopt a more optimal strategy for the deployment of attention
across the three elements of the digit trigram when that is the only kind of
stimulus presented, but not when digit trigrams and dot trigrams are
intermixed, because the participant does not know which type of stimulus
will appear in any given visual field on any given trial.
The main finding of Experiment 6 is that substantial bilateral gain can
be achieved even when the two stimuli presented to the RVF/LH and LVF/RH
are physically distinctive and in a different format (one a digit trigram and the
other a dot trigram), but lead to the same response. However, even more
gain can be achieved when the two stimuli are physically identical (and thus
in the same format).
GENERAL DISCUSSION
When identical CVC stimuli are presented simultaneously to the
LVF/RH and RVF/LH, identification accuracy is substantially better than when
just one copy of the target CVC is presented to a single visual half-field and a
noise stimulus appears in the other visual half-field (Marks & Hellige, 1999).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
78
The primary purpose of the present series of experiments was to investigate
the contribution of the physical identity of bilateral redundant stimuli to the
production of this bilateral redundancy gain and to observe the effect of
manipulating physical identity on measures of hemispheric strategy
differences. A second purpose was to discover whether similar performance
patterns would be observed with non-linguistic stimuli, including numbers
represented as Arabic digits or dot patterns, as have been observed for
CVCs. Each of these topics will be taken up in turn.
Effects of Physical Identity
When each hemisphere has access to the visual information needed
to produce a response, interhemispheric communication and cooperation
could occur at a variety of stages of information processing, beginning with
basic visual sensory and perceptual processes and continuing through
higher levels of stimulus coding and identification to the final selection and
production of a response. If bilateral redundancy gain depends exclusively or
in large part on the repetition of lower level sensory information in the two
hemispheres, then varying the physical characteristics of the two copies of
the target should reduce or eliminate bilateral gain. However, if more
abstract information processing accounts for a significant portion of the
bilateral gain, the gain should remain even when the two copies of the target
vary in their physical characteristics but lead to the same response.
In Experiments 1 to 3, the physical appearance of the stimuli was
manipulated by varying the case or the case and font of the CVC trigrams. In
Experiment 6, the physical appearance of the stimuli was manipulated by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
79
varying the format of presentation of the number trigrams (as digit trigrams
or dot pattern trigrams). In all four of these experiments, the two copies of the
trigram presented on bilateral trials could be either physically identical
(Bilateral Consistent trials) or physically different (Bilateral Inconsistent
trials). Although it did not always reach the levels predicted by the statistical
summation formula, substantial bilateral redundancy gain (ranging from
12.3% to 14.2%) was realized in all four of these experiments.
To test whether physical identity is important for bilateral redundancy
gain, the crucial comparison was of accuracy of trigram identification on
Bilateral Consistent trials with that on Bilateral Inconsistent trials. For
Experiments 1 and 2, the accuracy results for Bilateral Inconsistent trials
were virtually indistinguishable from those for Bilateral Consistent trials. In
Experiment 3, there was even a small but reliable advantage in favor of
Bilateral Inconsistent trials. The accuracy results of Experiments 1 to 3 using
CVC trigrams thus offer no support for the hypothesis that physical identity of
stimuli is required for bilateral redundancy gain.
Experiment 6 represented an even stronger test of the hypothesis,
because the two copies of the target were not only different in appearance,
but different in format. Again, both Bilateral Consistent and Bilateral
Inconsistent trials produced substantial bilateral redundancy gain, but in this
experiment, there was a reliable difference in favor of Bilateral Consistent
trials, suggesting that physical identity of the stimuli may have made some
contribution to the enhanced performance on Bilateral Consistent trials.
However, even in this experiment, it is clear that physical identity of the two
copies of the target was not required to produce bilateral redundancy gain,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
80
because Bilateral Inconsistent trials produced a gain of 11.2% compared to
RVF/LH trials. Although this level of gain is not as high as that for Bilateral
Consistent trials (16.3%), it is nonetheless substantial.
The present experiments did not directly address the source of the
bilateral redundancy gain observed even when the two copies of the
stimulus were not physically identical. Nonetheless, it may be worthwhile to
consider the possibilities and suggest avenues of further research. The
identification of an alphabetic stimulus like the letters used in Experiments
1-3 may involve the activation of abstract letter identities, which encode the
identity of letters independent of such visual variations as case and font (e.g.,
Besner et al., 1984; Evett & Humphreys, 1981). Given two copies of the
stimulus, whether or not they are physically identical, the appropriate
abstract letter identity may be more strongly activated. A similar process may
hold for number identities, through the activation of abstract digit identities
(McCloskey & Macaruso, 1995), or through the activation of some amodal
number representation (McCloskey, Macaruso, & Whetstone, 1992), or
through spreading activation among various numerical codes (Dehaene,
1992; Koechlin, Naccache, Block, & Dehaene, 1999).
Another possibility is that stimulus identification and response
selection proceed in parallel in the two hemispheres. Information from these
two independent processes then converges on a single response selection
and production process. Since the output is verbal naming in both CVC and
number trigram identification experiments, this process is probably localized
in the left hemisphere. Of course, performance enhancement may result
from a combination of mechanisms.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
81
If redundancy gain in the identification of trigrams relies on
interhemispheric cooperation, it should be abolished when the corpus
:
callosum is severed. In a lexical decision study ( Mohr, Pulvermuller,
Rayman, & Zaidel, 1994) a split-brain subject did not show any redundancy
gain compared to RVF/LH presentation. The researchers concluded that the
gain shown by normals may be due to excitatory transcallosal connections
within interhemispheric cell assemblies corresponding to words. A similar
study using CVC or number identification could establish whether or not
interhemispheric communication via the corpus callosum contributes to
bilateral redundancy gain.
Processing of Non-Linguistic Trigrams
One of the questions to be answered by Experiments 4 to 6 was
whether the patterns of performance (particularly an RVFA in accuracy of
identification and lower QE scores for RVF/LH trials than for LVF/RH trials)
which have typically been found for CVC trigrams would be replicated with
number trigrams. In Experiment 4, using digit trigrams, the pattern of
performance was remarkably like that found for CVC trigram identification in
Experiments 1 and 2 and in the previous CVC experiments whose results
are summarized in Table 1. It is perhaps unsurprising that we found an
RVFA, because digits are essentially verbal stimuli (Boles, 1986) and
previous researchers have found an RVFA for digit identification, particularly
when more than one digit is presented at a time (e.g., Dimond & Beaumont,
1971; Hatta & Dimond, 1980). The RVFA for digit identification was replicated
for the digit trigrams in Experiment 6. In addition, the level of bilateral
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
82
redundancy gain found for digit trigrams in Experiments 4 and 6 was not
significantly different than that predicted by the statistical summation
formula. This result was similar to that found for CVC trigrams in Experiment
3 and in Marks & Hellige (1999) and is further evidence that digit trigrams are
processed similarly to CVC trigrams.
It was somewhat less expected that RVF/LH QE scores were found to
be lower than LVF/RH QE scores for digit trigrams in Experiment 4. Notably,
this did not replicate for the digit trigrams in Experiment 6, which showed no
visual field differences for QE scores. The context in which stimuli appear
may influence the processing strategy that is adopted. It is possible that the
left hemisphere is able to distribute attention rapidly over the three elements
of a digit trigram when only digit trigrams are being presented. However,
when dot pattern trigrams are intermixed with digit trigrams, a more serial
deployment of attention may be induced, because dot pattern trigrams
appear to require a slower or more serial decoding. Evidence for this
possibility is seen in the relatively higher RVF/LH QE scores for dot pattern
trigrams in both Experiments 5 and 6 compared to RVF/LH QE scores for
either CVC or digit trigrams. Overall, the patterns of performance found for
digit trigrams in Experiments 4 and 6 tend to replicate those found in CVC
experiments.
The same is not true for dot pattern trigrams. First, there is no RVFA
for the identification of dot patterns. The total error percentage is the same
for LVF/RH and RVF/LH trials for dot pattern trigrams in Experiments 5 and 6.
Second, as noted above, the RVF/LH QE scores were higher than LVF/RH
QE scores for dot pattern trigrams in Experiment 5; in Experiment 6 there
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
83
was no difference between the QE scores for the two unilateral visual field
conditions. These results offer no evidence for more holistic processing of
dot trigrams by the left hemisphere, in contrast to the findings for CVCs and
for digits when presented alone (not intermixed with dot patterns). Overall,
the patterns of performance found for dot pattern trigrams in Experiments 5
and 6 are quite different from those found in CVC experiments and from
those found for digit trigrams in Experiments 4 and 6.
The Consistency Effect
One of the questions raised by the results of this series of
experiments is how we might explain the finding of a significantly lower error
rate for Bilateral Consistent than for Bilateral Inconsistent trials in
Experiment 6. When two digit trigrams or two dot pattern trigrams were
presented (Bilateral Consistent trials), the mean error rate was 35.2%.
However, when one digit and one dot pattern trigram were presented
(Bilateral Inconsistent trials), the error rate was 40.3%. This represents an
improvement in performance for trials where the stimuli were physically
identical compared to trials where the stimuli were physically non-identical
(a consistency advantage). No such improvement was observed for Bilateral
Consistent trials in Experiments 1 to 3.
One difference between the experiments which failed to show a
consistency advantage and Experiment 6, which did show such an
advantage, is that the two stimuli presented on inconsistent trials in
Experiment 6 differed not only in physical appearance but in format. The
uppercase and lowercase CVCs used in Experiments 1 to 3 were different in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
84
physical appearance, but both were alphabetic and presumably were
processed in the same brain areas using similar computational algorithms.
On the other hand, perhaps the two different notations in which numbers
were represented in Experiment 6 were processed in different brain areas
using different computational algorithms.
According to some researchers, there are fundamental differences in
the ways the brain deals with numbers presented in different notations (for
review see Campbell & Clark, 1992; Dehaene, 1992). Dehaene’s triple-code
model posits three different codes for the representation of numbers: a
verbal word form, an Arabic number form, and an analogue magnitude code.
The structured dot-pattern trigrams used in the present experiments involve
the recognition of the numerosity of small sets of dots, presumably using the
analogue magnitude code (Koechlin et al., 1999). The fast process
responsible for the recognition of the numerosity of very small sets is called
subitizing. Current usage generally restricts the meaning of this term to the
range 1-3 or 1-4 (Koechlin et al., 1999), but it is possible that with practice
and the presentation of the dots in a canonical configuration (Mandler &
Shebo, 1982), our participants were able to use subitizing to identify the dot
patterns 1-6 used in the trigrams. Evidence that Arabic digits and dot
patterns invoke different mental representations comes from a recent
priming study (Koechlin et al., 1999) using a number comparison task.
Cross-notation quantity priming was observed for Arabic digits and number
words, but not for Arabic digits and random dot patterns. Koechlin and his
colleagues conclude that digits and dot patterns invoke different internal
quantity representations, one specialized for numerical symbols like Arabic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
85
digits and the other specialized for analogue representations like dot
patterns.
One possible explanation of the consistency effect (more bilateral
redundancy gain for physically identical than non-identical trials) is that
bilateral redundancy gain in trigram identification depends partially on neural
coactivation via the corpus callosum. A similar explanation has been
proposed to partially account for bilateral gain observed for redundant
targets in visual search (e.g., Fournier & Eriksen, 1990; Miller, 1982).
Perhaps such neural coactivation is more effective when the two stimuli are
processed in a similar manner. One explanation would be that homologous
brain areas in the right and left hemisphere are involved in the processing of
similar stimuli, and activation spreads across the corpus callosum through
excitatory transmissions leading to homotopic activation (Horwitz, Duara, &
Rapoport, 1986; Horwitz, Grady, Schlageter, Duara, & Rapoport, 1987).
Processing of digit trigrams and dot pattern trigrams is likely to occur in
different brain regions. The very different patterns of performance found for
these two types of stimuli in the present experiments argues that different
processes are involved in the recognition of digits and dot patterns. Koechlin
et al.’s (1999) failure to find cross-notational quantity priming for Arabic digits
and dot patterns is additional evidence that different brain areas may be
involved in the representations of these two formats. There is not much
evidence from brain imaging studies to shed light on this question, although
a recent study (Dehaene, Spelke, Pinel, Stanescu, & Tsivkin, 1999) indicates
that approximate arithmetic draws on analogue processing and recruits
bilateral areas of the parietal lobes involved in visuospatial processing.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
86
Arabic digit processing may be more closely allied to verbal processing, as
Koechlin et al. suggest to account for their finding of cross-notational priming
between digits and number words (1999), and thus may both involve
language areas of the brain. It may be possible to test this hypothesis by
carefully designed brain imaging studies.
Conclusion
The primary purpose of the present series of experiments was to
investigate whether or not physical identity of the stimuli presented on
bilateral redundant trials is necessary to produce bilateral redundancy gain.
The clear answer is that it is not. Substantial bilateral redundancy gain is
realized both when the two stimuli are physically identical and when they are
physically different but lead to the same response. This is true even when
the two stimuli are different not only in appearance but in format, as in
Experiment 6. The presentation of the same three-digit number as an Arabic
digit trigram to one hemisphere and as a structured dot pattern trigram to the
other hemisphere produces greater accuracy of identification than the
presentation of only one trigram in either format to only one hemisphere.
Therefore, bilateral redundancy gain must result at least partially from
processes other than low-level visual sensory processes. To clarify what
these processes are will require further research. The consistency
advantage observed in Experiment 6 suggests that a fruitful line of
investigation may be functional imagery of interhemispheric processes
contrasting the cases in which the two bilateral redundant stimuli are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
87
identical in appearance and format, different in appearance but the same in
format, or different in both appearance and format.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8 8
References
Allen, D. (1968). Spatial summation in the evoked cortical response to
auditory stimuli. Perception and Psychophysics, 4, 355-356.
Banich, M. T. (1995). Interhemispheric processing: Theoretical
considerations and empirical approaches. In R. J. Davidson & K.
Hugdahl (Eds.), Brain asymmetry (pp. 427-450). Cambridge, MA: MIT
Press.
Banich, M. T. (1998a). Integration of information between the cerebral
hemispheres. Current Directions in Psychological Science, 7, 32-37.
Banich, M. T. (1998b). The missing link: The role of interhemispheric
interaction in attentional processing. Brain and Cognition, 36, 128-
157.
Banich, M. T., & Karol, D. L. (1992). The sum of the parts does not equal the
whole: Evidence from bihemispheric processing. Journal of
Experimental Psychology: Human Perception and Performance, 18,
763-784.
Banich, M. T., & Shenker, J. I. (1994). Investigations of interhemispheric
processing: Methodological considerations. Neuropsychology, 8, 263-
277.
Belger, A., & Banich, M. T. (1998). Costs and benefits of integrating
information between the cerebral hemispheres: A computational
perspective. Neuropsychology, 12, 380-398.
Berger, J. M., & Landolt, M. (1990). Bilateral processing of numbers: Parallel
processing versus functional unity. Cortex, 26, 555-566.
Berlucchi, G., Brizzolara, D., Marzi, C. A., Rizzolatti, G., & Umilta, C. (1979).
The role of stimulus discriminability and verbal codability in
hemispheric specialization for visuospatial tasks. Neuropsychologia,
17, 195-202.
Besner, D., Coltheart, M., & Davelaar, E. (1984). Basic processes in reading:
Computation of abstract letter identities. Canadian Journal of
Psychology, 38, 126-134.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
89
Biederman, I., & Gerhardstein, P. C. (1993). Recognizing depth-rotated
objects: Evidence and conditions for three-dimensional viewpoint
invariance. Journal of Experimental Psychology: Human Perception
and Performance, 19, 1162-1182.
Boles, D. B. (1986). Hemispheric differences in the judgment of number.
Neuropsychologia, 24, 511-519.
Boles, D. B. (1992). Fast visual generation: Its nature and chronometrics.
Perception and Psychophysics, 51, 239-246.
Brown, W. S., & Jeeves, M. A. (1993). Bilateral visual field processing and
evoked potential interhemispheric transmission time.
Neuropsychologia, 31, 1267-1281.
Campbell, J. I. D., & Clark, J. M. (1992). Cognitive number processing: An
encoding-complex perspective. In J. I. D. Campbell (Ed.), The nature
and origins of mathematical skills (pp. 457-491). Amsterdam,
Netherlands: North-Holland.
Cherry, B. J., Hellige, J. B., & McDowd, J. M. (1995). Age differences and
similarities in patterns of cerebral hemispheric asymmetry.
Psychology and Aging, 10, 191-203.
Cohen, G. (1975). Hemispheric differences in the effects of cuing in visual
recognition tasks. Journal of Experimental Psychology: Human
Perception and Performance, 1, 366-373.
Coney, J. (1985). Dual hemispheric processing in a letter matching task.
Perception and Psychophysics, 38, 331-342.
Dehaene, S. (1992). Varieties of numerical abilities. Cognition, 44, 1-42.
Dehaene, S., Spelke, E., Pinel, P., Stanescu, R., & Tsivkin, S. (1999).
Sources of mathematical thinking: Behavioral and brain-imaging
evidence. Science, 284, 970-974.
Dimond, S. J., & Beaumont, G. (1971). Use of two cerebral hemispheres to
increase brain capacity. Nature, 232, 270-271.
Eng, T. L., & Hellige, J. B. (1994). Hemispheric asymmetry for processing
unpronounceable and pronounceable letter trigrams. Brain and
Language, 46, 517-535.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
90
Evett, L. J., & Humphreys, G. W. (1981). The use of abstract graphemic
information in lexical access. Quarterly Journal of Experimental
Psychology: Human Experimental Psychology, 33A, 325-350.
Eviatar, Z., Hellige, J. B., & Zaidel, E. (1997). Individual differences in
lateralization: Effects of gender and handedness. Neuropsychology,
11, 562-576.
Fournier, L. R., & Eriksen, C. W. (1990). Coactivation in the perception of
redundant targets. Journal of Experimental Psychology: Human
Perception and Performance, 16, 538-550.
Hasbrooke, R. E., & Chiarello, C. (1998). Bihemispheric processing of
redundant bilateral lexical information. Neuropsychology, 12, 78-94.
Hatta, T. (1978). Visual field differences in a mental transformation task.
Neuropsychologia, 16, 637-641.
Hatta, T., & Dimond, S. J. (1980). Comparison of lateral differences for digit
and random form recognition in Japanese and Westerners. Journal of
Experimental Psychology: Human Perception and Performance, 6,
368-374.
Hellige, J. B. (1990). Hemispheric asymmetry. Annual Review of Psychology,
41, 55-80.
Hellige, J. B. (1993). Unity of thought and action: Varieties of interaction
between the left and right cerebral hemispheres. Current Directions in
Psychological Science, 2, 21-25.
Hellige, J. B., Bloch, M. I., Cowin, E. L., Eng, T. L., Eviatar, Z., & Sergent, V.
(1994). Individual variation in hemispheric asymmetry: Multitask study
of effects related to handedness and sex. Journal of Experimental
Psychology: General, 123, 235-256.
Hellige, J. B., & Cowin, E. L. (1996). Effects of stimulus arrangement on
hemispheric differences and interhemispheric interaction for
processing letter trigrams. Neuropsychology, 10, 247-253.
Hellige, J. B., & Marks, N. L. (in press). Are hemispheric strategy differences
independent of level of performance? Neuropsychology.
Hellige, J. B., & Scott, G. B. (1997). Effects of output order on hemispheric
asymmetry for processing letter trigrams. Brain and Language, 59,
523-530.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
91
Hellige, J. B., Taylor, A. K., & Eng, T. L. (1989). Interhemispheric interaction
when both hemispheres have access to the same stimulus
information. Journal of Experimental Psychology: Human Perception
and Performance, 15, 711-722.
Hellige, J. B., Taylor, K. B., Lesmes, L, & Peterson, S. (1998). Relationships
between brain morphology and behavioral measures of hemispheric
asymmetry and interhemispheric interaction. Brain and Cognition, 36,
158-192.
Hellige, J. B., & Yamauchi, M. (1999). Quantitative and qualitative
hemispheric asymmetry for processing Japanese kana. Brain and
Cognition, 40, 453-463.
Horwitz, B., Duara, R., & Rapoport, S. I. (1986). Age differences in
intercorrelations between regional cerebral metabolic rates for
glucose. Annals of Neurology, 19, 60-67.
Horwitz, B., Grady, C. L., Schlageter, N. L., Duara, R., & Rapoport, S. I. (1987).
Intercorrelations of regional cerebral glucose metabolic rates in
Alzheimer's disease. Brain Research, 407, 294-306.
Kee, D. W., Cherry, B. J., Neale, P. L., McBride, D. M., & Segal, N. L. (1998).
Multitask analysis of cerebral hemisphere specialization in
monozygotic twins discordant for handedness. Neuropsychology, 12,
468-478.
Kim, H. (1996). Qualitative hemispheric differences for processing trigrams.
Brain and Cognition, 30, 205-214.
Kimura, D. (1966). Dual functional asymmetry of the brain in visual
perception. Neuropsychologia, 4, 275-285.
Klein, R. M., & Mclnnes, J. (1988). Visual field differences in the processing
of numerical stimuli. Brain and Cognition, 7, 247-256.
Koechlin, E., Naccache, L., Block, E., & Dehaene, S. (1999). Primed
numbers: Exploring the modularity of numerical representations with
masked and unmasked semantic priming. Journal of Experimental
Psychology: Human Perception & Performance, 25, 1882-1905.
Lepore, F., Ptito, M., & Jasper, H. H. (Eds.). (1986). Two hemispheres, one
brain: Functions of the corpus callosum. New York: Alan R. Liss.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
92
Levy, J., Heller, W., Banich, M. T., & Burton, L. A. (1983). Are variations among
right-handed individuals in perceptual asymmetries caused by
characteristic arousal differences between hemispheres? Journal of
Experimental Psychology: Human Perception and Performance, 9,
329-359.
Liederman, J. (1998). The dynamics of interhemispheric collaboration and
hemispheric control. Brain and Cognition, 36, 193-208.
Liederman, J., & Meehan, P. (1986). When is between-hemisphere division
of labor advantageous? Neuropsychologia, 24, 863-874.
Liederman, J., Merola, J., & Martinez, S. (1985). Interhemispheric
collaboration in response to simultaneous bilateral input.
Neuropsychologia, 23, 673-683.
Luh, K. E., & Levy, J. (1995). Interhemispheric cooperation: Left is left and
right is right, but sometimes the twain shall meet. Journal of
Experimental Psychology: Human Perception and Performance, 21,
1243-1258.
Luh, K. E., & Wagner, D. A. (1997). Cerebral asymmetries in processing
strategies for letter and symbol trigrams. Brain and Language, 60,
464-488.
Mandler, G., & Shebo, B. J. (1982). Subitizing: An analysis of its component
processes. Journal of Experimental Psychology: General, 111, 1-22.
Marks, N. L. (2000). Interhemispheric interaction and the corpus callosum: A
review. Unpublished manuscript, University of Southern California,
Los Angeles
Marks, N. L., & Hellige, J. B. (1999). Effects of bilateral stimulation and
stimulus redundancy on interhemispheric interaction.
Neuropsychology, 13, 475-487.
McCloskey, M., & Macaruso, P. (1995). Representing and using numerical
information. American Psychologist, 50, 351-363.
McCloskey, M., Macaruso, P., & Whetstone, T. (1992). The functional
architecture of numerical processing mechanisms: Defending the
modular model. In J. I. D. Campbell (Ed.), The nature and origins of
mathematical skills (pp. 493-537). Amsterdam, Netherlands: North-
Holland.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
Miller, J. (1982). Divided attention: Evidence for coactivation with redundant
signals. Cognitive Psychology, 14, 247-279.
Mohr, B., Pulvermuller, F., Mittelstadt, K., & Rayman, J. (1996). Multiple
simultaneous stimulus presentation facilitates lexical processing.
Neuropsychologia, 34, 1003-1013.
Mohr, B., Pulvermuller, F., Rayman, J., & Zaidel, E. (1994). Interhemispheric
cooperation during lexical processing is mediated by the corpus
callosum: Evidence from the split-brain. Neuroscience Letters, 181,
17-21.
Mohr, B., Pulvermuller, F., & Zaidel, E. (1994). Lexical decision after left, right
and bilateral presentation of function words, content words and non
words: Evidence for interhemispheric interaction. Neuropsychologia,
32, 105-124.
Mordkoff, J. T., & Yantis, S. (1991). An interactive race model of divided
attention. Journal of Experimental Psychology: Human Perception and
Performance, 17, 520-538.
Norman, W. D., Jeeves, M. A., Milne, A., & Ludwig, T. (1992). Hemispheric
interactions: The bilateral advantage and task difficulty. Cortex, 28,
623-642.
Oldfield, R. C. (1971). The assessment and analysis of handedness: The
Edinburgh inventory. Neuropsychologia, 9, 97-113.
Polk, T. A., & Farah, M. J. (1997). A simple common contexts explanation for
the development of abstract letter identities. Neural Computation, 9,
1277-1289.
Reuter-Lorenz, P. A., Stanczak, L., & Miller, A. C. (1999). Neural recruitment
and cognitive aging: Two hemispheres are better than one, especially
as you age. Psychological Science, 10, 494-500.
Schuez, A., & Preissl, H. (1996). Basic connectivity of the cerebral cortex and
some considerations on the corpus callosum. Neuroscience and
Biobehavioral Reviews, 20, 567-570.
Taylor, K. B. (1999). The relationship of corpus callosum morphology to
performance on lateralized tasks and to cerebral structural laterality.
(Doctoral dissertation, University of Southern California, 1998).
Dissertation Abstracts International, 60, 2381B.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
94
Weissman, D. H., & Banich, M. T. (2000). The cerebral hemispheres
cooperate to perform complex but not simple tasks. Neuropsychology,
14, 41-59.
Weissman, D. H., Banich, M. T., & Puente, E. I. (2000). An unbalanced
distribution of inputs across the hemispheres facilitates
interhemispheric interaction. Journal of the International
Neuropsychological Society, 6, 313-321.
Zaidel, E., & Rayman, J. (1994). Interhemispheric control in the normal brain:
Evidence from redundant bilateral presentations. In C. Umilta & M.
Moscovitch (Eds.), Attention and performance XV: Conscious and
nonconscious information processing, (pp. 477-504). Cambridge, MA:
MIT Press.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Asset Metadata

Creator Marks, Nancy Lee (author)

Core Title Interhemispheric interaction in bilateral redundancy gain: Effects of physical similarity

Contributor Digitized by ProQuest (provenance)

School Graduate School

Degree Doctor of Philosophy

Degree Program Psychology

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

Tag OAI-PMH Harvest,psychology, cognitive,psychology, psychobiology

Language English

Advisor Hellige, Joseph (committee chair), Lavond, David (committee member), McClure, William (committee member), Thompson, Richard (committee member), Walsh, David A. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-102326

Unique identifier UC11327006

Identifier 3027746.pdf (filename),usctheses-c16-102326 (legacy record id)

Legacy Identifier 3027746.pdf

Dmrecord 102326

Document Type Dissertation

Rights Marks, Nancy Lee

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 au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA