Affective information processing and the assessment of anxiety

Journal of Behavioral Assessment, Vol. 6, No. 4, 1984 Affective Information Processing and the Assessment of Anxiety Peter J. Lang ~ and Bruce N. Cuthbert ~ Accepted December4, 1984 A conceptualization of anxiety as comprising three loosely coupled response systems of overt behavior, verbal report, and physiological activation has proven useful in clinical and theoretical work. With this framework as a starting point, an information-processing approach to the study of emotion is described. Emotions are conceived as affective programs within the brain, with information coded as propositions organized into associative networks. Affective expression results when such a network is accessed and processed, which can occur when a sufficient number of propositions are activated by environmental stimuli and~or internal associations. It is hypothesized that information about the expressive physiology is an integral component of the associative structure, and that processing of the network accordingly results in measurable psychophysiological response. Data from studies of emotional imagery, as well as other areas of research, are reviewed in support of these theories. The utility of this approach for the assessment of anxiety disorders is discussed, and results of clinical studies are presented to suggest that individual differences in accessing and processing emotional information may bear significant implications for prognosis and treatment selection. It is speculated that differences among the anxiety disorders could be interpreted in terms of the degree of cognitive organization of the network, leading to potential refinement of current diagnostic categories. In conclusion, cognitive psychology paradigms are discussed in terms of their application to the assessment and treatment of anxiety disorders. KEY WORDS: anxiety; information processing; assessment. The authors' research is supported in part by N I M H Grant M H 37757. Portions of this paper were presented at the N I M H Conference on Anxiety, Tuxedo, NY, September 1983. Several of its themes were developed there at more length, and this material may be found in the publication of the conference proceedings (Tuma & Maser, 1985). ~Department of Clinical Psychology, University of Florida, Gainesville, Florida 32610. 369 0164-0305/84/12000369503.50/0 @ 1984 Plenum Publishing Corporation 370 Lang and Cuthbert INTRODUCTION Over the past decade our research program has applied concepts and methods from the information-processing literature to the psychophysiological study of emotion. This work has had a major focus on the normal and pathological phenomena of fear. Our approach has been to conceive emotions as affective programs within the brain, to attempt the specification of their data base and logical organization, and to describe how and when such programs might be run. This paper presents some of the theories and results of this project, describing their ramifications for behavioral assessment. It is argued that psychophysiological responses are integral to the expression of clinical anxiety and that a consideration of their role in affective information processing can facilitate an understanding of the development and modification of anxiety disorders. In support of this view, research in the areas of state-dependent learning, excitation transfer, and emotional imagery is reviewed and evaluated. Some speculations are made regarding ways in which this approach can be used to interpret differences among states of fear and anxiety (i.e., according to their cognitive organization), with ensuing potential nosological refinement. In conclusion, specific suggestions are offered for the application of paradigms from cognitive psychology to research on anxiety and anxiety-related disorders. THE R A W D A T A OF A N X I E T Y In view of the wide range of contemporary approaches to the analysis and assessment of anxiety, it is perhaps best to begin with a perspective on anxiety and fear as clinical entities. That is, what constitutes the raw data base which must be accounted for by a theoretical system? For most patients the basic datum of anxiety is their feeling state, i.e., a direct experience or internal apprehension, requiring no further definition. In some cases even the most careful clinical inquiry may elicit from the patient no more elaborate a description of his or her disorder. Nevertheless, feeling states are completely private and represent a Poor data resource for the clinician preparing to undertake treatment. Their unavailability to the community of observers appears to deny any possibility of scientific investigation. However, most patients show peculiarities of behavior and will themselves describe a variety of symptoms associated with their feeling of fear or dread which are more yielding to objective analysis. Several investigators have undertaken the factor analysis of these symptom reports (e.g., Hamilton, 1959; Buss, 1962). Buss concluded that observed and reported symptoms of anxiety fell into two primary domains of complaints [originally proposed by Eysenck (1961)]: (1) autonomic overreactivity-physiological events such as sweating, flushing, and shallow breathing and subjective reports of heart palpitations, Information Processing and Anxiety Assessment 371 intestinal discomfort, and aches and pains; (2) conditioned anxietyrestlessness, worry, and muscular tension, in a subsequent study of manifest anxiety test items, Fenz and Epstein (1965) and Fenz (1967) reported three factors in anxiety, of which autonomic overreactivity again accounted for the majority of the variance. The Eysenck-Buss conditioned anxiety factor was divided into two subfactors: symptoms of striate muscle tension (including aches and pains, twitching, and shaking) and an additional factor called feelings of anxiety, which contained inability to concentrate, worry, compulsive mannerisms, insecurity, sleep and relaxation disturbances, and feelings of fear or panic. Such studies have led to the proposal (Lang, 1964, 1968, 1978; Rachman, 1974) that anxiety is a construct comprised of measurable responses which fall into three general categories. (1) Verbal reports of distress, i.e., reports of anxiety, fear, dread, panic, and associated complaints of worry, obsessions, inability to concentrate, insecurity, and the like: At this primary data level, no assumptions are made about the reality of underlying "states of feeling." However, the absence of a "feeling" premise should not be taken to devalue verbal reports as a data source. As for other complaints, the patient is not treated here as an observer (in this case, of a phenomenon which has no objective, validating referent); rather the reports themselves are considered to be part of the primary response of anxiety. (2) Fear-related behavioral acts: Included here are the observable acts of anxiety and f e a r avoidance, escape, hypervigilance, dysfunctional immobility, compulsive mannerisms, and deficits in attention, performance, and control. (3) Patterns o f visceral and somatic activation: For the most part, anxiety and fear are associated with sympathetic arousal, and the physiological data of anxiety are events such as heart-rate and blood-pressure increases, sweating, and generalized muscle tension. However, individual patterns of response are highly varied, and in some phobic reactions, e.g., fear of blood and mutilation, autonomic tonus may be predominantly vagal rather than sympathetic (Sattory, 1981, p. 217). The three data subsystems in anxiety are only loosely coupled and change desynchronously (Rachman & Hodgson, 1974), presenting a variable configuration within and between subjects and fear contexts. Thus, it is important at the present state of our knowledge that all three systems are sampled. The assumption that a single measure of fear is sufficient for diagnosis and analysis is currently not tenable. Accordingly, in addition to the sampling of behavioral and verbal outputs, the measurement of physiological responses deserves equal prominence in assessment. Taking this framework as a starting point, the goal of the proposed information-processing approach is eventually to explicate the entire spectrum of anxiety and stress disorders. As a model system for the development of theory, recent research in our laboratory has focused on the study of fear, particularly its pathological manifestation in phobia. The specifici- Lang and Cuthbert 372 ty of context and response in phobic disorder is great, and cognition and behavior remain relatively stable in the absence of any real nociceptive events. Thus, considering the transient character of most emotional states (the fact that affects must be, so to speak, caught on the fly, if indeed they are to be examined at all), phobia is a valuable, relatively stable preparation for the study of pathological emotion. The aim of this work has been to define the data structure in long-term memory which forms the information base for phobic behavior and to understand the conditions under which this information is or is not accessed and processed and comes to prompt efferent productions. COGNITION IN EMOTION We propose (Lang, 1977, 1979, 1983, 1985) that emotion information is coded in memory in the form of propositions and that these propositions are organized into associative networks of the general sort first described by Quillian (1966) for semantic knowledge and later adapted to accommodate other types of information (e.g., Anderson & Bower, 1974; Kieras, 1978). Our working theory presumes the emotion memory structure to contain three categories of information: (1) information about prompting external stimuli and the context in which they occur; (2) information about responding in this context, including expressive facial or verbal behavior, overt acts of approach or avoidance, and the visceral and somatic events that support attention and action; and (3) information which elaborates or defines the meaning of the stimulus and response data. The information network of an emotion is a sort of prototype or schema, which, when a critical number of propositions is accessed (through a match to environmental stimuli, internal association, or both), is processed as a unit. Phobic expression is produced as a result of the processing of such a network. As a working hypothesis we currently presume that the probability of a phobic production is determined by the number of matching propositions, irrespective of their taxonomic category (stimulus, response, meaning), which are present in short-term memory. Obviously, network activation is most likely when the phobic individual is confronted by the actual phobic object, which presumes a near-perfect stimulus match. However, phobic emotions may also be elicited by degraded i n p u t - p i c t u r e s of the phobic object, verbal descriptions of thecontext, and the like. It is important to remember that the information stored in memory is presumed to be conceptual and not iconic, and that conscious intention is not a necessary condition for its activation. A variety of media can be the vehicles of information input to the network and the production system. Furthermore, degraded stimulus matches may be as effective in prototype activation as exposure, Information Processing and Anxiety Assessment 373 if other propositions (response or meaning) are simultaneously instigated. For example, reduced stimulus cues will prompt a full phobic reaction in a subject who is otherwise aroused [e.g., as in the Schachter and Singer (1962) experiment, where the prior administration of epinepherine to subjects activated affect-concordant response propositions]. The probability that minimal stimulus cues will occasion an affective display also increases when contextrelevant meaning propositions are also activated (e.g., a film about snakes is more likely to prompt a fear response when viewed in a jungle clearing than when seen in a Manhattan apartment). While stimulus and meaning information are important in the phobic network, the representation of information about responding is crucial in the storage and expression of anxiety and, indeed, of emotions in general. Our view is consistent- with Sperry's (1952) and Weimer's (1977) phylogenetic analysis, holding that the survival path in brain evolution has been to organize action and response (and n o t to provide, for example, a gratuitous ability for self-contemplation). Comparisons of brain and computer are misleading if they encourage us to view internal information transfers as ends in themselves, creating a new but equally restricting subjectivism. The key importance of the human brain's increased processing capability is in the variety, flexibility, and power of its output functions. Cognitive processing in emotion is fundamentally the execution of programs for efferent expression. The task is to investigate the ways in which these programs are developed, accessed from storage, and modified. AFFECTIVE MEMORY Evidence supporting the associative network theory of emotional processing has appeared from a variety of paradigms and investigators. Some of the most relevant work on the organization of emotions in memory has been accomplished by Gordon Bower and his associates (e.g., Bower, 1981; Bower & Cohen, 1982). An experimental paradigm used by these investigators involved the induction of an emotional state through hypnotic instructions to relive a previous affective experience (either happy or sad) and the subsequent learning and recall of lists of unrelated words. Thus, in a typical study subjects learned two lists of words - first, List A, then List B - and were subsequently tested for recall of List A. Different subject groups experienced different pre-word-task emotions according to prearranged patterns (e.g., happy, A; sad, B; happy, recall A; o r happy, A; sad, B; sad, recall A). It was determined that subjects who were in congruent moods during learning (List A) and recall (List A) remembered the words better than if acquisition and remembrance occurred in different affective states. The poorest memory scores were obtained when the mood at recall of A was congruent with the 374 Lang and Cuthbert emotional experience during List B learning and different from that experienced at List A acquisition. In this "interference c o n d i t i o n . . , recall of the target List A suffers because the recall mood evokes memories of the wrong List B rather than the target List A" (Bower, 1981, p. 131). Bower interprets these results in terms of a semantic network, whose nodes include representations of emotional states. Due to the associative value of these emotion nodes, activation of a congruent emotion at recall is seen as facilitating access to the rest of the network, which was linked to the emotion at the time of original learning. While there is great merit in the network approach, it is not clear that there are indeed emotion nodes that mediate the observed state-dependent memory effects and which must be activated before emotional behavior occurs. His formulation implies that some number of emotional states are discrete, separate entities and, furthermore, that their representations in working memory map on traditional affective labels (e.g., love, anger, pride, etc.). Although efforts to identify such states have occupied philosophers and scientists for hundreds of years (e.g., see Panksepp, 1982), there is still no specific agreement as to their number or taxonomy. As an alternative to the concept of discrete affective states, some investigators view emotion as varying along certain basic parameters or dimensions of behavior. Coincidences of behaviors at specific strengths on each dimension are what give the impression of substantive emotional states. Support for this view comes from work with the semantic differential, employing bipolar scales with opposing qualities at either end (e.g., good-bad, hot-cold) rather than searching for stable individual factors. Russell and Mehrabian (19"77) utilized such a scheme to study subjects' ratings of a variety of situations. They reported, as had earlier investigators (Osgood, Suci, & Tannenbaum, 1957), that with this method two basic factors accounted for most of the variance: one was associated with valence (pleasure-displeasure) and the other with activation (arousal-quiescence). A third factor, variously called potency (Osgood et al., 1957) or dominance-submission (Russell & Mehrabian, 1977), was found to account for a smaller, but nevertheless palpable proportion of the variance. These parameters are appealing for their simplicity and relevance to all possible emotional situations. In addition, there is evidence, beyond the scope of the present discussion, that these same three dimensions may reflect fundamental components of behavioral and physiological response systems as well (see Lang, 1983), i.e., that the phylogenetic history of affect is to be understood in terms of approach and avoidance (valence), behavioral intensity (arousal), and phenomena of control such as response sequence disruption or relative social dominance. It is not, however, necessary to make a prejudgment about the organization of affect. We propose instead that Information Processing and Anxiety Assessment 3"/5 researchers pursue three tiers of analysis in the study of emotion: (1) general response parameters of valence, arousal, and dominance; (2) organized action programs, consistent with the concept of an emotional state; and (3) specific response subroutines, as for facial expression, vigilance posture, ~adrenergic visceral physiology, etc. It is as yet unclear whether all levels are represented and active in the deep structure of associative memory. That is, we do not know which type of concept is the prime mediator in emotional state-dependent learning and memory. More systematic manipulation and controlled study will be necessary to dissect their interrelationships and relative importance in this regard. With respect to Bower's data, it may be that it is the central representation of specific action programs (verbal expressions, overt acts, or physiological patterns) which determines the observed effects. The concept of an emotional feeling state may be represented only semantically. Thus, rather than depending on the activation of an emotion node, it is then the response concept or concepts themselves which are the individual a n d / o r collective mediators of the state-dependent effects. These, in turn, may bear some functional relationship to the superordinate parameters. If so, then mood congruence effects would be more likely to be expressed across a dimension of responding (arousal, valence, and perhaps control) than confined to popular notions of specific emotional states. MEDIATION AND RESPONSE INFORMATION With respect to this hypothesis, recent research by Laird, Wagener, Halal, and Szegda (1982) suggests that a program for facial expression could be part of such a mediating code in memory that might account for some state-dependent memory effects. In these experiments subjects first read either an "anger" story (about the killing of dolphins) or a happy story (humor by Woody Allen). They were later instructed to assume various positions of the facial muscles, corresponding to their configuration in emotional expressions, i.e., either a frowning or a smiling visage, and recall trials were conducted. On the basis of an independent test, subjects were divided into groups depending on whether, in an ambiguous situation, their self-reports of affect were more consistent with external emotional cues (labels on abstract pictures) or their own facial expression at the time of viewing. The external-cue group failed to show any interaction between the manipulated facial expression at recall and the story content. However, the group that had been shown previously to be more congruent in facial expression and affective report showed significantly better recall of the hu- 376 Lang and Cuthbert morous story while smiling, and better recall of the "anger" story while frowning. This study is provocative and certainly encourages the thesis that facial expression is one of the mediators of mood-congruent memory, at least within a subsample of expression-sensitive subjects. However, other factors have also been implicated. For example, Riskind (1983) provides evidence that overall posture, "slumped" or upright, may facilitate recall of sad or happy material, respectively. It must also be noted that, in the Laird study, no effort was made to control for sympathetic arousal, which could well have been different for both contents and expression and could have accounted for the specificity noted within negative affective states (e.g., anger and sadness). Indeed, as shown below, other experiments suggest that patterns of arousal may in themselves mediate otherwise unrelated memories. Thus, mood-congruence effects could be facilitated either by valence responses such as facial expression, by specific visceral arousal patterns, or perhaps by an interaction of these parameters. Clark and associates (Clark, Milberg, & Ross, 1983) have recently undertaken a series of experiments designed to determine whether physiological arousal is the critical mediator of mood-congruent memory effects. They propose "that information about changes in autonomic arousal that accompany moods may be part of how a mood is stored in memory and that changes in arousal that reoccur with subsequent moods may be part of what primes affectively-toned material stored earlier." In one test of the hypothesis, subjects first learned a list of unrelated phrases while experiencing enhanced arousal and then learned a second list while relaxing in a normal state. Later they were given a recall test for both lists under ostensible conditions of visual distraction. The arousal condition which preceded the learning of one list involved stepping up and down repeatedly on a cinder block (which produced palpable heart-rate and blood-pressure increases). The relaxation task which preceded the other study list simply required subjects to rest in a lounge chair (with no cardiovascular pressor effect). Orders of lists and arousal tasks were counterbalanced. All subjects were asked to recall both lists. According to random subgroup assignment, recall was preceded by one of two conditions: arousal-subjects viewed a sexually explicit film; and relaxation-subjects viewed a film about a chimpanzee learning sign language. The results were as predicted: when subjects were aroused by the sex film at recall, they best remembered the list which was learned after the cinder-block step test. Furthermore, when subjects recalled the lists after viewing the sign-language film, memory was better for the phrases originally learned after relaxation than it was for the list previously learned after exercise arousal. This effect is interesting, particularly because of the marked difference in the arousal task from learning to recall Information Processing and Anxiety Assessment 377 phases (e.g., step test and sexual stimulation). This change in induction procedure would appear to greatly reduce the possibility that commonalities in stimulus information could account for these data. The research cited to this point suggests that affect information (about arousal o r about valence) may mediate otherwiseunrelated memories over fairly long time spans. A related line of investigation also merits consideration in the context of anxiety studies: there is the possibility that the experience and expression of different emotions are themselves modified by close sequential evocation. In the experiment considered from the Clark group, sexual arousal appeared to be linked to (i.e., contained overlapping information with) exercise arousal, as attested to by the obtained memory effects. Another phenomenon predictable from the hypothesis of a shared response structure in memory is that emotions with such information-base commonalities will be synergistic or mutually augmenting in effect. Zillmann (1980, 1983) has reported extensive research on this theme. His work suggests, first, that residual excitation from physical exercise can intensify subsequently induced anger and anger behavior (Zillmann & Bryant, 1974; Zillmann, Katcher, & Milavsky, 1972) or sexual excitement (Cantor, Zillmann, & Bryant, 1975). Thus, like Clark, Zillmann and colleagues specifically implicate a general arousal factor. Furthermore, between-emotion transfer has been shown. Both Zillmann's group (Zillmann, Bryant, Comisky, & Medoff, 1981) and Donnerstein and Hallam (1978) report that prior sexual arousal can potentiate aggression. Such excitation transfer appears to be independent of hedonic tone. Thus, Mueller and Donnerstein (1981) have demonstrated that sexual arousal can also facilitate positive social behavior, and Zillmann (1980, 1983) reports that both positive arousal and disgust at sexual material can prime the enjoyment of music, humor, or drama. This research implies that valence and arousal manipulations can affect not only the learning of neutral material, but subsequent emotional behavior which varies along one or more dimensions from the original. There are some indications that constraints exist on the direction and extent of such mediation (e.g., Hoon, Wincze, & Hoon, 1977). Overall, however, the evidence to date is very strong that propositions regarding valence and arousal are part of affective networks, that commonalities exist among networks, and that a thorough sampling of response systems is necessary in order to achieve a comprehensive assessment of the emotions. T H E E M O T I O N A L IMAGE AS A N I N F O R M A T I O N P R O T O T Y P E Imagery has proved to be a useful concept in exploring the nature and organization of these associative networks. Emotional imagery is widely 378 Lang and Cuthbert remarked in dreams, in anticipation of affective events, as aesthetic experience, and as a vehicle of therapeutic change (e.g., Singer, 1974). In contrast to the traditional phenomenological accounts of imagery, positing images as sensory events presented to the "mind's eye," recent accounts have viewed the image in terms of the processing of a propositional structure (e.g., Pylyshyn, 1973). Subjects code response information into an associative memory network during the original perception. Subsequent image processing involves the accessing of the entire network (including both the stimulus and the response information). While the image is a cognitive production, processing of response information initiates associated motor programs. In imagery, the final action commands of these programs are gated out or inhibited. However, there is always a certain amount of efferent leakage. Thus, a subovert pattern of motor responses apes many of the original sense organ and postural adjustments of the specific stimulus orienting task (e.g., Shaw, 1940; Deckert, 1964; Jacobson, 1930). It is our view that this same motor regeneration process underlies emotional imagery. In this case, however, the efferent pattern is even more elaborate, representing the prototype of the emotion action set. In a series of experiments we have explored the effects of inducing emotional imagery through textual descriptions of affect-arousing scenes (Lang, Kozak, Miller, Levin, & McLean, 1980; Lang, Levin, Miller & Kozak, 1983; Miller, Levin, Kozak, Cook, McLean, & Lang, 1985). On the assumption that the text or image script can be a cue for prototype concept matches, we have varied the propositional content of these materials. One manipulation involves presenting an auditory description of an encounter with the phobic object and including or excluding information about the responses which occur in that context: "Your heart is pounding as you began to speak," and "Your muscles tense; you press backward." Another procedure utilizes training programs which encourage subjects to focus on either stimulus or response information in their imagery. The results of these experiments have been very clear: the inclusion of a response emphasis encourages response processing, resulting in a psychophysiological pattern of efferent activity during imagery which duplicates in topography (albeit at lower amplitude) the affective responses observed with actual stimulus confrontation. Phobic subjects imaging phobic scenes show significant increases in heart rate, skin conductance, and respiratory rhythm, in a pattern similar to what they actually do when faced with the real phobic context. In general, we have found that subjects are most likely to develop an affective response to imagery instructions (1) when response processing is encouraged (as described above), (2) when self-reported good imagers are used as subjects, and (3) particularly when networks of high associative coherence are tapped (as with phobics). Contrary to the emphasis on stimulus information prompted by subjective Information Processing and Anxiety Assessment 379 theory, emotional imagery has been found to be much more of a response process, and in this it is consistent with the notion both that affect is basically an action set and that memory representations of psychophysiological events figure importantly in cognition. This is not to say that propositions regarding the stimulus field are wholly irrelevant or insufficient for prototype evocation. The view presented above holds that the number of input matches to network concepts from any source determines the probability of an affective production. This hypothesis predicts that improved stimulus matches should be as effective as the response matches encouraged in the above studies of emotional text processing in evoking the psychophysiology of affect. To test this assumption, McLean (1981) presented subjects with realistic emotional playlets that they were to use as a prompt to an imagery experience. The scenes included, for example, an actor playing a clumsy laboratory attendant, who removed a live, ostensibly poisonous snake from a cage a few feet from the subject and then struggled, almost unsuccessfully, to reconfine it. In these circumstances the conceptual match of stimulus information to affective view prototype is nearly perfect. Even though the subjects were clearly preinstructed that all aspects of the scenario were staged and not real (the snake was harmless and the actor skilled), we nevertheless observed the same context-specific physiology of emotion that we had previously noted with response concept matching in our studies of emotion-inducing text. These results demonstrate that stimulus materials of sufficient potency can activate the phobic prototype. This is not inconsistent with the emphasis upon response enhancement mentioned above. In fact, if the quantity of matching propositions is the key to prototype access, it is expected that response training would have the same enhancing effect on efferent responding regardless of the input medium. Kozak (1982) recently explored this hypothesis, administering both stimulus and response training to subjects and examining the effects of these procedures on different stimulus materials. Subjects were administered scripted text, as well as being exposed to some of the actual physical stimuli presumed to be part of the stimulus structure of a relevant fear prototype (e.g., for speech anxiety, a view of an audience). For both the physically presented stimuli and the text, response training prompted the enhanced physiological fear pattern predicted; stimulus training occasioned a significantly smaller physiological response to the objective stimulus than it did to the scripted rendering of the same material. Thus, consistent with the theory, both stimulus information and response information serve as effective matching inputs to associative networks. An exploration of meaning propositions, the third putative class, could be achieved by deceiving the subjects, i.e., that the situation is "real" and not a staged event. While this manipulation has not yet been attempted, we 380 Lang and Cuthbert presume that consonant meaning information would have a further enhancing effect. The data are, at this point, consistent with the view that any concept match increases the probability of network activation. CLINICAL STUDIES These findings from the basic research program have been applied to studying the imagery of anxious and phobic patients in the context of clinical assessment and treatment. The procedures employed have evolved in response to a desire to assess each patient as comprehensively as possible and to compare the organization of prototypes across patients. Each patient in the clinic participates in an imagery assessment session prior to therapy, at an intermediate point in therapy, just after termination, and 6 months after the completion of therapy. Physiological responses are measured while the patient visualizes a series of 11 scenes. Five of the scenes are constructed by the patient in collaboration with the therapist at the first interview. Two scenes represent clinical fear material, while the remaining three scenes are of a pleasant or happy situation, an unpleasant or sad experience not related to the clinical material, and a personal neutral scene. The remaining six trials are standard scenes given to all patients to facilitate between-subject comparisons and include two neutral scenes, two action scenes, and two fear scenes. The latter involve dental and speech scenes, fears common in the general population. Each of the 11 trials is comprised of four parts: a 30-sec initial resting period when baseline physiological measures are taken, a 50-sec period during which the scene script is presented (of which only the last 30 sec is analyzed), a 30-sec visualization period, and a 30-sec recovery period. Patients are initially given brief relaxation training, then instructed to imagine each prerecorded scene as it is presented. They are told to continue imagining the scene until a tone sounds indicating the start of the recovery period, at which time they are to stop imagining the scene and concentrate once again on muscle relaxation. For physiological analysis, change scores are calculated by deviating scores for the presentation, visualization, and recovery periods from the resting baseline of each trial. Self-Assessment Mannikin (SAM). Following each trial, patients make ratings of their preceding imagery using four scales. Three of these scales are based upon the three emotional dimensions reviewed above, i.e., valence, arousal, and dominance. To optimize the three-factor model for clinical use, we developed a cartoon-like figure dubbed SAM (self-assessment mannikin; Lang, 1980). There is one set of figures for each of the three dimensions, implemented as an interactive computer graphics program (see Fig. 1). The figure changes its characteristics according to the subject's manipulation of Information Processing and Anxiety Assessment 381 PLEASURE AROUSAL DOMINANCE Fig. 1. The self-assessment mannikin (SAM) is shown at five evenlyspaced positions (including extreme values) for each of the three dimensions of pleasure, arousal, and dominance. As implemented on the graphics computer, each figure can range over 30 different values controlled by the patient's ratings knob; the final value is retained by the computer. a rating knob attached to a potentiometer; pressing a switch terminates the display and causes the computer to record the last reading of the knob as the subject's rating. In order to validate this method, two independent samples were drawn and the ratings for each scene, averaged across subjects, were compared to those reported by Russell and Mehrabian (1977). For the two samples, the average correlations between the SAM ratings and the earlier semantic differential data were .94 for the valence dimension, .92 for activation, and .68 for dominance, all highly significant. Correlations between the factors were low except for that between valence and dominance, which shared more variance for these test materials; this result was also observed by Russell and Mehrabian. This indicated that SAM yielded estimates of factorial structure nearly identical to that provided by the semantic differential, but in a manner more rapid and accessible for patient use. A third study was performed without providing subjects any information about the definitions of the SAM displays; they were told simply to read the scenes and rate them using the figures. The intercorrelation matrix was again very similar to the original 382 Lang and Cuthbert data. Thus, SAM appears to be a reliable, relatively language-free methodology with which to assess affective reports of patients and other subjects. The fourth rating scale is for the vividness of the patient's image. It is a simple analog scale composed of a horizontal line, along which the patient moves a pointer (controlled by the rating knob) to indicate his or her rating of imagery vividness. Using these assessment procedures, we have studied the imagery of patients as a function of diagnostic grouping and change over time. The information-processing view has prompted us to develop the concept of functional imagery. A functional image is held to be one that can mediate a response change in the target reality context described by the image, and is most likely to occur when generated optimally. That is, it is clear that subjects can respond to an image script with alternative processing modes• A subject may be instructed simply to report what events the text describes or even just to comment on its lexical properties, a n d unless he or she is severely phobic, no affective response will be observed• Imagery implies, in our view, a different processing mode, one which includes the regeneration of response code, i.e., total network production• We already have data that the patients who profit from imagery therapies (that is, become less anxious or frightened after treatment) are those patients who can be shown to generate functional images, i.e., appropriate affective response topographies are observed during image induction in therapy (Lang, Melamed, & Hart, 1970). Furthermore, our recent results suggest that it is specifically fear patients who regenerate response code during a pretherapy assessment who have the best prognosis for treatment (Levin, Cook, & Lang, 1982; see Fig. 2). It is not yet clear how the imagery experience comes to modify emotional behavior• However, the fact that regeneration and at least partial processing of response code are critical elements suggests that subjects could be accomplishing "offline" processes of extinction or reconditioning which normally occur only in objective settings of response evocation. In any event, these results show that imagery procedures derived from the present model may be extremely useful both in distinguishing among different pathological states of anxiety and in providing a firmer basis for estimates of prognosis and treatment selection. • THE \ ANXIETY DISORDERS: DIFFERENCES MEMORY S T R U C T U R E IN AFFECTIVE These assessment methodologies are well suited to the investigation of the various anxiety disorders. In fact, the results to date adumbrate potential refinements in nosological approaches to these diagnostic categories. Our basis for these views lies in a consideration of the differences between pho- InformationProcessingandAnxietyAssessment oooo o 383 SUCCESSES H ~ 4.01 vw 2.0 I<~ r,. v- (E O.O- LIJ "I" - 2 . 0 - <3 -4.0 i i * NEUTRL , • i i * i i ACTN 'ST'ND' CLiNI(~ PdS.' NEG. FEAR AFFECT STANDARD IMAGERY SCRIPTS PERSONAL IMAGERY SCRIPTS Fig. 2. A comparison of pretreatment heart-rate change scores for patients who successfully completed therapy (N = 19) and those who discontinued therapy prior to successful outcome (N = 5). These patients were seen at the Psychology Department Clinic, University of Wisconsin. Patients were provided scripts and instructed to imagine vividly six scene types: neutral situations (e.g., relaxing at home in an easy chair); action scenes that were nonaffective but included vigorous activity (e.g., bike riding); standard fear scenes (e.g., being locked in a sauna); scripts prepared with the patients' collaboration describing the clinical fear context; and two other personal scripts, one referring to an outstanding positive life experience (pos.) and the other to an unpleasant but nonclinical fear context (neg.). The heart-rate data are change scores from the prescene test. Three data points are shown for each scene: the 30-sec period during which the scripts were orally presented, the subsequent 30 sec of imagery, and a 30-sec recovery time which began immediately after the patients were told to stop imagining the scene. In the average patient, neutral material prompted little change in heart rate. Action and negative affects occasioned acceleration, while patients tended to decelerate during positive-affect scenes. The latter effects were dramatically larger in patients who responded positively to behavior therapy intervention (e.g., flooding, desensitization, coaching, and modeling). [From Levin et al. (1982).] bias a n d t h e o t h e r a n x i e t y d i s o r d e r s . F i r s t , t h e s t i m u l u s i n f o r m a t i o n is explicit a n d r e l i a b l e ; s e c o n d , t h e r e s p o n s e p a t t e r n a p p e a r s t o be b u i l t a r o u n d a c t i v e a v o i d a n c e . I n a r e c e n t e x p e r i m e n t c o m p a r i n g f o c a l p h o b i c s a n d socialiy a n x i o u s subjects ( L a n g e t al., 1983), o v e r t a v o i d a n c e t e n d e d to o c c u r m o r e f r e q u e n t l y in f o c a l p h o b i c s f a c i n g t h e i r o w n f e a r o b j e c t t h a n in socially a n x i o u s s u b j e c t s s i m i l a r l y c o n f r o n t i n g t h e i r p r i m a r y stress. D i f f e r e n c e s between these fear types were even clearer when physiological responses were e x a m i n e d . L i t t l e d i s t i n c t i o n was f o u n d b e t w e e n t h e a r o u s a l p h y s i o l o g i e s o f s o c i a l l y a n x i o u s s u b j e c t s a n d t h o s e o f p h o b i c s w h e n b o t h g a v e s p e e c h e s to an u n f a m i l i a r a u d i e n c e ( b o t h g r o u p s s h o w e d m a r k e d visceral a c t i v a t i o n d u r i n g 384 Lang and Cuthbert this social performance task); however, when these same two groups viewed phobic fear stimuli, socially anxious subjects stayed relatively calm, while the specific phobics showed a progressive increase in heart rate the closer they were to their phobic o b j e c t - a n increase paralleled by a higher probability of avoidance. Thus, the psychophysiologyof phobia looked very much like preparation for flight, while the arousal response of socially anxious subjects, although quite high in amplitude, was more general in pattern and less explicit in affective meaning. These group differences in psychophysiological response were also apparent in their fear imagery and in anticipation of threat. Overall, severe phobics showed more three-systems (affective report, physiological, and behavioral) concordance across contexts of emotional instigation than did comparably distressed socially anxious subjects. The differences observed between socially anxious subjects and focal phobics have reappeared with some consistency across experiments, but the effect has often been too subtle for statistical significance and seldom dramatically large. This may be because socially anxious subjects were selected according to their degree of speech fear. We inadvertently included in the sample both explicit speech phobics (who may be no different from other phobics in imagery) and members of a more broadly, socially anxious population (who are presumed to have a relative imagery deficit). Thus, in contrast to the current Diagnostic and Statistical Manual o f Mental Disorders, 3rd ed. (DSM-III), categories, our research indicates that a circumscribed and focused fear of public speaking may more properly belong diagnostically with the specific phobias. More pervasive deficits would characterize the socially anxious, such as difficulty in a wide range of interpersonal situations. We infer from the above phenomena that the information network of a phobia is, in general, more stable and coherent than that of social anxiety. Furthermore, we believe that this property of network coherence is related to a general ability to imagine vividly both new and previously experienced events. In several studies focal phobics have reported images to be more vivid, to involve more movement, and to be more closely related to content in arousal response than was observed for socially anxious subjects (Lang et al., 1970; Weerts & Lang, 1978; Bardach, E., & Weerts, T. C., Unpublished manuscript, University of Iowa). We are not suggesting here that phobics are better imagers than normals, hut only that, among the anxiety disorders studied, they generate the most stable, vivid, and representative imagery. Furthermore, as we have previously argued (Lang, 1979), what is reported subjectively as an image has much more to do with the organization of responses than it has to do with pictures in the mind's eye. The phobic image is a highly focused psychophysiological memory structure and a prototype affective response disposition. It may be that the effect of high network coherence in phobia is to restrict the association of its related arousal response processes Information Processing and Anxiety Assessment 385 and thus limit their influence on other memory content, relative to the spread or transfer of such excitation in other anxiety disorders. In recent clinical applications of our imagery assessment technique, we have found even more dramatic differences in imagery ability between focal phobics and a population of agoraphobics. Perhaps even more than the socially anxious (with whom they are sometimes diagnostically confused), agoraphobics show much less affective network coherence and considerably less context specificity of the consequent anxious behaviors. While some stimulus and response information is recurrent across fear induction settings, the associative generality of the arousal response pattern is again much more broad than that of focal phobia. It is expected that these patients will tend to report less vivid images, to show a less responsive physiology in imagery, to be less concordant across a three-systems assessment, and to be less consistent in the anxiety-induction contexts to which they respond. Furthermore, as the associative structure is more diffuse and less accessible to any therapeutic intervention, agoraphobia is more persistent and the prognosis is poorer than for focal phobia. Preliminary results of a psychophysiological imagery assessment of agoraphobic and other phobic patients are presented in Fig. 3 (McNeil, Melamed, Cuthbert, & Lang, 1983). It will be noted that the physiological response to affective imagery is significantly less for the agoraphobics than for the other phobics. Agoraphobics also report themselves to image less vividly and report lower levels of experienced arousal to clinical fear material (Fig. 4). Agoraphobics also have a higher incidence of dropping out from therapy, while focal phobics more frequently continue therapy to a successful resolution. Research accomplished by other investigators (Barlow, Mavissakalian, & Schofield, 1980) supports this general view, suggesting that agoraphobics have a less concordant three-systems structure and that this may be related to this group's poorer prognosis. Epidemiological studies (Agras, Sylvester, & Oliveau, 1969) are also consistent with this analysis, showing that focal phobias are relatively short-lived, while agoraphobia persists in the population, with the peak prevalence late in adult life. AN ANXIETY-DISORDER C O N T I N U U M It is proposed that the anxiety disorders may be distributed along a continuum of affective memory organization. The continuum is defined by the degree to which arousing, negatively valent responses (and perhaps also disruption of control) are linked associatively to coherent affect networks or, viewed from the other direction, the degree to which these affective response dispositions float in memory and are prompted by many stimuli, transferring their excitation to a great variety of other memory structures. Proceed- 386 Lang and Cuthbert IMAGERY RESPONSE AFTER SCRIPT PRESENTATION "IMAGE" PERIOD I0- 8- • Simple Phobics(n=9) [] Other Anxiety Disorders (i]=5) [ ] Agorophobics(n=5) ~ 6I<c bJ ~_ 4- , = °! 0 ::2:0 . . . . TZ~-- <~ -~= . . . . . . ~ --'ml- -2-! -4- I | I I CLINICAL FEAR STANDARD PERSONAL STANDARD SPEECH DANGER ACTIONFEAR KITE FLYING Fig. 3. A comparison of pretreatment heart-rate change scores for patients grouped according to the diagnosis of agoraphobia, simple phobia, or other anxiety disorders (including obsessive-compulsive). These patients were seen at the Fear Clinic, University of Florida. The data presented are change scores from the prescene baseline, for the 30-sec imagery period. Agoraphobics exhibit markedly diminished responding, compared to specific phobics, for all three affective scene types of clinical fear, standard speech fear, and personal danger. ing from most to least along the continuum of network coherence, the order of nosology is as follows: focal phobia, obsessions and compulsions, social anxiety, agoraphobia, generalized anxiety states. Generalized anxiety would appear to represent the condition of m a x i m u m associative fluidity of affective response structures. It is for these patients that the research on statedependent learning may be most relevant. Their behavior seems consistent with the notion of an underlying affective response program, which mediates associated anxious thoughts, worries, and the like across disparate contexts and environments. Such a mechanism could account for Gentil and Information Processing and Anxiety Assessment 387 AROUSAL Simple Phobics In=9) [[]--'l--I Other Anxiety Disorders (n=5) High 29Arousal (~--C) Agorophobics (n=5] 25- 20J co 15- \ n~ IO- 5- Calm 0 i CLINICAL FEAR l l PERSONAL STANDARD DANGER SPEECH FEAR i STANDARD ACTION KITE FLYING Fig. 4. Posttrial ratings of subjective arousal obtained with the self-assessment mannikin (SAM), obtained from the University of Florida patient sample. Agoraphobic patients reported less arousal for clinical fear material, while indicating arousal levels highly similar to those of the other two patient groups for personal danger, standard fear, and standard action scenes. Lader's (1978) report that anxiety patients have more frequent dreams of being attacked or Butler and Mathews' (1983) finding that anxiety patients overestimate the probability of many threatening events (both social and physical). The presence of the proposed mediating physiology already has general documentation: Lader and Wing (1966) and Hart (1974) are among the many who have shown that anxiety patients generate more spontaneous sympathetic responses, even at rest. We have already noted the high frequency of visceral and somatic symptom reports in this population (Hamilton, 1959; Buss, 1962). Furthermore, the evidence suggests that the response program of an emotion can be either highly differentiated or general in its representation, It can serve either as a focus for specific, context-bound acts (as in phobia) or as the mediator of a spectrum of syndromal behaviors that may appear in any setting. Thus, in the future diagnostic efforts may be enhanced, and/or 388 Lang and Cuthbert defined, by a more quantitative assessment of the degree of such affective ubiquity. PERSPECTIVES We have repeatedly emphasized that the meaning of emotion and cognition is in the organization and production of behavior. However, it now appears that in focusing too narrowly on overt action and ignoring patient differences in cognitive processing characteristics, early behavioral assessment neglected important nosological issues. A more balanced view permits us to come to grips with a complex clinical and theoretical literature (e.g., Mischel, 1979; Bandura, 1977). Indeed, cognitive behavior modification has already responded to the challenge (Meichenbaum, 1977; Mahoney, 1974). However, the current concept of action programs departs from the latter approach as well, in at least two substantial respects. First, a variety of clinicians tends to view cognitive and physiological response mechanisms as separate domains of symptomatology and as independent treatment targets (e.g., Meichenbaum, 1977; Ellis, 1971). However, our own studies of affective imagery and the research on emotional statedependent learning argue something quite different. They suggest that the physiology of emotion is represented in associative memory and is, in this sense, no less cognitive than a verbal report. It is important here to remind ourselves that response information does not need to exert a central influence through interoceptive or exteroceptive feedback from peripheral afferents. We do not need to assume a homunculoid self that assesses acts after they occur. In various embodiments, this notion has been widespread across a wide variety of theoretical systems. For example, S. Schachter and his associates (Schachter & Singer, 1962) proposed that the physiology of emotion was marked by a unidimensional sympathetic arousal; differences between emotional states were assumed to be the result of a cognitive labeling dependent upon appraisal of the activation in the context of current environmental circumstances. This view assumes that the physiological response is a relatively independent phenomenon, which requires a secondary perceiving and labeling process for an emotion to occur. From the present perspective, however, there is every reason to believe that the action program itself (or a particular subroutine, such as a pattern of visceral activation) is associatively linked to other information in the brain. While subsequent feedback may support, confirm, or modify ongoing emotional processing, it is not necessary to the instigation or persistence of affective responses or their associations. The second, related point is that cognition denotes much more than the natural language manipulations emphasized by the cognitive behavior Information Processing and Anxiety Assessment 389 modification approach. H u m a n information processing should not be construed as isomorphic with the cogito of Descartes. We must be careful not to embrace a crypto-subjectivism, however attractively it may be festooned with the novel terminology of the computer metaphor. It is fundamental that the mechanistic premise of natural science be retained regarding causality and the admissibility of evidence for scientific discourse. Thus, the data reviewed above suggest that the activation of cognitive networks, including those pertaining to affective material, is largely outside of awareness and is not readily explained by volitional concepts. However, this should not invite the resurrection of the volitional unconscious. Automatic processes (Shiffrin & Schneider, 1977) are conceived very differently from those invoked by the dynamic psychologies of the past. As Laird et al. (1982) said in concluding their paper on facial expression and state-dependent recall, "Perhaps many of the e m o t i o n / m e m o r y relationships that have previously been thought to reflect self-protective or self-aggrandizing purposes may instead arise from essentially undirected organizational properties of memory" (p. 656). The emphasis on the real-time monitoring of cognitive events through psychophysiological measurement (as opposed to a reliance on retrospective narratives) is an important key to the current approach. Van den Bergh and Eelen (1983) aptly summarized this view of cognitive psychology in assessment vis-&-vis behavioristic and subjectivistic approaches: "By removing conscious or verbal reports as necessary indicators of cognition, cognitive psychology provides a more solid theoretical framework to understand . . . clinical phenomena for which the psychodynamic clinician used the concept of 'the Unconscious' and for which behavior therapists used the concept of conditioning" (p. 5). FUTURE DIRECTIONS The assessment of fear and anxiety is still in its earliest stages in terms of cognitive psychophysiology. We are just beginning to see the merger of older viewpoints with contemporary information-processing paradigms and techniques. We can only sketch here some of the approaches which appear promising with respect to their applications to the anxiety domain. (1) Continued investigation of affective networks will provide a better understanding of their organizational properties. One paradigm that appears promising in this regard has been termed priming. This refers to a choice reaction-time (RT) task in which the latency of a decision regarding some lexical stimulus (e.g., previously seen/not seen) is shorter when the stimulus is preceded (or primed) by a semantically or logically related word. For example, Ratcliff and McKoon (1981) reported that the RTs to report whether a target word had appeared in a previously presented paragraph set were 390 Lang and Cuthbert reduced as a function of the logical proximity of the priming word. We are now using this paradigm in our laboratory to explore networks with affecrive components. This work already suggests large differences in priming phenomena during recall of neutral and affective text. We now hope to determine whether affective propositions are incorporated across the dimensions of emotion described above; whether the presence of affective propositions facilitates (e.g., Zajonc, 1980) or, conceivably, delays access to a network; and whether the presence of emotional material affects access across different networks. (2) Another literature relevant to anxiety is that of attention, with a set of interrelated concepts such as filtering (e.g., Broadbent, 1958; Posner & Snyder, 1975) and automatic vs. controlled attention (Shiffrin & Schneider, 1977). For example, Segal and Fusella (1970) instructed subjects to imagine either auditory or visual material and concurrently presented low-level auditory or visual stimuli. They found that signal detection thresholds were higher when the stimulus modality matched the image modality, suggesting that the ability to detect the stimulus was reduced by the allocation of attentional capacity to the image. Thus, measures of attentional allocation could be studied as potential on-line markers of the vividness of imagery. Dependent variables that might be employed to study this possibility include signal detection, as in Segal and Fusella's work; reaction times; the magnitude of autonomic reactions (orienting) to an intruding stimulus; slow cortical potentials; and evoked potentials shown to be related to attentional and decision processes, e.g., the P300 (Donchin, 1981). (3) The study of state-dependent learning, as typified by the work cited above by Bower, Clark, Laird, and others, has produced some of the best evidence to date regarding the importance of response processes in emotional expression. The study of emotion-dependent learning and excitation transfer bears direct and immediate application to anxiety patients. The arguments made here suggest that the mediation of memories through affective state may be more generalized in anxious subjects than in normals. It is important that we confirm the evidence that imagery is best in phobics and that it is reduced in quality as the hypothesized network coherence lessens and the diagnosis of generalized anxiety becomes more appropriate. We should test the hypothesis that excitation transfer occurs more easily in anxiety, that valence may be less significant in inhibiting generalizations between arousal states in anxious patients than in normals, and that positive affective arousal as well as aversive activation may potentiate worry and negative self-reference in high trait-anxious subjects. In terms of therapeutic issues, Butler and Mathews (1983) have speculated that anxiety patients should be introduced to the coping instructions of choice during a period of high anxiety, rather Information Processing and Anxiety Assessment 391 than in a relaxed state as is usually done, so as to render them more accessible during periods of anxiety outside the clinic. (4) We need to know more about the general dimensions of emotion, i.e., valence, arousal, and dominance. For example, in studying valence, should we be considering attentional sets such as orienting and defensive reflexes (Sokolov, 1963)? Arousal also continues to be a confusing and poorly defined concept. We know already that marked differences exist in the peripheral physiology of various emotional acts (Obrist, 1981); these must be accounted for in developing a comprehensive account of activation's role in emotion. Even within presumably similar patterns of response, Zillmann (1983) has presented interesting data suggesting that excitation transfer occurs best when the preceding emotion has had some time to dissipate; this phenomenon must also be accounted for. We also need more information on how control-dominance may be represented in m e m o r y - f o r example, whether this is truly an independent dimension of affect with its own psychophysiology or whether it can be understood, as some argue (Russell, 1980), as "meaning" information that can be associated with various interactions of valence and arousal. The significance of valence, arousal, and control as parameters of anxiety is extensively documented in research already reviewed. However, it is worth considering that these dimensions also provide a natural classification of methods used in the treatment of anxiety disorders. That is, therapists either attempt to reduce arousal through drugs or relaxation training (Wolpe, 1958; Klein, 1981); try to modify the valence of negative contexts by, for example, providing success experiences in the aversive context and by reinterpreting the meaning of negative situations (e.g., Ellis, 1971; Meichenbaum, 1977); or focus on training the subject in self-control, efficacy, and competency, b.y teaching skills and modeling (Bandura, 1977; McFall & Twentyman, 1973). It may well prove that the most effective therapy will not be similarly unimodal but will depend on careful consideration of all three affective parameters. In conclusion, basic research on memory and affect is already significantly advanced, and the principles of information processing and contemporary cognitive psychology have been shown to be germane to issues regarding emotional behaivor. Parallel progress in the study of pathological emotion can be achieved if these paradigms and strategies are applied in research on anxiety. In this effort it will be necessary to monitor carefully the psychophysiology of patients, as the evidence is now strong that the representation of physiological events figures prominently in cognitive life. The integration of psychophysiological propositions with other elements of the cognitive programs for emotion represents an advance both in theory 392 development cations. Lang and Culhbert and in the possibilities for assessment and treatment appli- REFERENCES Agras, W. S., Sylvester, D., & Oliveau, D. (1969). The epidemiology of common fears and phobias. Comprehensive Psychiatry, 10, 151-156. Anderson, J. R., & Bower, G. H. (1974). A propositional theory of recognition memory. Memory & Cognition, 2(3), 406-412. Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191-215. Barlow, D. H., Mavissakalian, M., & Schofield, L. (1980). Patterns of desynchrony in agoraphobia. Behaviour Research and Therapy, 441-448. Bower, G. H. (1981). Mood & Memory. American Psychologist, 36, 129-148. Bower, G. H., & Cohen, P. R. (1982). Emotional influences in memory and thinking: Data and theory. In S. Fiske & M. Clark (Eds.), Affect and social cognition. Hillsdale, N J: Lawrence Erlbaum Associates. Broadbent, D. E. (1958). Perception and communication. London: Pergamon Press. Buss, A. H. (1962). Critique and notes: Two anxiety factors in psychiatric patients. Journal of Abnormal and Social Psychology, 65, 426-427. Butler, G., & Mathews, A. (1983). Cognitive processes in anxiety. Advances in Behaviour Research & Therapy, 5, 51-62. Cantor, J. R., Zillmann, D., & Bryant, J. (1975). Enhancement of experienced sexual arousal in response to erotic stimuli through misattribution of unrelated residual excitation. Journal of Personality and Social Psychology, 32, 69-75. Clark, M. S., Milberg, S., & Ross, J. (1983). Arousal cues, arousal-related material in memory: Implications for understanding effect of mood on memory. Journal of Verbal Learning and Verbal Behavior, 22, 633-649. Deckert, G. H. (1964). Pursuit eye movements in the absence of moving visual stimulus. Science, 143, 1192-1193. Donchin, E. (1981). Surprise!...Surprise? Psychophysiology, 18, 493-513. Donnerstein, E., & Hallam, J. (1978). Facilitating effects of erotica on aggression against women. Journal of Personality and Social Psychology, 36, 1270-1277. Ellis, A. (1971). Emotional education in the classroom: The living school. Journal of Clinical Child Psychology, 1, 19-22. Eysenck, H. J. (1961). The handbook of abnormal psychology. Basic Books: New York. Fenz, W. D. (1967). Specificity in somatic responses to anxiety. Perceptual and Motor Skills, 24, 1183-1190. Fenz, W. D., & Epstein, S. (1965). Manifest anxiety: Unifactorial or multifactorial composition? Perceptual and Motor Skills, 20, 773-780. Gentil, M. L. F., & Lader, M. (1978). Dream content and daytime attitudes in anxious and calm women. Psychological Medicine, 8, 297-304. Hamilton, M. (1959). The assessment of anxiety states by rating. British Journal of Medical Psychology, 32, 50-59. Hart, J. D. (1974). Physiological responses of anxious and normal subjects to simple signal and non-signal auditory stimuli. Psyehophysiology, 11, 443-451. Hoon, P., Wincze, J., & Hoon, E. (1977). A test of reciprocal inhibition: Are anxiety and sexual arousal in women mutually inhibitory? JournalofAbnormalPsychology, 86, 65-74. Jacobson, E. (1930). Electrical measurements of neuromuscular state during mental activities. IlI. Visual imagination and recollections. American Journal of Physiology, 95, 694-702. Kieras, D. (1978). Beyond pictures and words: Alternate information processing models for imagery effects in verbal memory. Psychological Bulletin, 5, 532-554. Klein, D. F. (1981). Anxiety reconceptualized. In D. F. Klein & J. Rabkin (Eds.), Anxiety: New research and changing concepts. New York: Raven Press. Information Processing and Anxiety Assessment 393 Kozak, M. J. (1982). The psychophysiological effects of training on variously elicited imaginings, Unpublished doctoral dissertation. Madison: University of Wisconsin-Madison. Lacey, J. I. (1959). Psychophysiological approaches to the evaluation of psychotherapeutic process and outcome. In E. A. Rubinstein & M. B. Parloff (Eds.), Research in psychotherapy (Vol. 1). Washington, DC: American Psychological Association. Lader, M. H., & Wing, L. (1966). Physiological measures, sedative drugs, and morbid anxiety. New York: Oxford Press. Laird, J. D., Wagener, J. J., Halal, M., & Szegda, M. (1982). Remembering what you feel: Effects of emotion on memory. Journal of Personality and Social Psychology, 42, 646-657. Lang, P. J. (1964). Experimental studies of desensitization psychotherapy. In J. Wolpe (Ed.), The conditioning therapies. New York: Holt, Rinehart & Winston. Lang, P. J. (1968). Fear reduction and fear behavior: Problems in treating a construct. In J. M. Shlien (Ed.), Research in psychotherapy (Vol. III, pp. 90-103). Washington, DC: American Psychological Association. Lang, P. J. (1977). Imagery in therapy: An information processing analysis. Behavior Therapy, 8, 862-886. Lang, P. J. (1978). Anxiety: Toward a psychophysiological definition. In H. S. Akiskal & W. L. Webb (Eds.), Psychiatric diagnosis: Exploration of biological predictors (pp. 365-389). New York: Spectrum. Lang, P. J. (1979). A bio-informational theory of emotional imagery. Psychophysiology, 16, 495-512. Lang, P. J. (1980). Behavioral treatment and bio-behavioral assessment: Computer applications. In J. Sidowski & T. Williams (Ed.), Technology in mental health care delivery systems. Norwood, N J: Ablex. Lang, P. J. (1983). Cognition in emotion: concept and action. In C. Izard, J. Kagan, & R. Zajonc (Eds.), Emotion, cognition, and behavior. New York: Cambridge University Press. Lang, P. J. (1985). The cognitive psychophysiology of emotion: Fear and anxiety. In A. H. Tuma & J. D. Maser (Eds.), Anxiety and the anxiety disorders. Hillsdale, N J: Lawrence Erlbaum Associates (in press). Lang, P. J., Melamed, B. G., & Hart, J. D. (1970). A psychophysiological analysis of fear modification using an automated desensitization procedure. Journal of Abnormal Psychology, 31, 220-234. Lang, P. J., Kozak, M. J., Miller, G. A., Levin, D. N., & McLean, A., Jr. (1980). Emotional imagery: Conceptual structure and pattern of somato-visceral response. Psychophysiology, 17, 179-192. Lang, P. J., Levin, D. N., Miller, G. A., & Kozak, M. J. (1983). Fear behavior, fear imagery, and the psychophysiology of emotion: The problem of affective response integration. Journal of Abnormal Psychology, 92, 276-306. Levin, D. N., Cook, E. W., III,& Lang, P. J. (1982). Fear imagery and fear behavior: Psychophysiological analysis of clients receiving treatment for anxiety disorders. Psychophysiology, 19(5), 571-572 (abstr.). Mahoney, M. J. (1974). Cognition and behavior modification. Cambridge, MA: Ballinger. McFall, R. M., & Twentyman, C. T. (1973). Four experiments on the relative contributions of rehearsal, modeling, and coaching to assertion training. Journal of Abnormal Psychology, 81, 199-218. McLean, A. (1981). Emotional imagery: Stimulus information, imagery ability and patterns of physiological response, Doctoral dissertation. Madison: University of WisconsinMadison. McNeil, D. W., Melamed, B. G., Cuthbert, B. N., & Lang, P. J. (1983). Emotional imagery and psychophysiological responsivity in simple phobia and agoraphobia. Psychophysiology, 20, 459 (abstr.). Meichenbaum, D. (1977). Cognitive-behavior modification: An integrative approach. New York: Plenum Press. Miller, G. A., Levin, D. N., Kozak, M. J., Cook, E. W., III, McLean, A., & Lang, P. J. (1985). Emotional imagery: Effects of individual differences in perceptual imagery on physiological response (manuscript submitted for publication). 394 Lang and Cuthbert Mischel, W. (1979). On the interface of cognition and personality: Beyond the person-situation debate. American Psychologist, 34, 740-754. Mueller, C. W., & Donnerstein, E. (1981). Film-facilitated arousal and prosocial behavior. Journal of Experimental Social Psychology, 17, 31-41. Obrist, P. A. (1981). Cardiovascular psychophysiology: A Perspective. New York: Plenum Press. Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. (1957). The measurement of meaning. Urbana: University of Illinois Press. Panksepp, J. (1982). Toward a general psychobiological theory of emotions. The Behavioral and Brain Sciences, 5, 407-422. Posner, M. I., & Snyder, C. R. R. (1975). Attention and cognitive control. In R. L. Solso (Ed.), Theories in information processing. Hilldsale, N J: Lawrence Erlbaum Associates. Pylyshyn, Z. W. (1973). What the mind's eye tells the mind's brain: A critique of mental imagery. Psychological Bulletin, 80, 1-24. Quillian, M. R. (1966). Semantic memory. In M. L. Minsky (Ed.), Semantic information processing. Cambridge, MA: MIT Press. Rachman, S. (1974). The meaning of fear. Middlesex: Penguin Books. Rachman, S., & Hodgson, R. I. (1974). Synchrony and desynchrony in fear and avoidance. Behavior Research and Therapy, 12, 311-318. Ratcliff, R., & McKoon, G. (1981). Does activation really spread? Psychological Review, 88(5), 454-462. Riskind, J. H. (1983). Feedback effects of physical posture appropriateness: Or, where people "slump to conquer." Presented at the Nag's Head Conference on Stress, Conflict, and Emotion. Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 38, 1161-1178. Russell, J. A., & Mehrabian, A. (1977). Evidence for a three-factor theory of emotions. Journal of Research in Personality, 11, 273-294. Sartory, G. (1981). Some psychophysiological issues in behavioral psychotherapy. Behavioral Psychotherapy, 9, 215-230. Schachter, S., & Singer, J. E. (1962). Cognitive, social, and psychological determinants of emotional state. Psychological Review, 69, 379-399. Segal, S. J., & Fusella, V. (1970). Influence of imagined pictures and sounds on detection of visual and auditory signals. Journal of Experimental Psychology, 83, 458-464. Shaw, W. A. (1940). The relation of muscular action potentials to imaginal weight lifting. Archives of Psychology, 23, 380-389. Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human information processing. II. Perceptual learning, automatic attending, and a general theory. Psychological Review, 84, 127-190. Singer, J. L. (1974). lmagery and daydream methods in psychotherapy and behavior modification. New York: Academic Press. Sokolov, Y. N. (1963). Perception and the conditioned reflex. New York: Macmillan. Sperry, R. (1952). Neurology and the mind-brain problem. American Scientist, 40, 291-312. Tuma, A. H., & Maser, J. D. (eds.) (1985). Anxiety and the anxiety disorders. Hillsdale, N J: Lawrence Erlbaum Associates (in press). Van den Bergh, O., & Eelen, P. (1983). Unconscious processing and emotions. Paper presented at the World Congress on Behavior Therapy/AABT 17th Annual Convention, Washington, DC. Weerts, T. C., & Lang, P. J. (1978). The psychophysiology of fear imagery: Differences between focal phobia and social-performance anxiety. Journal of Consulting and Clinical Psychology, 46, 1157-1159. Weimer, W. B. (1977). A conceptual framework for cognitive psychology: Motor theories of the mind. In R. Shaw & J. Bransford (Eds.), Perceiving, acting, and knowing. Hillsdale, NJ: Lawrence Erlbaum Associates. Wolpe, J. (1958). Psychotherapy by reciprocal inhibition. Stanford CA: Stanford University Press. Zajonc, R. B. (1980). Feeling and thinking. American Psychologist, 35, 151-175. Zillmann, D. (1980). Anatomy of suspense. In P. H. Tannenbaum (Ed.), The entertainment functions of television. Hillsdale, N J: Lawrence Erlbaum Associates. Information Processing and Anxiety Assessment 395 Zillmann, D. (1983). Treatment of excitation in emotional behavior. In J. T. Cacioppo & R. E. Petty (Eds.), Social psychophysiology. New York: Guilford Press. Zillmann, D., & Bryant, J. (1974). Effect of residual excitation on the emotional response to provocation and delayed aggressive behavior. Journal of Personality and Social Psychology, 30, 782-791. Zillmann, D., Katcher, A. H., & Milavsky, B. (1972). Excitation transfer from physical exercise to subsequent aggressive behaivor. Journal of Experimental Social Psychology, 8, 247-259. Zillmann, D., Bryant, J., Comisky, P. W., & Medoff, N. J. (1981). Excitation and hedonic valence in the affect of erotica on motivated intermale aggression. European Journal of Social Psychology, 11, 233-252.