AN ENVES‘EIGATEQN Qé‘i ?HE RELATfiQNSEfl? {SERVE-ER WFEL SEZE AND EMOTEQNAL WGEDg Thests For the Degree of DE. D. IRQECEESRH SHTE UHWERSETY James F. Guinan 1966 THESYS ’J LIBRA: :1“ Michigan Sta; 3 .~ University "L V J This is to certify that the thesis entitled AN INVESTIGATION ON THE RELATIONSHIP BETWEEN PUPIL SIZE AND EMOTIONAL "NOA‘DS presented by James F. Guinan has been accepted towards fulfillment of the requirements for _P_:1_D_-____ degree in M '3 S .Y Major professor Date May 13, 1966 0-169 ABSTRACT AN INVESTIGATION ON THE RELATIONSHIP BETWEEN PUPIL SIZE AND EMOTIONAL WORDS James F. Guinan A series of recent investigations have suggested a relationship between pupil size and emotional components of visual stimuli. A review of the literature revealed that while a physiological basis for a relationship between pupil size and emotional excitation could be stated, no investigator had yet demonstrated that emo- tional and neutral stimuli have differential effects upon pupil size. Although it is generally accepted that the pupil can be conditioned (Kimble, 1961), attempts to condition pupillary constriction or dilation (which would demonstrate that the pupil reSponds to psychological components of stimuli) have been equivocal. These studies were discussed and possible eXplanations of the negative findings were suggested. The hypothesis of the present study was that present- ation of emotional words would result in significantly different pupil reactivity (constriction or dilation) than would presentation of neutral words to the same James F. Guinan subjects. The words utilized were selected from published lists of words which were shown to affect behavior as measured by GSR and the Semantic Differential, and presented to a group of students. The three words with the highest and lowest "emotionality scores" were then chosen. Twenty-eight Ss were presented each of the three emotional and three neutral words for five seconds while motion picture recordings were being taken of their pupils. It was found that overall mean pupil size of 27 out of 28 Ss was larger to emotional than to neutral words. Analysis of variance demonstrated that emotion- ality did have a significant effect on pupil size, and that there was also a significant interaction effect of emotionality and time (intervals). When the data for the first and last 2.5 seconds were analyzed separately, results showed that pupillary size during the first 2.5 seconds was not significantly different for emo- tional than neutral words. However, during the second 2.5 seconds of stimulus presentation the results clearly demonstrated that: pupil size was significantly larger to emotional than to neutral words. Pupil dilation was discussed as a measure of a generalized autonomic response to emotionally meaningful James F. Guinan stimuli. Numerous research implications were suggested, both in the investigation of the characteristics of the pupil response itself, and in the utilization of the pupil response in investigating the emotional properties of visual or sensory stimuli. Chairman a... Q 3‘? MM \ AN INVESTIGATION ON THE RELATIONSHIP BETWEEN PUPIL SIZE AND EMOTIONAL WORDS By .fi 1“ (\L' James F: Guinan A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1966 DEDICATION To Marcia ACKNOWLEDGMENTS The author wishes to express his appreciation to his committee chairman, Dr. Bill Kell, and to his committee members, Dr. William Mueller, Dr. David Baskin, and Dr. Norman Abeles whose individual and collective counsel and advice proved invaluable in completion of this thesis. A very special thanks is also extended to Mr. Edward McCoy of the Michigan State University Department of Audio-Visual Aids, to Mr. William Hughes, Motion Picture Photographer with the Michigan Department of Health, and to Mr. Maurice Strahl, Photographer for Michigan Tourist Council, whose help and consultation made construction and Operation of the apparatus possible. Mr. Hughes served as camera operator during collection of the data. TABLE ACKNOWLEDGMENTS . LIST OF TABLES . . LIST OF FIGURES . LIST OF APPENDICES INTRODUCTION . . . METHOD . . . . . . Subjects Apparatus Materials Procedure RESULTS . . . . . DISCUSSION . . . SUMMARY . . . . . BIBLIOGRAPHY . . . APPENDICES . . . OF CONTENTS Page ii iv vi ll 11 11 13 15 18 25 32 34 4O LIST OF TABLES Table Page 1 Words used as stimuli and their corresponding scores of "judged emotionality . . . . . . . . . . . . . . 15 2 Analysis of Variance for emot. vs. neutral words with 3:28 subjects and 1:20 intervals for each word . . . . . . . . . . . . . . l9 3 Duncan's New Multiple Range Test applied to the differences between mean pupil size in response to W:6words . . . . . . 2O 4 Analysis of Variance for emot. vs.neutra1 words with 3:28 subjects and 1:10 intervals: first 10 intervals 0 O O O O O O O O O O 22 5 Analysis of Variance for emot. vs.neutra1 words with 5:28 subject and I-10 intervals: second 10 intervals . . . . . . . . . . . 22 6 Duncan's New Multiple Range Test applied to the differences between mean pupil size in response to W:6 words: second 10 intervals . . . . . . . . . . . 23 7 Sums and correlations of rankings of 28 Se on emotionality of words and rankings of mean pupil size in response to words . . . . . . . . . . . . 24 iv LIST OF FIGURES Figure Page 1 Experimental Apparatus . . . . . . . . 12 2 Approximation of mean actual pupil size for 28 Se during 5-second stimulus presentation . . . . . . . . l8 LIST OF APPENDICES Appendix A Judgments of emotionality of 30 words by 58 students . . . . B Order of presentation of verbal Stimuli 0 O O O O O O O O O 0 O C Rankings of the six exPerimental words in order of mean pupil size in response to those words . . . . . . . . . . . . . D Rankings of S of emotionality of eXperimental words . . . . . E Recorded measurements of pupil size of 28 $3 to six words at 20 intervals . . . . . . . . . . F Mean recorded pupil size of 28 $3 to each of six words . . . . vi Page 41 43 44 45 46 52 INTRODUCTION A series of recent publications by Hess and Polt (Hess and Polt, 1965; Hess and Polt, 1964; Hess and Polt, 1960) offer evidence that the pupillary response may be utilized as a measure of affective states. In these studies the authors presented a number of visual stimuli to their subjects and then measured the diameter of each subject's pupil in relation to the manifest content of the visual stimuli. In one report (Hess and Polt, 1964) they measured pupil size of 83 during simple problem-solving procedures. The results of these studies demonstrated a relationship between pupil size and psychological components of the stimuli and were suf- ficiently consistent to lead the authors to conclude that the pupil response can provide quantitative data on the psychological effects of visual and other stimu- lation, on complex mental activities, on interests and attitudes, and on states of emotional excitation. Hess, Seltzer, and Schlien, (1965) reported that they were able to differentiate homosexual from heterosexual males by measuring the pupil response to visual stimuli. A number of criticisms may be directed at the above studies. On a procedural level, Hess and his associates have made no attempt to differentiate or even Operationally define such variables as ”interest", "emotionality", or "attitude." To the contrary, they 1 have hypothesized that certain visual stimuli would arouse a specified affective state, e.g. interest, and then report that the changes in pupil size were due to that particular affect. Secondly, the above reports have been more descriptive than eXperimental. The procedures have been loosely controlled and not reported in sufficient detail so as to allow replication. The results have been presented only in terms of direc- tion and percentage of change, and the application of precise statistical techniques has been conSpicuously absent. It is the belief of this author that the nature of the relationship of the pupil reSponse to definable visual stimuli has not yet been established. If the pupil response is to be considered "a new tool with which to probe the mind. . . and to establish personality dif- ferences" (Hess and Polt, 1965, p. 54), then it merits a more systematic experimental investigation. The present study proposes to empirically validate the contention that pupil reactivity is related to affective states. More Specifically, the purpose of this experiment is to establish whether there is a measureable relationship between pupil size and certain emotionally meaningful words. While physiologists have repeatedly noted that the pupil response is affected by emotional stimulation (Morgan, 1965; Lowenstein and Loewenfeld, 1962; Lowen- stein and Loewenfeld, 1952; Brown and Page, 1939; Ury and Gillhorn, 1939; and Ferre and Bond, 1933), a review of the literature revealed that only one investigator, prior to Hess and Polt, had attempted to critically examine this phenomenon. Bender (1933) photographed S's pupils which were simultaneously eXposed to light and to emotional stimuli. "Emotional stimuli" were defined as a white rat held in front of Ss' face or a pin prick. When these photographs were compared with photographs of the pupils of the same 33' when only the light was flashed, it was found that the measured absolute size of the pupil was larger, and that the amount of time for the pupil to return to normal after the flash of light was longer when emotional stimuli were present. While statistical treatment was not applied to these data. the results were consistentin the hypothesized direction, and Bender concluded that a purely psychic stimulus may cause the pupil to dilate. There have been a number of investigators who have attempted to condition the pupil reaponse. The relevance of these studies for the present research lies in the fact that a pupil response to emotional components of stimulus is a learned response. Furthermore, the conditioning studies have contributed important informa- tion about the nature of the pupil response. Watson (1916) obtained a conditioned pupil constriction to sound, but noted that pupillary conditioning was not | a satisfactory way to study conditioning principles, because it was such a difficult response to obtain. Since the Watson report, there has been a continued controversy as to whether a pupil response could be conditioned at all, and if so, what experimental procedures were necessary to obtain a conditioned pupillary response. Cason (1922); Hudgins (1933); Harlow and Stagner, G933); Hudgins, (1935); Baker, (1938); Metzner and Baker, (1939); Harlow, (1940); Girden, (1942); and Crasilneck and McCranie, (1956) were able to condition a pupillary response to either sound or shock. Stickle and Brenshaw, (1934); Stickle, (1936); Wedell, Taylor and Skolnick, (1940); Hilgard, Miller, and Ohlson, (1941); Stern, (1948); Hilgard, Dutton and Helnick, (1949); and Young, (1954) reported experimental procedures which failed to obtain condi- tioning even when the conditions reported by others were exactly replicated. Other investigators (Young, 1958 and 1965; Sampson and Boslow, 1957; Gerall and Obrist, 1955; Young and Biersdorf, 1954; Girden, 1942) have found that pupillary conditioning could be obtained but that the conditioned response was not stable or easily elicited when changes in light intensity or sound were used as the conditioned stimuli. One possible explanation for this that seems not to have been considered is that the pupil response to light itself may have some of the characteristics of a conditioned response. Lowenstein and Loewenfeld (1952) found the pupil response to changes in light intensity was not difficult to ex- tinguish. The studies cited above have all utilized the pupillary response to light as an unconditioned response, and it is quite possible that the repeated presentation of light may have resulted in extinction of the unconditioned response before conditioning occurred. Moreover, these studies have not adequately controlled the homogeniety of the visual field. Typically, the pupil was observed by use of a telescope, and the presence of a lens resulted in different intensities of reflected light across the Ss' visual field. Interestingly enough, there have been no published reports of failure to obtain conditioning when the unconditioned stimulus was electric shock or pain. §32h, stimuli are usually described as "emotignal," (Young, 1962; Hilgard, Dutton, and Helnick, 1949; and Girden, 1942). In attempting to eXplain the negative findings of other investigators, Girden (1942) discussed components of the pupillary response as consisting of both simple dilation and constriction, and of a "hippus" or disturb: ance reSponse. The hippus refers to the continuous oscillation of the size of the pupil. Typically, the pupil changes size at a rate of 20 to 60 oscillations per minute and that this hippus may account for a pupil- size change as much as one millemeter. Crasilneck and McCranie (1956) also discuss the hippus effect and report that because of this continuous oscillation, direct observation of the pupil, even with use of a telescope or pupilograph is inefficient and inaccurate. It should be noted that all of the pupillary conditioning studies prior to 1950 utilized telescopic or pupilographic apparata. Lowenstein and Loewenfeld (1962) have described the physiological mechanisms of the pupillary response. They state that the iris.is a representative of all smooth muscle structures that are reciprocally innervated by the sympathetic system, while the pupillary sphincter (pupil constriction) is innervated by parasympathetic activity. Physiologically, there is thus reason to believe that pupil reactivity should be associated with stimuli that are emotionally laden. Bender, (1933); Brown and Page, (1939); and Ury and Gellhorn, (1939) have also described the reciprocal relation of sympathetic and parasympathetic systems to pupillary dilation and constriction. However, there has not been sufficient evidence to suggest a directionality factor in pupil reactivity. Hess and Polt (1965) state that with reapect to visual stimuli, there is a range of pupil responses from extreme dilation for pleasing stimuli to extreme ,constriction for unpleasant stimuli. This is contrary to Bender's finding that painful and unpleasant visual stimuli resulted in pupil dilation. It is also incon- sistent with Young's (1965) report that conditioning of pupillary constriction is not possible because pupillary constriction is not part of the generalized unconditioned autonomic response to psychic stimuli. It has been suggested by a number of authors that the pupillary response is representative of a geralized autonomic response to noxious or emotional stimuli. YOung (1965) stated that such a generalized autonomic response is modified during conditioning so that changes in the size of the pupil represent changes that are occurring within the autonomic nervous system. Girden (1942) reported that pupillary dilation is a function of generalized response of the organism to noxious stimuli. McGinnies (1956) Operationally defined "emotionality" as generalized autonomic activity, as measured by the galvanic skin response, without regard to presence or absence of phenomenological content. Thus, while it may be hypothesized that pupil reactivity is related to emotionality, dilation or constriction in response to specifically "toned" words (pleasant or unpleasant) respectively, remains speculatory. As has been stated, inadequate definition and control of the visual stimulus has been a deficiency in all of the studies relating pupil size to affective stimuli. Bender did not control the visual field or light intensity. Hess did not attempt to measure or control physical' complexity or psychological meaningfulness of the stimuli used. In the proposed study words that have been judged as emotionally laden will be used. That simple words can elicit emotional responses has been repeatedly demonstrated. Generally, at least one of three methods has been employed in demonstrating that the perception of words elicits affective arousal. Research studies on perception have used judge's ratings of emotionality. Typically, words judged as emotional have shorter or longer recog- nition thresholds than words judged to be neutral, (Jenkins, Russell, and Suci, 1958; Singer, 1956; Postman, Bruner, and McGinnies, 1948; Bruner and Postman, 1947). A second procedure has been utilized in the studies on meaningfulness. The Semantic Differential or word- association techniques have been administered to groups in order to establish the meaningfulness of frequently used words. Noble (1958) has noted that emotionality is one attribute of meaningfulness. Heise (1965) recently published a "dictionary" of affective content of words as measured by the Semantic Differential. Thirdly, galvanic skin response (GSR) measures have been shown to be associated with the emotional or affective connotation of words. Although McCleary (1950); Venables, (1955); and Watson, (1957) have questioned the used of the GSR as a measure of emotionality, Silverman, Cohen, and Shanavonian (1959) have shown that in an adequately controlled procedure, GSR varies directly with affective responses as measured by clinical inter- views. McGinnies (1949) showed that emotionality as measured by GSR was significantly greater for selected "critical" words when compared with neutral words. Eajonc (1962) compared 12 taboo words with 12 neutral words and found that on every trial the taboo words exceeded the neutral words in eliciting a GSR. Noble (1958) also utilized the GSR to assign weights to words that had been judged neutral, pleasant, or unpleasant. He concluded that "judged emotionality is a reliable lO attribute of verbal stimuli" (Noble, 1958, p. 16). Cohen, Silverman, and Barch (1956) studied the effects of neutral and "charged" words and also found consistent GSR's associated with emotionally-laden words. They reported that the GSR responsivity to a word stimulus is a function of the affective connotation of the stimulus and level of arousal of the subject in relation to that stimulus. In summary, it has been shown that on a physiological basis, pupil reactivity should be related to emotional or effective components of visual stimuli. Secondly, it has been shown that words used as visual stimuli can result in the elicitation of affective states. It is therefore the hypothesis of this investigation that pupil reactivity, as measured by average pupil size, will be significantly greater when emotionally-laden words are used as stimuli than when neutral words are used as stimuli. METHOD Subjects. The subjects were 30 male college students from an introductory psychology course who volunteered for this experiment as partial fulfillment of the requirements of the course. Apparatus. The apparatus (see Fig. 1) was designed and constructed following Hess and Polt (1965) with certain modifications and improvements made on the prototype. The present apparatus consisted of a rectangular box 18” high, 16" wide, and 32" long. On one end was a viewing aperture such that when S's head was in place, the right eye was directly in front of the aperture. On the Opposite end there was mounted a 6" x 8" projection screen. The visual target was projected on this screen by a 500 watt 35mm. Kodak Carusel slide projector. The projector was placed 18" from the projection screen. Inside the rectangular box a 9" x 20" chromium- coated, one-way vision screen was placed at a 450 angle across the S's line of vision between aperture and projection screen. This mirror reflected the image of S's eye directly into the lens of a 16mm. Eolair 1611 motion picture camera which was mounted on the side of the box. The camera was fitted with an Angenioux 12- 120mm. Zoom lens and a +2 diapter close-up lens. The distance along the visual axis from the viewing aperture 11 Slide Projector L_J If 18 F ‘6- w ' I«—8—-4 Rear 7 ‘ A? vision screen Panel Camera One-way mirror Aperture Fig. 1. Experimental Apparatus: top view. Scale: 1/8" = 1". l2 13 to the mirror was 9". The camera lens was a 90mm. Kilfitt Macro Kilar, with a shutter Speed set at .25 second at 4 frames per second. Illumination was furnished entirely by the SOC-watt projector. During projection of the eXperimental slides, reflected light was measured at 10.6 foot candles allowing a shutter opening of F4. The camera was loaded with Kodak XXX panchromatic 16mm. film. Also within the apparatus, 20" from the viewing aperture, was a &" plywood panel in which a 4" x 5.3" rectangular hole was cut. This panel was placed such that when looking through the viewing aperture, only the surface of the 6" x8" projection screen was visible. The interior area of the apparatus, from the aperture to the panel was painted a flat black; the remainder of the interior was painted an enamel white. Materials. The visual targets utilized in this exPeri- ment consisted of six 35mm. slides. Slides were made by imprinting Paratype symbols No. 3 upon clear plastic slides. A Pentex light meter, placed at the viewing aperture of the apparatus registered less than 300 millilamberts difference in light intensity between each word and its background. This is in accordance with Bender's (1933) finding that the reactivity 14 threshold of the pupil to changes in light intensity ranges from 300 to 500 millilamberts. "Emotionality" of the words used in this study was determined in the following manner. From the published lists of words that have been shown to affect behavior, (Heise, 1965; Solarz, 1963; Zajonc, 1962; Eriksen, Azuma and Hicks, 1959; Alexander, 1938; Cohen and Silverman, 1956; Howe and Solomon, 1951; Bruner and Postman, 1947) and that have been judged to be of either affective or neutral connotation, E selected an original sample of 30 words. These 30 words consisted of three groups of three-letter, four-letter, and five- letter words, with each group containing five neutral and five emotional words. These 30 words were then presented in a random order to a group of 58 students in an introductory psychology course who were asked to sort the words into two groups; one group containing those words which were thought to be emotional, or which would be expected to elicit emotional reSponses; and one group containing those words that were thought to be neutral or without emotional connotation. The judg- ments of these students--none of whom served as Se in the experiment--were then summed. Emotionality of the stimuli used in this study was thus describable on a scale ranging from zero (None of the 58 students judged 15 it to be emotional.) to 58 (All of the 58 students judged it to be emotional.) The scores of the thirty words that were originally chosen are presented in Appendix A. Controlling for frequency of usage and word-length (Rosenzweig and McNeil, 1962), the three words with the highest, and lowest scores, were then selected as the stimuli for this study. Table l is a presentation of these words and their corresponding scores 0 Table 1. Words used as stimuli and their corresponding scores of "judged emotionality." Score indicated the number of students out of 58 who judged the word to be emotional. Emotional Neutral Words Segre Words Score Sex 55 The 0 Kiss 54 Card 10 Vomit 53 Shelf 0 Mean Emotionality Mean Emotionality it 24.2 Procedure. To control possible effects due to order of presentation of stimuli, a program of order of presenta- tion for each S was devised. The program (see Appendix B) 16 was constructed by assigning words arranged in alphabetical order, numbers from 1 to 6, and then con- sulting a table of random numbers (Edwards, 1962). Data was collected in the following manner. Ss were seated before the apparatus and given the following instructions: "Please place this eyepatch over your left eye and look into the box. We are going to project a series of slides onto the viewing screen which you are now looking at. Please try to keep your eye on the screen at all times and just react to whatever you see. We would like you to remain silent while reacting to the slides." The E, and assistant, and a camera Operator were present during the eXperiment. After S was given the instructions, the camera was checked for focus, and the projector was activated. After a ten-second "orientation" period, the first stimulus was presented. Each visual target was automatically presented for five seconds in the pre-programmed order for that S. The speed of the slide projector was such that there was .5 seconds of relative darkness between target exposures. Thus there were collected twenty photographs of each S's pupil in response to each visual target. After all slides were presented, each S was given a packet of 3" x 5” cards on which were printed the words they had just viewed. 17 They were then asked to rank order the words in order of their emotional meaningfulness. The 3600 photographs were examined by use of a Baush and Lomb Scale Projector which projected the image of the photographed pupil onto a surface magnified 22 times. A standard mm. ruler was used to measure the size of the pupil on each frame. RESULTS The data of S 1 and 15 were discarded because they closed their eyes and measurement was not possible for over 25% of the photographs. The .5 second inter-target interval resulted in shading on the film which made presentation and termination of a given target easily recognized. Recorded measurements were transformed into an approximation of actual pupil size in relation to emotional and neutral stimuli and plotted in Fig. 2. The recordings were at 3 times actual pupil size. Pupil size in mm. 0\ to 01 Emotional Neutral 1"!t"'!3"r'12'"'io'r"2o Fig. 2. Intervals Approximation of mean actual pupil size for 28 Se during 5-second stimulus presentation. 18 19 Complete analysis of variance (Lindquist, 1953) of the data is presented in Table 2. Emotionality had a significant effect on recorded pupil size, as did words, time (intervals), and Ss. With the exception of EXS, the interaction effects were also significant. Table 2. Analysis of Variance for emotional vs. neutral words with 5:28 Subjects and 1:20 intervals for each word. Sgurce of Variance SS d,f, MS F Emotionality(E) 12.97 1 12.97 12.09 ** Words (W) 54.86 2 27.43 2.881* Intervals(I) 390.71 19 20.56 5.24 ** Subjects(S) 20,114.70 27 744.98 275.86 ** E X S 29.13 27 1.08 W X S 514.21 54 9.52 3.53 ** I X S 2,112.66 513 3.92 1.45 * E X W 83.17 2 41.59 2.81 * E X W X S 792.53 54 14.67 5.43 ** E x I 729.76 19 36.41 5.45 ** E X I X S 3,425.22 513 6.67 2.47 * W X I 250.61 38 6.59 3.27 ** W X I X S 2,065.00 1,026 2.01 E X W X I 501.27 38 13.19 4.88 ** E X W X I X S 2,g£2,75 i,026 2170 iotai 22' 2,55 2.252 *P.< .05 **P.< .005 To test for significant differences between pupil size in relation to each of the words a Duncan's New Multiple Range Test (Edwards, 1962) was applied to the differences in mean pupil size of all measurements of 20 each word. Results presented in Table 3 indicate that pupil size in two of the emotional words (kiss and vomit) was significantly larger than any of the neutral words but not significantly different from each other. Pupil size in response to "sex” was significantly larger than two of the neutral words (the and card) but not shelf. Table 3. Duncan's New Multiple Range Test applied to the differences between mean pupil size in response to W:6 words. Means the card shelf sex kiss vomit shortest sig. ranges at P=.05 the 18.664 .055 .096 .314 .512 .519 .224 card 18.719 .041 .259 .457 .463 .240 shelf 18.761 .218 .417 .423 .248 sex 18.979 .198 .205 .254 kiss 19.177 .007 .259 vomit 19.184 1 2 3 21_ 5A 6 _- Any two words not underscored by the same line are significantly different P