OVERDUE FINES ARE 25¢ PER DAY PER ITEM Return to book drop to remove this checkout from your record. THE DETERMINANTS OF THE STIMULUS-VALUE OF MENTAL IMAGERY By James L. Pretzer A THESIS Submitted to Michigan State University in partial fuifillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1979 ABSTRACT THE DETERMINANTS OF THE STIMULUS-VALUE OF MENTAL IMAGERY By James L. Pretzer The use of the stimulus-value of mental imagery (SVMI) in psychotherapy offers many advantages. However, studies investi- gating the role of SVMI in psychotherapy have produced mixed results. This study tests a theoretical model of the determinants of SVMI in hopes of providing a means of understanding these mixed results. Undergraduate 55 completed measures of personality vari- ables and imaged twenty scenes, rating both SVMI and the charac- teristics of the image while experiencing the image. Half of the trials were conducted before three weeks of progressive relaxation training and half after. The hypothesized model was tested through causal analysis and path analysis. While nineteen of twenty-two predicted correlations sup- ported the hypotheses, the path analysis and causal analysis pro- duced partially conflicting results. It was concluded that this study provides support for basic assumptions involved in the use of imagery in psychotherapy but does not provide an empirically sup- ported model of the processes involved. ACKNOWLEDGMENTS I would particularly like to thank the members of my thesis committee: Dr. Dozier Thornton who provided sound advice, encour- agement, and support; Dr. Gordon Williams who raised important methodological and theoretical questions; and Dr. Ralph Levine who worked closely with me on the statistical analyses. Through them this thesis was transformed from a rite of passage to a valuable experience. In addition, it would like to thank Dr. Auke Tellegen for providing me with the Differential Personality Questionnaire, Dr. Alan Richardson for providing me with information on several measures of image quality. and Dr. F. D. McGlynn for sharing some unpublished data with me. Their cooperation has greatly facili- tated this research. Most importantly I would like to thank my friends and colleagues who provided support, encouragement, and advice and who helped me to persevere with a complex, time-consuming study. My special thanks go to Barb, Howard, and Cheryl who supported me throughout the project. ii TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES . STATEMENT OF THE PROBLEM . RELATED LITERATURE . The Stimulus Properties of Mental Imagery . . The Stimulus- Value of Mental Imagery in Psychotherapy The Determinants of the Stimulus- Value of Mental Imagery . . . . . . Summary HYPOTHESES METHOD . Overview of the Procedure Subject Selection . Settings and Apparatus Measures Tendency to Avoid Aversive Stimuli Imagery Ability. . Stimulus- Value of the Stimulus Imaged Demographic Information . Subject' 5 Perception of the Study. Personality Data . Initial Level of Relaxation or Anxiety and Level of Relaxation or Anxiety. Relaxation Ability. Image Quality . . . Stimulus- Value of the Experienced Image . Relaxation Instructions . . Experimental Procedure Session One Session Two Session Three Session Four . Session Five . Page vi RESULTS Effects of Subject Attrition Equivalence of Groups . Equivalence of Assignment to Experimental Cdnditions Reliability of Keyboard Ratings . Reliability of the Relaxation Rating Checklist Correlations Between Variables . Causal Analysis . DISCUSSION REFERENCES APPENDICES A. GJ'TII'T‘IUOW Outline of Presentation Used in Subject Recruitment . Stimulus-Value Survey Schedule . Personal Data Sheet Participant's Evaluation Form Relaxation Rating Checklist . Experimental Instructions Used in Sessions Two and Five. Relaxation Monitoring Card iv 107 110 114 118 124 126 139 Table LIST OF TABLES T-Tests for the Effects of Subject Attrition T-Tests for the Effects of Subject Attrition One Way ANOVA for Differences Between Groups on DPQ Unfriendly World Scores . . One Way ANOVA for Differences Between Groups on DPQ Inconsistency Scores . Reliability of Subjects' Keyboard Responses Inter-rater Reliabilities for the Relaxation Rating Checklist . . . . . . . . . . . Correlations Between Variables . Page 75 76 7B 79 81 83 84 Figure OOWNO‘U'IDOJN d d d d d d d d \l 05 UT -b m N -‘ C O . C . O . LIST OF FIGURES Hypothesized Causal Model Hypothesized Direct Effects Correlations Predicted on the Basis of Direct Effects . Setting for Relaxation Training and Imagery Trials . Subject Response Keyboard Rating Scale Used for ILRX and LRX Timing of Ratings or LRX Rating Scale fur Image Vividness . Rating Scale for Image Controlability The Range of Values for IMO Rating Scale for SVEI Outline of the Procedure Timing of Instructions During Imagery Trials . Sequence of Relaxation Instructions . Causal Analysis Path Analysis Combined Results of Path Analysis and Causal Analysis . vi Page 40 4l 42 45 47 56 57 6O 61 63 64 67 69 71 87 88 9O STATEMENT OF THE PROBLEM The successful use of mental imagery in a variety of psychotherapeutic techniques holds potential for increasing the effectiveness of psychotherapy (Beck, l970; Wilkins, 1974). Imagery techniques have been used with a wide range of psychological and behavioral problems (Wilkins, 1974) and are often reported to be more effective and more efficient than verbal techniques (Sheikh & Panagiotou, l975). Though there are some difficulties in using imagery techniques, the advantages of these techniques are often believed to outweigh the disadvantages (Cautela, l973). These reported advantages include greater experience of self-control by clients, increased flexibility due to decreased reliance on equip- ment, freedom to work with situations which cannot be dealt with practically in vivo, and avoiding a high drop-out rate and occa- sional overgeneralization when using aversive stimuli (Cautela, l973). Mental imagery is also reported to provide a more accurate recreation of a situation than a verbal description does (Gordon, l972) and to evoke a stronger emotional response (Kosab, 1974). The enthusiasm of proponents of imagery techniques is often based on clinical experience rather than more controlled investiga- tion. Despite the number of hypotheses and theories which have been generated, much remains to be investigated concerning the function of mental imagery in psychotherapy (Wilkins, 1974; Singer, 1973). Psychotherapeutic techniques which use mental imagery can be divided into those techniques which rely on the stimulus proper- ties of imagery and those which rely on the symbolic properties of imagery. Stimulus-oriented techniques such as sytematic desensiti- zation, covert conditioning, and covert modeling are typically based on a behavioristic approach to psychopathology and treat the mental images as interchangable with real stimuli. Techniques using the symbolic properties of imagery such as Reyher's Emergent Uncover- ing Therapy (Reyher, 1963) and Jung's "active imagination" are often based on a psychodynamic approach to behavior and imagery is seen as a product of the individual's personality which can be used as a source of insight or information. Stimulus-oriented imagery techniques are more widely used than symbolic techniques and are both more amenable to scientific investigation and more thoroughly investigated. Research into the nature and function of the stimulus properties of mental imagery has supported the assertion that mental imagery is potentially use- ful in psychotherapy and has shown that mental imagery is superior to verbal description for recreating a stimulus or experience in some situations (Gerst, 1969). However, research into psychotherapy techniques using the stimulus properties of imagery has not con- sistently supported the hypothesis that imagery is a crucial vari- able in the effectiveness of these techniques (Beere, 1971). It has been suggested (Beere, 1971) that the characteristics of imagery experienced during psychotherapy are multiply determined and that an understanding of the determinants of the stimulus value of mental imagery experienced in clinical settings would resolve the discrep- ancy between the imagery research which supports the use of imagery techniques and the psychotherapy research which fails to support the importance of imagery. This study investigates the determinants of the stimulus value of mental imagery. RELATED LITERATURE The Stimulus Properties of Mental Imagery_ Mental imagery has consistently been defined by contemporary researchers in terms of sensation or perception in the absence of the stimuli which customarily evoke such sensations or perceptions (Sheehan, 1972). In this study a definition of mental imagery is used which is based on the definition used by Gordon (1972). The term mental image (image) is defined as perception in any modality in the absence of the appropriate stimulus. Often no distinction is made between images and hallucinations; when such a distinction is made images are seen as being less vivid and more controlled than hallucinations. There is no commonly accepted criterion available which reliably discriminates between vivid, spontaneous images and hallucinations. A variety of typographies for imagery have been proposed, ranging from Richardson's (1969) discussion of four types of imagery to Horowitz's (1970) complex, four-fold categorization of imagery based on vividness, context, interaction with perception, and con- tent. This study is focused on imagery produced upon request by subjects (55) either from memory or from imagination and, unless otherwise indicated, the term imagery refers specifically to this phenomena. Obviously imagery is a subjective experience which cannot be measured directly. ,The actual data available consists of self- reports of imagery, performance measures of imagery, and physio- logical correlates of imagery. For the sake of brevity the term image is used and we will speak of images "occurring" or “being Iexperienced" and of scenes or objects "being imaged“ without ignor- ing the source of the data and its limitations. Though systematic use of mental imagery as a mnemonic device antedates Aristotle, the scientific study of imagery dates from Sir Francis Galton's early survey of the varieties of imagery experience. Early psychologists considered the study of imagery an important part of the study of the mind and used a variety of introspective methods to collect data. However, both the Anssage method used by Titchener at Cornell and the Ausfrage method used by the Wurzburg school reached the limits of their usefulness in the famous contro- versy over the existence of imageless thought (Holt, 1964). The two new approaches to psychology which developed in the years preceding World War I, behaviorism and psychoanalysis, both focus primarily on behavior, through their viewpoints differ con- siderably, and both assume that the contents of consciousness explain little. In the reaction against introspectionism which accompanied the rise of behaviorism in academic circles, imagery, attention, states of consciousness, and other subjective phenomena were rejected as mentalistic and received little scientific atten- tion (Hebb, 1960). While the theoretical bias against imagery research con- tinued at least into the 19605, a number of factors pressed for the re-emergence of imagery research. The practical concern which developed when spontaneous, vivid images were involved in accidents by radar operators, long-distance truck drivers, and jet pilots could not be dismissed by the argument that talk of images is "mentalistic." The subsequent discovery of "experimental halluci- nations“ during sensory deprivation~(Hebb, 1960) both spurred interest in this area of imagery research and demonstrated that images could be studied in the laboratory. Among advances in neuropsychological research which encouraged imagery research were Pennfield's studies in direct stimulation of the cortex (discussed in Holt, 1964). This research demonstrated that experiences were recorded in unexpected detail and provided a neurological basis for understanding eidetic imagery and memory imagery. Research suggesting a relationship between EEG alpha waves and visual imagery raised the hope that imagery research could be made objective by the development of a physiological measure of the occurrence of imagery. Subsequent research has shown that the rela- tionship between the occurrence of imagery and various physiologi- cal measures is complex (Morishige & Reyher, 1975; Sheehan, 1973) but studies in this area have further developed methods for imagery research. The emergence of information theory, high-speed computers, and artificial intelligence has provided models for relating cognitive processes to behavior. "Cognitive behaviorism“ with its recognition of the impact of expectations, verbal mediation, and other cognitive processes on behavior has provided a place for imagery in contemporary theoretical psychology. A powerful, non-scientific force in the re-emergence of imagery research is the pattern of socio-cultural change which has led to an upsurge in interest in mysticism, drug-induced exper- iences, meditation, para-psychology, encounter groups, and similar phenomena. Psychologists, being human, have reflected this trend to some degree in changing research interests. In responding to psychologists who argue that the study of imagery cannot be scientific, Sheehan (1966) distinguishes between subjective research where introspective reports are assumed to be valid and objective research which examines the relationship between variations in introspective reports and variations in behavior. An example of the objective approach would be Marks' (1972) statement* that research into the function of vividness of visual imagery is directed towards the question, "In what ways does the behavior of a man who says he experiences vivid visual imagery differ from that of another who says that his imagery is vague and dim?" Certainly this approach to research can be methodologically rigorous and scienti- fic. Of the many properties of imagery which have been the sub- ject of research or speculation, one property which is directly relevant to many imagery techniques is the functional similarity between imaging and perceiving. Research on the stimulus properties of imagery has focused on the relationship between imaging and per- ception of stimuli, the similarities between responses to images and responses to stimuli, and the use of images as stimuli in per- forming a variety of tasks. Much of the research on the relationships between imaging and perception has been based on the Perky phenomenon. Perky, a student of Titchener, demonstrated (Perky, 1910) that if a subject imagined an image projected on a screen and the experimenter (E) projected a faint image of the imagined object on the screen the S was unable to detect the projected image. Segal (1972) has carried out extensive work on this phenomenon exploring the variables influ- encing its occurrence. She has shown that this is a genuine effect, replicable with naive Ss and that it is modality specific, i.e., that visual images affect only visual perception while auditory images affect only auditory perception. Similar conflicts between imagery and perception in the same modality have been found by Antrobus, et a1. (1970) in a study of daydreaming during signal detection tasks, Brooks (discussed in Neisser, 1972) in cognitive tasks where the response involved a conflict between imagery and perception, and Fusella (1973) in a visual signal detection task during which 55 were asked to experience visual imagery. Though, for normal 85, imagery can only block perception of weak stimuli, the demonstration of modality specific interference between imagery and perception is seen as evidence that imaging is a quasi- perceptual process (Neisser, 1972) or that the same neurological process is involved in both phenomena (Hebb, 1968). Clear evidence of the stimulus properties of imagery has come from research on the similarity between physiological responses to stimuli and physiological responses to images of stimuli. After a thorough review of psychophysiological studies of systematic desensitization (SD) and related procedures, Matthews (1971) con- cluded that the research had demonstrated that the physiological effects of phobic imagery are consistently different from the effects of neutral imagery. The consensus is that physiological and subjective responses to phobic imagery are qualitatively simi- lar to responses to actual stimuli but are less intense. Evidence that $5 respond physiologically to phobic imagery in the same way as they respond to phobic objects has been reported by Wolpe and Flood (1970), Van Egeren (1970), Van Egeren, Feather, and Hein (1971), Marks and Huson (1973), Haney and Euse (1976), and May (1977). Craig (1968) demonstrated that this relationship held for other types of imagery by showing that physiological response to an imagined cold stressor was qualitatively similar to 35 response to the actual stressor. A sophisticated demonstration that physiological and sub- jective responses to imagery discriminates between phobic and neutral imagery was provided by Chapman and Feather's (1971) sensory decision theory analysis of the data reported by Van Egeren, Feather, and Hein (1971). This study in conjunction with the other research shows that responses to imagery resemble responses to actual stimuli and that responses to imagery reliably discriminate between types of images. 10 A striking example of the stimulus properties of imagery is the performance of Stromeyer's eidetiker "Elizabeth" (Stromeyer, 1970; Stromeyer & Psotka, 1970). Elizabeth was able to summon up a sufficiently clear image of a 10,000 dot random stereogram seen the day before to combine it with the second stereogram of the pair and report the figure formed by their stereoscopic combination. She was also able to duplicate the Land Color Phenomenon by superimposing an eidetic image of a picture projected through a red filter on the picture projected through a green filter and perceiving the picture in full color with remarkable accuracy. These tests do not rely on memory and cannot be faked easily. A less unusual demonstration of the use of imagery as a stimulus is the use of imagery by normal 55 to improve performance on a variety of cognitve tasks. A number of studies which have investigated hypotheses related to Betts' statement, "In all places where we would welcome the percept but cannot have it, the image may serve as a very acceptable substitute" (Betts, 1909, p. 93) are discussed by Sheehan (1972b) in his examination of the use of imagery in unexpected recall. -In a number of these studies imagery orient- ing instructions improved incidental learning, concrete (high imagery-evoking) noun pairs were more easily recalled, and 55 high in imagery ability performed better in unexpected recall tasks. Bugelski (1970) found that 55 who were instructed to image but not to learn were unable to prevent learning. I In Sheehan and Neisser's (1969) study of the use of imagery in reconstruction of a pattern of geometric figures two Kohs Block 11 Design patterns were used as an intervening task to control S activity. When Ss were unexpectedly asked to recall the Kohs design and were asked whether they had employed imagery in doing so most 55 reported that imagery was intimately involved in their recall process (Sheehan, 1972b). This finding spurred a more systematic examination of the role of imagery in incidental learning. In a series of experiments based on a verbal learning paradigm (Sheehan, 1971a, 1971b, 1972c) Sheehan found that when high imagery-arousing and low imagery-arousing nouns were equated for meaningfulness and frequency, high imagery-arousing nouns were more frequently recalled in an unexpected recognition task and that, for high imagery nouns, incidental learning was not significantly less efficient than inten- tional learning (Sheehan, 1972b). In a series of studies involving a "pattern construction" task in which Ss were asked to duplicate patterns made of geometric forms Sheehan (1966) found that 55 made fewer errors when asked to duplicate the pattern by using a mental image of the pattern as a stimulus than when they were asked to duplicate the pattern without imaging the pattern. This relationship held if 55 inspected the elements of the pattern before the stimulus pattern was presented but the opposite result was found when 55 were not allowed to inspect the elements of the pattern. This apparent inconsistency was explained in terms of the vividness of the imagery experienced. Inspection of the elements of the pattern resulted in reports of more vivid imagery and Ss who reported vivid images made fewer errors than 55 who reported dim images or 55 who were asked to 12 use memory without imagery. In a second portion of this study Sheehan (1966a) investigated the possibility that differential demand characteristics confounded his findings and once again found that reproduction of geometric patterns by 55 who reported vivid images was more accurate than the performance of 55 who reproduced the pattern from memory without using imagery. Further research into the role of image vividness in this type of task and into the pro- cesses involved has supported the belief that images can be used to reproduce stimuli and that this use of imagery results in a signifi- cant change in performance (Marks, 1972). The Stimulus-Value of Mental Imagery in Psychotherapy The psychotherapy techniques which rely on the stimulus properties of imagery share the common assumption that mental imagery effectively recreates stimuli and is governed by the same laws which govern overt behavior (Cautela, 1973). These techniques assume that imaged stimuli, behavior, and consequences have the same effects as if they had actually occurred. Although research into the stimulus properties of imagery offers partial support for these assumptions, studies directly investigating the role of imagery in these psycho- therapy techniques have not offered consistent support for the assump- tion that imagery is a functional component of these techniques. Research into the role of imagery as a stimulus in psycho- therapy has used two basic approaches--comparing a procedure using imagery with a procedure using real stimuli and comparing the effects of imagery which accurately recreates the desired stimulus with the 13 effects of imagery which does not accurately recreate the desired stimulus. The studies which compare imagery procedures with procedures using real stimuli cannot investigate the role of imagery directly. A comparison showing that imagery techniques are as effective as other techniques does not prove that imagery is a functional com- ponent of the technique because it cannot discount the alternative hypothesis that some other variable is responsible for the effective- ness of both procedures. Similarly, a comparison showing differ- ences in effectiveness can investigate the role of imagery only if all other variables are controlled, a circumstance which is diffi- cult to achieve. Studies investigating the relationship between image quality and outcome of treatment rest on the assumption that images which are very similar to the perception of the desired stimulus will be more effective than images which are dissimilar to the desired stim- ulus either in terms of quality or content. These studies directly investigate the role of imagery in imagery techniques and make it possible to systematically control other variables which may influ- ence treatment outcome. The psychotherapy technique in which the role of imagery has been most thoroughly investigated is systematic desensitization (SD). As Paul (1966) has pointed out, 50 is not one technique but a collec- tion of similar techniques based on Wolpe's original procedure. Despite the extensive debate concerning the process or processes responsible for the effects of $0, the effectiveness of SD is not 14 questioned. SD is probably the most thoroughly researched psycho- therapeutic technique and has been empirically demonstrated to be effective. Yates (1975) presented an increasingly common view when he wrote: The original technique of systematic desensitization was precisely, if complexly, specified. Over the years, how- ever, each and every one of the components of systematic desensitization have been shown to be neither necessary nor sufficient--like the Cheshire cat left with only its smile, systematic desensitization seems to work, but there seems to be no component parts that cannot be removed, and the technique will then fail or be significantly reduced in its efficiency. . . . (can) we then continue to accept that we are dealing with a form of therapy that can be meanin ful called systematic desensitization . . .? IA) major reappraisal would appear to be required. During the past five years the theoretical underpinnings of SD have been under a continuous assault and increasingly authors have suggested that it may be no more than a singularly effective placebo. However, in a recent review of the role of non-specific treatment factors in SD, Kazdin and Wilcoxon (1976) discuss a number of methodological flaws common to much of the SD research which raise questions concerning the validity and generalizability of many of the findings. As is true of much psychological research, research on SD has generally relied on undergraduate college students as 55 because of their availability and convenience. Unfortunately, it has been discovered that undergraduates with fears of small animals, tests, and public speaking bear little resemblance to persons with clinical phobias (Olley & McAllister, 1975) and there is no evidence to 15 support the common practice of generalizing findings based on undergraduate S5 to the treatment of persons with clinical phobias. Also much research on SD has attempted to control for placebo effects by comparing SD with a pseudotherapy or attention placebo. A series of studies (Borkovec & Nau, 1972; Nau, Caputo, & Borkovec, 1974; McGlynn & McDonnel, 1974) has demonstrated that SD, pseudotherapies, and attention placebos are not equally cred- ible and that $0 is consistently more credible than most procedures used to control for placebo effects. These findings suggest that placebo effects were not adequately controlled and that superior outcomes of SD could be due to placebo effects. Since the credi- bility of treatments was typically not assessed it is impossible to judge the importance of this flaw without further research. Another methodological flaw which is present in many studies but which is rarely discussed is the unwitting use of control groups which may have therapeutic effects. An example of this appears in a study by McGlynn and McClaren (1975) where the treatment designed to control for non-specific effects consisted of visualization of pleasant imagery. It was assumed that, since the treatment included neither phobic imagery nor relaxation train- ing, it was inert. However, research has demonstrated that visual- ization of pleasant images can be used to reduce pain, discomfort, and anxiety (Horan, Layling & Pursell, 1976; Horan, 1973). Thus though McGlynn and McClaren (1975) interpreted the lack of a signif- icant difference in outcome between treatment groups as evidence .that placebo effect were responsible for decreases in avoidance 16 behavior it is quite possible that all treatments had specific, though different, therapeutic effects. This type of flaw leads to misinterpretation of results primarily when there is no significant difference between control and treatment groups. The three methodological flaws discussed could result in a lack of generalizable findings, false positive results, and false negative results. Though the research on SD is voluminous much of it is of doubtful value and is best regarded as merely suggestive (Kazdin & Wilcoxon, 1976). Little is definitely known about $0 but that it works for a range of problems. The reciprocal inhibition theory of SD postulates that muscle relaxation or some other inhibitory response is essential for $0 to be effective and early studies supported this view (Yates, 1970, 1974). More recently, SD has been found to be effective with or without relaxation training with both undergraduate subjects (Crowder & Thornton, 1970; Water, McDonald, & Koresko, 1972; Waters & McDonald, 1973; Henkel & Bastine, 1972) and clinical phobics (Agras, et al., 1971; Craighead, 1973). Some studies with under- graduate Ss have found that muscle tension is an effective as muscle relaxation (Proctor, 1969; Sue, 1972) but other studies have suggested that, while muscle relaxation is not essential, it facilitates SD (Nawas, Welsh, & Fishman, 1970; Matthews, 1971). Muscle relaxation, as used in SD, resembles the conditions of reduced sensory stimulation which have been found to increase the vividness of imagery in other situations (Singer, 1973) and it has been suggested that relaxation functions in this way in SD. 17 Wolpe and Flood (1970), Van Egeren, Feather, Hein (1971), Van (Egeren (1970), and Chapman and Feather (1971) found that relaxed 55 had stronger physiological responses to phobic imagery than non-relaxed S5 and this may indicate that the images were more vivid or realistic fOr relaxed Ss. Wolpin and Kirsh (1974) found that relaxation affected the quality of the imagery by making the images more benevolent and increasing the 55' feeling of involve- ment but a previous study (Henkel & Bastine, 1972) found that relaxation did not affect image quality. The role assigned to construction of a hierarchy of phobic scenes was central in Wolpe's original fbrmulation of 50 because it was theoretically important that the client not experience anxiety intense enough to overcome the inhibitory effect of relax- ation. The research in this area clearly implies that the construc- tion of a hierarchy is unnecessary (Yates, 1975) but the research is inconclusive because all of the studies reviewed by Yates (1975) used undergraduate 55 rather than clinical phobics. The research. shows that hierarchies are unnecessary with undergraduates with subclinical fears but generalization to clinical populations is of doubtful validity. Clinically, proceeding through a graduated hierarchy of phobic scenes minimizes clients' negative emotional reactions to treatment and decreases the probability of their terminating therapy prematurely (Bandura, 1969). With the mounting evidence that the theory of reciprocal inhibition could not adequately explain SD and that none of the components of SD were necessary or sufficient, it has been 19 visualize hierarchy scenes do not respond to SD (Lazarus, 1964), and difficulty experiencing vivid imagery has been considered a major cause of failure of SD (Darwin & McBreaty, 1969; Richardson, 1972; Wolpe and Lazarus, 1966). In his discussion of the treat- ment of phobias Marks (1969) suggests that inability to obtain images, dissociation of anxiety from phobic imagery, or dilution or intensification of the phobic value of imagery will result in failure of SD. Clinical reports that some clients who are unable to experience phobic imagery and thus do not respond to SD respond well to "in vivo" desensitization (Bonem, 1976) suggest that imagery does serve a functional role in 50. However, many of the studies which have directly investigated the role of imagery (Davis, McLemore, & London, 1970; McLemore, 1971; Beere, 1971; Hyman, 1973) have not supported this hypothesis. In the first reported empirical research on this topic Davis, McLemore, and London (1970) investigated the relationship between a measure of "visual imagery ability" and outcome of $0. This study found a non-significant relationship between "visual imagery ability" and a behavior change score with pre-therapy performance controlled; however, a number of methodological flaws make the results uninterpretable. The "visual imagery ability" measure used actually measured the relative dominance of imagery in the visual modality over imagery in other sensory modalities rather than the ability to experience vivid visual imagery on request. Also, the subjects were self-selected from a larger pool of 55 who had previously undergone SD and it is possible 12 use memory without imagery. In a second portion of this study Sheehan (1966a) investigated the possibility that differential demand characteristics confounded his findings and once again found that reproduction of geometric patterns by 55 who reported vivid images was more accurate than the performance of 55 who reproduced the pattern from memory without using imagery. Further research into the role of image vividness in this type of task and into the pro- cesses involved has supported the belief that images can be used to reproduce stimuli and that this use of imagery results in a signifi- cant change in performance (Marks, 1972). The Stimulus-Value of Mental Imagery in Psychotherapy The psychotherapy techniques which rely on the stimulus properties of imagery share the common assumption that mental imagery effectively recreates stimuli and is governed by the same laws which govern overt behavior (Cautela, 1973). These techniques assume that imaged stimuli, behavior, and consequences have the same effects as if they had actually occurred. Although research into the stimulus properties of imagery offers partial support for these assumptions, studies directly investigating the role of imagery in these psycho- therapy techniques have not offered consistent support for the assump- tion that imagery is a functional component of these techniques. Research into the role of imagery as a stimulus in psycho- therapy has used two basic approaches--comparing a procedure using imagery with a procedure using real stimuli and comparing the effects of imagery which accurately recreates the desired stimulus with the 13 effects of imagery which does not accurately recreate the desired stimulus. The studies which compare imagery procedures with procedures using real stimuli cannot investigate the role of imagery directly. A comparison showing that imagery techniques are as effective as other techniques does not prove that imagery is a functional com- ponent of the technique because it cannot discount the alternative hypothesis that some other variable is responsible for the effective- ness of both procedures. Similarly, a comparison showing differ- ences in effectiveness can investigate the role of imagery only if all other variables are controlled, a circumstance which is diffi- cult to achieve. Studies investigating the relationship between image quality and outcome of treatment rest on the assumption that images which are very similar to the perception of the desired stimulus will be more effective than images which are dissimilar to the desired stim- ulus either in terms of quality or content. These studies directly investigate the role of imagery in imagery techniques and make it possible to systematically control other variables which may influ- ence treatment outcome. The psychotherapy technique in which the role of imagery has been most thoroughly investigated is systematic desensitization (SD). As Paul (1966) has pointed out, SD is not one technique but a collec- tion of similar techniques based on Wolpe's original procedure. Despite the extensive debate concerning the process or processes responsible for the effects of $0, the effectiveness of SD is not 14 questioned. SD is probably the most thoroughly researched psycho- therapeutic technique and has been empirically demonstrated to be effective. Yates (1975) presented an increasingly common view when he wrote: The original technique of systematic desensitization was precisely, if complexly, specified. Over the years, how- ever, each and every one of the components of systematic desensitization have been shown to be neither necessary nor sufficient--like the Cheshire cat left with only its smile, systematic desensitization seems to work, but there seems to be no component parts that cannot be removed, and the technique will then fail or be significantly reduced in its efficiency. . . . (can) we then continue to accept that we are dealing with a form of therapy that can be meaningful called systematic desensitization . . .? . . . (A) major reappraisal would appear to be required. During the past five years the theoretical underpinnings of 50 have been under a continuous assault and increasingly authors have suggested that it may be no more than a singularly effective placebo. However, in a recent review of the role of non-specific treatment factors in SD, Kazdin and Wilcoxon (1976) discuss a number of methodological flaws common to much of the SD research which raise questions concerning the validity and generalizability of many of the findings. As is true of much psychological research, research on 50 has generally relied on undergraduate college students as 55 because of their availability and convenience. Unfortunately, it has been discovered that undergraduates with fears of small animals, tests, and public speaking bear little resemblance to persons with clinical phobias (Olley & McAllister, 1975) and there is no evidence to 15 support the common practice of generalizing findings based on undergraduate $5 to the treatment of persons with clinical phobias. Also much research on 50 has attempted to control for placebo effects by comparing SD with a pseudotherapy or attention placebo. A series of studies (Borkovec & Nau, 1972; Nau, Caputo, & Borkovec, 1974; McGlynn & McDonnel, 1974) has demonstrated that $0, pseudotherapies, and attention placebos are not equally cred- ible and that SD is consistently more credible than most procedures used to control for placebo effects. These findings suggest that placebo effects were not adequately controlled and that superior outcomes of SD could be due to placebo effects. Since the credi- bility of treatments was typically not assessed it is impossible to judge the importance of this flaw without further research. Another methodological flaw which is present in many studies but which is rarely discussed is the unwitting use of control groups which may have therapeutic effects. An example of this appears in a study by McGlynn and McClaren (1975) where the treatment designed to control for non-specific effects consisted of visualization of pleasant imagery. It was assumed that, since the treatment included neither phobic imagery nor relaxation train- ing, it was inert. However, research has demonstrated that visual- ization of pleasant images can be used to reduce pain, discomfort, and anxiety (Horan, Layling & Pursell, 1976; Horan, 1973). Thus though McGlynn and McClaren (1975) interpreted the lack of a signif- icant difference in outcome between treatment groups as evidence -that placebo effect were responsible for decreases in avoidance 16 behavior it is quite possible that all treatments had specific, though different, therapeutic effects. This type of flaw leads to misinterpretation of results primarily when there is no significant difference between control and treatment groups. The three methodological flaws discussed could result in a lack of generalizable findings, false positive results, and false negative results. Though the research on SD is voluminous much of it is of doubtful value and is best regarded as merely suggestive (Kazdin & Wilcoxon, 1976). Little is definitely known about 50 but that it works for a range of problems. The reciprocal inhibition theory of SD postulates that muscle relaxation or some other inhibitory response is essential for $0 to be effective and early studies supported this view (Yates, 1970, 1974). More recently, SD has been found to be effective with or without relaxation training with both undergraduate subjects (Crowder & Thornton, 1970; Water, McDonald, & Koresko, 1972; Waters & McDonald, 1973; Henkel & Bastine, 1972) and clinical phobics (Agras, et a1., 1971; Craighead, 1973). Some studies with under- graduate Ss have found that muscle tension is an effective as muscle relaxation (Proctor, 1969; Sue, 1972) but other studies have suggested that, while muscle relaxation is not essential, it facilitates SD (Nawas, Welsh, & Fishman, 1970; Matthews, 1971). Muscle relaxation, as used in SD, resembles the conditions of reduced sensory stimulation which have been found to increase the vividness of imagery in other situations (Singer, 1973) and it has been suggested that relaxation functions in this way in SD. 17 Wolpe and Flood (1970), Van Egeren, Feather, Hein (1971); Van Egeren (1970), and Chapman and Feather (1971) found that relaxed 55 had stronger physiological responses to phobic imagery than non-relaxed Ss and this may indicate that the images were more vivid or realistic for relaxed Ss. Wolpin and Kirsh (1974) found that relaxation affected the quality of the imagery by making the images more benevolent and increasing the 55' feeling of involve- ment but a previous study (Henkel & Bastine, 1972) found that relaxation did not affect image quality. The role assigned to construction of a hierarchy of phobic scenes was central in Wolpe's original formulation of SD because it was theoretically important that the client not experience anxiety intense enough to overcome the inhibitory effect of relax- ation. The research in this area clearly implies that the construc- tion of a hierarchy is unnecessary (Yates, 1975) but the research is inconclusive because all of the studies reviewed by Yates (1975) used undergraduate 55 rather than clinical phobics. The research. shows that hierarchies are unnecessary with undergraduates with subclinical fears but generalization to clinical populations is of doubtful validity. Clinically, proceeding through a graduated hierarchy of phobic scenes minimizes clients' negative emotional reactions to treatment and decreases the probability of their terminating therapy prematurely (Bandura, 1969). With the mounting evidence that the theory of reciprocal inhibition could not adequately explain SD and that none of the components of 50 were necessary or sufficient, it has been 18 increasingly suggested that behavior change fellowing SD is due to non-specific (placebo) effects, specifically the client's expec- tancy of change. Much research has been conducted in this area but Kazdin and Wilcoxon (1976) found that almost all of the studies used control procedures which are less credible than SD and/or used undergraduate 55 who may be more susceptible to placebo effects than clinical 55. The conclusion they reached after an extensive review was: The most parsimonious explanation of the results would seem to be that systematic desensitization includes a specific therapeutic ingredient which accounts for change. . . . However, when the different investigations of desensitization are examined, it appears that the results still do not unambiguously rule out non-specific treatment effects as a rival hypothesis (Kazdin & Wilcoxon, 1976). Of the various components, procedural variations, and theoretical explanations of 50 which have been investigated, none have been found to be consistently related to the outcome of 50. Indeed, the at least occasional successes of implosive therapy and flooding procedures which resemble SD only in their use of phobic imagery argue that the effective component of SD is imagery, a conclusion reached by Wilkins (1971, 1972) and supported by Singer (1973). All of the theories concerning specific treatment effects of SD treat phobic imagery as interchangeable with real stimuli. The ability to experience vivid images of phobic stimuli on request is considered necessary for SD to be successful (Lazarus, 1964; Paul, 1966). Clients who report that they are unable to 19 visualize hierarchy scenes do not respond to SD (Lazarus, 1964), and difficulty experiencing vivid imagery has been considered a major cause of failure of SD (Darwin & McBreaty, 1969; Richardson, 1972; Wolpe and Lazarus, 1966). In his discussion of the treat- ment of phobias Marks (1969) suggests that inability to obtain images, dissociation of anxiety from phobic imagery, or dilution or intensification of the phobic value of imagery will result in failure of 50. Clinical reports that some clients who are unable to experience phobic imagery and thus do not respond to SD respond well to "in vivo" desensitization (Bonem, 1976) suggest that imagery does serve a functional role in SD. However, many of the studies which have directly investigated the role of imagery (Davis, McLemore, a London, 1970; McLemore, 1971; Beere, 1971; Hyman, 1973) have not supported this hypothesis. In the first reported empirical research on this topic Davis, McLemore, and London (1970) investigated the relationship between a measure of "visual imagery ability" and outcome of SD. This study found a non—significant relationship between “visual imagery ability" and a behavior change score with pre-therapy performance controlled; however, a number of methodological flaws make the results uninterpretable. The "visual imagery ability" measure used actually measured the relative dominance of imagery in the visual modality over imagery in other sensory modalities rather than the ability to experience vivid visual imagery on request. Also, the subjects were self-selected from a larger pool of S5 who had previously undergone SD and it is possible 20 that the sample was biased in unknown ways. Finally, even if the findings were valid, since the S5 were undergraduates with sub- clinical fears, the findings may not generalize to clinical 50. In an effort to overcome the problems caused by the use of an unvalidated measure of imagery and a possibly biased sample of Ss, McLemore (1971) conducted another study with undergraduate Ss in which he used four subscales from Sheehan's (1967) short form of Betts' Questionnaire Upon Mental Imagery (QMI) and Gordon's Control of Visual Imagery Questionnaire (CVIQ) (Richardson, 1969) as measures of imagery. Though he did not find a significant relationship between imagery measures and outcome of SD it is extremely doubtful that these findings can be generalized to any population because the 55' fears were so mild that they responded well to two "therapy" sessions totaling 60 minutes of taped SD. Also, the "trait" scores obtained from the QMI and CVIQ are not good predictors of imagery experienced during SD (Beere, 1971). Thus the lack of a significant relationship between the ”trait“ measures and outcome of so does not imply lack of a relationship between imagery experienced during SD and outcome. In a test of London's hypothesis that the crucial variable in both so and implosive therapy is the elicitation of vivid imagery, Beere (1971) examined the relationship between image vividness and controlability and outcome of $0. In this well controlled study he found no significant relationship between either QMI and CVIQ scores or the reported vividness of imagery during 50 and the outcome of SD. 21 Beere found that QMI scores predicted vividness of neutral images during SD but did not predict vividness of phobic images. He hypothesized that reported vividness of imagery was a function of ability to image vividly and the anxiety-potential of the images requested, that $5 capable of imaging vividly avoided anxiety by producing less vivid images of phobic scenes. While this could explain some of Beere's findings it is puzzling that reported image vividness was significantly related to reports of anxiety but not significantly related to outcome of SD. It is possible that, since this was an analog study, non-specific treatment effects masked specific treatment effects, but it is also possible to interpret Beere's findings as indicating that imagery is not functionally related to the outcome of SD. Gaupp (1972) conducted a study which was virtually a replication of Beere's (1971) but found that pretest QMI scores did predict outcome of $0. In an analog study with $5 with an intense fear of crawling insects he found that 55 with high image vividness scores showed more behavioral, physiologica1,'and cogni- tive change than $5 with low image vividness scores following SD. Vividimagers found it more difficult to visualize phobic imagery than neutral images but consistently experienced less difficulty visualizing and controlling stimuli than weak imagers. It is not at all clear why Gaupp's (1972) findings were positive while Beere's (1971) were negative. The procedures used were not ident- ical and the Ss differed in type of fear so a direct comparison between the two studies is not possible. 22 In a comparison of the relative effectiveness of SD and $0 plus focusing instructions on math anxiety in high-school students Hyman (1973) administered an image vividness rating scale following each treatment session. She reports a near-significant correlation between reported image vividness and outcome of SD. She also found that S0 S5 reported visualization of extraneous images and reports a non-significant trend for the occurrence of extraneous images to interfere with treatment. In a direct test of the hypothesis that image vividness is directly related to the outcome of 50 but is not related to the outcome of "in vivo“ desensitization, McSweeney (1975) used five pretest measures of image vividness and controlability with volunteer $5 with public speaking anxiety. No significant rela- tionship was found between any one imagery measure and outcome, but a significant cannonical correlation was found between all five imagery measures and outcome measures. This suggests that while no one measure was a good predictor of outcome of SD a linear combination of several measures is a good predictor. This may possibly indicate that neither image vividness nor image controlability alone is the crucial variable in SD but that the two, in combination, are crucial. Imagery also serves as a stimulus in the covert condition- ing techniques--covert reinforcement (COR) (Cautela, 1970b), covert sensitization (Cautela, 1967), covert extinction (Cautela, 1971), covert negative reinforcement (Cautela, 1970a), covert modeling (Cautela, 1976) and covert response cost (Cautela, 1976). 23 These techniques are based theoretically on principles of operant conditioning which have been demonstrated in the laboratory and rest on the assumption that appropriate imagery is an adequate substitute for overt stimuli, behavior, and consequences (Cautela, 1973). The assumption that imaging of appropriate stimuli is necessary and sufficient for behavior change has been tested in research with COR. The procedure of COR involves selection of the response to be increased, selection of appropriate reinforcers, and imaginal presentation of the response to be increased followed by imaginal presentation of a reinforcer (Cautela, 1970b). It is assumed that the parameters which are important in positive reinforcement such as the temporal relationship between response and reinforce- ment, magnitude of reinforcer and schedule of reinforcement play the same role in COR as in operant reinforcement. Reinforcement is made immediately contingent upon imaging of the response. several of the most powerful available reinforcers are used to avoid satiation, and a continuous reinforcement schedule is used. Research on the role of imagery in COR can be divided into three general types: analog studies in which imagery is used to reinfbrce overt behavior, analog studies in which imagery is used to reinforce covert behavior, and clinical analog studies in which imagery is used to reinforce covert behavior, and clinical analog studies in which imagery is used to reinforce approach towards phobic stimuli or relaxation in phobic situations. Since all of these studies are analog studies using either analog tasks 24 or sub-clinical behavior problems~the generalizability of the results can be questioned. Cautela, Steffan, and Wish (cited in Scott and Rosensteil, 1976) used COR to reinforce overestimates or underestimates in a circle size estimation task. COR resulted in the predicted changes in size estimates while there was no change without reinforcing images or with non-contingent covert reinforcement. However, the word "reinforcement" alone also resulted in significant behavior change. Tondo and Cautela (1974) used COR with the same circle estimation task and found that pretest scores on the Imagery Survey Schedule predicted both ratings of imagery during COR and differential outcome of COR. Steffan (1971) used COR to investigate the Greenspoon effect in hospitalized patients diagnosed as schizophrenics. In this well controlled study he found that COR of plural nouns resulted in a significant increase in the number of plural nouns. Neither non-contingent covert reinforcement or contingent presen- tation of "scene," the cue for reinforcement, without reinforce- ment resulted in an increase in significant change. In a similar study Ascher (1973) found that COR produced a significant increase in the probability of use of a reinforced pronoun which varied directly with the number of times each pronoun was reinforced. Asher (1973) also found that, during an extinction period, the probability of reinforced pronouns decreased but did not return to baseline. 25 Krop, Messinger, and Reiner (cited in Scott and Rosensteil, 1976) used COR to increase eye contact during an anxiety arousing interview by having 55 image a pleasant scene following five or more seconds of eye contact. The COR group differed significantly from non-contingent reinforcement and no reinforcement control groups immediately post—treatment but did not differ from the other two groups on a one-week follow up. Baron (1975) investigated the importance of quality and duration of imagery in COR using a two-choice key pressing task where signals for reinforcement were presented on a VI-lO schedule for each key. The results indicated that higher quality images produced significantly higher response rates and that longer image durations produced significantly higher response rates. Overt reinforcement and overt plus covert reinforcement were equally effective and both were significantly more effective than the signals for reinforcement alone. In the studies which have investigated the effect of COR on covert responses the experimenters have attempted to modify attitudes towards the mentally retarded and to modify self-concept in institutionalized Ss. Cautela, Walsh, and Wish (1971) had 55 imagine a "mentally retarded" person and then imagine a pleasant scene. They found a significant positive change in experimental Ss' attitudes towards the mentally retarded in comparison with S5 who imagined a mentally retarded person without any pleasant imagery. It is difficult to explain these results in terms of operant conditioning (Scott and Rosensteil, 1976) because Ss were 26 reinforced simply for imaging a retarded person, not for expression of positive attitudes. An alternative explanation is that classi- cal conditioning occurred due to the pairing of the image of a retarded person with a pleasant scene and resulted in the attitude change. It is also quite possible that the objective of the exper- imental manipulation was apparent to Ss and that this biased the results. Krop, Calhoon, and Verrier (1971) and Krop, Perez, and Beaudoin (1973) have used COR to modify the "self-concepts“ of institutionalized Ss. Krop et a1. (1971) used children with behavior disorders as $5 and compared COR with overt reinforce- ment and no reinforcement while Krop et a1. (1973) used male psychiatric patients as $5 and compared COR with overt reinforce- ment and non-contingent pleasant imagery. In both studies the COR and overt reinforcement Ss were reinforced following "positive self-concept" responses to items from a self-concept scale. Though positive results were obtained in both studies it is appar- ent that overt responses to self-concept scale items were rein- forced and that any link between these behaviors and the covert responses which constitute self-concept is purely hypothetical. The majority of COR clinical analog studies have simply been outcome studies or have tested theoretical conceptualiza- tions of COR and do not bear directly on the role of imagery in COR. A number of studies have been reported which question the effects of COR and challenge the simple operant conditioning paradigm of COR (Bajtelsmit and Gershman, 1976). However, 27 those studies which have investigated the role of imagery in COR have supported the assumption that imagery serves a functional role in COR. The other therapy techniques which rely on the stimulus properties of imagery have been supported by clinical case studies and some outcome research (Cautela, 1976, 1970a, 1971, 1976). However, the role of imagery in these techniques has not been investigated directly. The assumption that imagery can substitute for stimuli in psychotherapy techniques receives mixed support in research on SD and COR and has not been tested with other therapy techniques. 0f the eight studies which have investigated the role of imagery, four have found a significant relationship between image quality and treatment outcome (Gaupp, 1972; Tondo & Cautela, 1974; McSweeney, 1975; Baron, 1975) and four studies have failed to find such a relationship (Davis, McLemore, & London, 1970; McLemore, 1971; Beere, 1971; Hyman, 1973). Two of the studies with negative findings (Davis, McLemore, & London, 1970; McLemore, 1971) suffer from methodological flaws which may invalidate their results. However, the failure of Beere (1971) and Hyman (1973) to find a significant relationship suggests that the relationship between image quality and behavior change is not robust. 28 The Determinants of the Stimulus-Value of Mentallmagery The variables which have been most commonly seen as determ- inants of the stimulus-value of imagery are the stimulus-value of the stimulus being imaged and the vividness and controlability of the image experienced. The stimulus-value of the stimulus being imaged is an obvious determinant of the stimulus-value of the image. A number of studies reported earlier (Matthews, 1971; Wolpe & Flood, 1970; Van Egeren, 1970; Van Egeren, Feather, & Hein, 1971; Chapman & Feather, 1971; Marks & Huson, 1973; Haney & Euse, 1976; May, 1977; Craig, 1968) have found that 55' subjective and physiological responses to images are qualitatively similar to 55' responses to the actual stimuli. However, in all of the studies investigat- ing the role of imagery as a stimulus in psychotherapy, the stimuli imaged have been standardized and S differences in response to the stimuli being imaged have not been investigated. Of the variables which may influence the stimulus-value of imagery, the most thoroughly investigated has been individual differences (105) in the ability to experience imagery. IDs occur both in the type of imagery experienced and in the characteristics of the images experienced. The types of imagery used in psychotherapy are commonly classified as memory images and imagination images and, of the many dimensions along which images may vary, the dimensions which have been most 29 thoroughly researched are sensory modality, vividness, and con- trolability. During the long search for a suitable typology for classi- fying persons acCording to their experiences of imagery, IDs in the sensory modality in which imagery was experienced were long considered important. 105 in the predominant modality in which imagery is experienced have been reported consistently and a, typology of habitual visualizers versus habitual verbalizers has been proposed and supported by some research (Richardson, 1969). However, Sheehan's finding (1967) that ratings of image vividness in various sensory modalities were highly correlated within sub- jects has cast doubt on the value of this categorization. Lindauer's conclusion that repOrts of predominant modality are based on preference rather than ability and that such a typology is irrelevant to tasks such as SD where a specific image is requested (Sheehan, 1972) is widely accepted. The most heavily researched area of 105 in imagery has been 105 in image vividness, the degree to which the intensity of an image matches the intensity of the stimulus imaged. 105 in reports of image vividness have been related to performance on a number of cognitive and perceptual tasks, sometimes with mixed results. For example, Fusella (1973) found that "inner-acceptant" 55 who reported high image vividness and scored high on two cognitive variables were significantly more susceptible to the Perky effect than Ss who scored low on all three variables but 30 that ratings of the vividness of individual images did not corre- late significantly with accuracy of signal detection. This contra- dicts the findings of a number of studies of the use of imagery to increase accuracy of reconstruction of geometric patterns (Sheehan, 1966, 1972; Sheehan & Neisser, 1969; Neisser, 1972; Marks, 1972). These studies found significant correlations between ratings of image vividness and accuracy within 55 but not between Ss. Neisser (1972) has suggested that, in many tasks, the image may serve as a discriminative stimulus and that, in these cases, the vividness of the imagery is irrelevant as long as the image is perceptible. Marks (1973), on the other hand, suggests that Neisser's conclu- sions are based on studies which produce atypical results because the task used was low in meaningfulness, interest, and affect and because a score for image vividness across modalities was used with a purely visual task. In his research using a picture recall task somewhat higher in meaningfulness, interest, and affect than the geometric figures used previously, he found that persons scoring high in vivid visual imagery performed consistently better than $5 scoring low on vividness. In an investigation of physiological responses to phobic imagery, Lang, Melamed, and Hart (1970) found significant corre- lations between reported image vividness and physiological responses to phobic imagery. These correlations, ranging from .52 to .88, not only were significant but also accounted for a sizeable portion, 1/4 to 1/2, of the variance of the physiological 31 responses. This suggests that vivid images function more effec- tively as phobic stimuli than dim images. The investigation of 105 in image controlability, the degree to which a stimulus can be imaged at will, maintained in consciousness, and purposely modified, was begun by Gordon (1949). Her research on the relationship between image controlability and ethnic stereotypes and her investigation of the relationship between imagery control and perceptual control (Gordon, 1950) demonstrated the importance of I05 in image controlability. A series of studies has shown that 55 who report voluntary control over mental imagery report greater control over the rate of reversal of a Necker cube and some studies have found that con- trolled imagers performed significantly better on the Stroop-Color Word Test (Richardson, 1972). * Many researchers investigating mental practice have sug- gested that 105 in image vividness and controlability play a part in explaining 105 in improvement under mental practice conditions. In a study designed to investigate this hypothesis Richardson and Start (1964) had male 55 practice a gymnastic movement mentally using visual and kinesthetic imagery and measured actual perform- ance after a week of mental practice. It was found that $5 with vivid, controlled imagery performed significantly better than $5 with vivid, uncontrolled imagery and that the relationship obtained could not be explained in terms of gymnastic ability. Unfortu- nately, the possibility of differential motivation between groups was not ruled out completely so that while this research supports 32 the relevance of 105 in imagery control to uses of imagery, it is not conclusive (Richardson, 1972). Though image vividness and image controlability are theo- retically distinct dimensions of ID's in imagery experience they are not empirically distinct (Lane, 1974). Reports of image vivid- ness and image controlability tend to correlate significantly in normal 55 (Sheehan, 1972) and pretest measures of ability to experience vivid imagery and ability to experience controlled imagery correlate significantly both because of this relationship and because of test characteristics (Lane, 1974). For this reason pretest measures of image vividness and image controlability were considered as joint predictors of image quality in this study and S ratings of the vividness and controlability were combined into a composite report of image quality. In seven of the eight studies investigating the stimulus- value of imagery in psychotherapy, questionnaires which assess the 55' ability to experience vivid or controlled imagery have been used as a predictor of image quality (Davis, McLemore, & London, 1970; McLemore, 1970; Beere, 1971; Gaupp, 1972; Hyman, 1973; Tondo & Cautela, 1974; McSweeny, 1975). Both Beere (1971) and Hyman (1973) found that these "trait" measures were good predictors of the quality of neutral or pleasant imagery but were not good predictors of the quality of aversive imagery. Several authors (Beere, 1971; Wilkins, 1971; Reyher, 1976) have suggested that variables other than ability to exper- ience vivid, controlled imagery also determine the image quality. 33 The following variables may be determinants of the image quality and thus be determinants of the stimulus-value of imagery: tendency to avoid aversive stimulation, level of relaxation or anxiety, sex of S, complexity of the stimulus being imaged, amount of imagery practice, trait anxiety, neuroticism, introversion, and openness to imagery and related phenomena. Of these variables the two which have been suggested most often as important determinants of image quality are tendency to avoid aversive stimulation and degree of relaxation or anxiety. Beere (1971) has suggested that 55 image aversive stimuli less vividly to avoid aversive stimulation, and Reyher (1976) has suggested that 55' defense mechanisms may block imagery, decrease image vividness, transform imagery, or disassociate imagery from affect. Both views suggest that S differences in response to aversive stimulation may influence image quality. On a clinical level it is obvious that individuals respond differently to aversive situations. Some persons avoid them strenuously, some tolerate them but experience great subjective discomfort, and some persons seek out aversive situations in "counterphobic" behavior. On an empirical level Rona et al. (1976) and LaZarus and Alfert (1969) have found that pretest measures of 55' tendenCy to respond to aversive stimulation by cognitively approaching or avoiding the stimulation predicted subjective and physiological responses to a stressor. Other research has shown that this measure, Byrne's Repression-Sensitization Scale (Byrne, 1961), 34 predicts differences in magnitude of subjective estimates of the magnitude of aversive stimulation (Lazarus & Alfert, 1969), and 55 position on the repressor-sensitizer dimension predicts dif- ferences in recall of stimuli associated with aversive stimulation (Lazarus & Longo, 1953). The hypothesis that $5 respond to aversive imagery in the same way as they respond to aversive stimuli leads to the conclusion that S differences in response to aversive imagery should correspond to S differences in response to aversive stimuli. This conclusion is supported by a study of physiological responses to positive, negative, and neural imagery by Haney and Euse (1976). The authors interpret their finding that 55 reported positive and neutral imagery to be clearer than negative imagery as supporting the hypothesis that anxiety is a mediator of image clarity (Haney & Euse, 1976; Euse & Haney, 1975). If, as Wolpe (1970) hypothesizes, relaxation inhibits subjective and autonomic responses to aversive images it is pos- sible that relaxation moderates the hypothesized tendency to avoid aversive images. While this possibility has not been investigated empirically, the common finding that relaxed 55 experience stronger responses to aversive stimuli than non-relaxed Ss (Wolpe & Flood, 1970; Van Egeren, 1970; Van Egeren, Feather, & Hein, 1971; Chapman & Feather, 1971) is compatible with the hypothesis that relaxation suppresses avoidance of imagery. Bandura (1977) suggests a model of the relationship between level of arousal and response to arousing stimuli which combines the hypothesis that relaxation inhibits arousal with the hypothesis 35 that relaxation inhibits avoidance of the arousing stimulus. He predicts a curvilinear relationship where a high level of arousal leads to behavioral and/or cognitive avoidance of an arousing stimulus and thus to little additional arousal, where a moderate level of arousal leads to little avoidance and a large increase in arousal, and where a low level of arousal also leads to little avoidance of the stimulus but leads to a smaller increase in arousal. This suggests that it may be possible to maximize or minimize response to arousing stimuli by manipulating the 55 level of arousal. It has been hypothesized that relaxation influences the vividness and controlability of imagery directly. Singer (1973) reports that muscle relaxation produces conditions similar to the conditions of reduced sensory stimulation which have been found to increase image vividness. The previously reported finding (Wolpe & Floor, 1970; Van Egeren, 1970; Van Egeren, Feather, & Hein, 1971; Chapman & Feather, 1971) that relaxed Ss experience stronger physiological responses to aversive imagery could also be explained by the hypothesis that relaxation resulted in more vivid, controlled images. While some form of relaxation is conmonly used with many imagery techniques, the effects of S differences in depth of relaxation have not been investigated directly. Three variables clearly should influence the level of relaxation or anxiety experienced by 55 during therapeutic pro- cedures using mental imagery: Ss' ability to relax, relaxation instructions, and the stimulus-value of the stimulus being imaged. 36 Relaxation is commonly conceptualized as a skill which can be learned through guided practice (Bernstein & Borkovec, 1973). Wide individual differences in the depth of relaxation both before and after training have been noted (Bernstein & Borkovec, 1973) but have not been fully investigated. These ID's will directly influence the level of relaxation or anxiety experienced by each S during imagery techniques. Since relaxation is conceptualized as a behavior under Ss' voluntary control, the instructions 55 receive concerning relaxa- tion should directly influence the level of relaxation or anxiety. The effects of relaxation instructions within 55 has not been clearly investigated but the assumption that $5 relax, insofar as possible, when instructed to relax is commonly made. The assumption is commonly made that relaxation and anxiety are mutually inhibitory. It has been clearly shown that relaxation decreases anxiety and autonomic arousal (Bernstein & Borkovec, 1973) and clinical experience suggests that intense anxiety or autonomic arousal inhibit relaxation. Since phobic imagery has been found to result in physiological and psychologi- cal arousal it is possible that this arousal inhibits Ss' relaxa- tion. Summar Research into the stimulus properties of mental imagery has supported the hypothesis that mental images can function as stimuli in a variety of tasks and settings. Physiological and 37 subjective responses to mental images are qualitatively similar_ to responses to the stimuli imaged, responses to mental images reliably discriminate between images of phobic stimuli and images of neutral stimuli, and images of complex patterns of stimuli can be used to facilitate reconstruction of the stimulus pattern. However, studies directly investigating the role of imagery in psychotherapy techniques based on the stimulus properties of mental imagery have not produced clear support for the hypothesis that mental images are a functional component of these therapy techniques. Theoretically the vividness and controlability of imagery should be directly related to the stimulus-value of the images and thus to treatment outcome yet tests of this hypothesized relationship have produced mixed results. It has been suggested that the lack of clear support for the importance of imagery is a result of the stimulus-value of mental imagery being multiply determined, that the pretest measures of image vividness and controlability which have commonly been used are not good predictors of the stimulus-value of mental imagery. Investigation of determinants of the stimulus-value of mental imagery is needed to clarify the role of mental imagery in these psychotherapy techniques. HYPOTHESES This study investigates the hypothesis that the stimulus- value of mental imagery is multiply determined. More specifically it is hypothesized that: 1. The stimulus-value of a mental image is determined by the stimulus-value of the stimulus imaged, the quality of the experienced image, and the level of relaxation or anxiety. The quality of the experienced image is determined by imagery ability, level of relaxation or anxiety, tendency to avoid aversive stimuli, and the stimulus- value of the experienced image. The level of relaxation or anxiety while imaging a stimulus is determined by the initial level of relaxation, relaxation ability, and the stimulus- value of the experienced image. In the absence of stimulation the experienced level of relaxation is determined by the initial level of relaxation, relaxation ability, and relaxation instruction. In this study Relaxation Ability, Imagery Ability, the Stimulus-Value of the Stimulus Imaged, and Tendency to Avoid Aversive Stimuli were measured before imagery trials. The level of relaxation was measured three times before imaging a stimulus, twice while imaging a stimulus, and twice after imaging a stimulus. Image Quality and the Stimulus Value of the Experienced Image were measured while imaging the stimulus and Relaxation Instructions before and after imaging the stimulus were varied from trial to 38 39 trial. The hypothesized pattern of relationships between these variables in shown in Figure l. The operationalization and measure- ment of these variables is discussed in the subsequentheasures section. These hypotheses predict a specific pattern of causal relationships which implies a pattern of correlations between var- iables. The hypothesized pattern of causal relationships is shown in Figure 2 and the implied pattern of correlations are shown in Figure 3. 4o .quoz Femsmu uonmmcuoqxz .F weaned mecp to new on , cowuowawm co Pm>m4 1 mxmu xuwpmao momEH 1 czH we mcowpozcumcH cowumeme 1 NHm NHa mcwzoFFoc mace? co pomco H* mcowuuscumcfi cowummemm 1 HHm cowumchmm co Fm>w4 1 exmu acwzoppoc cowuexupmm co Fm>w4 1 Nxmo ummmEH mzpzewum once? we mapzswum mcwmmsw mo mzpo>1marzewum 1 Hm>m coruecpELmu mcouma mcocmg cowuaxerm co Fw>m4 1 Fxmu __:ewum w>wmcm>< :owumxe_mm mo Fm>m4 1 mxm; HHm mCAzoFFoc uwo>< op zucwucwh 1 m<
m4 1 oxmu pr—_n< acmmoea 1 (gm
co m=~n>1mapaemum 1 Hm>m cowumxmpwm we Fm>m4 Fewuecfi 1 me4 zuw_vn< coeuexerm 1