University of Southern California Dissertations and Theses

Autonomic and cognitive indices of semantic conditioning and generalization

(USC Thesis Other)

Autonomic and cognitive indices of semantic conditioning and generalization

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content AUTONOMIC AND COGNITIVE INDICES OF SEMANTIC CONDITIONING AND GENERALIZATION by Terry Tang A D issertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In P artial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Psychology) January 1973 INFORMATION TO USERS This dissertation was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this docum ent have been used, th e quality is heavily dependent upon the quality of the original subm itted. The following explanation of techniques is provided to help you understand markings or patterns which may appear on this reproduction. 1. The sign or "target" for pages apparently lacking from the docum ent photographed is "Missing Page(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting thru an image and duplicating adjacent pages to insure you com plete continuity. 2. When an image on the film is obliterated with a large round black mark, it is an indication th at the photographer suspected th at the copy may have moved during exposure and thus cause a blurred image. You will find a good image of the page in the adjacent frame. 3. When a map, drawing or chart, etc., was part of the material being p h o to g rap h e d th e photographer followed a definite m ethod in "sectioning" the material. It is customary to begin photoing at the upper left hand corner of a large sheet and to continue photoing from left to right in equal sections with a small overlap. If necessary, sectioning is continued again - beginning below the first row and continuing on until complete. 4. The m ajority of users indicate that the textual content is of greatest value, however, a somewhat higher quality reproduction could be made from "photographs" if essential to the understanding of the dissertation. Silver prints of "photographs" may be ordered at additional charge by writing the Order Department, giving the catalog number, title, author and specific pages you wish reproduced. University Microfilms 300 North Zeeb Road Ann Arbor, Michigan 48100 A Xerox Education Company 73-9327 i TANG, T erry, 1942- AUTONO M IC A N D COGNITIVE INDICES OF SEMANTIC CONDITIONING A N D GENERALIZATION. U n iv e r s ity o f Southern C a lif o r n ia , Ph.D., 1973 P s y c h o lo g y , exp erim en tal University Microfilms, A X E R O X Company, Ann Arbor, Michigan (C o p y r ig h t by TERRY T A N G 1973 THIS DISSERTATION HAS BEEN MTf!Rf)FTIAIPn EYAfTf.Y AS BFPPTVPn UNIVERSITY O F SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA 9 0 0 0 7 T his dissertation, written by IlXXZ-TSfl*....................................................................... under the direction of h .ia ... Dissertation C om m ittee, and approved by all its members, has been presented to and accepted by T he Graduate School, in partial fulfillm ent of requirements of the degree o f D O C T O R O F P H IL O S O P H Y D tan Date. October 19, 1972 DISSERTATION COMMITTEE Chairman PLEASE NOTE: Some p ag es may have in d is tin c t p rin t. Filmed a s r e c e iv e d . University Microfilms, A X erox Education Company ACKNOWLEDGMENTS I wish to express my appreciation to members of my d is se r ta tio n committee, Professors Stephen A. Madlgan, William W. Grlngs, and Alexander W . McEachern, for t h e i r very helpful In s tr u c tio n s , suggestions, and assistance 1n the completion of th is research. I thank the National I n s t i t u t e of Mental Health for Tra1neesh1p Grant M H 10554, which made th is d is se rta tio n possible and the State of California for financial a s s i s tance. I also wish to express my gratitude to Dr. Grlngs for the use of his fine library and laboratory. This d is s e r t a t i o n is dedicated to w riters whose words have fascinated me. 1 i TABLE OF CONTENTS Page ACKNOWLEDGMENTS..................................................................................... 1i LIST OF TABLES..................................................................................... v LIST OF FIGURES v1 i Chapter I INTRODUCTION..................................................................... 1 The Theoretical I s s u e ..................................... 1 E a rlier Investigations of the Issue. . . 2 The Present Experimental Approach. . . . 8 The 0ne-Tr1al Acquisition Procedure. . . 10 II EXPERIMENTAL HYPOTHESES AND EMPIRICAL PROPOSITIONS................................................................... 15 Experimental Hypotheses................................ 15 Empirical P r o p o s i t i o n s ................................ 17 III METHOD................................................................................... 22 D e s i g n ................................................................................ 22 Stimulus Variable M a te r ia ls .............................. 23 Apparatus........................................................................... 26 S u b j e c t s ........................................................................... 27 Procedure........................................................................... 28 IV RESULTS..................................................................................... 29 Dependent Measures .................................................. 29 Stimulus Word Type and GSR Responding. . 31 Stimulus Word Type and Recognition Responding................................................................. 37 Stimulus Word Type and GSR and Recognition Responding................................... 38 Recognition Response Parameters.................... 50 i i i Table of Contents - - con tin u e d Chapter Page V DISCUSSION 58 The Dependent Measure ............................................. 60 GSR and Recognition Responding......................... 61 Converging Operations and Convergent T h e o r i z i n g ................................................................. 65 Verbal Recognition Responding ......................... 70 SUMMARY.............................................................................................................. 73 REFERENCES........................................................................................................ 76 Appendlx A Stimulus Word L i s t s ............................................................. 82 B Instructions to the S u b ject.......................................... 86 C Effects of Word Type, Input and Test P osition, and Association Strength on GSR R esponding...................................................................... 88 D Operational Definitions of Good and Poor Recognition Performance Subjects and GSR A n a l y s e s ............................................................................................100 E Operational Definitions of Good, Average, and Poor Discriminative Stimulus Recogni tion P e rf o rm a n c e ........................................................................107 F Operational Definitions of Good, Poor, and No GSR Discrimination Conditioning and Generalization S ubjects.........................................................108 iv LIST OF TABLES Table Page 1 Word Type and 1-1 GSR Magnitude Means, Standard Deviations, and F Max for all Subjects, N = 72.......................................................................... 33 2 Analysis of Variance of Word Type and 1-1 Magnitude GSRs for all Subjects, N=72 . . . . 34 3 Neuman-Keuls S t a t i s t i c s on 1-1 Magnitude Changes to Stimulus Words for all Subjects, N = 72.................................................................................................... 35 4 Means and Standard Deviations of Recogni tion Responses to Stimulus Words for all Subjects, N= 72.......................................................................... 39 5 Analysis of Variance of Word Type and "Old" Recognition Responses for all Sub j e c t s , N= 7 2 ................................................................................ 40 6 Neuman-Keu'.s S t a t i s t i c s on "Old" Recogni tion Responses to Stimulus Words for all Subjects, N= 72........................................................................... 41 7 Word Type x Recognition Response Frequency for all Subjects, N = 72....................................................... 43 8 Word Type x Recognition Response Contin gency Tables: Weighted GSR Magnitude Mean of Mean Per Tri a l ................................................................. 44 9 Means and Standard Deviations of Latency and Confidence of Correctness of Recogni tion Responses to Stimulus Words for all Subjects, N= 72........................................................................... 52 10 Analysis of Variance of Word Type and Con fidence of Correctness of Recognition Res ponses for all Subjects, N = 7 2 .................................. 53 11 Neuman-Keuls S t a t i s t i c s on Confidence of Correctness of Recognition Responses to Stimulus Words for all Subjects, N=72 . . . . 54 v L i s t o f Tables - - c o n ti n u e d Table Page 12 Stimulus Word Type x Recognition Response Contingency Tables: Recognition Latency and Confidence of Correctness for all Sub je cts* N=72............................................................................................55 vi LIST OF FIGURES Figure Page 1 S-R paradigm of semantic generalization. . . 20 2 Bar graph of mean GSRs e l i c i t e d by five stimulus word types during t e s t t r i a l s . . . 32 3 Bar graph of weighted mean GSRs e l i c i t e d by five stimulus word types as a function of stimulus recognition response during t e s t t r i a l s ................................................................................ 45 4 Bar graph of weighted mean confidence of correctness of recognition response scores as a function o f recognition response and stimulus word t y p e ............................................................ 57 vi i CHAPTER I INTRODUCTION The Theoretical Issue The basic concern of this d is s e r ta tio n was the nature and number of processes involved in the acquisition and expression of c l a s s i c a l l y conditioned autonomic responses in normal human subjects (Ss). S11mulus-response (S-R) theories ( e . g . , Hull, 1943) propose that conditioned r e s ponse acquisition (CRA) is due to a sin g le , d ire c t S-R association whereas cognitive theories ( e . g . , Tolman, 1948) propose th a t CRA is the re s u lt of a cognitive process such as re la tio n a l learning (RL). A th ird proposal is that both associative (S-R) and cognitive (RL) processes are Involved ( e . g . , Kimble, 1962; Grlngs, 1965). Psychophysiologists in v estig atin g this issue have been concerned with whether or not human discrimination conditioning of autonomic responses is dependent upon S/s awareness of CS-UCS contingencies or RL ( e . g . , Grlngs, 1965; Dawson, 1970). This question has been studied be cause a generally accepted operational d efin itio n of c l a s s ical conditioning -- response modification a t t r i b u t a b l e solely to the temporal pairing of a r e la tiv e neutral stimu lus (CS) with a potent, response-producing stimulus (UCS) 2 (Hllgard and Marquis, 1940) -- can produce in human Ss two measureable phenomena: conditioned response a cq u isitio n (CRA) and the a b i l i t y to verbalize the re la tio n sh ip between the two stimuli ( r e la tio n a l learning - RL). In studying this issu e, Dawson and Grlngs (1966, 1968) reviewed the l i t e r a t u r e on classical conditioning and relational learning and observed that there are cur rently three major views, (a) CRA and RL are d if f e r e n t and somewhat opposing processes of learning, (b) RL is one of the determiners of CRA, though not an e ss e n tial one, and (c) RL is a necessary condition for CRA. Another aspect of the problem is i f classica l con ditioning is defined in terms of certain operations, i t is not clear whether th e o r i s t s would r e s t r i c t such learning to those operations where RL learning can be excluded, or whether they would regard cognitive-perceptual variables as Independent variables determining classica l conditioning (Grlngs, 1965). Theoretical and clinical implications of each of these views have been discussed by several w riters ( e . g . , Kimble, 1964; Grings, 1965; Dawson, 1967). E a rli e r Investigations of the Issue Various experimental procedures have been employed for the purpose of demonstrating CRA 1n the absence of RL. E a r li e r studies assumed that i f the CS is below the t h r e s hold of verbalized perception and the temporal pairing of such a stimulus with an UCS resu lts in response modifica tio n , pure CRA would be demonstrated. It was reasoned that i f the CS cannot be consciously perceived, then aware ness of CS-UCS contingency is absent. Subliminal condi tioning, however, has not been s a t i s f a c t o r i l y demonstrated (Eriksen, 1960; Galto, 1964; Dawson, 1967). S im ilarly, other studies have reasoned that response modification resu ltin g from Internal presentations of CS and UCS (Interoceptive conditioning) would be evidence of conditioning without awareness (Razran, 1961). The r e su lts of such studies were Inconclusive and the reasoning employed was questionable because the presupposition that internal events are unconscious is controversial ( e . g . , Armstrong, 1962; Uno, 1966; Dawson, 1967). Inherent in these e a r l i e r studies was the problem of determining whether or not S^ was aware of the experimental stimuli presented. In l a t e r studies in which all c r i t i c a l stimuli were e a s ily discernable, the problem was to d e t e r mine whether or not was aware of the relatio n sh ip between s t i m u l i . To determine awareness of stimulus contingencies Ss were typically asked to express t h e i r f e e lin g s, thoughts, and expectancies concerning aspects of the experiment during or a f t e r the experiment. Common procedures for in ferrin g awareness were questionnaires asking open-ended 4 and multiple choice items, UCS expectancy r a tin g s , and semantic d i f f e r e n t i a l sca le s. These procedures, however, have been c r i t i c i z e d for both logical and experimental reasons ( e . g . , Eriksen, I960; Dawson, 1967; Maltzman, 1968; Page, 1970; Dixon and Moulton, 1970). B riefly , experimental criticism s focused on the fa c t th a t the introduction of the awareness measuring i n s t r u ment during the conditioning procedure also introduces uncontrolled variables such as expectation and s e n s i t i z a tion which have been shown to a ffe c t the acquisition pro cess ( e . g . , Orne, 1962; Dixon and Moulton, 1970). Logical criticism s involve unresolved problems concerning absolute unawareness and r e la tiv e awareness, temporal e f f e c t s be tween stimulus relationships and the dependent measure, and the task of lo g ically in fe rr in g the r e la tio n s h ip be tween CRA and RL ( e . g . , Eriksen, 1960; Nagel 1961; Dawson, 1967). The d i f f i c u l t task of logically in f e r r in g the r e l a tionship between CRA and RL is i l l u s t r a t e d in the following example. Given hypothetical data indicating a positive corre la tio n between two types of responses, verbalized RL and autonomic CRA, at le a s t three conclusions can be drawn: r e la tiv e RL mediated CRA, CRA mediated r e l a t i v e RL, and r e l a t i v e RL and CRA were mediated by some t h i r d variable. In lig h t of such problems in measuring RL, other experimenters sought to control i t . This approach has been 5 verbalized by Grlngs (1965). I f the verbal-perceptual behavior involved 1s not considered a part of what is la b elled 'condi t i o n i n g , ' some a tten tio n must be paid to the matter of control in order to prevent verbal and perceptual variables from confounding ex p eri ments designed to study ' t r u e ' c la s s ic a l con ditioning. (p. 72) The major methods employed for co n tro llin g RL are the man ip u latio n of Interstimulus interval ( I S I ) and stimulus sequence variables, the use of d i f f e r e n t i a l information to e sta b lis h or to a ffe c t RL, and the employment of RL masking tasks. The manipulation of stimulus variables for the pur pose of c ontrolling RL was suggested by evidence indica ting th a t d i f f e r e n t learning processes may occur with short and long CS-UCS intervals ( I S I ) ( e . g . , Lockhart, 1966; Prokasy and Ebel, 1967; Maltzman, 1968; Martin and Levey, 1969). Such evidence and the presupposition that a long ISI is a necessary condition for RL -- " I t is obvious that i f is to be able to hypothesize about the contingency, he must have s u f f i c i e n t time and opportunity" (Koneuni and Slamecka, 1972, p. 248) -- led to experiments which varied the ISI in backward conditioning and reversal learning paradigms ( e . g . , Jones, 1962; Champion, 1966; Shean, 1968; Zeiner and Grings, 1968; Furedy, 1970; Carey, Schell, and Grings, 1971). Assuming th a t r e la tio n a l learning requires a long ISI, a demonstration of backward conditioning using a short ISI would be an example of S-R learning in the ab sence of RL learning. When the ISI is sh o rt, the uncon ditioned response (UCR) will s t i l l follow the CS thereby permitting a S-R association to form. When the ISI is long, backwards conditioning via a S-R process would not be possible because the UCR will occur p rio r to the CS. When the ISI is long, cognitive RL processes can function within the ISI so th a t backward conditioning can occur. Results of experiments employing such reasoning have not been consistent in demonstrating backward conditioning in the absence of RL when the ISI was less than one second ( e . g . , Trapold, Homzie, and Rutledge, 1964; Furedy, 1967; Zimney Stern, and Fjeld, 1967). Conditioned response ac q u is itio n with RL has been demonstrated with an ISI of .750 msec, however (Ziener and Grings, 1968). In the reversal learning paradigm, the stimulus that is contingent with the UCS and the stimulus which is not contingent with the UCS are reversed. If cognitive RL learning is not possible when the ISI is short, then r e versal learning would not be a function of number of ac q u is itio n t r i a l s preceding the reversal. I f tr a n s p o s i tional e ffe c ts are observed when the acquisition ISI is sh o rt, then the presupposition that a long ISI is a 7 necessary condition for RL must be rejected. The limited research 1n this area tend to show that reversal learning is dependent upon cognitive RL processes, and the issue on the dependence of RL on a long ISI is s t i l l unresolved ( e . g . , Hebert and Krantz, 1965; Ziener and Grlngs; 1968; Carey, Schell, and Grings, 1971). The CRA-RL issue has been studied with the use of RL masking tasks and d i f f e r e n t i a l information. Subjects receive CS-UCS pairings simultaneously with a masking task. If no CR occurs on subsequent te sts t r i a l s but do occur 1n Ss who l a t e r receive RL i n s t r u c t i o n s , then CS-UCS pairings alone are not s u f f i c i e n t for CR development; RL is a neces sary variable for CRA ( e . g . , Dawson and Grings, 1968; Fuhrer and Baer, 1969). Although Dawson (1970) used this approach in an ex periment which demonstrated th a t the masking task did not i n h i b i t discrimination performance in the GSR in terms of some response-interference property of the masking task, his controls did not eliminate the p o s s i b i l i t y that the masking task in h ib ited CRA for some reason other then its RL prevention p ro p e rtie s . As Dawson (1970) suggested, the e f f e c t of the masking task might have been the in h ib itio n of pure CRA i t s e l f . Pavlov (1927), Razran (1961), and Lynn (1966) ob served th a t the orienting reflex (OR) in h ib its the expres sion of CRs when the OR is evoked j u s t p r io r to or within 8 the temporal interval in which the CR would have otherwise occurred. Since ORs are readily e l i c i t e d by changes in stimulation ( e . g . , Berlyne, 1960; Sokolov, 1963), i t is possible for a masking task to e l i c i t ORs which i n h i b i t the expression of CRs. The mere operation of presenting a r e l a t i v e l y neutral perceptible stimulus (CS) with a potent, response-producing stimulus (UCS) is not an adequate operational d e fin itio n of classical conditioning as Dawson (1967) presupposed, and investigators using the masking task procedure have t a c i t l y assumed ( e . g . , Dawson and Grings, 1968; Shean, 1968; Dawson, 1970). Other conditions ( e . g . , masking, fa tig u e , drugs) are capable of overriding the e ffe c ts of paired CS-UCS stimulation. The Present Experimental Approach The above discussion pointed out various problems in using stimulus manipulations for controlling verbal- perceptual variables during c lassical conditioning. The present experiment therefore used a d i f f e r e n t method of control. Subject's recognition of t e s t stimuli was moni tored during the t e s t phase of a classical discrimination GSR conditioning experiment. Stimulus recognition was investigated in order to determine i f this cognitive-perceptual variable 1s part of a class of determining variables subsumed under c la s s ic a l 9 conditioning, or i f 1t is a limiting variable lying outside the conditioning system. The dependence of classical discrimination GSR conditioning on the awareness of CS- UCS contingency was also studied by examining the r e l a t i o n ship between the stimulus presented, the stimulus recogni tion response, and the autonomic response (GSR) e l i c i t e d . Whether S-R or RL or both forms of learning are Involved 1n CRA was studied with a semantic conditioning and g eneralization procedure and verbal learning task. During acquisition (Input) Ss liste n ed to a 11st of words. Some of the words were paired with noise (UCS) and other words were not. In the t e s t phase, some of the Input words were again presented. Generalization stimulus words and novel, neutral words were also presented. While GSRs were being recorded, Ss attempted to remember which words were previously presented with noise, previously presented without noise, or not previously presented. This ex p eri ment therefore consisted of two procedures occurring con currently: c las s ica l discrimination GSR conditioning and p aired -asso ciate learning. Stimulus-GSR response s p e c i f i c i t y , instances in which autonomic response patterns vary with the kinds of stimulus which e l i c i t them, was employed as a criterio n indicating that CRA performance is the r e s u l t of a s in g le, direct a sso ciativ e process. Likewise, recognition res- ponse-GSR response s p e c i f i c i t y , instances in which 10 autonomic response patterns vary with the s u b j e c t 's i n t e r pretation of the stimulus rather than with the physical stimulus i t s e l f , was employed as a c r ite r io n ind icatin g that CRA is the re s u lt of a cognitive form of learning. I f a d ire c t relatio n sh ip is observed between the stimulus presented and the autonomic response e l i c i t e d , then the proposal that CRA is the resu lt of a single associative (S-R) process is supported. If such a stimulus-autonomic response relationship is not observed and instead, a d i r e c t re la tio n sh ip is observed between the stimulus recognition response and the autonomic response e l i c i t e d , then the proposal that CRA is the re s u lt of a cognitive (RL) process is supported. The proposal that both S-R and RL learning can be involved in an i n t e r relatio n sh ip during CRA would be supported by data i n d ic a ting a d ir e c t stimulus-autonomic response relatio n sh ip under one set of conditions and data indicating a d i r e c t stimulus recognition response-autonomic response r e l a t i o n ship under a d i f f e r e n t set of conditions. These conditions are described in the next chapter. The One-Trial Acquisition Procedure The functional operational definition of c las s ica l conditioning was GSR modification a t t r i b u t a b l e solely to the temporal pairing of a r e la tiv e neutral stimulus word with a loud, response-producing noise (UCS). Verbal 11 learning was denoted by the c o rre ct i d e n t i f i c a t i o n of t e s t words as previously presented with noise, previously pre sented without noise, or not previously presented. Discriminative GSR conditioning was said to occur when words previously presented with noise (CS+ words) e l i cited a larger mean GSR than words previously presented without noise (CW- words) (Lockhart and Grings, 1963). Semantic generalization of the GSR was said to occur when words re la ted by association to words previously presented with noise (GS+ words) e l i c i t e d a larg er mean GSR than words r e la ted by association to words previously presented without noise (GS- words) (Maltzman, 1968). Semantic generalization of the recognition response was said to occur when novel t e s t words related by association to pre viously presented input words (GS+ and GS- words) were more frequently id e n tifie d as "old" input words than novel t e s t words not related by association to previously presented input words (NW words) (Kimble, 1968). These operational d efin itio n s were applied within and between S s . To t e s t for the above phenomena the independent v a r i able, stimulus word type, consisted of five c l a s s i f i c a t i o n of words: 1. CS+: words followed by UCS and presented during the input (acquisition) and t e s t phase of the experiment. 2. CW-: words not followed by UCS and presented during the input and t e s t phase of the experi ment. 12 3. GS+: pairs of semantically related words,* each followed by UCS and one member of the pair was presented during the input (acqui sitio n ) and the other during the t e s t phase of the experiment. 4. GS-: pairs of semantically related words* of which one member was presented during the input and the other during the t e s t phase of the experiment. 5. NM: novel words presented only once during the experiment and only during the t e s t phase. With the exception of word pairs being re la ted by a s s o c i atio n , none of the input or t e s t words were semantically re la ted to each other by logical relationships (synonyms, antonyms, subordinates, and superordinates) or by phonic s i m i l a r i t y . Each input word was presented only once during acquisition and each t e s t word received only one t e s t t r i a l . The acquisition procedure used in this research has not been reported in the l i t e r a t u r e . Classical d iscrim in ation GSR conditioning experiments usually employ one or a few reinforced and not reinforced stimuli for about a dozen acquisition t r i a l s at a constant or varying i n t e r - t r i a l interval ( IT I) of e ig h t seconds or more ( e . g . , Hart man, 1965; Maltzman, 1968). In the present experiment, there were 16 reinforced and 16 not reinforced stimulus words presented during acqusition at an ITI of three * Words related by association (Palermo and Jenkins, 1964). 13 seconds. Although differences e x i s t between the "usual" and the present experimental procedures, both f u l f i l l the generally accepted stimulus requirements of c la s s ic a l con ditioning: response modification a t t r i b u t a b l e solely to the temporal pairing of a r e l a t i v e l y neutral stimulus (CS) with a potent, response-producing stimulus (UCS) (Hilgard and Marquis, 1940; Grings, 1963). As discussed e a r l i e r , the fu lf illm e n t of c las s ica l conditioning stimulus requirements does not necessarily produce c las s ica l conditioning. Variables such as o r i e n t ing response, fatig u e, and drugs may i n h i b i t or override the effe cts of paired CS-UCS stim ulation. Traditional S-R theories would predict th a t GSR conditioning would not oc cur when using a three-second ITI because the expression of a complete GSR requires more than three seconds. Con temporary S-R th e o rie s, however, no longer maintain that the conditioning of a response is dependent upon the ex pression of the response during a cquisition ( e . g . , Gold s t e i n , ejt al_. , 1965). Also mentioned e a r l i e r , i f an acquisition procedure produces many orienting responses (ORs) and a high level of physiological arousal, Pavlov's ( 1927) theory would predict th a t c la s s ic a l conditioning would not occur. His theory stated th a t strong investigatory responses and an aroused physiology would in h i b i t conditioning. Contemporary orienting reflex and autonomic conditioning theories ( e . g . , 14 Lynn, 1966; Martin and Levey, 1969; Sokolov, 1969), how ever, s t a t e th a t ORs and high arousal i n h i b i t the expres sion of conditioned responses (CRs) but need not i n h i b i t the conditioning process i t s e l f . Since the t e s t t r i a l ITI was t h i r t y seconds 1n this experiment, there was ample time for GSRs to return to base level during these t r i a l s . The long ITI would permit CRs to be e l i c i t e d according to contemporary OR th eo ries. The o n e - tr ia l acquisition procedure employed enabled t r i a l two to be the t e s t t r i a l for denoting conditioning and generalization. If RL learning is exhibited on the t e s t t r i a l ( t r i a l two), the inference is th a t RL learning can occur during c las s ica l autonomic conditioning involving more than one acquisition t r i a l . Whether or not pure S-R learning or classica l d i s crimination conditioning without RL learning can occur can not be determined here. The view that "awareness" or RL is a necessary condition for the e l i c i t a t i o n of c l a s s ic a lly conditioned autonomic responses was in v e stig ate d , however, with recognition performance used as a c r i t e r i o n of know- lege of CS-UCS contingencies or RL. CHAPTER II EXPERIMENTAL HYPOTHESES AND EMPIRICAL PROPOSITIONS Experimental Hypotheses To determine i f the o n e -tria l acq u isitio n procedure can simultaneously produce classica l discrimination GSR conditioning (CRA) and re latio n al learning (RL), the f o l lowing two experimental hypotheses were tested: 1. A stimulus followed immediately with an UCS on i t s f i r s t presentation (CS+ word) will e l i c i t a conditioned GSR on its second pre sen ta tio n . 2. A stimulus followed immediately with an UCS on i t s f i r s t presentation (CS+ word) will on i t s second presentation be i d e n t i f i e d as having been previously presented with an UCS. Conditioning theory and data ( e . g . , Grlngs, 1963) p redict that Hypothesis 1 would be accepted. Common sense suggests th a t Hypothesis 2 would be accepted because i f a recognition t e s t t r i a l is presented immediately a f t e r the a cq u isitio n t r i a l the presupposition is that most Ss would be able to recognize whether or not the t e s t stimu lus was followed with noise. If Hypothesis 1 1s re je cte d , the inference 1s that the effe cts of CS-UCS stimulation was in h ib ited or e x t i n guished p r io r to the t e s t t r i a l . Likewise, 1f Hypothesis 15 16 2 is re jected , the inference is that memory of stimuli presented was inhibited or extinguished p rio r the t e s t t r i a l . The te s t in g of these two hypotheses provided an indication of the effects of one set of stimulus conditions on two d i f f e r e n t response systems. The extents to which stimuli not followed with an UCS can e l i c i t CRA-I1ke responses within the autonomic system and RL-11ke responses within the cognitive system were tested with four additional experimental hypotheses: 3. A stimulus re la ted by association to a stimulus previously presented with an UCS (GS+ word) will e l i c i t a GSR having the appearance of a con ditioned GSR. 4. A novel stimulus related by association to a stimulus previously presented (GS+ or GS- word) will be erroneously i d e n tif ie d as having been previously presented more frequently than a novel stimulus not related by association to a previously presented stimulus (NW word). 5. A stimulus not re la ted by association to a stimulus previously presented with an UCS (CW-, GS-, or NW word) will e l i c i t a GSR having the appearance of a conditioned GSR. Semantic generalization theory and GSR data ( e . g . , Maltzman, 1968) predict th a t Hypothesis 3 would be ac cepted, and Hypothesis 5 would be rejected. Semantic generalization theory and somatic response data ( e . g . , Kimble, 1968) predict th a t Hypothesis 4 would be accepted, and Hypothesis 6 would be rejected. The te stin g of these six hypotheses provided autonomic and cognitive Indices of semantic conditioning and generalization. 17 Empirical Propositions There are currently three major views on the issue of the nature and number of processes involved in the ac q u is itio n of c l a s s i c a l l y conditioned autonomic responses (CRA) (Dawson and Grings, 1966; 1968). One view is th a t CRA is a peripheral refle c tio n of a cognitive relational learning (RL) process, therefore RL is a necessary condi tion for CRA (Woodworth, 1958). This contention was tested by the following empirical proposition: 1. A CS (CS+ word) will e l i c i t a c la s s i c a l l y conditioned GSR only when the CS (CS+ word) is correctly recognized as having been pre viously presented with an UCS. The other two views maintain that c lassical condi tioning is separate and d i f f e r e n t from cognitive RL learning. These views contend th a t CRA can be the r e s u lt of a singular associative (S-R) process and can be the r e s u l t of the S-R process in te r a c t in g with the RL process. These views d i f f e r in that one regards CRA and RL as invol ving somewhat opposing processes of learning (Razran, 1955; 1961) whereas the other view regards RL as one of the determiners of CRA, though not an e ss e n tial one (Kimble, 1962; 1964). The following empirical propositions were te st e d in order to determine i f the "opposing processes" view can be supported: 2. A CS (CS+ word) correctly recognized will e l i c i t a larger GSR CR than a CS (CS+ word) erroneously recognized. 18 3. When a stimulus re la ted by association to the CS (GS+ word) is co rrectly recognized the GSR e l i c i t e d will be larg er than that e l i c i t e d under erroneous recognition. The rejection of Propositions 2 and 3 will support the "opposing processes" view of learning because this position contends that awareness of contingent events and expectation of an UCS lead to proprioceptive stimulation which i n i t i a t e s operant responses th a t have a c o n tro llin g influence on "somatic and visceral reactions" (Razran, 1961, p. 135). The outcome of these propositions however, will not a ffe c t the s tatu s of the "one of the determiners of CRA" pos1t1on. The viewpoint th a t RL is one of the determiners of CRA, that CRA can occur without RL, and that RL is not a s u f f i c i e n t condition for CRA (Kimble, 1962; 1964) was examined with empirical propositions designed to i n v e s t i gate the extent to which RL affe cts CRA. The propositions were: 4. A previously presented neutral stimulus (CW- word) erroneously recognized as being a CS (CS+ word) will e l i c i t a GSR having the appear ance of a GSR CR e l i c i t e d by a correctly recognized CS (CS+ word). 5. A novel neutral stimulus (NW word) erroneously recognized as being a CS (CS+ word) will e l i c i t a GSR having the appearance of a GSR CR e l i cited by a co rre ctly recognized CS (CS+ word). 6. A stimulus r e l a t e d by association to a p r e viously reinforced stimulus (GS+ word) erron- enously recognized as being a CS (CS-*- word) 19 will e l i c i t a GSR having the appearance of a GSR CR e l i c i t e d by a correctly recognized CS (CS+ word). 7. A stimulus re la ted by association to a p re viously presented neutral stimulus (GS- word) erroneously recognized as being a CS (CS+ word) will e l i c i t a GSR having the appearance of a GSR CR e l i c i t e d by a correctly recognized CS (CS+ word). The re jectio n of Propositions 4 and 5 will support the view th a t the expectancy of an UCS or the presence of cognitive RL processes is not a s u f f i c i e n t condition for CRA. Both S-R mechanisms as i l l u s t r a t e d in Figure 1, and/ or RL processes may be responsible for semantic g e n eraliza tion ( e . g . , Razran, 1961; Hartman, 1965; Maltzman, 1968). The acceptance of Proposition 6 will indicate th a t semantic generalization of the GSR can occur a f t e r one acquisition t r i a l . The rejection of Proposition 7 will support the view that such generalization is the r e s u l t of associative S-R effects overriding cognitive RL e f f e c t s . In order to examine i f CRA and RL performances are mediated by a subject variable which affects S-R and RL processes in a sim ila r or co rrelated manner, the following two empirical propositions were examined. 8. Subjects with b e t t e r discriminative stimulus recognition performance ex h ib it b e t t e r c l a s sical discriminative GSR conditioning than subjects with poorer discriminative stimulus recognition performance. Stage 1 - Subject Variable Word»d0g» IARnc a t », I AR»pUppy , etc. C S« j . - ii IA R M + . ii, IARm h i etc. dog ^ cat puppy \ \ Stage 2 - Conditioning ^ \ UCS-------------------=*GSR Stage 3 - Generalization Test»cat n_______ — . . . G S R Figure 1. S-R paradigm of semantic generalization. Solid lines are unconditioned connections. Dashed lines are conditioned connections. In Stage 1, S ^ brings to the experiment an im p licit cognitive set such th a t when the word, "dog" is per ceived, the im plicit associative response (IAR) "cat" is evoked. In Stage 2, the conditioned stimulus (CS) "dog" and i t s IAR "cat" are c la s s i c a l l y conditioned to e l i c i t a conditioned galvanic skin response (GSR). In Stage 3, the generalization t e s t word, "cat," e l i c i t s the previously conditioned GSR ( a f t e r Kimble, 1961, and Brotsky, 1964). ro o Subjects with b e t t e r semantic generalization of the recognition response e x h ib it b e t t e r semantic generalization of the GSR than subjects with poorer semantic generalization of the recognition response. CHAPTER I I I METHOD Desi gn This research was concerned with autonomic and cog n itiv e Indices of semantic conditioning and generalization. Stimulus conditions I d e n tifie d by stimulus word type c l a s s i f i c a t i o n s provided experimental manipulations for demon s t r a t i n g conditioning and generalization. Since temporal and sequential variations a f f e c t con ditioning and learning ( e . g . , Kimble, 1961), input and t e s t word positions were controlled variables. Another con t r o l l e d variable was the association strength or a s s o c i ative verbal habits (determined from word association norms) between input and t e s t words re la ted by a sso ciatio n . Researchers have demonstrated that semantic generalization of autonomic and cognitive responses are affected by such associative habits ( e . g . , Maltzman, 1968; Cramer, 1970). The experimental design was a between and within Ss repeated measure 3 x 5 x 2 x 2 f a c t o r i a l : association strength x stimulus word type x input word position x t e s t word position. Association strength was a between Ss con t r o l l e d variable, word type was a within £ independent 22 23 variable, and input and t e s t word positions were within !S controlled variables. The dependent variables were GSR and verbal recognition responses. Stimulus Variable Materials Associ ation S trength. Two sets of pairs of seman t i c a l l y related GS words were obtained from the Palermo and Jenkins (1964) word association norms. Set 1 con s iste d of words having stronger association strength be tween words within pairs than Set 2. Word pairs selected from Set 1 were assigned to experimental treatment Assoc-1. Word pairs selected from Set 2 were assigned to treatment Assoc-3. Treatment Assoc-2 consisted of an equal number of pairs of semantically related words selected from Set 1 and Set 2. In t o t a l , then, there were three levels of association strength between GS word pairs for the associa tion strength controlled variable: Assoc-1, Assoc-2, and Assoc-3. Stimulus Word Type. The stimulus word independent variable consisted of five c l a s s i f i c a t i o n s of words: CS+, CW-, GS+, GS-, NW . These word types were operationally defined in Chapter I. There were six replications of stimulus materials or stimulus word l i s t s . Within each stimulus word 11st, there were eight d i f f e r e n t CS+ and CW- words, eig h t d iff er en t 24 pairs of GS+ and GS- words, and eight novel or N W words. During the input (acq u isitio n ) phase, the eight CS+, eight CW-, and one member of each of the eight pairs of GS+ and eight pairs of GS- words were presented at a rate of one word every three seconds. The UCS immediately followed the o f f s e t of CS+ and GS+ words. Thirty seconds a f t e r the presentation of the l a s t input word, the t e s t phase of the experiment began. There were 40 t e s t t r i a l s . Eight N W words, eight CS+, eig h t CW-, and the other member of each of the eight pairs of GS+ and eight pairs of GS- words were presented at a rate of one word every th i r t y seconds. Forty d iff e r e n t words were, th e re fo re , presented to each S ^ during the GSR e l i c i t a t i o n and recognition t e s t . During the t e s t phase the UCS followed twenty seconds a f t e r CS+ and GS+ word onset for the purpose of maintaining activation and conditioning e f f e c t s . The CW-, GS-, and N W t e s t words were not followed with UCS. Condi tioning and generalization were distinguished from s e n s i t i zation effects with a within S an tic ip ato r y , d i f f e r e n t i a l responding c r ite r io n ( e . g . , Lockhart and Grings, 1963; Maltzman, 1968). To demonstrate discriminative condition ing, mean GSR e l i c i t e d by CS+ words must be s ig n i f i c a n t l y larger than mean GSR e l i c i t e d by CW- words. To demonstrate gen eralizatio n , mean GSR e l i c i t e d by GS+ words must be s ig n i f i c a n t l y larg er than mean GSR e l i c i t e d by GS- words. Generalization of the recognition response occurs when novel GS+ and GS- t e s t words are erroneously i d e n tif ie d more frequently than N W words ( e . g . , Kimble, 1968). The extent which the GSR was affected by novel neutral t e s t words was determined by measuring GSRs e l i c t e d by N W words. Input and Test Word Posi ti on. In order to monitor position and temporal effects of word p re sen ta tio n , the input phase of the experiment was divided into early (Block 1) and la te (Block 2) word presentation orders. During input, h a lf of the words within each stimulus word c l a s s i f i c a t i o n (CS+, CW-, GS+, GS-) were presented during Block 1, and the remaining h alf were presented during Block 2 with a random procedure which limited reinforced (CS+, GS+) and not reinforced (CW-, GS-) words to a maxi mum run length of three. During t e s t t r i a l s 1 to 20, half of the words within each stimulus word c l a s s i f i c a t i o n in Block 1 and Block 2 plus four N W words were randomly pre sented with a maximum run length of three reinforced items. During t e s t t r i a l s 21 to 40, the remaining half of the words within each stimulus word c l a s s i f i c a t i o n in Block 1 and Block 2 and the remaining four N W words were randomly presented with a maximum run length of three reinforced items. The monitoring of word position and 26 temporal effects can therefore be described with a 2 x 2 table: e a r l y - l a t e input position x e a r l y - l a t e t e s t position. Stimulus L i s t s . Each 11st of stimulus materials contained an adaptation phase of eight random word-numbers (e .g , eleven, s ix ty , two, e t c . ) , an acquisition or stimulus input phase, and a t e s t phase. The adaptation phase was identical on each of the six l i s t s used in the experiment. All of the stimulus words and stimulus word type positions were d if f e r e n t between l i s t s 1, 2, and 3. Lists 4, 5, and 6 counterbalanced l i s t s 1, 2, and 3, respectively in terms of reinforcement ( e . g . , reinforced words in l i s t 1 were not reinforced in l i s t 4). L ist 1 and 4 contained GS words from Assoc-1, l i s t s 2 and 5 from Assoc-2, and l i s t s 3 and 6 from Assoc-3. Stimulus l i s t s are shown 1n Ap- pendi x A. Apparatus The GSR was picked up by 12 mm x 16 mm s i l v e r e le c trodes attached by adhesive tape to the ventral surface of the f i r s t and th ir d f in g e rtip s of the l e f t hand. Skin resistance was measured continuously by a Physiograph M K II GSR Preamplifier. The output was fed d ire c tly into one channel of a Physiograph CA - 200 Amplifier and recorded by a Physiograph Four-A M K Polygraph. Amplification was set at a level which permitted detection of GSR changes 27 as small as 500 ohms. Paper speed was 5 mm/sec. A ground plate was attached to the volar surface of the S>'s l e f t w rist. NASA lo t #E 40-4 Electrode Paste was used at all skin-electrode or ground plate junctions. All experimental stimulus words and the UCS (a one second white noise produced by a Grason-Stadler 901 B Noise Generator) were presented by a Wollensak 1520 AV Tape Recorder, amplified by a Fisher 80-AZ am p lifier, and delivered to S^ through a s e t of Grason-Stadler D 30 16- ohm ear phones. The JE monitored the auditory presenta tions over another set of ear phones in p a rallel c i r c u i t . A General Radio Company Type 1 565-A Sound-Level Meter was used to c a li b r a t e the auditory stim u li. Verbal stimuli were approximately 70 db, UCS was 100 db, and the contin uous background white noise was 55 db. Subjects T h irty -e ig h t male and t h i r t y - f o u r female volunteers from introductory psychology classes at the University of Southern C alifornia served as Ss in the experiment. All Ss had normal hearing a b i l i t i e s , used English as t h e i r primary language, and had GSR base resistance levels lower than 100,000 ohms. Subjects were randomly assigned to one of six t r e a t ment groups (3 levels of association strength x counter balancing reinforced and not reinforced items = 6 groups). 28 Eighteen additional Ss were terminated or t h e i r data were discarded for one of the following reasons: English was not t h e i r native language (7), did not have normal hearing a b i l i t i e s (2), had GSR base resistance levels higher than 100,000 ohms (5), equipment f a ilu r e (2), experimenter e rro r (1), asked to be released from the experiment (1). Procedure The subject s a t in an easy chair facing away from the apparatus console. The experimenter stated the i n s tru c tio n s (Appendix B) which informed £ that the task 1s to remember Input words and whether or not they were f o l lowed with noise, th a t the recognition t e s t contained some input and some new words, and that in this t e s t , input words followed with noise would again be followed with noise and some new words may be followed with noise. These in stru ctio n s also informed £ on how to Indicate S's recognition of t e s t Items as previously presented with noise, previously presented without noise, or not pre viously presented, and how to indicate £ 's confidence in the correctness of the recognition answer given. The £ then answered questions, connected £ to the experimental apparatus, and monitored the experiment from behind £. The apparatus console separated £ from £ but permitted a back view of £ from the neck up. GSR's were recorded continu ously during the adaptation, input ( a c q u isit io n ) , and t e s t phases of the experiment. CHAPTER IV RESULTS Dependent Measures GSR magnitude change was defined as the la rg e st drop in skin resistance which occurred within a 0-10-second Interval ( I -1) and a 10-20-second interval (1-2) a f t e r t e s t word onset.* To calculate the magnitude change score, the skin resistance at the onset of the GSR response (B) and the point of maximum deflection (A) were determined and subjected to the square root conductance transforma tion, AC*5 = (106 (1/A - 1 / B)) •5. Number of GSR changes of 500 ohms or larg er within each of the two response i n t e r vals co n stitu ted the GSR frequency measure. The GSR de pendent measures were, th erefo re, magnitude change and frequency scores occurring in 1-1 and 1-2. In addition, the larg est GSR magnitude and frequency scores occurring in e i t h e r 1-1 or 1-2 were defined as Interval L (I-L) s c o r e s . * GSR's to adaptation words were not analyzed because no more than two Ss within each association strength level group f a i l e d to habituate to word presentation by the sixth adaptation t r i a l . 29 30 Recognition dependent measures were number of co r rect recognition responses and statements of "more" or "less" indicating confidence of correctness of recognition answer. Recognition response latency -- whether recogni tion response onset occurred during 1-1 or 1-2 -- was also recorded. A completely randomized m ulti-factor analysis of variance was conducted to determine the e ffe c ts of input word position (e a rly , l a t e ) , t e s t word position (early , l a t e ) , and association strength (3-levels) on the GSRs e l i c i t e d by the independent v ariab le, stimulus word type. This analysis was conducted on each of the six GSR depen dent measures: I -1, 1-2, and I-L (larg est score within 1-1 and 1-2) magnitude and frequency scores. The results of these analyses (Appendix C) indicated that only stimulus word type had a s i g n i f i c a n t and consis tent e f f e c t on the GSR. For this reason, additional i n f o r mation on the input and t e s t position and association strength variables are presented in Appendix C whereas additional analyses of stimulus word type effe cts are pre sented here. Also shown by these analyses was that of all the GSR dependent measures employed, the 1-1 magnitude score was the most affected by the Independent variable and the l e a s t affected by the controlled variables. Testing of experimental hypotheses and empirical propositions was 31 conducted predominately with the 1-1 magnitude data, a l though reference to the other GSR measures were made when informatl ve. Stimulus Word Type and GSR Responding Experimental Hypothesis 1_. Hypothesis 1 s tates that a stimulus followed Immediately with an UCS on its f i r s t presentation (CS+ word) will e l i c i t a conditioned GSR on Its second presentation. Table 1 shows the mean 1-1 GSR magnitude and standard deviations e l i c i t e d by all word types. Figure 2 1s a graphic presentation of mean 1-1 and 1-2 GSRs e l i c i t e d by the five word types. To determine i f Hypothesis 1 1s sup ported by this data, a one-factor repeated measure analysis of variance (Table 2) and Neuman-Keuls s t a t i s t i c s (Table 3) were conducted.* These analyses indicated a s i g n if ic a n t word type treatment e ffe ct. Magnitude GSRs e l i c i t e d by CS+ words were s i g n i f i c a n t l y larg er than GSRs e l i c i t e d by CW- words (p < .01). Classical discrim inative GSR condi tioning was therefore demonstrated and Hypothesis 1 ac cepted. Responses e l i c i t e d by CS + words were also larger than responses e l i c i t e d by GS+, GS-, and N W words (p < .01). Experlmental Hypothesis 2. Hypothesis 3 states that a stimulus r e la ted by association to a stimulus previously * Analysis of variance in Appendix C did not include res ponses to N W words and did not account for S variance. 22T Figure 2 cv/- NW WOft.0 T Y P t Bar graph of mean GSRs e l i c i t e d by five stimulus word types during t e s t t r i a l s . Unshaded bars indicate 1-1 responses. Shaded bars indicate 1-2 responses. N=72. U > ro 33 TABLE 1 W ORD TYPE AND 1-1 GSR MAGNITUDE MEANS, STANDARD DEVIATIONS, AND F M AX FOR ALL SUBJECTS, N=72 Stimulus Word Type Item CS + CW- GS+ GS- N W F-Max+ Mean .851 .639 . 704 .585 .5 70 1.855 S.D. .410 .345 .356 .301 .331 Note - (S/s total score per word type/8) = S_'s score per stimulus word type. + H artley's F-max. The df upon which each variance is based was assumed to to be 60 for significance te stin g . The obtained F-max was not s i g n i f i c a n t , p > .05. 34 TABLE 2 ANALYSIS OF VARIANCE OF W ORD TYPE AND 1-1 MAGNITUDE GSRs FOR ALL SUBJECTS, N = 72 Source of Variation SS df "MS" .... F~ Between subjects 32. 88 71 Within subjects 14.49 288 Word type (CS+, CW-, GS+, GS-, NW ) 3. 77 4 0.94 24.94** Res 1 dual 10.73 284 0.04 Total 47. 37 359 ** p < .01 35 TABLE 3 NEUMAN-KEULS STATISTICS ON 1-1 MAGNITUDE CHANGES TO STIMULUS WORDS FOR ALL SUBJECTS, N = 72 Word Type Stimulus Word Type CW - GS+ GS- N W CS + 9.28** 6.46** 11.63** 12.31** CW - 2.82* 2. 36 3.03 GS + 5.18** 5.85** GS- 0.68 * p < .05 ** p < .01 36 presented with an UCS (GS+ word) will e l i c i t a GSR having the appearance of a conditioned GSR. The above analyses (Tables 1, 2, 3, and Figure 2) also indicated th a t GS+ words e l i c i t e d s i g n i f i c a n t l y larger GSRs than CM- words (p < .05) and GS- and N W words (p < .01). Responses to GS+ words were not CRs because GSRs e l i c i t e d by GS+ words were s i g n i f i c a n t l y smaller than GSRs e l i c i t e d by CS+ words (p < .01). Semantic generalization of the GSR did occur, however. The GSRs e l i c i t e d by GS + words were re lia b ly larger than GSRs e l i c i t e d by GS- words (p < .01). In this sense, GS+ words e l i c i t e d CR-like re sponses. Subjects responded d i f f e r e n t i a l l y to t e s t words related by association to input words paired and not paired with the UCS. Hypothesis 3 was supported. Although GS+ words were followed with the UCS during the t e s t phase, th is did not condition generalized GSR CRs to be e l i c i t e d by all t e s t words having strong associative relationships with Input phase words. Test phase GS- words had strong associations with Input GS- words and did not e l i c i t GSRs as large as GSRs e l i c i t e d by G S-*- t e s t words during the twenty-second interval a f t e r t e s t word onset. Experimental Hypothesis 5.. Hypothesis 5 s ta t e s th a t a stimulus not r e la ted by association to any stimulus pre viously presented with an UCS (CM-, GS-, or N W words) will e l i c i t a GSR having the appearance of a conditioned GSR. 37 As discussed in the re su lts above, GSRs e l i c i t e d by CW-, GS - , and N W words were not s i g n if ic a n tly d iff e r e n t from one another but were s i g n i f i c a n t l y d iff e r e n t from GSRs e l i c i t e d by CS+ and GS+ words. Hypothesis 5 was therefore rejected. Stimulus Word Type and Recognition Responding Experimental Hypothes1s 2. Hypothesis 2 states that a stimulus followed Immediately with an UCS on its f i r s t presentation (CS+ word) will on its second presentation be id e n t i f i e d as having been previously presented with an UCS. Table 4 gives the proportion of correct recognition responding to the various stimulus word types. Over h alf of all recognition responses to CS+ words were erroneous. Hypothesis 2 was not supported. This suggests that cogni tive RL e ffe c ts were not active or present during h a lf of the t e s t t r i a l s in which GSRs to CS+ words were e l i c i t e d . Likewise, over h alf of all recognition responses to CW- words were erroneous. This suggests the absence of RL e f fects during most of the t r i a l s for determining the d i s criminative GSR conditioning e f f e c t produced by the one- t r i a l acquisition procedure. Experi mental Hypothesis 4. Hypothesis 4 s tates that a novel stimulus related by association to a stimulus pre viously presented (GS+ or GS- word) will be erroneously 38 id e n tif ie d as having been previously presented more f r e quently than a novel stimulus not related by association to a previously presented stimulus (NW word). Semantic generalization of the recognition response occurs when a new word is mistaken for an old, previously presented stimulus word on the basis of some semantic r e l a tio n s h ip , and the response to the old word is e l i c i t e d by the new word (Underwood, 1965). If a 6S+ or GS- word presented in the t e s t phase is mistaken for a semantically related Input GS+ or GS- word presented e a r l i e r , and the unrelated N W word is not mistaken for an input word, Hypothesis 4 would be accepted. Semantic generalization of the recog nition response would be demonstrated. The frequency of instances in which Ss id e n t i f i e d various t e s t words as "old" or previously presented is shown in Table 4. These mean frequencies were s i g n i f i cantly d iff e r e n t (p < .01) as indicated by analysis of v a r i ance in Table 5. Neuman-Keuls s t a t i s t i c s (Table 6) and means (Table 4) indicated that the frequency in which N W words were recognized as "old" words was smaller (p < .01) than the frequency in which GS+ or GS- words were recog nized as "old" words. Hypothesis 4 was supported. Stimulus Word Type and GSR and Recognition Responding Empirical Proposition 1_. Proposition 1 s ta t e s that a CS (CS+ word) will e l i c i t a c la s s ic a lly conditioned GSR 39 TABLE 4 MEANS AND STANDARD TO STIMULUS DEVIATIONS WORDS FOR OF RECOGNITION RESPONSES ALL SUBJECTS, N=72 §timulus Word Type Item C§+ i CW- i GS+ i GS- 1 NW Proportion of Correct Recognition Responding Mean S.D. .491 .229 .450 .217 .536 .209 .516 .263 .587 .247 Frequency of "(31 d" Recog nition Res pondi ng Mean 4.057 3.889 3. 703 3. 703 1.891 S.D. 1.026 1 .057 1.080 1 .063 0.530 Note - Correct recognition response = 1, incorrect recog nitio n response = 0; "old" recognition refers to identifying a t e s t item as previously presented ("old") with or without noise. Each "old" response * 1. (S/s total score per word type/8) = Si's mean score per stimulus word type. 40 TABLE 5 ANALYSIS OF VARIANCE OF W ORD TYPE AND "OLD" RECOGNITION RESPONSES FOR ALL S U B J E C T S , N = 72 Source of Variation "' s $ ...... w - "■■MS' F Between subjects 314.55 71 Within subjects 247.34 288 Word type (CS+, CW-, GS+, GS-, NW ) 224.72 4 56.18 705.45 Res i dual 22.62 284 0.08 Total 561.89 359 Note - "old" recognition refers to id e n tify in g a t e s t item as previously presented ("old") with or without noise. Each "old" response = 1. (S/s to ta l score per word type/8) = S/s score per word type. 41 TABLE 6 NEUMAN-KEULS STATISTICS ON "OLD" RECOGNITION RESPONSES TO STIMULUS WORDS FOR ALL SUBJECTS, N=72 Word Type Stimulus Word Tvoe CW - GS + GS- N W CS + 5.06* 10.65* 10.65* 65.15* CW- 5.59* 5.59* 60.08* GS + 0.00 54.50* GS- 54.50* Note - "old" refers to recognition responses indicating previously presented words. * p < .01 42 only when the CS (CS+ word) is correctly recognized as having been previously presented with an UCS. The frequency of various recognition responses e l i c i t e d by the five stimulus word types is shown in Table 7. These frequencies were used in the p a r titio n i n g of the GSR data in order to obtain weighted GSR means. Table 8 presents weighted GSR means as a function of word type and recognition response. Figure 3 is a graphic presentation of this data. These results were replicated with the GSR frequency dependent measures. Proposition 1 was tested by determining confidence limits for obtained mean of mean GSRs to CW- words, the control stimuli for discrimination conditioning. If CS + words e l i c i t mean responses f a l l i n g within or below the CW- confidence i n t e r v a l , then CR e l i c i t a t i o n by these CS+ words did not occur. Stated 1n the p o s itiv e , mean responses e l i c i t e d by CS+ words f a l l i n g above the CW- confidence interval are CRs (the acquisition of CRs was demonstrated in the "Stimulus Word Type and GSR responding" section of this Chapter). Table 8 indicated that both correctly and Incor rectly recognized CS+ words e l i c i t e d magnitude means o u t side the confidence interval defining responses as not CRs. Regardless of the recognition response given, CS+ words e l i c i t e d GSR means having values within the limits defining 43 TABLE 7 W ORD TYPE X RECOGNITION RESPONSE FREQUENCY FOR ALL SUBJECTS, N=72 Stimul us Recoqn1t1on Response Word Type Yes no New CS + 283 204 89 (T o tal) 576 CW- 186 259 131 5 76 GS+ 79 188 309 5 76 GS- 79 200 29 7 576 N W 62 176 338 576 (Total) 6 89 1027 1164 2880 TABLE 8 W ORD TYPE X RECOGNITION RESPONSE CONTINGENCY TABLES: WEIGHTED GSR MAGNITUDE M EAN OF MEAN PER TRIAL Stimul us Recognition Response Row Word Type Yes 1 No 1 New LL Mean UL* Interval 1-1 CS+ .912 .762 .864 .755 .851 cw- .699 .625 .581 .639 .720 GS + .631 .670 .743 .620 .704 GS- .688 .575 .564 .585 .656 N W .579 .554 .576 .570 Column Mean .768 .6 39 .640 Interval 1-2 CS+ .367 .429 .457 .335 .403 cw- .315 .275 .255 .283 .335 GS+ .459 .410 .414 .347 .419 GS- .367 .361 .286 .323 .384 N W .303 .304 .242 .268 Column Mean .431 .352 .317 Note - see Table 7 for number of observations per c e ll . * LL = lower lim it , UL=upper lim it , of a 2 - ta il t^-test 95% confidence interval for means. Recognition: 1 = previously presented with noise 2 = previously presented without noise 3 = not previously presented * = correct recognition response * xd 1 A <2> .1 IKO^NITIOM w o r d T r r « \ \ \ | \ \ \ \ \ \ \ \ \ \ \ \ I* I 3 c s * \ \ \ \ ' \ \ I » C W - / / / 1 I * 3* OS* / ✓ 1/ / / / ✓ / ✓ / / / / s / / / / / / / / / / < / / / / / V / / / / / I 1 39 G S - I t »* N W Figure 3. Bar graph of weighted mean GSRs e l i c i t e d by five stimulus word types as a function of stimulus recognition responses during t e s t t r i a l s . Unshaded bars indicate 1-1 responses. Shaded bars indicate 1-2 responses. N=72. 46 responses as CRs. Proposition 1 was not supported. A CS need not be correctly recognized as having been previously presented with an UCS in order for that CS to e l i c i t a con ditioned GSR. Empi rical Proposi tion 2 and 3. Proposition 2 s tates that a CS (CS+ word) correctly recognized will e l i c i t a larger GSR CR than a CS (CS+ word) erroneously recognized. Proposition 3 s tates that when a stimulus related by association to the CS (GS+ word) is correctly recognized, the GSR e l i c i t e d will be larger than that e l i c i t e d under erroneous recognition. As shown in Table 8 and Figure 3, Proposition 2 was supported by the 1-1 GSRs but not by the 1-2 GSRs. Pro position 3 was supported by the 1-2 GSRs but not by the 1-1 GSRs. Empi ri cal Propositions 4 and 5_. Proposition 4 s ta t e s that a previously presented neutral stimulus (CW- word) erroneously recognized as being a CS (CS+ word) will e l i c i t a GSR having the appearance of a GSR CR e l i c i t e d by a cor re c tly recognized CS (CS+ word). Proposition 5 s tates that a novel neutral stimulus (NW word) erroneously recognized as being a CS (CS+ word) will e l i c i t a GSR having the appearance of a GSR CR e l i c i t e d by a correctly recognized CS (CS+ word). Empirical Propositions 4 and 5 were not supported by the data in Table 8. The expectation of an UCS as 47 Indicated by recognition responses did not lead to the e l i c i t a t i o n of CR-I1ke GSRs when the t e s t stimulus was a neutral previously presented (CW- word) or a neutral novel (NW word) stimulus. Expectatlng an UCS was r e la ted to larger GSR responding to these neutral words, however, than not expecting an UCS (Figure 3). Empirical Proposition 6 and 7. Proposition 6 s tates that a stimulus r e la ted by association to a previously reinforced stimulus (GS+ word) erroneously recognized as being a CS (CS+ word) will e l i c i t a GSR having the appear ance of a GSR CR e l i c i t e d by a correctly recognized CS (CS+ word). Proposition 7 s tates that a stimulus related by association to a previously presented neutral stimulus (GS- word) erroneously recognized as being a CS (CS+ word) will e l i c i t a GSR having the appearance of a GSR CR e l i c i t e d by a co rrectly recognized CS (CS-*- word). Propositions 6 and 7 were not supported by the 1-1 GSR data in Table 8. Test stimuli related by association to previously presented stimuli (GS+ or GS- words) recog nized as CS+ words did not e l i c i t 1-1 GSR means having values within the confidence interval identifying CRs. The 1-2 GSR data, however, supported these propositions. Regardless of the recognition response given, GS+ words e l i c i t e d a la rg e r GSR mean than GS- words during both 1-1 and 1-2 i n t e r v a l s . In this sense, stimulus effe cts 48 override recognition response e ffe c ts on the dependent GSR measures. Table 8 and Figure 3 suggest a complex i n t e r action between the stimulus presented and S's recognition of that stimulus and the GSR. Expecting or not expecting an UCS (indicated by recognition responses "yes" and "no," respectively) appeared to have affected the GSR e l i c i t e d by GS+ and GS- words d i f f e r e n t l y . For example, GS+ words e l i c i t e d a smaller 1-1 GSR mean when the UCS was expected than when i t was not expected whereas GS- words e l i c i t e d a larg er 1-1 GSR mean when the UCS was expected than when i t was not expected. To determine i f the re su lts in Table 8 were affected by a recognition a b i l i t y variab le, Ss were separated in terms of overall recognition performance. Twenty-six Ss were i d e n tif ie d as good recognition scoring Ss and 21 Ss were id e n t i f i e d as poor recognition scoring Ss (see Ap pendix D). Mean 1-1 and 1-2 GSR magnitudes as a function of stimulus word type for these two subsamples are shown in Appendix D. Analysis of variance conducted on these scores indicated s i g n i f i c a n t word type treatment effects and Neuman-Keuls s t a t i s t i c s revealed that s i g n i f i c a n t discrim inative GSR conditioning occurred on these measures for both good and poor recognition scoring subjects (Appendix D, p < .01). Semantic generalization of the GSR was s i g n i fic a n t only in 1-2 responses, however, and was more 49 r e lia b le in the good recognition subsample (p < .05) (Ap pendix D). These findings suggest that the relationships shown in Table 8 were due to the pooled re su lts of all sub jects and not simply a function of a subsample of Ss showing good or poor overall recognition performance. Empi ri cal Proposi ti ons 8 and £. Proposition 8 states that subjects with b e t t e r discriminative stimulus recogni tion performance ex h ib it b e t t e r c la s s ic a l discriminative performance. Proposition 9 s tates that subjects with b e tte r semantic generalization of the recognition response ex h ib it b e t t e r semantic generalization of the GSR than subjects with poorer semantic generalization of the recog nition response. These propositions were evaluated by identifying Ss in terms of good, average, and poor discriminative stimulus recognition performance (Appendix E). Subjects were also id e n tif ie d in terms of good, poor, and no discriminative GSR conditioning and GSR generalization (Appendix F). Chi square (x2) analyses were then conducted to determine the re la tio n sh ip between CS+ and CW- word recognition perform ance and discriminative GSR conditioning. Such analyses were also conducted on GS+ and GS- word recognition per formance and GSR general 1z a t i o n . A lack of dependence between CS+ and CW- word recognition a b i l i t y and classical discrim inative GSR 50 conditioning was shown by chi square (x2) analyses: xdf=4 = 7 «50 for GSRs ( p > -05) anc* xdf=4 = 8.85 for 1-2 GSRs (p > .05). Proposition 8 was not supported. Ch1 square analysis of three levels of 1-1 and 1-2 magnitude conditioning p er formance also suggested an independence between such r e s ponding within Ss, x j f _4 = 0.79. The e l i c i t a t i o n of c l a s s i c a l l y conditioned GSRs by semantic generalization t e s t words associatively related to acquisition words occurred in both good and poor stimu lus recognition scoring Ss (Appendix D). Chi square analyses also indicated th a t Ss exhibiting good semantic generalization of the conditioned GSR did not necessarily ex h ib it good GS+ and GS- word recognition a b i l i t y : xdf=4 = 2.84 for 1-1 GSRs (p > .05) and xdf=4 = 4.03 for 1-2 GSRs (p > .05). Proposition 9 was not supported. Chi square analysis of three levels of 1-1 and 1-2 magnitude generalization also suggested an independence between such responding within Ss, x jf=4 = 0.89. Recognition Response Parameters Confidence of C orrectness. Recognition response answers were rated by S ^ in terms of "confidence of C orrect ness." Table 9 shows mean confidence of correctness scores 51 given to the various word types. Analysis of variance (Table 10) indicated that Ss degree of confidence in the correctness of th e ir recognition answer was s i g n i f ic a n tly d i f f e r e n t (p < . 0 1 ) for the various word types regardless of whether the recognition answer was rig h t or wrong. Neuman-Keuls s t a t i s t i c s (Table 11) revealed that Ss were more confident of the correctness of answers given to CS+ words than of answers given to all other stimulus word types (p < .01). Confidence of correctness of recognition answers to CW- words was also s i g n i f i c a n t l y higher than th a t expressed for answers given to GS+ and to N W words (p < .05). Subjects were l e a s t confident of the correctness of t h e i r recognition responses to N W words. In other words, Ss expressed g reater confidence of correctness for answers given to previously presented or more fa m iliar stimuli (CS+ and CW- words) than to new or novel stimuli (GS+, GS-, and N W words) (Table 9) although the frequency of correct recognition responding was larg er for the new stimuli than for the previously presented stimuli (Table 7). The inference here is that a time variable affected correct recognition responding and con fidence of recognition response in an inverse manner. A stimulus word type x confidence of correctness of recognition response contingency table (Table 12) in d i cated th a t t e s t words correctly recognized were given 52 TABLE 9 MEANS AND STANDARD DEVIATIONS OF LATENCY AND CONFIDENCE OF CORRECTNESS OF RECOGNITION RESPONSES TO STIMULUS WORDS FOR ALL SUBJECTS, N=72 Stimulus Word type I tem CS + CW- GS + GS- N W Confidence of Correctness of Recognition Responding Mean 1.559 1.469 1.399 1.420 1.392 S.D. .230 .197 .210 .209 .233 Laitency of Recogni tion Respond!nc Mean 1.201 1 .248 1.273 1 .245 1.253 S.D. .207 .226 .250 .216 .232 Note - More confident of correctness of recognition r e s ponse = 2 , less confident = 1 ; recognition response occurring during 1-1 = 1 , that occuring during 1-2 = 2 (_S' s total score per word type/ 8 = S/s score per word type. 53 TABLE 10 ANALYSIS OF VARIANCE OF W ORD TYPE AND CONFIDENCE OF CORRECTNESS OF RECOGNITION RESFffNSl ---------- F U T F A lL T uB 'JE 'C T S , N= 72----------------------- Source of Variation -------■ W " "M S -------- F........... Between subjects 8.29 71 Within subjects 9.67 288 Word type (CS+, CW-, GS+, GS-, NW ) 1. 37 4 .34 11.70** Res i dual 8 . 30 284 .03 Total 1 7.96 359 Note - more confident = 2, less confident = 1. (§_' s total score per word t y p e / 8 ) = S/s score per word type. ** p < .01 54 TABLE 11 NEUMAN-KEULS STATISTICS ON CONFIDENCE OF CORRECTNESS OF RECOGNITION RESPONSES TO STIMULUS WORDS FOR ALL SUBJECTS, N = 72 Word Type Stimulus Word Type CW- GS + GS- N W CS + 4.48* 7.9 3* 6.89* 8.27* CW- 3.45** 2.41 3.79** GS + 1.03 0.34 GS- 1.38 * p < .01 ** p < .05 55 TABLE 12 STIMULUS W ORD TYPE X RECOGNITION RESPONSE CONTINGENCY TABLES: RECOGNITION LATENCY AND CONFIDENCE OF CORRECTNESS FOR ALL SUBJECTS, N=72* Stlmulus Recognition Response Word Type Yes no New Confldence Weighted Mean/Trial (Row Mean) CS + 1.696 1.441 1.39 3 1.559 CW - 1.495 1.506 1. 359 1.469 GS + 1. 329 1.239 1.551 1.399 GS- 1. 304 1.320 1.519 1.420 N W 1.242 1.26 7 1 .485 1.39 2 (Total) 1. 514 1 .36 7 1.480 Latency Weighted Mean/Trial (Row Mean) CS+ 1.134 1.260 1.281 1.201 cw- 1.188 1.286 1.260 1.248 GS + 1. 354 1 .362 1.197 1.273 GS- 1. 342 1. 320 1.168 1.245 N W 1.371 1. 364 1.175 1.25 3 (T o tal) 1.219 1 .315 1.197 * Weighted mean of mean score per t r i a l . Larger values Indicate longer latency or more confl dence. See Table 9 for scoring procedure. 56 higher ratings of confidence of correctness than t e s t words not correctly recognized. These re la tio n s h ip s, also shown in Figure 4, supported s im ila r findings obtained in verbal learning experiments involving recognition and con fidence ( e . g . , Underwood and Freund, 1968). The confidence of correctness measure suggested that the recognition r e s ponse (yes, no, or new) was a truncated measure of r e l a tional learning (RL). To determine i f the c rite rio n em ployed for denoting RL was truncated and biased, questions concerning absolute awareness and r e la tiv e unawareness ( e . g . , Erikson, 1960) and questions concerning presolution learning and absolute learning ( e . g . , Dixon and Moulton, 1970) must f i r s t be answered. Latency of Recognition Response. There was s i g n i fic a n t variation in the latency of the recognition r e s ponse (F4 J 0 = 2.51, p < .05) among the various stimulus word types. Table 9 indicated th a t recognition response latencies were longest for GS+ words and s h o r te st for CS+ words. Neuman-Keuls s t a t i s t i c s revealed that of all pos s ib le stimulus word type comparisons on the latency mea sure only the GS+ and CS+ means were sig n if ic a n tly d i f ferent (q = 4.31, p < .05). A stimulus word type and recognition response contingency table of latency data (Table 12) showed th a t the latencies of correct recogni tion responses were s h o rte r than the latency of incorrect recognition responses except for responses to CM- words. 1 .7 It Recogni ti on: 1 2 3 * previously presented with noise previously presented without noise not previously presented correct recognition response I.V • 1 4 14 14 WMI TVM I * 2 3 CS+ I I* 3 CW- I l F G S* I 2 3 * G S - t I 3 * N W gure 4. Bar graph of weighted mean Confidence of Correctness of Recognition Response^,, Scores as a function of recognition response and stimulus word type. N=72. CHAPTER V DISCUSSION In his discussion of verbal-perceptual factors in the conditioning of autonomic responses, Grings (1965) s tated th a t "the researcher in autonomic conditioning can at the present time neither handle these variables well conceptually nor can he control them adequately by such means as disguising his experimental situ a tio n " (p. 85). The question of the occurrence of human discrimination conditioning of autonomic responses (CRA) in the absence of awareness of CS-UCS contingencies (RL) was, therefore, investigated in this d is s e r ta tio n by a conditioning and 's t i m u l u s recognition procedure designed to monitor both autonomic and cognitive indices of semantic conditioning and generalization simultaneously. One major experimental concern was the relationship between the e l i c i t a t i o n of c l a s s i c a l l y conditioned GSRs and stimulus recognition pro cesses in normal human subjects. The re su lts obtained by this experiment demonstrated that r e lia b le c la s s ic a l discriminative GSR conditioning (CRA) and the semantic generalization of such conditioning did occur with the one t r i a l acquisition procedure 58 59 employed. The p a r t i t i o n i n g of the data on the basis of recognition response indicated that the e l i c i t a t i o n of conditioned GSRs and the semantic generalization of con ditioned GSRs occurred independently of whether the t e s t word was co rrectly or erroneously recognized as having been previously presented with the UCS. Also, no s i g n i f i c a n t re la tio n s ip was observed between the e l i c i t a t i o n of con ditioned GSRs and S's overall and sp ec ific discriminative stimulus recognition performance. The assumption th a t stimulus recognition performance could be employed as a c r ite r io n of relational learning was questioned when correctly recognized t e s t items were given higher confidence ratings of correctness than items erroneously recognized. This finding suggested that the recognition response might have been a truncated measure of RL. Before this question can be answered however, the issues of absolute awareness and r e la tiv e unawareness and of absolute learning and presolution learning must f i r s t be s e ttle d . Even i f the recognition response is a truncated measure of RL, th is does not necessarily mean th a t the results obtained were biased in terms of a ffe ctin g the te stin g of hypotheses and propositions. Assuming that RL is a continuous v a ria b le, the use of a q u a l i t a t i v e depen dent measure means a loss of p recision, hence, a loss in the strength of the conclusions. There is no reason for 60 saying th a t this loss is greater or le ss, higher or lower, or b e tte r or worse for one response category than for an other (Guilford, 1965). The Dependent Measures Several sources of a r t i f a c t need to be considered before in te rp re tin g this data because variables other than perceptual-memory variables may influence the verbal report of stimulus recognition. Most important of these potential biasing a r t i f a c t s were the forced-choice stimu lus recognition method employed, JJ's response to the f a ilu r e of correctly identifying recognition t e s t s tim u li, and S/s in te r p r e ta tio n s and expectations of the experiment and each given t e s t t r i a l . Although randomization procedures were used, there was no guarantee th a t they affected the GSR and the verbal response system in an equivalent manner. Sources of errors t h a t occur in GSR recording whether due to the equipment or to the nature of the response i t s e l f may prevent a f a i r comparison of the two response systems. More crucial s t i l l , although e f f o r t s were made for the purpose of a t tempting to make the GSR and verbal response measures more comparable ( e . g . , confidence r a tin g s , multiple measures of the GSR), there is no indication that they are in fact comparable. The r e s u lts of this experiment do suggest, 61 however, th a t the two response systems do not or need not covary with one another as a function of stimulus condi tions . Subjects might have been able to receive some feed back indicative of t h e i r recognition performance because they were told that words previously followed with noise during acquisition would again be followed with noise and th a t some new words may be followed with noise during the recognition t e s t phase. The p o s s i b i l i t y of such a biasing e f f e c t on GSR magnitude responding to the various stimulus word types was discounted because analysis of variance revealed no s i g n i f i c a n t interaction (p > .05) between word type and the t e s t position variable. A tendency was ob served, however, for discriminative GSR conditioning and semantic GSR generalization to be b e t t e r on l a t t e r t e s t t r i a l s . Standard deviations for both GSR and recogni tion responding indicated t h a t the sample te st e d was homo geneous on these measures. GSR and Recognition Responding The c r it e r i o n for indicating the presence of r e l a tional learning (RL) during the t e s t t r i a l s was correct recognition responses to CS + and CW- t e s t words. I f RL was present, £ would have been able to d i f f e r e n t i a t e be tween which of these two word types were presented during input with or without noise. 62 Given e i t h e r a CS+ or CW- word on a t e s t t r i a l , the probability of making a correct recognition response on the basis of chance alone was .33 because there were three possible types of recognition responses: "yes" or previously presented with noise, "no" or previously pre sented without no ise, and "new" or not previously pre sented. For correct recognition responses to indicate the presence of RL, the proportion of correct recognition responding for each word type must occur above .33, the proportion of correct recognition responding expected on the basis of chance alone. Eighteen Ss had proportions of correct recognition responding above .33 on both CS+ and CW- t e s t items, 19 S^s had proportions above .33 on one of these t e s t items, and 35 had proportions a t or below .33 on both CS+ and CW- t e s t items. Using this c r i t e r i o n , these findings i n d i cated that a minority of the Ss (18 or 25%) were aware of a CS-UCS contingencies a f t e r one acquisition t r i a l . A stimulus followed immediately with an UCS on its f i r s t presentation was observed to e l i c i t a conditioned d i s criminative GSR on i t s second presentation in 58 Ss (81 %) (Appendix F) . After one acquisition t r i a l then, more s demonstrated r e lia b le CRA than re lia b le RL in this exper iment. Although RL did occur in a minority of S/s and confidence ratings suggested the presence of presolution 63 RL learning effe cts in some S.'s, this experiment can be regarded as a demonstration of GSR CR e l i c i t a t i o n in the absence of absolute RL because the majority of the subjects did not e x h ib it absolute RL when they exhibited re lia b le CRA. The investigation of the extents to which stimuli not followed with an UCS can ex h ib it CRA-like responses and RL-like responses indicated that semantic generaliza tion of both GSR and recognition responses can occur a f t e r one acquisition t r i a l . The GSR was also observed to be more related to stimulus conditions than to stimulus re cognition conditions. The inference is that stimulus-GSR s p e c i f i c i t y overrides recognition response-GSR s p e c i f i c i t y in the e l c i i t a t i o n of GSR-CRs acquired a f t e r one acq u isi tion t r i a l . The view that CRA is a peripheral r e f le c tio n of a cognitive RL process and th a t RL is a necessary con dition for CRA was therefore rejected. It is suggested that data supporting the "peripheral reflection" view may be accounted for by the fact that such data ( e . g . , Woodworth, 1958) were obtained from experiments employing more than one acquisition t r i a l . The suggestion is that as the number of acquisition t r i a l s increase, the probability of RL occurring also increases. An on e -tria l acquisition experiment does not neces s a r ily present an accurate description of CRA and RL pro cesses however. For example, i t is presupposed that i f a stimulus recognition t e s t t r i a l Is presented immediately a f t e r the acquisition t r i a l , most Ss would be able to recognize whether or not the t e s t stimulus was followed with an UCS. I t is suggested here that a memory variable was responsible for the rejection of Experimental Hypo- theis 2, a stimulus followed immediately with an UCS on its f i r s t presentation (CS+ word) will on i t s second presentation be i d e n t i f i e d as having been previously pre sented with an UCS. There is no reason for s ta t in g that RL will not occur a f t e r one acquisition t r i a l . There are reasons, as the re jectio n of Empirical Proposition 1 (a CS (CS+ word) will e l i c i t a c l a s s i c a l l y conditioned GSR only when the CS (CS+ word) is correctly recognized as having been previously presented with an UCS) in d icated , to s t a t e th a t GSR CRs can occur in the absence of absolute RL. The empirical propositions were evaluated with res- ponse-response analyses. Conclusions concerning Proposi tions 2 to 7 could not be made because propositions sup ported by the 1-1 GSR data were not supported by the 1-2 GSR data and vice versa. The re su lts obtained, however, were compatible with the view th a t CRA can occur without RL, that RL is one of the determiners of CRA, and that RL is not a s u f f i c i e n t condition for CRA (Kimble, 1962, 1964). The results also suggested th a t the semantic generalization of the GSR obtained by the o n e - t r i a l acquisition procedure was related more to the e f f e c t s of CS-UCS pairing than to the e ffe c ts of expecting an UCS. Proposition 8 (Ss with b e t t e r discriminative stimulus recognition performance e x h ib it b e t t e r c las s ica l d is c r im i native GSR conditioning than Ss with poorer discrim inative stimulus recognition performance) and Proposition 9 (Ss with b e t t e r semantic generalization of the recognition response e x h ib it b e t t e r semantic generalization of the GSR than Ss with poorer semantic generalization of the recogni tion response) were not supported. These negative resu lts suggest th a t the e l i c i t a t i o n of conditioned GSRs and recog n itio n responding are not mediated by a subject variable which affects asso c iativ e S-R and cognitive RL processes in a sim ilar or co rre la ted manner. The present experiment did not demonstrate th a t RL learning was absent during the acquisition t r i a l i t s e l f . Cognitive RL processes might have been active during the acquisition t r i a l or during CRA but l a t e r became inactive or absent during the e l i c i tation of GSR-CRs. The next section discusses these implicati ons. Converging Operations and Convergent Theorizing There are several implications of the preceding findings which indicated that the e l i c i t a t i o n of the GSR CR is not dependent upon the correct recognition of the CR e l i c i t i n g stimulus. The most dramatic might perhaps be the suggestion that certain types of aversive learning 66 are more e f f i c i e n t l y acquired or are more r e s i s t a n t to extin ctio n within the autonomic nervous system as exhibited by the GSR than within the central nervous system as reported by verbal responses. Paraphrases of this suggestion, however, commonly appear in the clin ic al l i t e ra tu re in expressions l i k e , "I feel afraid whenever I see a spotted dog, but I do not remember ever having been hurt by o n e ." The same suggestion can be stated with the emphasis placed on a memory process rather than a learning process. With this o r ie n t a t i o n , storage and re triev al constructs have important functional roles. Given this o r ie n t a t i o n , the experimental question in this d is se r ta tio n would then involve the storage of auditory information and its subse quent re trie v a l and manifestation as a function of the GSR and verbal response systems. For example, the ex p eri ment conducted here may be called classical conditioning or paired-associate learning depending upon the response system and theory u t i l i z e d or emphasized. As a r e s u l t of this kind of formulation, the c l a s sical discriminative autonomic conditioning phenomena can s t i l l be investigated but without the necessity of d e t e r mining "what is awareness" or "how much of i t 1s neces sary?" Such a formulation is deemed reasonable because c las s ica l conditioning e x ist s only in terms of a p a r t i c ular s e t of experimental conditions or operations (Spence, 67 1956; Kimble, 1961; Grlngs, 1963). The conditioning pro cedure described in this d i s s e r ta tio n is proposed as a new approach for the investigation of autonomic conditioning phenomena and verbal and cognitive factors within the conceptual framework of e i t h e r learning theories or memory models or both. H is to ric a l ly , from the tenets of operationism, classica l conditioning is not distinguishable from the experimental procedures on which i t is based and therefore CRA and RL are not distinguishable from the responses which indicate t h e i r existence and character. That i s , "The concept is synonymous with the corresponding set of opera tio n s," (Bridgman, 1927, p. 5) but this does not mean that any operation can produce a concept. Bridgman (1945) l a t e r emphasized this d is tin c tio n : Operational d e f i n i t i o n s , in s p ite of t h e i r pre cisio n , are in application without significance unless the situ a tio n s to which they are applied are s u f f i c i e n t l y developed so that at le a s t two methods are known of getting to the terminus. (p. 248) The CRA-RL issue resulted from observations in d ic a ting that a set of experimental procedures called classical conditioning ( e . g . , paired auditory stimulation) can produce resu lts which define the CRA concept within one response system ( e . g . , GSR) and can also simultaneously produce results which define the RL concept within another response system ( e . g . , verbal re p o rts). I t is proposed 68 that additional properties of both response systems can and must be considered as a function of various converging operations distinguished as c lassical conditioning. Again, "The distinguishing feature of classica l con ditioning is a p a r t i c u l a r set of experimental conditions or operations. Two stimuli are presented in a temporal relation (p a irin g ) , and changes in behavior with reference to one of the stimuli are produced" (Grings, 1963, p. 495). This operational d e f in itio n of classical conditioning r e quires only that "changes in behavior with reference to one of the stimuli" is produced. No s t ip u la tio n or limiting condition 1s placed on the "change in behavior with reference to one of the stimuli" requirement defining classica l conditioning. This requirement, however, may be used to indicate the presence of divergent concepts like CRA and RL, depending upon the response system of the de pendent variable. At issue here is not whether or not there is a con tingency relationship between stimuli and responses; this is a given by d e f in itio n . At issue is the nature or con ceptualization of this contingency relationship between stimuli and responses and, hence, whether or not the nature or conceptualization of this contingency r e l a t i o n ship is dependent upon the s p ec ific response system in volved. Classical conditioning is defined independent of response modalities. I f CRA and RL are also defined as concepts independent of s p e c if ic response systems, there are rules or converging operations for inductively d e t e r mining the properties of these concepts. Since CRA and RL are also concepts u ti l i z e d for the purpose of explaining the classical conditioning phenomenon deductively, i t is necessary to r e la te these concepts to t h e i r theoretical origins which in this case began with the s t i l l unresolved systematic issue of "Cognitive versus Stimulus - Response Theories of Learning" and questions concerning what is learned ( e . g . , Goldstein, e t aH_. , 1965). The methodical o rie n ta tio n proposed is that i f con cepts having th e o re tic al origins related to unresolved systematic issues are to be employed for explanatory purposes, data from at l e a s t two dependent measures should be obtained. A form of "convergent theorizing," analogous to Bridgman's "converging operations," is suggested for in te rp re tin g such data. Convergent theorizing as conceptualized here is the study of two or more concurrent dependent measures with concepts from more than one theory. Concepts from separate theories may then be u t i l i z e d for explaining the resu lts of each dependent measure. These concepts may have theoretical origins re la ted to unresolved systematic issues. These concepts, however, must be logically and operationally compatible across all dependent variables studied. The concepts from d i f f e r e n t theories may then 70 be related with coordinating d e f in itio n s . In this manner, unresolved systematic issues may be studied inductively. Converging operations have been used h e u r i s t i c a l l y to define concepts in the history of experimental psycho logy. It is suggested here th a t when given a behavioral phenomenon such as c las s ica l conditioning, "convergent theorizing" would aid in the formulation and selectio n of concepts employed for explaining the data observed. For example, autonomic responses may be analyzed within the framework of conditioning th e o rie s, verbal responses may be studied in terms of memory models, and these two con ceptual systems may be lo g ically related by coordinating definitions or some form of convergent theorizing. At this stage in the development of psychology as a science and as a body of knowledge, there is no reason to expect that all behavioral responses across all e f f e c t o r systems e l i c i t e d by one stimulus can be explained within one con ceptual system, esp e c ially since subjects have varying a b i l i t i e s to e d i t t h e i r responses in the various response systems. This d i s s e r ta tio n presents a conditioning proce dure which enables the r e s u l t a n t behavioral changes to be studied simultaneously within and between response systems. Verbal Recognition Responding The verbal recognition results obtained in this c l a s sica l autonomic conditioning experiment replicated findings frequently observed in tr a d i t i o n a l verbal learning and memory and signal detection experiments: 1) False recognition is more frequent for novel t e s t words a sso ciati vely related to acquisition words than for novel t e s t words not associa- tiv e ly re la ted to acquisition words ( e . g . , Table 4; Kimble, 1968). 2) A loud noise or tone presented contingent with a stimulus f a c i l i t a t e s the correct recogni tion of th a t stimulus ( e . g . , Table 4; Thet- ford, ejt aj_. , 1968). 3) Stimulus f a m ilia riz a tio n decreases verbal response latency* ( e . g . , Table 9; Hall, 1971). 4) More fa m iliar (repeated) items are given g reater confidence of correctness of recog nition answers than less fam iliar (novel) items ( e . g . , Table 9; Hall, 1969). 5) Items co rrectly recognized are given higher confidence of correctness ratings than items erroneously recognized ( e . g . , Table 12; Under wood and Freund, 1968). 6 ) Latency of correct verbal reports is s h o rte r than th a t for in c o rre c t verbal reports ( e . g . , Table 12; Hall, 1971). The above observations therefore indicated that classica l autonomic conditioning contemporaneously con ducted during a verbal learning and recognition task did not influence the direction of the l a t t e r ' s resu lts in this experiment. Also, since r e lia b le autonomic condi tioning occurred in this experiment in sp ite of the con current recognition task, i t is again suggested th a t two * Latency of recognition responding was s h o rte r for "old previously presented items, than for “new," not pre viously presented items in this experiment. response systems may be studied simultaneously as a func tion of one set of experimental conditions. An In te rp r e ta tio n of this research applied by othe (see Sternback, 1966) to t h e i r findings, has been appro p r ia te ly written by Chester W. Darrow (1967), one of the founding fathers of Psychophysiology: The domain of GSR is thus a hierarchy wherein the ' corti cal-most-hi gh ' speaks only from the secrecy of the cerebral sanctum sanctorum through his m inisters, and they through t h e i r emissaries, and they through t h e i r mouthpieces in the epidermis, Including the o r i f i c e s of the sweat glands. The 'voice from on high' is never heard except by the whim and the disposition of his minis t e r s , and of t h e i r em issaries, and of t h e i r mouthpieces, and they may sometimes be prompted to speak on t h e i r own. They may respond on t h e i r own i n i t i a t i v e to external stimulation without the word even getting through to the cortical-m ost-high. And by the same token, silence of the emissaries or of t h e i r mouth pieces is no proof that the 'most-high' has not spoken. Thus those of us who watch the tapes for the wrlteout of the word from the cortical-most- hlgh must always i n t e r p r e t the leakings via the mouthpieces, 1n case what seems to come from above is only some more menial but inde pendent agent speaking, (p. 394) SUMMARY This d is se rta tio n was concerned with the processes Involved In the acquisition and expression of c la s s ic a lly conditioned GSRs 1n human S s . Stimulus-response theories propose th a t conditioned response acquisition (CRA) 1s due to a single S-R association. Cognitive theories propose that CRA Is the re su lt of a cognitive process such as re lational learning (RL). A th ird proposal Is that both pro cesses are Involved. These proposals were studied by examining the r e l a tionship between the stimulus presented, the stimulus r e cognition response and the GSR e l i c i t e d . Stlmulus-GSR s p e c i f i c i t y was the c r i t e r i o n indicating th a t CRA is due to a S-R process. Recognition response-GSR s p e c i f i c i t y was the c r i t e r i o n Indicating th a t CRA is due to a RL process. Whether S-R or RL or both are involved was studied with a semantic conditioning and generalization procedure and re cognition task. Seventy-two Ss liste n ed to Input words presented at a rate of one per three-seconds and to t e s t words p r e sented at one per thirty-seconds. The independent variable was five word types (8 words per type). CS+ words were paired with noise (UCS) and presented 1n the Input 73 74 (acquisition) and t e s t phase. CW- words were presented in the input and t e s t phase. GS words were pairs of seman t i c a l l y related words of which one was presented during input and the other during the t e s t phase. Of these GS words, GS+ word pairs were followed with noise and GS- pairs were not. N W words were presented only during the t e s t phase. No word appeared more than once 1n the input or t e s t phase. After a cq u isitio n , Ss id e n tifie d words as previously presented with noise, without noise, or not previously presented. Subjects also rated th e ir confidence in the correctness of each answer. Reliable discriminative GSR conditioning and semantic generalization of GSR and r e cognition responses were demonstrated. P a rtitio n in g the GSR data on the basis of recognition performance strongly suggested the independence of GSR responding and correct stimulus recognition. The GSR was as large when Ss co r rectly recognized CS+ words as when CS+ words were errone ously id e n tif ie d . GSRs to CW- words were not s ig n if ic a n tly greater when S mistakenly i d e n t i f i e d them as CS+. Simil arly , GS+ words e l i c i t e d larger GSRs than GS- regardless of the recognition response. With recognition performance as a c r it e r i o n of "awareness," these data suggested th a t knowledge of CS-UCS contingencies (RL) is not necessary for the e l i c i t a t i o n of GSR CRs . 75 The confidence r a tin g s , however, suggested that the recognition response was a truncated measure of RL and that some RL effe cts were present. Questions concerning re la tiv e RL were not answered but were handled by proposing that they can be resolved by studying learning and memory processes concurrently ( i . e . , conditioning can be viewed as the storage of information and its subsequent r e tr iv a l as a function of various response systems). In this manner conditioning can be studied without the need to determine "what is awareness" or "how much of i t is necessary?" This d i s s e r t a t i o n 's methodology was proposed as a new approach for studying autonomic conditioning and verbal-perceptual factors within the conceptual framework of e i t h e r learning theories or memory models or both. REFERENCES Armstrong, D. M. Bodily Sensations. New York: Humanties Press, 1962. Berlyne, D. C onflict, Arousal, and C u rio sity . New York: McGraw-Hi11, 1960. Bridgman, P. W . The Logic of Modern Physics. New York: MacMillan, 1927. Bridgman, P. W. Some general principles of operational analysis. Psychological Review. 1945, 5_2, 246-249. Brotsky, S. J. Classical Conditioning of the Galvanic Skin Response to Verbal Concepts. Unpublished doctoral d i s s e r t a t i o n . University of C alifo rn ia, Los Angeles, 1964. Brotsky, S. J. Classical conditioning of the galvanic skin response to verbal concepts. Journal of Experi - mental Psychology, 1968, 76, 244-253^ Carey, C., Schell, A., and Grings, W . Effect of ISI and reversal manipulations on cognitive control of the conditioned GSR. Paper presented at the 11th Annual Meeting of the Society for Psychophysiological Research, Clayton, Missouri, 1971. Champion, R. A. Review of Prokasy's (ed.) c las s ica l con ditioning: a symposium. Australian Journal of Psychology, 1966, JJJ, 85-87"! Cramer, P. Semantic generalization: Demonstration of an associative gradient. Journal of Experimental Psy chology., 1970, 84, 164-T77! Darrow, C. W . Problems in the use of the galvanic skin response (GSR) as an index of cerebral function: implications of the l a t e n t period. Psychophysiology. 1967, 2, 389-396. Dawson, M. E. Human GSR Classical Conditioning and Aware ness of the CS-UCS Relation1 ! (Doctoral d i s s e r t a t i o n , University of Southern C alifornia) Ann Arbor, Michigan: University Microfilms, 1967. No 67-13,019. 76 77 Dawson, M. Cognition and conditioning: Effects of masking the CS-UCS contingency on human GSR c las s ica l condi tioning. Journal of Experimental Psychology, 1970 , 85, 389-396. Dawson, M., and Grlngs, W . A comparison of classica l con ditioning and re la tio n a l learning. Proceedings of the 74th Annual Convention of the American Psychological A ssociation. 1^66. Dawson, M. E., and Grlngs, W. W . Comparison of classical conditioning and rela tio n a l learning. Journal of Experimental Psychology, 1968, 76, 227-231. Dixon, T. R., and Moulton, A. E. Effects of questioning unaware problem solvers 1n a "verbal conditioning" task. Journal of Experimental Psychology. 1970, 83, 431-434. ~ Eriksen, C. W . Discriminative and learning without aware ness: A methodological survey and evaluation. Psy- chologlcal Review, 1960, 6_7, 279-300. Fuhrer, M. J . , and Baer, P. E. Cognitive processes 1n d i f f e r e n t i a t e d GSR conditioning: Effects of a masking task. American Journal of Psychology 1969 , 82, 168- 180. ~ Furedy, J. J. CS and UCS Intervals and orders 1n human autonomic c las s ica l d i f f e r e n t i a l trace conditioning. Canadian Journal of Psychology. 1970 , 24, 41 7-426. Galto, J. Stages of perception. Journal of General Psychology, 1964, 70, 183-197. Goldstein, H., Drantz, D., and Rains, J. Controversial Issues in Learning. New York: Appleton-Centurv- 'Crofts, lV 6 5 .-------- Grlngs, W. W . Classical conditioning. In Marx, M. (Ed.) Theories in Contemporary Psvchology. New York: Mac- mlllan, 1963. Grlngs, W. W . Verbal-perceptual factors 1n the conditioning of autonomic responses. In W . Prokasy, (Ed.), Clas- s ica l Conditioning, New York: Appleton-Century, 1965. Guilford, J. P. Fundamental S t a t i s t i c s in Psychology and and Education" New York: McGraw-Hill, 1965. 78 Hall, J. F. Verbal Learning and Retention. New York: J. B. Lippincott Co., 1971. Hall, J. W . Effects of IAR occurrence during learning on confidence 1n judgments during recognition. Journal of Experimental Psychology, 19 6 9 , 75J, 578-580. Hartman, T. F. Dynamic transmission, e lec tiv e g en eraliz a tio n , and semantic conditioning. In W . F. Prokasy (Ed.), Classical Conditioning: A Symposium, New York: Appleton-Century-Crofts, 1965, 90-106. Hayes, W . L. S t a t i s t i c s for Psychologists. New York: Holt, Rinehart, and Wilson, 1^63. Hebert, J. A., and Krantz, D. L. Transposition: a re- evaluation. Psychological B u l l e t i n , 1965, 6_3, 244-257. Hilgard, E. R., and Marquis, D. G. Conditioning and Learni ng. New York: Appleton-Century-Crofts, 1940. Hull, C. Pri nclples of Behavior. New York: Appleton- Century-Crofts, 1943. Jones, J. E. Contiguity and reinforcement in relatio n to CS-UCS in terv als 1n c las s ica l aversive conditioning. Psychological Review, 1962, 69_. Kimble, G. A. Hilgard and Marouis1 Conditioning and Learning, 2nd e d . , New York: Appleton-Century-Crofts, I n c . , 1961. Kimble, G. A. Classical conditioning and the problem of awareness. In C. W . Eriksen (Ed.), Behavior and Awareness. Durham, N. C.: Duke University Press, 1962. Kimble, G. A. Attitudlnal factors 1n eyelid conditioning. Paper presented at American Psychological Association, Los Angeles, September 1964. Kimble, G. A. Mediating asso c iatio n s. Journal of Experi mental Psychology, 1968, 7JS, 263-266. Konecni, V. J . , and Slamecka, N. J. Awareness 1n verbal nonoperant conditioning: an approach through dlchotlc lis te n in g . Journal of Experimental Psychology. 1972 , 94, 248-254. Lockhart, R. Dominance and contiguity as In te r a c tiv e determinants of autonomic conditioning. Unpublished 79 doctoral d i s s e r t a t i o n , University of Southern C a l i f o rnia, 1966. Lockhart, R. and Grlngs, W . Comments on an "Analysis of GSR conditioning." Psychological Review, 1963, 70, 562-564. Lynn, R. Attention. Arousal, and the Orientation ReactioTT New York: Pergamon Press, 1966. Maltzman, I. Theoretical conceptions of semantic condi tioning and generalization. In T. Dixon & D. Horton (Eds.), Verbal Behavior and General Behavior Therapy. Englewood C lif f s : Prentice-Hal 1 , In c ., 1968. Martin, I . , and Levey, A. The Genesis of the Classical Conditioned Response. Oxford: Pergamon l^ress, 1969 . Nagel, E. The Structure of Science - Problems in the Logic of S c i e n t i f i c ExplanatiorT! New York: Harcourt, Brace, and World, Inc., Id6l. Orne, M. T. On the social psychology of the psychological experiment: with p a r tic u la r reference to demand ch aracteri s ti cs and t h e i r im plications. Ameri can Psychologists. 1962 , ]T_t 776-783. Page, M. M. Demand awareness, subject s o p h is ti c a tio n , and the effectiveness of a verbal "reinforcement." Journal of P e r s o n a lity . 1970, J8, 287-301. Palermo, D. S., and Jenkins, J. J. World Association Norms. Minneapolis: University Mlnnesota Press , 1964. Pavlov, I. Conditioned Reflexes. Oxford: Clarendon Press, Prokasy, W., and Ebel, H. Three components of the c l a s s i c a l l y conditioned GSR in human su b jec ts . Journal of Experimental Psychology. 1967, 73. 247-256. Razran, G. Conditioning and perception. Psychological Review, 1955, 62, 83-95. Razran, G. The observable unconscious and the in ferab le conscious in current soviet psychophysiology: Intro- ceptive conditioning, semantic conditioning, and the o rie n tin g re fle x . Psychological Review. 1961, 68, 1-81. 80 Shean, G. D. Vasomotor conditioning and awareness. Psy- cophysioloqy, 1968, 5_, 22-30. Sokolov, E. N. Perception and the Conditioned Reflex. New York: Pergamon Press, 1963. Sokolov, E. The modeling properties of the nervous system. In M. Cole and I. Maltzman, A Handbook of Contemporary Soviet Psychology. New York: Basic Books, 1969. Spence, K. W. Behavior Theory and Conditioning. New Haven: Yale University Press, 1956. Sternbach, R. Principles of Psychophysiology. New York: Academic Press, 1966. Thetford, P. E., Klemme, M. E., and Spohn, H. E. Skin p o t e n t i a l , heart r a te , and the span of immediate memory. Ps.ychophysi ol oqy, 1968, 5, 166-1 77. Tolman, E. L. Cognitive maps in rats and men. Psycho- loqical Review, 1948, 5j>, 189-208. Trapold, M. A., Homzie, M., and Rutledge, E. Backward conditioning and UCR latency. Journal of Experi mental Psychology, 1964, £7, 387-391. Underwood, B. J. False recognition produced by Im plicit associative responses. Journal of Experimental Psychology. 1965, 70, 122-129. Underwood, B. J . , and Freund, J. S. Errors 1n recognition learning and reten tio n . Journal of Experimental Psychology, 1968, 78, 55-FTi Uno, T. I n h ib ito ry -e x citato r y in te ra ctio n of conditioned autonomic responses through interoceptive and e x t e r oceptive conditioning. Unpublished Ph.D. d i s s e r t a tio n , University of Southern C alifo rn ia, 1966. Woodworth, R. S. Dynamics of Behavior. New York: Holt. 1958. Ziener, A., and Grlngs, W. W. Backward conditioning: A rep lic atio n with emphasis on conceptualization by the subject. Journal of Experimental Psychology. 1968. 76, 232-235"! -------------------------- 81 Zimny, G. H. , Stern, J. A., and Fjeld, S. P. Effects of CS and UCS relatio n sh ip s on electrodermal response and heart rate. Journal of Exoerimental Psychology, 1 9 6 6 , 72. 1 7 7 - 1 8 1 : ----------------------1--------- APPENDIX A STIMULUS W ORD LISTS List 1 Input 1. GS + bed 2. CW- moisture 3. GS- table 4. GS+ b 1 tter 5. CW- carri age 6. GS- black 7. CW- devi ce 8. CS + concern 9. GS + boy 10. CS+ amount 11. GS- long 12. CW - meeting 13. CS + famous 14. GS + slow 15. GS- scissors 16. CS+ v1 tal 17. CW- sentence 18. GS- b u tter 19. GS+ 20. CW- kl ng bu lle t 21. GS- k1ttens 22. GS + lamp 23. CS+ label 24. CW- motor 25. CW- feather 26. CS+ ta len t 27. GS + hard 28. GS- sell 29. CS + quarrel 30. CS+ aissolve 31. GS- you 32. GS+ blossoms Test 1. GS- cut 2. GS + sleep 3. CS+ guarrel 4. GS+ fas t 5. CW - motor 6. N W de v11 7. CW - feather 8. CS+ famous 9. CS+ concern 10. GS- buy 11. CW - carri age 12. CS+ label 13. N W we 1 come 14. CW- moi s ture 15. GS+ 1 i gh t 16. N W d1 rty 17. GS- chai r 18. N W laugh ter 19. GS+ queen 20. GS- cats 21. GS+ flowers 22. N W c1 ty 23. N W mai den 24. CS+ t a le n t 25. GS- short 26. CS+ amount 27. CW- sentence 28. N W double 29. GS- whi te 30. GS+ so ft 31. GS- me 32. CS+ v1 tal 33. CW- devi ce 34. CW- meetlng 35. CS+ dissol ve 36. CW- bul l e t 82 L i s t 1 ( c o n t i n u e d ) Tes t 37. N W control 38. GS+ girl 39. GS- bread 40. GS+ sweet List 2 Input Test 1 . CW - mixture 1. CW- ti ny 2. GS- blossoms 2. GS- nai Is 3. CS + navy 3. GS+ smoke 4. cw- golden 4. CW- mons t e r 5. GS- ki ng 5. GS- chai r 6. GS + k1ttens 6. CS+ c o lle c t 7. CS + wonder 7. N W di sease 8. GS- 1 amp 8. CS+ package 9. GS+ bread 9. cw- ol i ve 10. GS- needle 10. N W cheerful 11. CW - provi de 11 . GS + cats 12. CS + col 1ar 12. GS- thread 13. CS + mi ss1on 13. GS- f 1 owers 14. CW - olive 14. CS + col 1ar 15. CS* tobacco 15. GS + cut 16. GS + sour 16. N W splendi d 17. GS- bed 17. CS + navy 18. CW - tiny 18. N W express 19. GS + black 19. GS + water 20. GS- hammer 20. CW- provi de 21. CS + pai nter 21. N W mistake 22. cw- ti mber 22. CS+ wonder 23. GS + sci ssors 23. GS- queen 24. CW - monster 24. CS+ pa in te r 25. CS + language 25. GS- sleep 26. GS- table 26. GS- t a l l 27. CS + c o lle c t 27. GS + wh1 te 28. GS + thi rsty 28. CW- spi ri t 29. CW- spi ri t 29. GS + b u t t e r 30. CS + package 30. N W fortune 31. GS + buying 31. CW- golden 32. GS- short 32. CW- mi xture 33. GS + s e l l i n g 34. GS- l i g h t 84 L i s t 2 ( c o n t i n u e d ) Test 35. N W arrow 36. CW- timber 37. CS+ mission 38. CS+ language 39. N W pattern 40. GS+ sweet List 3 Input Test 1. CS + pupi 1 2. GS- always 3. CS+ pressure 4. GS + hammer 5. GS + short 6. CW - manage 7. CW - tr a n s f e r 8. CS+ cover 9. CW- hones t 10. GS+ needle 11. GS- s a l t 12. GS+ thi rsty 13. CS+ harness 14. GS- tabacco 15. GS- us 16. CW - s i gnal 17. CS + armor 18. CW - a r r e s t 19. GS+ bread 20. CW - sparrow 21. CS+ wri te r 22. CW - mus cle 23. CS+ funny 24. CS+ machine 25. GS- buyi ng 26. GS+ dark 27. GS- younger 28. GS + dogs 29. CW - echo 30. GS- dream 31. GS- swi f t 32. GS+ sour 1. GS- smoke 2. GS + cats 3. CW- echo 4. GS- pepper 5. NW oni on 6. CS+ pressure 7. CS + cover 8. NW campai gn 9. GS- ol der 10. GS + b u tte r 11. NW i ncli ne 12. CW- muscle 13. GS + t a l l 14. GS- s leep 15. NW noti ce 16. CS + machi ne 17. CS + funny 18. cw- si gnal 19. GS + nai Is 20. CW- hones t 21. CS + wri te r 22. CS + harness 23. NW weddi ng 24. CW- arres t 25. GS+ 1 i gh t 26. NW condense 27. GS + thread 28. CW- manage 29. GS- fact 30. CS + armor 31. CW- trans fer 32. GS- we 33. CW- sparrow 85 L i s t 3 ( c o n tin u e d ) Test 34. GS+ sweet 35. N W supreme 36. CS+ pupi 1 37. GS- never 38. GS+ water 39. N W gal 1 ant 40. GS- s e l l i n g APPENDIX B INSTRUCTIONS TO THE SUBJECT "This experiment is concerned with memory and the way your body responds to various words and loud, but harmless, noise. These recording electrodes will indicate how your autonomic nervous system is responding, but you do not have to be concerned with t h i s , and you will not feel anything through them. Your task is to try to remember which words were followed with noise and which words were not. F i r s t , eight warm-up numbers will be presented at a rate of once every ten seconds. You are to say each number out loud a f t e r you have heard i t , and you are not required to remem ber i t . Ten seconds a f t e r hearing the l a s t warm-up number, thirty-tw o d i f f e r e n t words will be presented at a rate of one word per three seconds. Sixteen of these words will be followed immediately with noise and 16 of them will not. Those words with noise and those without noise will be pre sented in random order, and your task is to remember which words were followed with noise and which were not. To t e s t your memory, 30 seconds a f t e r the l a s t word- to-be-remembered is presented, some of the same words pre sented e a r l i e r will again be presented randomly with new words or words not previously presented through the 86 87 earphones. Words previously followed with noise will again be followed with noise and some new words may be followed with noise. This time, however, the noise will not follow words immediately, but will follow them 20 seconds la te r. All words will be presented at a rate of one word every 30 seconds during this word recognition t e s t phase. I f you hear a word that you recognize as having been followed with noise during the f a s t p re sen ta tio n , say 'yes, more1 out loud a f t e r i t to indicate that you are more con fid en t than not of the correctness of your answer. If you are less confident of your answer, say, 'yes, l e s s . ' If you hear a word th a t you recognize as having been pre viously presented but not followed with noise, say 'no, more' a f t e r i t to indicate you are confident of the cor rectness of your answer or say, 'no, l e s s , ' i f you are not confident of your answer. If you hear a word th a t you r e cognize as not having been previously presented through the earphones, say 'new, more,' i f you are more confident than not of the correctness of your answer or say 'new, l e s s , ' i f you are less confident of your answer. Try to give all of your recognition responses within 10 seconds a f t e r you hear the word. A diagram of your recognition task appears on the blackboard in front of you. Guess i f you are not sure of an answer, but always give an answer. Do you have any questions? Also, i f you wish to terminate the experi ment, simply raise your r ig h t hand." APPENDIX C EFFECTS OF W ORD TYPE, INPUT AND TEST POSITION, AND ASSOCIATION STRENGTH ON GSR RESPONDING Tables 1 to 6 of Appendix C indicate stimulus word type was s i g n i f i c a n t (p < .01) and input word position was not s i g n if ic a n t (p > .05) on all GSR measures. Signi f i c a n t t e s t word position effe cts were observed only on the 1-1 and I-L frequency measures (p < .01 and p < .05, r e s p ec tiv e ly ). The e f f e c t of association strength was s i g n i f i c a n t (p < .05) only during I-L magnitude and 1-1 frequency responding. A s i g n i f i c a n t in teractio n e f f e c t was observed involving all four main effects on the 1-1 frequency measure. The above analyses suggest that semantic condi tioning and generalization of the GSR may be affected by the t e s t word position and association strength but not by the input word position. Mean GSR differences between discrim inative stimulus words, as shown in Table 7, sug gest, however, that early stimulus input position re s u lted in b e t t e r discriminative GSR conditioning and gen e r a l i z a t i o n than late input po sitio n . Also shown in Table 7 was the tendency for discriminative GSR condi tioning and generalization to be b e t t e r on l a t t e r t e s t 88 89 t r i a l s than on e a r l i e r ones, and th a t frequency and mag nitude GSRs to unreinforced CW- and GS- words decreased from early to la te t e s t t r i a l s therefore suggesting an habituation e f f e c t . The association strength e f f e c t was s i g n i f i c a n t (p < .01) on the I-L magnitude measure. As association strength between input GS word and t e s t GS word decreased, mean magnitude responses during I-L to word types (CS+, CW-, GS+, GS-) increased from .803 for the strong and medium association levels to a .887 for the weak a ss o c i ation level. Lists containing GS word pairs having weak association strength e l i c i t g reater GSR magnitudes than l i s t s containing GS word pairs having strong association strength. A s i g n i f i c a n t association strength e f f e c t was also obtained in the 1-1 frequency measure (p < .05). No systematic re la tio n sh ip was discernable, however, be tween frequency scores as a function of association strength. Several explanations are suggested for the observa tion th a t a given set of experimental manipulations did not a f f e c t these GSR measures in the same manner. Prokasy and Ebel (1967) stated that although the frequency measure correlates with the magnitude measure in c lassical GSR conditioning, the same may not be said for discrimination learning. Their research also indicated that in order for conclusions to be made about the re la tio n sh ip between 90 various measures of the GSR (magnitude, p r o b a b ility , am p l i t u d e ) , the analyses of these measures must "be con ducted on identical Ss over identical t r i a l blocking" (p. 249). In the present experiment, there were s ix r e p l i c a tions of stimulus materials and six d i f f e r e n t groups of S s. Also, t r i a l blocking involving 20 t r i a l s ra th e r than the "usual" eight or less. Tables 1 to 7 were obtained from the pooled r e s u lts of all six groups. For these reasons, conclusions about the relationships between the various interval measures of GSR magnitude and frequency cannot be made. The observation that the association strength e f fect was not s i g n i f i c a n t on the 1-1 and 1-2 GSR magnitude measure but was s i g n i f i c a n t on the I-L measure (Tables 1, 2, and 3) may be c l a r i f i e d with "set terminology" (Hayes, 1965). The I-L measure consisted of the la rg e st magnitude score occurring in e i t h e r 1-1 or 1-2. In "set termin ology," I-L scores or Subset I-L resulted from the i n t e r section of Set 1-1 and Set 1-2. Since Subset I-L con tained the larg er 1-1 or 1-2 score, the mean score of Subset I-L would be larger than the means of Set 1-1 or Set 1-2. Also, more zero scores and lower scores would be in these sets than in the subset. If the 91 association strength gradient e f f e c t on GSR magnitude is dependent upon a minimum score and is independent of GSR latency, then this e f f e c t would more likely be de tected with the I-L measure than with the 1-1 or 1-2 meas u r e s . In summary, this Appendix i l l u s t r a t e s that a given set of manipulations may a f f e c t various components of the GSR d i f f e r e n t i a l l y . The effects of the independent variable, stimulus word type, was co n sisten t across all dependent GSR measures, however. Although some of these dependent measures may be redundant, the importance of employing several measures of the GSR have been emphasized in autonomic research ( e . g . , Sternbach, 1966; Prokasy and Ebel, 1967; Maltzman, 1968). 92 Appendix C -- con tin u e d TABLE 1 ANALYSIS OF VARIANCE FOR 4 X 2 X 2 X 3 FACTORIAL EXPERIMENT HAVING 48 1-1 GSR MAGNITUDE SCORES PER CELt Source of Variation SS ' " '"d'f " - ' M S' F ... A (Word Type: CS+, CW-, GS+, GS-) 22.92 3 7.64 22.17** B (Input Position) 0.01 1 0.01 0.02 C (Test Position) 0.06 1 0.06 0.16 D (Association Strength) 1.01 2 0.50 1 .46 AB 1.56 3 0.52 1.51 AC 0.67 3 0.22 0.63 AD 2.01 6 0.34 0.97 BC 0.15 1 0.15 0.42 BD 0.25 2 0.12 0.36 CD 0.40 2 0.20 0.58 ABC 1.42 3 0.47 1 . 38 ABD 0.70 6 0.12 0.34 ACD 0. 77 6 0.13 0.37 BCD 0.54 2 0.27 0.78 ABCD 1.58 6 0.26 0.77 Within cell 777.25 2256 0.35 Total 811 .27 2 30 3 ** p < .01 93 Appendix C - - con tin u e d TABLE 2 ANALYSIS OF VARIANCE FOR 4 X 2 X 2 X 3 FACTORIAL EXPERIMENT HAVING 48 1-2 GSR MAGNITUDE SCORES PER CELl Source of Variation "SS df " M S r A (Word Type: CS+, CW-, GS+, GS-) 7.24 3 2.41 8.51** B (Input Position) 0.57 1 0.57 2.00 C (Test Position) 0.24 1 0.24 0.83 D (Association Strength) 0.62 2 0.31 1 .08 AB 0.34 3 0.11 0.40 AC 1.63 3 0.54 1.92 AD 1.67 6 0.28 0.98 BC 0.13 1 0.13 0.45 BD 0.09 2 0.05 0.16 CD 0.08 2 0.04 0.14 ABC 0.67 3 0.23 0.80 ABD 1 .49 6 0.25 0.87 ACD 1.65 6 0.27 0.97 BCD 0.44 2 0.22 0.77 ABCD 0.27 6 0.05 0.16 Within cell Total 639.23 656.33 2256 230 3 0.28 ** p < .01 94 Appendix C -- c ontinue d TABLE 3 ANALYSIS OF VARIANCE FOR 4 X 2 X 2 X 3 FACTORIAL EXPERIMENT HAVING 48 I^L GSR MAGNITUDE SCORES PER CELl Source of Variation sS d f ' M s F A (Word Type: CS+, CW-, GS+, GS-) 19.41 3 6.47 18.59** B (Input Position) 0.07 1 0.07 0.21 C (Test Position) 0.27 1 0.28 0.80 D (Association Strength) 3.59 2 1 .80 5.16** AB 1.15 3 0.38 1 .10 AC 1.49 3 0.50 1 .43 AD 2.41 6 0.40 1.16 BC 0.01 1 0.01 0.02 BD 0.39 2 0.19 0 .55 CD 0.17 2 0.08 0.24 ABC 0. 72 3 0.24 0.69 ABD 2.09 6 0.35 1 .00 ACD 1.20 6 0.20 0.57 BCD 0.49 2 0.25 0.71 ABCD 1.18 6 0.20 0.56 Within cell T otal 785.22 8 1 9 . 8 1 ' 2256 2303 0.35 ** p < .01 95 Appendix C -- c o n tin u e d TABLE 4 ANALYSIS OF VARIANCE FOR 4 X 2 X 2 X 3 FACTORIAL EXPERIMENT HAVING 48 1-1 GSR FREQUENCY SCORES PER CELt Source of Variation ■ ~rs " ' - d? ’ U S ' _ p A (Word Type: CS+, CW-, GS + , GS-) 15.24 3 5.08 14.56** B (Input Position) 0.07 1 0.07 0.21 C (Test Position) 3. 75 1 3. 75 10.76** D (Association Strength) 2.10 2 1.05 3.01* AB 1.11 3 0.37 1 .06 AC 0.17 3 0.06 0.16 AD 3.14 6 0.52 1 .50 BC 0.42 1 0.42 1 .20 BD 0.42 2 0.21 0.60 CD 0.26 2 0.13 0.37 ABC 1.81 3 0.60 1 .73 ABD 1.05 6 0.18 0.50 ACD 1.40 6 0.23 0.67 BCD 0.02 2 0.01 0.03 ABCD 5.32 6 0.89 2.54** Within cell Total 786.98 823.26 2256 2303 0.35 * p < .05 ** p < .01 96 Appendix C -- c ontinue d TABLE 5 ANALYSIS OF VARIANCE FOR 4 X 2 X 2 X 3 FACTORIAL EXPERIENCE HAVING 48 1-2 GSR FREQUENCY SCORES PER CELt Source of Variation ... _ " dT" W 5- ' F" " A (Word Type: CS+, CW-, GS+, GS-) 6.54 3 2.18 6.41** B (Input Position) 1.04 1 1.04 3.06 C (Test Position) 1.22 1 1.22 3.58 D (Association Strength) 0.64 2 0.32 0.94 AB 0.79 3 0.26 0.77 AC 1.22 3 0.41 1.20 AD 2.06 6 0.34 1.01 BC 0.05 1 0.05 0.16 BD 0.14 2 0.07 0.20 CD 0.24 2 0.12 0.35 ABC 1.53 3 0.51 1.50 ABD 2. 32 6 0.39 1.14 ACD 2.90 6 0.48 1.42 BCD 0.38 2 0.19 0.55 ABCD 1.29 6 0.22 0.63 Within cell Total 767.56 789.92 2256 2303 0.34 ** p < .01 97 Appendix C -- continued TABLE 6 ANALYSIS OF VARIANCE FOR 4 X 2 X 2 X 3 FACTORIAL EXPERIMENT HAVING 48 I-L GSR FREQUENCY SCORES PER CELl Source of Variation SS <Tf"' ' " mst ... F - A (Word Type: CS+, CW-, GS+, GS-) 13.23 3 4.41 14.44** B (Input Position) 0.34 1 0.34 1.11 C (Test Position) 2.01 1 2.01 6.57* D (Association Strength) 0.77 2 0.39 1.26 AB 0.43 3 0.14 0.48 AC 0.32 3 0.11 0.35 AD 1.61 6 0.27 0.90 BC 0.00 1 0.00 0.03 BD 0.12 2 0.06 0.20 CD 0.03 2 0.01 0.04 ABC 1.26 3 0.42 1.37 ABD 0.98 6 0.16 0.53 ACD 2. 35 6 0.39 1.28 BCD 0.09 2 0.04 0.14 ABCD 3.65 6 0.61 1.99 Within cell Total 688.75 715.92 2256 230 3 0.31 * p < . 0 5 ** p < .01 Appendix C -- c ontinue d TABLE 7 EFFECTS OF STIMULUS W ORD INPUT POSITION AND TEST W ORD POSITION ON SEMANTIC CONDITIONING AND GENERALIZATION OF THE GSR FOR ALL SUBJECTS, N=72* Conditioning Stimuli Generalization Stimuli Posi ti on 1-1 1 1-2 r T -T “ I 1-2 GSR Magnitude Scores CS + CW - di f CS+ " c n - di f GS+ GS- di f GS+ 6S- dif Input Early Late .86 .86 .60 .68 .256 .170 .41 .41 .27 . 30 .135 .104 .73 .68 .61 .56 .120 .117 .42 .43 .29 .36 .122 .070 Test Early Late .87 .85 .66 .62 .203 .222 .32 .45 .31 .25 .045 .195 .68 .73 .60 .57 .077 .160 .41 .43 .31 .33 .09 3 .099 VO 00 Table 7 -- c ontinue d Conditioninq Stimuli Generalization Stimuli Posi ti on< 1-1 1 1-2 1-1 1 1-2 GRS FREQUENCY SCORES CS + CW - di f CS + CW - di f GS + GS- di f GS + GS- dif Input Early Late .94 .98 . 76 .81 .174 .163 .52 .51 .36 .41 .158 .101 .87 .87 .77 .71 .097 .159 .49 .53 .38 .48 .108 .045 Test Early Late 1.01 .90 .82 .75 .188 .149 .51 .52 .44 . 33 .069 .189 .90 .84 .78 .70 .121 .135 .52 .50 .46 .40 .056 .098 * dif= di fference score (S/s total score per wood type/8) = S/s score per word type. VO VO APPENDIX D OPERATIONAL DEFINITIONS OF GOOD AND POOR RECOGNITION PERFORMANCE SUBJECTS AND GSR ANALYSES Good Recognition: correctly recognizing 50% or more of all words presented within at l e a s t four of the five word type categories (CS+, CW-, GS+, CS-, NW ) and 25% or more of all words presented in the category containing the smallest number of cor rectly recognized items. Poor Recognition: correctly recognizing 20% or less of all words presented in at le as t two word type categori e s . Average Recognition: recognition a b i l i t y not defined by good and poor recognition. 100 Appendix D -- c ontinue d TABLE 1 MEANS, STANDARD DEVIATIONS, AND F MAXs OF MAGNITUDE GSRs TO STIMULUS W ORDS IN GOOD AND POOR RECOGNITION SCORING SUBJECTS Sti mul us Word Type Item CS + CW - GS+ GS- N W F Max+ 26 Sub.iects with Good Recoanition Performance Interval Mean S.D. 1 .900 .39 3 .6 76 .331 .721 .333 .621 .626 .278 .361 1.996 Interval Mean S.D. 2 .413 .339 .307 .233 .415 .330 .304 .237 .232 .1 88 3.251* 21 Subjects with Poor Recognition Performance Interval Mean S.D. 1 .800 .324 .608 .321 .668 .288 .591 .583 .335 .295 1.352 Interval Mean S.D. 2 .39 3 .200 .300 .187 .429 .316 .326 .307 .230 .246 2.869 * p < .05 + H a r t l e y ' s F Max o 102 Appendix D -- continued TABLE 2 ONE FACTOR ANALYSIS OF VARIANCE OF STIMULUS W ORD TYPE AND 1-1 MAGNITUDE GSRs IN. GOOD REC0GNITI0N~TC0RING SUBJECTsT n=26 Source of Variation SS df M S F Between subjects 12. 33 25 Within subjects 3.60 104 Word type (CS+, CW-, GS+, GS - , NW ) 1. 36 4 0.34 15.11** Res i dual 2.24 100 0.02 Total 15.93 129 ** p < .01 103 Appendix D -- continued TABLE 3 ONE FACTOR ANALYSIS OF VARIANCE OF STIMULUS W ORD TYPE AND 1-1 MAGNITUDE GSRs IN POOR RECOGNITION SCORING SUBJECTS, n=21 Source of Variation SS df M S F Between subjects 8.42 20 Within subjects 2.08 84 Word type (CS+, CW-, GS+ , GS-, NW ) 0.69 4 0.17 9.82** Resi dual 1.40 80 0.02 T otal 10.50 104 ** p < .01 104 Appendix D -- continued TABLE 4 ONE FACTOR ANALYSIS OF VARIANCE OF STTMULUS W ORD TYPE AND 1-2 MAGNITUDE GSRs IN GOOD REC0GNITI0N~5C0R1NG SUBJErTsT n=26 Source of Variation SS df M S F Between subjects 6.59 25 Within subjects 3.20 104 Word type (CS + , CW-, GS+, GS-, NW ) 0.62 4 0.15 5.99** Resi dual 2.59 100 0.03 Total 9.79 129 ** p < .01 105 Appendix D -- continued TABLE 5 ONE FACTOR ANALYSIS OF VARIANCE OF STIMULUS W ORD TYPE AND N2 MAGNITUDE GSRs IN. POOR RECOGNITION SCORING SUETErTsT n=2l Source of Variation Between subjects 4.04 20 Within subjects 2.00 84 Word type (CS + , CW-, GS+, GS-, NW ) 0.27 4 0.07 3.16* Res i dual 1 .72 80 0.02 Total 6.04 104 * p < .05 106 Appendix D - - c o n tin u e d TABLE 6 NEUMAN-KEULS STATISTICS ON MAGNITUDE CHANGES TO STIMULUS WORDS IN GOOD AND POOR RECOGNITION SCORING SUBJECTS Sti mulus Word Type I I tem CW- 1 GS+ 1 GS- 1 N W Good Recognition Ss (n=26) Interval 1 CS + 7.61** 6.08** 9.48** 9.31** CW- 1.53 1.87 1. 70 GS + 3.40 3.24 GS- 0.17 Interval 2 CS + 3. 36* 0.09 3.44* 5.57** CW- 3.45* 0.07 2.21 GS + 3.52 5.66** GS- 2.13 Poor Recognition Ss (n=21) Interval 1 CS + 6.67** 4.50** 7.27** 7.53** CW- 2.09 0.60 0.86 GS + 2.69 2.95 GS- 0.26 Interval 2 CS + 2.90 1.12 2.10 2.67 CW- 4.02* 0.80 0.23 GS + 3.22 3 .8 0 * GS- 0.57 * p < .0 5 ** p < .01 APPENDIX E OPERATIONAL DEFINITIONS OF GOOD, AVERAGE, AND POOR DISCRIMINATIVE STIMULUS RECOGNITION PERFORMANCE Good CS Discrimination Recognition: co rrectly recognizing a l 1 CS+ and CW- words. Average CS Discrimination Recogni ti on: co rrectly recogni zing at l e a s t 50% of all CS+ and of all CW- words. Poor CS Discrimination Recognition: recognition a b i l i t y not defined by good and average CS discrimination recogni ti on. Good GS Discrimination Recognition: correctly recogni zing at le as t 53% of all GS+ and GS- words. Average GS Discrimination Recogni ti on : co rrectly recog nizing 50-52% of all GS+ and GS- words. Poor GS Discrimination Recognition: Recognition a b i l i t y not defined by good and average GS discrimination recogni ti on. Thirty Ss were i d e n tif ie d as having good, 20 Ss as having average, and 22 as having poor CS recognition ab i 1 i t y . Twenty-seven Ss showed good, 18 Ss showed average, and 27 Ss showed poor GS word recognition a b i l i t y . 107 APPENDIX F OPERATIONAL DEFINITIONS OF GOOD, POOR, AND NO GSR DISCRIMINATION CONDITIONING AND GENERALIZATION SUBJECTS Good c las s ica l discriminative 1-1 GSR magnitude conditioning in a S is operationally defined as the mean response to CS+ words minus the mean response to CW- words is larg er than or equal to 0.20 AC*5. Poor conditioning is defined as such GSR values between 0.10 and 0.20 AC*5. No conditioning is defined as such GSR values less than 0.10 AC*5. Thirty-two Ss showed good conditioning, 26 showed poor conditioning, and 14 showed no 1-1 GSR magni tude conditioning. Good classica l discrim inative 1-2 GSR magnitude conditioning in a S ^ is operationally defined as the mean response to CS+ words minus the mean response to CW- words is larger than or equal to 0.10 AC*5, poor condi tioning as such values between 0.05 and 0.09 AC*5. Thirty-seven Ss showed good 1-2 magnitude conditioning, seven Ss showed average, and 28 Ss showed no such condi tioning. Good 1-1 GSR magnitude generalization in a S is operationally defined as the mean 1-1 magnitude response to GS+ words minus the mean such response to GS- words is 108 109 Appendix F - - c ontinue d larger than or equal to 0.11 AC’5, poor g eneralization as such GSR values between 0.05 and 0.10 AC*5, and no gen e r a liz a tio n as such GSR values less than 0.05 AC*5. Thirty -six Ss showed good, nine Ss showed poor, and 37 Ss showed no 1-1 GSR magnitude generalization. Good 1-2 magnitude generalization in a S is opera tio n a lly defined as the mean 1-2 magnitude response to GS+ words minus the mean such response to GS- words is larger than or equal to 0.10 AC*5, poor generalization as such GSR values between 0.05 and 0.09 AC*5, and no gen e r a liz a tio n as such GSR values less than 0.05 AC*5. Thirty-three Ss showed good, nine showed poor, and 30 showed no 1-2 GSR magnitude g e n e r a liz a tio n .

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Relationship Of Personality And Task Demands To Self-Disclosing Behavior And Psychological Health

PDF

Dominance And Contiguity As Interactive Determinants Of Autonomic Conditioning

PDF

A Cognitive Interpretation Of Acquired Reinforcement Effects

PDF

Human Performance As A Function Of The Joint Effects Of Drive And Incentive Motivation

PDF

Habituation Of The Multiple Unit Discharge Response To White Noise Stimulation In The Unanesthetized Rabbit

PDF

The Effects Of Prior Part-Experiences On Visual Form Perception In The Albino Rat

PDF

Classical Discrimination Conditioning As A Function Of Probability Of Reinforcement

PDF

Human Gsr Classical Conditioning And Awareness Of The Cs-Ucs Relation

PDF

City Manager, City Council Role Consensus And Its Effect On Municipal Performance

PDF

Inhibitory-excitatory interaction of conditioned autonomic responses through interoceptive and exteroceptive conditioning

PDF

An Evaluation Of A Communal Organization

PDF

Sex-Role Preferences Of Early Adolescents In Relation To Adjustment

PDF

A Factor Analysis Of The Semantic-Evaluation Abilities

PDF

Effects Of Success And Failure On Impulsivity And Distractibility Of Three Types Of Educationally Handicapped Children

PDF

Verbal Reports Of Emotional States And Onsets And Offsets Of Conditioned Stimuli

PDF

Transfer Of The Partial Reinforcement Extinction Effect Across Tasks In Normal And Retarded Boys

PDF

Alternatives In Research Design: The Measurement Of Goal Conflict

PDF

Participative Management And Industrial Democracy: Toward A Participative Theory Of Organization

PDF

Dental Student Selection Through Handwriting Analysis

PDF

Modality and serial position effects in immediate and delayed recall

Asset Metadata

Creator Tang, Terry (author)

Core Title Autonomic and cognitive indices of semantic conditioning and generalization

Degree Doctor of Philosophy

Degree Program Psychology

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

Tag OAI-PMH Harvest,psychology, experimental

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Madigan, Stephen A. (committee chair), Grings, William W. (committee member), McEachern, Alexander W. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-790849

Unique identifier UC11364462

Identifier 7309327.pdf (filename),usctheses-c18-790849 (legacy record id)

Legacy Identifier 7309327

Dmrecord 790849

Document Type Dissertation

Rights Tang, Terry

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