Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Psychological Bulletin 1985, Vol. 97, No. I, 3-17 Copyright 1985 by the American Psychological Association, Inc, 0033-2909/85/$00.75 Does Vicarious Instigation Provide Support for Observational Learning Theories? A Critical Review G i n a G r e e n a n d J. G r a y s o n O s b o r n e Utah State University The vicarious instigation literature since 1962 is critically reviewed for its adequacy as support for the observational learning theories of Aronfreed, Bandura, Berger, and Hygge. These theories are examined for conditions purported to be necessary to vicarious instigation. The different theories imply the possibility of research differentiating them; however, most of the literature does not test any of the theories in any direct way. In fact, only 3 of 30 studies reveal conditions necessary to vicarious instigation. Unfortunately, no one theory is unequivocally supported or diseonfirmed. It has been 22 years since the publication of an article entitled "Conditioning Through Vicarious Instigation" (Berger, 1962). This article promoted a considerable research effort in the areas of vicarious instigation and vicarious classical conditioning. It has been cited approximately 96 times since 1969 (Social Science Citation Index, 1969-1983) in theoretical support of social learning theory, particularly the emotional aspects of observational learning (e.g., Bandura, 1965, 1969, 1971a, 1971b, 1977; Hygge, 1976a, 1976b; Ohman & Hygge, 1977), in theoretical support of the sociopsychological concepts of sympathy, empathy, sadism, and masochism (e.g., Krebs, 1970; Schwartz & Shapiro, 1973), and attitude development (Kanekar, 1976). The speculative relation of vicarious processes to these phenomena then is of continuing interest (e.g., Aronfreed, 1969; Berger, 1962). The vicarious instigation literature includes a potentially confusing array of experiments and findings. Yet, with the exceptions of a tangentially related logical/philosophical critique of "hyphenated" reinforcement processes (Gewirtz, 197 l) and a single review of An earlier draft of this article was the basis of an invited address to the Association for Behavior Analysis in Milwaukee, Wisconsin, May 1982. The authors are grateful to Frank Ascione, William Greene, Richard Herrnstein, William Prokasy, Robert Rescorla, and Sebastian Striefel for their comments on an earlier draft of this article. Requests for reprints should be sent to J. Grayson Osborne, Department of Psychology--UMC 28, Utah State University, Logan, Utah 84322. vicarious reinforcement studies (Thelen & Rennie, 1972), there has been no critical review of the literature dealing with vicarious processes. The present article is an attempt to rectify this omission partially. Formally, it is our purpose to review and critique the theory and research on vicarious instigation to determine the use of this concept as a foundation o f the observational learning of emotional behavior. To accomplish this, the review will examine four theories o f vicarious instigation and the procedures that appear necessary to test the theories. We then examine the experimental evidence to determine its support for the several theories in the literature. We believe there is little agreement regarding the necessary conditions for vicarious instigation, and that many of the studies purporting to produce vicarious instigation do not contain research designs that allow unequivocal inferences regarding theory. Therefore, there is only limited support at this time for observational learning theories from the research in this area. The review is restricted to vicarious instigation because of the broad nature of the findings classified as vicarious. Because vicarious classical conditioning,~ vicarious reinforcement, and vicarious punishment appear to follow vicarious instigation, it was determined to focus initially on vicarious instigation. Thus, this review includes research in the area of vicarious instigation published since 1962. It does not include studies that examined only vicarious instrumental/operant 4 GINA GREEN AND J. GRAYSON OSBORNE conditioning (i.e., studies subsumed under the descriptors, vicarious reinforcement and vicarious punishment). We also excluded from this review experiments that mixed operant and classical conditioning procedures, because the procedures and the dependent variables in such experiments deviate considerably from the rest of the vicarious instigation/conditioning literature. These procedures (cf. Gormezano & Kehoe, 1975) use a conditioned suppression paradigm in which a classically conditioned stimulus is subsequently presented to the subject in an operant situation (e.g., Venn & Short, 1973). To acquaint the reader with vicarious instigation, the following definition is offered. (This is modified in later sections in relation to particular theoretical positions.) Vicarious instigation is the occurrence of an unconditioned emotional response of an observer dependent on the observer's inference (Hygge & Dimberg, 1982) or perception (Berger, 1962) of the unconditioned emotional response o f a model. Vicarious instigation is an important phenomenon in its own right because it describes how one individual is affected by another's emotions. Additionally, it is important because the resultant behaviors of the observer may be conditioned to formerly neutral aspects of the situation. That is, vicarious classical conditioning may follow from vicarious instigation. Vicarious instigation may also play a role in instrumental conditioning when perception of another's unconditioned emotional response leads to a vicariously instigated response o f the observer, which in turn reinforces the observer's prior behavior (Berger, 1962). For example, an observer's instrumental response, such as giving a gift, produces an unconditioned emotional response in the receiver. Theoretically, the unconditioned emotional response functions as an unconditioned stimulus to the observer, producing an unconditioned emotional response by the observer that may reinforce the observer's preceding instrumental response (i.e., the gift-giving; Berger, 1962, p. 453). Prototypic Experiment Most investigations of vicarious instigation (and vicarious classical conditioning) use sirn~lar procedures. The typical experiment is arranged as follows: One of two subjects is assigned the role of model, the other the role of observer, and both are instructed regarding participation in the study. One subject, the model, is actually the experimenter's confederate, whose behavior is trained. Both individuals are connected to one or more monitoring devices, ostensibly to allow measurement of some physiological (e.g., electrodermal) or skeletal (e.g., electromyographic) response, although usually the model's responses are not recorded. The model is then supposedly shocked faradically (i.e., the unconditioned stimulus to the model); however, the model rarely experiences an unconditioned stimulus in actuality. During acquisition trials, a hypothetically neutral stimulus (tone or light) precedes the alleged administration of the unconditioned stimulus to the model. Several unreinforced presentations of the neutral stimulus to the model are interspersed throughout the acquisition phase or are massed following acquisition trials to test for conditioning of the observer's emotional responses, which are usually indicated by the foregoing physiological responses. The observer's responses are recorded throughout the experimental session. The foregoing describes several events that may be functional stimuli for an observer in the vicarious instigation preparation. After briefly reviewing four theoretical positions on vicarious instigation, the logic underlying vicarious processes and the research comparisons implied are discussed. Observational Learning Theory and Vicarious Processes Historical Views Historically, the vicarious concept appears in social psychology under the rubric, sympathy, beginning as early as the writings of Adam Smith (cited in Allport, 1954). McDougall (1908) wrote that emotions could be produced by an object or event itself or the perception of emotion derived from observing the behavior of another individual. His theory relied on instinct for an explanation as to why this sympathetic induction occurred (Allport, 1924). Allport (1924) proposed instead that it was "the knowledge of the conditions affecting A CRITICAL REVIEW OF VICARIOUS INSTIGATION [the performance]," that is, "the whole situation rather than the perception of an emotion in another which aroused the emotion in us" (p. 235). In m o d e m terms, Allport invoked one's conditioning history with respect to fear-provoking situations and stimulus generalization to account for fear arousal at the sight of fear responses by others. Berger's Theory The m o d e m treatment of vicarious processes began with Berger's (1962) formulation, which implicated a conditioning model. An observer was emotionally aroused by perceiving the emotional response of a model, and the observer's emotional response was conditioned to contiguous neutral stimuli. Berger differentiated between vicarious instigation and direct instigation that involved a second organism that supplied in its overt behavior an unconditioned stimulus that directly instigated responding in an observer. His intent was to relegate the concept of vicariousness to emotional responses effected in an observer from the perceived emotional state of the model and no other event. To accomplish this, it was necessary to define experimental procedures that would separate direct (he called it pseudovicarious) instigation from vicarious instigation. This required separating observer responses to the unconditioned response of the model, observer responses to the unconditioned stimulus to the model, observer responses to combinations of these, and observer responses due to expectations that the observer would be treated like the model (e.g., shocked). Residual effects constituted vicarious instigation. Bandura' s Theory Vicarious processes play a central role in social learning theory where vicarious instigation and vicarious classical conditioning are viewed as important to the acquisition of emotional responses (Bandura, 1965, 1969, 1971a, 1971b, 1977). Bandura (1969, 1971a, 1977) noted that another person's emotional response conveyed through vocal, facial, and postural cues could arouse an emotional response in the observer, and this response could then be conditioned to environmental cues. Affective cues acquired their instigating value as a result of similarities between one's 5 experiences and the experiences of others, much as was suggested by Allport (1924). Shared or c o m m o n experiences made the consequences accruing to others predictive of outcomes of one's own experiences (Bandura, 1977). According to this theory, vicarious instigation occurred through an intervening self-arousal process (i.e., the consequences experienced by the model were imagined to occur to oneself) and an emotional response was generated in the observer (Bandura, 1969, 1977). Among several differences from Berger's (1962) view was the extent to which Bandura emphasized the overt unconditioned response of the model as the primary source of instigation for the observer. Recall that in Berger's (1962) formulation, any response on the part of the observer that was elicited by the model's overt actions was considered direct instigation. Bandura wrote that it was precisely because such social cues had acquired instigating properties t h a t the observer was emotionally aroused to begin with (Bandura, 1969, p. 168). From this viewpoint, the unconditioned response of the model was not only a legitimate source of vicarious instigation but the most important source: It is exceedinglyditi~cult to establish preciselythe stimulus sources of the observer'semotionalstate since the behavior of others, depending on its character, undoubtedly augments or reduces the effects of environmental eliciting stimuli. The most convincing demonstration of vicarious instigation is therefore provided under conditions where the observer's emotional responses are elicited entirely by the performer's affectiveexpressions. (Bandura, 1969, p. 169). Aronfreed's Theory Aronfreed also did not agree wholly with Berger's theory, although he did accept major portions of it. For example, Aronfreed (1969) differentiated between empathetic and vicarious experience, whereas Berger did not: Empathy may be used to refer to the child's affective experience when it is elicited by cues of a corresponding affective state [italics added] in the expressive behavior of another person. (p. 292) The term vicarious is more appropriate when the child's affective experience is elicited by its observation of the stimulus events [italics added] which impinge on another person. (p. 292) For Berger, both these sources of stimulation were defined as direct instigation. Bandura's 6 GINA GREEN AND J. GRAYSON OSBORNE position equates with Aronfreed's definition of empathy and not his definition of vicarious. Aronfreed (1969) essentially agreed with Berger regarding his third source o f direct instigation: The concepts of "empathetic'" and "vicarious" can be applied accurately only to those components of the child's affectivestates which are elicited by its perception of the affective experience of others--that is, to those components which are independent of the social stimuli which it perceivesas having a direct impact upon itself. (p. 293) Aronfreed's final statement appears more consonant with Berger's position. Empathic and vicarious stimuli are separated from social stimuli in the situation that could directly arouse the observer, and vicarious processes are due only to "affective experience" (read: responses?) of the model. One is left not knowing precisely what Aronfreed considered the necessary conditions for vicarious instigation, because in the former quote (p. 292) the stimulus to the model is important, whereas in the latter (p. 293) empathy and vicarious instigation appear to blend and the model's response appears important. Because of the clarity o f the difference between empathy and vicarious instigation in his first statement, we consider Aronfreed's position on vicarious instigation to define the unconditioned stimulus to the model as a necessary event. Hygge 's Theory Hygge (1976a) proposed that information to the observer regarding the model's circumstances is the sole necessary condition for vicarious instigation. From this information the observer perceives or infers the emotional response of the model. Events such as the unconditioned stimulus to the model or the model's unconditioned response constitute aspects of the necessary condition if they convey information. Where these events convey no information they do not so function. No independent measures of the presence or absence of information were presented by Hygge, so it is assumed that the definition of an event as containing information is post hoc. Moreover, his theory also implies the possibility of vicarious instigation in the absence of either an unconditioned stimulus to the model or a model's unconditioned response. The foregoing suggests a lack of clarity and agreement among these four theories as to the necessary conditions for vicarious instigation. Although it may be difficult to demonstrate necessary conditions experimentally, the theories imply this possibility. Next, we examine the procedures used to produce vicarious instigation and, by inference from each theory, the research contrasts necessary to conclude what produces experimental effects in this area. Logical Research Contrasts to Examine Vicarious Instigation The determination of the necessary conditions for vicarious instigation is restricted by the possible events in the paradigm and the ease with which they can be operationalized. The major problem is what actually constitutes the unconditioned stimulus to the observer. How can it be isolated? The researcher is restricted further by what each theorist purports are the necessary conditions. Thus, research designs will be dictated by theory. We discuss first the events that can be possibly construed as the unconditioned stimulus to the observer, suggest research contrasts that isolate particular factors, and then examine what particular research outcomes mean to each theory. Unconditioned Stimulus to the Model and the Model's Unconditioned Response In the vicarious instigation paradigm, the two events of consequence are the unconditioned stimulus to the model and the unconditioned response of the model. Additionally, there may be instructions conveyed to the subjects by the experimenter, and these may concern either the unconditioned stimulus to the model, the unconditioned response o f the model, or both. These seem to be the only overt, measurable events that can be operationalized by an experimenter and experienced by an observer in this preparation. There are, therefore, a finite number of contrasts possible among them, namely, all possible combinations of their presence or absence. Logical manipulation of the combinations of the unconditioned stimulus to the A CRITICAL REVIEW OF VICARIOUS INSTIGATION 7 model and unconditioned response of the dition 1 make significantly more instigated model (suspending consideration of instruc- responses than those who experience Conditions for the moment) results in the four tion 2, the instigation relates to the unconexperimental conditions listed in Table 1. ditioned stimulus to the model. If the converse The possible experimental contrasts between occurs, the instigation relates to the unconeach pair of experimental conditions allow a ditioned response of the model, If no differdetermination of two of the important factors ences result from such a contrast, then the in vicarious instigation. The same rationale effects of each event could be similar but no can be applied to the literature to determine conclusion is possible in the absence of adthose studies that identify specific events ditional contrasts (e.g., Contrasts 1 vs. 4 and causing instigation. Subsequently, it may be 2 vs. 4). The outcomes of these contrasts possible to state which of the several theories imply the functional significance of two of appear to explain vicarious instigation best. the events in the instigation preparation. In Table 1, as an example we examine the Question marks in the Functional Significance possible outcomes of an experimental contrast columns of Table 1 suggest intuitively unlikely between Condition 1 and Condition 2. This outcomes for stimulus-response (S-R) theory requires one experimental group to observe .(e.g., where the presence of a presumably an unconditioned stimulus to a m o d e l - - t o excitatory stimulus leads to decreased rewhich stimulus the model makes no re- sponding by the observer, that is, less instis p o n s e - a n d a second experimental group to gation). This logical analysis shows that there observe no unconditioned stimulus to a model are the same number of contrasts where an but the model responds anyway (e.g., Berger, outcome decides in favor of and against the 1962, Exp. II). The outcomes seem reasonably unconditioned response of the model and the clear: If the observers who experience Con- unconditioned stimulus to the model as effective conditions in vicarious instigation. Table 1 Combinations of the Unconditioned Stimulus to the Model (US.,) and Unconditioned Response of the Model (UR,~) in Vicarious Instigation Studies Functional significance Contrasts Outcome Pro Con 1 vs. 2 1> 2 1 < 2 US= UR= URn, US., lvs. 3 NSD 1>3 1 < 3 1 vs. 4 2vs. 3 2 vs. 4 3 vs. 4 NSD 1> 4 1<4 NSD 2>3 2<3 NSD 2> 4 2<4 NSD 3> 4 3<4 NSD ? UR= USm ? UR= UR= US., ? US= UP,= ? US=, UR*, ? ? US,, USm ? UR*, ? US=, URm Note. 1 = US= + no UP,=; 2 --- no US= + UP,=; 3 = US= + UR=; 4 = no USm + no UR*,. USm = unconditioned stimulus to the model; UR= = unconditioned response of the model;? = unexplained outcome;NSD = no significant difference. Theoretical Implications of Experimental Contrasts Berger. There are at least three difficulties with Berger's theoretical position that are fairly obvious, as they were to Berger himself (Berger, 1962, p. 465). First, a substantial number of contrasted conditions is necessary to ascertain vicarious instigation. Second, the theoretical unconditioned stimulus to the observer (i.e., the emotional response of the model) is not directly observable, seemingly buried as it is in the model. Third, the investigator is in the unenviable position of precluding many nonvicarious effects to observe any residual vicarious effects and cannot apparently produce any residual vicarious effects independently of the many nonvicarious effects. Berger's theory gains no support from any of the pair-wise contrasts in Table 1 taken one at a time, because these contrasts result in demonstrating the effectiveness of the unconditioned stimulus to the model, the unconditioned response of the model, or both of these events in the instigation preparation. The unconditioned stimulus to the model 8 GINA GREEN AND J. GRAYSON OSBORNE alone and the unconditioned response of the model alone must produce no effect to support Berger's theory unequivocally. Where they do produce some effect together, Berger theorized an additional effect such that the sum o f the effect of the unconditioned stimulus to the model and the unconditioned response of the model is less than the effect of the unconditioned stimulus to the model and unconditioned response o f the model when they occur together. This appears to require that studies supporting Berger's theory contain: (a) Contrast l versus 4 where the outcome is no significant difference; (b) Contrast 2 versus 3 where Contrast 2 < 3 or there is no significant difference; (c) Contrast 1 versus 3 where Contrast l < 3 or there is no significant difference; (d) Contrast 3 versus 4 where Contrast 3 > 4; and (e) a contrast in which the effect of the unconditioned stimulus to the model from Condition l and the effect of the unconditioned response of the model from Condition 2, if any, are s u m m e d and compared with the total effect from Condition 3. If the latter is greater than the former, then Berger's theory is supported. By extension, outcomes of Contrast l versus 3 where Contrast 1 > 3 or Contrast 2 versus 3 where Contrast 2 > 3 would be p r i m a facie evidence against Berger's theory. Bandura. Bandura's language makes it difficult to determine what he considered the necessary conditions for vicarious instigation, The overt behavior of the model is said to modulate the effects of "environmental eliciting stimuli." From this it can be inferred that some of the model's responses are not necessary to vicarious instigation. This theory appears to separate the model's instrumental responses from the model's affective responses. Seemingly, the escape response of the model from apparent shock is not a valid elicitor of instigation. However, a grimace or an "Ouch!" is, surmising that a grimace or an "Ouch!" is an affective expression. The escape response, however, affects how arousing the grimace or "Ouch!" is. The theory rules out the remaining events as necessary to vicarious instigation. As causal conditions, the theory excludes: (a) observer responses to the unconditioned stimulus to the model that are absent from model affective responses; (b) observer responses to the in- strumental response of the model that are absent from model affective responses; and (c) interactions between these two conditions. Bandura's theory also requires a demonstration that the model's affective responses are necessary to instigation (i.e., presence/absence of the model's affective reactions is congruent with presence/absence of the observer's emotional response). Because Bandura noted that the observer's emotional response can also be modulated by these other events, combinatorial effects should be sought between the observer's emotional response on one hand and the model's affective and instrumental responses on the other; between the unconditioned stimulus to the model and the model's affective responses on one hand and the observer's emotional response on the other; and between the unconditioned stimulus to the model, the model's instrumental response, and the model's affective responses on one hand and the observer's emotional response on the other. Bandura's position is methodologically more tenable than Berger's because the necessary condition for vicarious instigation is overt. However, Bandura's position raises additional questions. The implicit division o f the model's response into instrumental and affective classes necessitates a valid means of separating and defining the two classes of responses. The difficulties are compounded if the theory does not consider these responses to be mutually exclusive (i.e., to belong only to one of these classes and not the other). I f they are mutually exclusive, then independent methods to determine which response is which are needed. If the responses are not mutually exclusive, then research designs that separate effects due to the model's response are critical to the subsequent findings of vicarious instigation. In this vein, Bandura's theory would be at least partially supported by any enhancement of instigated responding due to the unconditioned response of the model (Table 1). This support would be weak if the studies from which it was inferred did not contain conditions that ruled out any enhancement due to the unconditioned stimulus to the model (e.g., Contrast 1 vs. 2, Table 1). Thus, a positive result for Contrast 1 versus 3 (i.e., a finding that l < 3) would not be unequivocal A CRITICAL REVIEW OF VICARIOUS INSTIGATION support for Bandura's theory because further contrasts would be required to show whether there was any effect due to the unconditioned stimulus to the model. Aronfreed. Although Aronfreed believed that empathetic and vicarious experiences often occur together, each implies different experimental tests for its examination. When vicarious instigation is examined from Aronfreed's position, it is necessary to show that the observer's emotional arousal is due only to the unconditioned stimulus to the model. This necessitates an experimental test that removes responding by the model (presumably both instrumental and affective). Thus, Aronfreed's theory of vicarious instigation gains support from any of the contrasts in Table 1 that favor an effect due to the unconditioned stimulus to the model. As with the Bandura theory, some of the outcomes are more equivocal than others. For example, Contrast 2 versus 3 in which the result is Contrast 2 < 3 does not rule out the possibility that the effects of the unconditioned response by the model contribute something to the result. The outcome of Contrast 1 versus 2 is much less equivocal. Hygge. Hygge's theory suffers not only from the lack of a method to predetermine informativeness (the necessary condition), but also from similar difficulties attendant to Berger's theory. Where there is an overt unconditioned stimulus to the model and/or an unconditioned response by the model, one or both of these stimuli could produce emotional arousal by the observer. If either event does so, it must be determined that the information these events carry is responsible. Otherwise, the instigation is direct. Thus, this theory also must separate the observed instigation into that which is direct and that which remains, and it must do so in the possible absence of any independent method of determining the presence or absence of information in the overt event. Further, direct instigation is implied where there is information conveyed to the observer in the unconditioned stimulus to the model or the model's unconditioned response but where the observer does not infer the model's emotional response. This aspect, internal to the observer, is also a difficulty with Bandura's and Aronfreed's theories, both of which re- 9 quire active cognitions by the observer that may be difficult to measure. Yet more difficult is the case in which there is no information to the observer, no unconditioned stimulus to the model, and no response by the model of any kind, but the observer infers an emotional response of the model (i.e., a kind of false positive error). This would appear to constitute vicarious instigation for Berger and Hygge but not for Bandura and possibly not for Aronfreed. As for Berger no one pair-wise contrast in Table 1 suffices to support Hygge's theory. Given the absence of a valid method to determine information conveyed independently of the event conveying it, no one or a combination of the contrasts in Table 1 can disprove the theory either. (Recall the latter was not the case for Berger.) A specific test of the Hygge theory must either not be possible or require something other than the manipulation of the events in Table 1. One possibility would be to convey information about the unconditioned stimulus to the model or the unconditioned response o f the model via instructions to the observer in the absence of either an actual unconditioned stimulus to the model or an observable unconditioned response of the model. Instructions to observers can constitute another set of potential instigating events. In most studies, verbal or written instructions about the unconditioned stimulus to the model, the unconditioned response of the model, or both were included in subject briefings prior to their participation in the experiment (e.g., Geer & Jarmecky, 1973). The possibility that instructions should be considered independent variables in their own right is suggested by Hygge's theory, if information at least partially constitutes instructions. Although all possible combinations of instructions with the unconditioned stimulus to the model and the unconditioned response of the model and contrasting conditions could be described, it is clear from the following analysis of the literature that most investigators did not consider instructions a major independent variable. In sum, experimental tests appear at least possible to determine the existence of vicarious instigation. For Berger, these tests must show an extra effect of the combination of l0 GINA GREEN AND J. GRAYSON OSBORNE the unconditioned stimulus to the model and the unconditioned response of the model. For Aronfreed, these tests must show the effect of the unconditioned stimulus to the model only. For Bandura, they must show the effect of the unconditioned response of the model only. For Hygge, these tests must show the effect of information only. Next we turn to the literature to examine the presence and absence of these experimental tests in an attempt to determine what constitutes the effective stimulus to the observer and whether or not one theory is more substantiated by the outcomes in the literature than others. Analysis o f the Evidence Presence in the Literature of Events Possibly Functioning as the Unconditioned Stimulus to the Observer At least one theorist (Berger) considered anticipation of the unconditioned stimulus an important source of direct, rather than vicarious, instigation for observers. In 19 of 30 studies, all observers were instructed specifically that they would not experience the unconditioned stimulus. In one study observers were told they would receive the unconditioned stimulus (Craig, 1968), and in two studies one group of subjects was told to expect the unconditioned stimulus, whereas another was informed they would not experience the unconditioned stimulus (Bandura & Rosenthal, 1966; Craig & Lowery, 1969). One study did not involve an unconditioned stimulus to any observer or model (Berger & Hadley, 1975), and in another observers saw slides ostensibly representing unconditioned stimuli for the model, but observers were not instructed beforehand that they would experience the unconditioned stimulus (Hygge & Ohman, 1978). For the remaining six studies, the presence or absence of such instructions could not be ascertained. Table 2 shows whether the unconditioned stimulus to the model and unconditioned response of the model in each study were indicated to observers in some way other than by instructions, and whether instructions about these events were delivered to observers. For the unconditioned stimulus to the model, an indicator of its delivery (such as a light signaling onset of shock) or the clear presence of an actual unconditioned stimulus to the model (such as a container o f ice for a cold pressor test) was considered evidence of an unconditioned stimulus to the model. In the case of the unconditioned response of the model, an observable movement (e.g., a r m jerk), facial expression, or vocalized response to the ostensive unconditioned stimulus qualified. There are m a n y possible combinations of events when instructions are included. Separation of the unconditioned stimulus to the model and/or the unconditioned response of the model in the instigating situation from instructions about those events may be necessary to determine empirical support for Hygge's theory. From Table 2 it can be seen that 12 studies did not confound an indicated unconditioned stimulus to the model or unconditioned response of the model with instructions about either. Included in that group are: one study involving instructions about the unconditioned stimulus to the model and no unconditioned response by the model, either indicated or instructed (Hygge & Ohman, 1976b); one study in which both events were indicated but not otherwise instructed (Craig & Lowery, 1969); two studies in which both events were indicated but no instructions about them were provided to observers (Bernal & Berger, 1976; Yamaguchi, Harano, & Egawa, 1978); and eight studies that provided instructions but no other indicator of the unconditioned stimulus to the model, while the model made an unconditioned response and there were no instructions about that event (Aniskiewicz, 1979; Bandura & Rosenthal, 1966; Brown, 1974; Ogston & Davidson, 1972; Silver & Greco, 1975; Sutker, 1970; Vaughan & Lanzetta, 1980). Two studies confounded the unconditioned stimulus to the model but either involved no unconditioned response of the model or provided the reader with no information about it (Craig, 1968; Hygge & Ohman, 1976a). In five studies the unconditioned stimulus to the model was both indicated and instructed, whereas the unconditioned response of the model was only instructed (Craig & W e i n stein, 1965; Hygge, 1976b, 1978; Hygge & Dimberg, 1982; Hygge & Ohman, 1978). One study confounded instructions about an in- A CRITICAL REVIEW OF VICARIOUS INSTIGATION dicator of the unconditioned stimulus to the model, whereas there was an observable unconditioned response by the model but no instructions about it (Vaughan & Lanzetta, 1981). For these eight studies, it cannot be determined whether potential effects of the unconditioned stimulus to the model are due to some indication of the delivery of that stimulus, instructions about it, or both. Effects of the unconditioned response of the model could be attributed cleanly to either an indi- 11 cator of its occurrence or instructions in six of these studies. This group of studies constitutes over one quarter (27%) of the vicarious instigation literature. For almost another one third of the studies in this area (9/30), the unconditioned stimulus to the model and the unconditioned response of the model were both indicated and instructed (Baron, 1970a, 1970b; Berger, 1962; DiLollo & Berger, 1965; Geer & Jarmecky, 1973; Kravetz, 1970, 1974). In these cases, Table 2 ,' Experimental Events Possibly Serving as the Unconditioned Stimulus to the Observer USm Study Indicator UR= Instructions Movement Instructions Studies with unconfounded events (indicators or instructions) Hygge & Ohman (1976b) Craig & Lowery (1969) Bernal & Berger (1976) Yamaguchi, Harano, & Egawa (1978) Bandura & Rosenthal (1966) Sutker (1970) Ogston & Davidson (1972) Brown (1974) Silver & Greco (1975) Aniskiewicz (1979) Vaughan & Lanzetta (1980; Exp. 1) Vaughan & Lanzetta (1980; Exp. II) X X X X X X X X X X X" X X X X X X X X X X Studies confounding indicated events and instructions Craig (1968) Hygge & Ohman (1976a) Craig & Weinstein (1965) Hygge (1976b) Hygge (1978) Hygge & Ohman (1978) Hygge & Dimberg (1982) Vaughan & Lanzetta ( 1981) Berger (1962; Exp. I) Berger (1962; Exp. II) Berger (1962; Exp. III) DiLollo & Berger (1965) Baron ( ! 970a) Baron (1970b) Geer & Jarmecky (i 973) Kravetz (1974) Kravetz (1970) X × × X X X X X × × × X X X × × X X X × xb Xb × X X × × × X × × ×b X X X X X X X X X X X × X X X × X × X × X × × X ×d ×e Note. For Hygge and Ohman (1976b), there was no independent indicator of U S m delivery, but model actually received US. USm = unconditioned stimulus to model; URm = unconditioned response of model. • Observers in one condition were instructed about model's movement, but that movement was not a URm (no US preceded it). b Instructions described different intensities of US~ for each of two or more experimental groups, c~bserversweret~dt~n~t~model~sfacialresponses~butweregivenn~furtherinstructi~nsab~utURm, dOne group of observers was told that model's response was due to shock; another group was told that it was the result of model's viewing slides of nonsense syllables. • One group of observers was told that model's response was due to shock; another group (no shock--heart rate change) was told that model's response was the result of exercise. 12 GINA G R E E N A N D J. GRAYSON OSBORNE outcomes cannot be isolated as due to either the unconditioned stimulus to the model, the unconditioned response of the model, or instructions about either event. Thus, over half of the studies in this literature could not separate effects due to the unconditioned stimulus to the model from instructions about it, and almost one third could not make the analogous distinction for the unconditioned response of the model. What is the Unconditioned Stimulus to the Observer? Table 3 summarizes studies with designs that permit at least one of the contrasts described in Table 1, findings of statistical significance for the contrasts examined, and theoretical inferences, if any, possible from each finding. Studies are classified as to whether they confounded an indicated unconditioned stimulus to the model and/or unconditioned response of the model with instructions about either event (refer to Table 2). Unconditioned response of the model. Resuits from any contrasts indicating that the unconditioned response of the model is the effective stimulus for vicarious instigation would support Bandura's theory, at least partially. A Contrast 1 versus 3 finding in which 1 < 3 implies an effect of the unconditioned response of the model but permits no inference about the role of the unconditioned stimulus to the model. This is because it compares presence and absence of the unconditioned response of the model while holding presence of the unconditioned stimulus to the model constant. Six studies reported such findings (see Table 3), but two of those (Baron, 1970a; Yamaguchi et al., 1978) also reported no significant differences between Conditions 1 and 3, a result that suggests no effect of the unconditioned response of the model. Two studies by Kravetz (1970, 1974) also resulted in no significant differences between Conditions 1 and 3, contrary to Bandura's position. Note that six of the eight studies reporting results of Contrasts 1 versus 3 confounded indicated events and instructions (Baron, 1970a, 1970b; Berger, 1962, Exp. II; DiLollo & Berger, 1965; Kravetz, 1970, 1974). Stronger tests of Bandura's theory are found in studies that support effects of the uncon- ditioned response of the model and simultaneously rule out effects due to the unconditioned stimulus to the model. Significant findings from Contrast 1 versus 2 where the outcome is 1 < 2 or Contrast 2 versus 4 where the outcome is 2 > 4 would accomplish this. Alternately, a finding from Contrast 1 versus 3 where the outcome is 1 < 3 (evidence for an effect of the unconditioned response of the model) in combination with a nonsignificant finding from either Contrast 1 versus 4 or Contrast 2 versus 3 (suggesting no effect of the unconditioned stimulus to the model) in the same study would also constitute strong evidence for Bandura's position. No study reported an outcome of 1 < 2, but Berger (1962, Exp. II) reported outcomes of 2 > 4 as well as 1 < 3 (pro unconditioned response of the model) and no significant difference in Contrast 1 versus 4 (con unconditioned stimulus to the model). DiLollo and Berger (I 965) likewise obtained an outcome of 1 < 3 and no difference in Contrast 1 versus 4. These studies appear to have demonstrated effects of the unconditioned response of the model in the absence of effects of the unconditioned stimulus to the model in support of Bandura's theory. However, both studies confounded actual occurrence of both events with instructions, producing interpretation problems discussed earlier in this article. Unconditioned stimulus to the model. Partial support for Aronfreed's position comes from studies contrasting Conditions 2 and 3. Presence/absence of the unconditioned stimulus to the model is manipulated, whereas presence/absence of the unconditioned response of the model is held constant. This contrast does not factor out effects due to the unconditioned response of the model. All studies reporting results of Contrast 2 versus 3 confounded indicated events and instructions, so although five of six appear to support Aronfreed's position (Berger, 1962, Exps. I, II, III; DiLollo & Berger, 1965; Kravetz, 1970), no conclusions are possible as to the relative contribution of the actual unconditioned stimulus to the model, instructions about the unconditioned stimulus to the model, or both. The strongest case for Aronfreed's position obtains where effects of the unconditioned stimulus to the model are found, whereas 13 A CRITICAL REVIEW OF VICARIOUS INSTIGATION Table 3 Experimental Contrasts and Theoretical Evidence In Cerred Study/findings Theoretical evidence Studies with unconfounded events (indicators or instructions) Craig & Lowery (1969) Equiv. Bandura, Aronfreed 3 > 4a Bernal & Berger (1976) 1 < 3b Pro Bandura 1> 4 Pro Aronfreed 3> 4 Equiv. Bandura, Aronfreed Hygge & Ohman (1976b) 1 > 4c Pro Aronfreed, Hygge Yamaguchi et al. (1978) 1 <3 d 1 vs. 3 (NSD) Vaughan & Lanzetta (1980; Exp. I) 3 > 4c Vaughan & Lanzetta (1980; Exp. II) 3>4 • Equiv. Bandura Equiv. Bandura, Aronfreed Equiv. Bandura, Aronfreed Studies confounding indicated events and instructions Berger (1962; Exp. I) 2 < 3f Berger (1962; Exp. II) 1 < 3B 1 vs. 4 (NSD) 2 < 3s 2>4 s 3>4 s Berger (1962; Exp. III) 2<3 Craig & Weinstein (1965) 1 vs. 4 (NSD) Pro Aronfreed Pro Bandura Con Aronfreed Pro Aronfreed Pro Bandura, con Aronfreed Equiv. Bandura, Aronfreed Pro Aronfreed Con Aronfreed Note. NSD = no significant difference; Equiv. Study/findings Theoretical evidence Studies confounding indicated events and instructions (continued) DiLollo & Berger (1965) 1 vs. 2 (NSD) h 1< 3 1 vs. 4 (NSD) 2 < 3 2 vs. 4 (NSD) h 3 > 4b Baron (1970a) 1 <3 i 1 vs. 3 (NSD) i Baron (1970b) 1 < 3iJ Kravetz (1970) 1 vs. 2 (NSD) 1 vs. 3 (NSD) 2 < 3 Kravetz (1974) 1 vs. 2 (NSD) h 1 vs. 3 (NSD) 2 vs. 3 (NSD) Hygge (1976b) 3 > 4k Hygge & Ohman (1976a) 1 vs. 4 (NSD) Hygge (1978) 3> 4 Hygge & Ohman (1978) 3 > 4' 3 vs. 4 (NSD) 1 Vaughan & Lanzetta (1981) 3 > 4m Hygge & Dimberg (1982) 3 > 4f None Pro Bandura Con Aronfreed Pro Aronfreed Con Bandura, pro Aronfreed Equiv. Bandura, Aronfreed Equiv. Bandura Pro Bandura None Con Bandura Pro Aronfreed None Con Bandura Con Aronfreed Equiv. Bandura, Aronfreed Con Aronfreed Equiv. Bandura, Aronfreed Equiv. Bandura, Aronfreed Equiv. Bandura, Aronfreed Equiv. Bandura, Aronfreed equivocal finding. " F o r USm, contrast was between subjects in one group that was instructed there was a USm, and a second group that was instructed there was no US,~; neither group saw an actual US or indicator of US delivery. Both groups saw URm. b Significant for instigation only; no significant difference for conditioning, c Instructions about USm but no indicated USm; no URm, either instructed or indicated, d Significant for one dependent measure (pulse amplitude); no significant difference for second dependent measure (heart rate change). • In differential conditioning~ observers were instructed that one conditioned stimulus was followed by USm. There was no indication of USm delivery. All observers saw URm. Conditioning effects significant for skin conductance response in first of three intervals (in long interstimulus interval conditioning procedure) only; not significant for electromyographic level, f Significant for test trials only; no significant difference for adaptation and conditioning trials, s Significant for conditioning trials only; no significant difference for adaptation and test trials, hNonsignificant finding inferred; data were not directly reported, i Contrast 1 vs. 3 confounded because groups under the two conditions i'eceived different instructions about USm. Effects could be due t o US m instructions, indicated URm, or both. One experimental group was in Condition I (US~ + URm), whereas two groups were in Condition 3 (USm + URm). Contrast 1 vs. 3 was significant for only one group. J Presence of UR~ (Condition 3) appeared to decrease observer's conditionability relative to subjects in Condition 1 (USm + no URm) only if observer's task was independent of US~ delivery, k Significant differential conditioning only for groups who were instructed that model's response to the purported US was strong (rather than weak), l Significant conditioning for first interval anticipatory responses; not significant for second interval anticipatory responses, m Significant differential conditioning (Contrast 3 vs. 4) reported for all observers, but effect due primarily to data for one group that was instructed to amplify their facial responses to the model's pain display. 14 GINA G R E E N A N D J. GRAYSON OSBORNE effects of the unconditioned response of the model are precluded within the same study. Either of two outcomes--Contrast 1 versus 2 where 1 > 2 or Contrast 1 versus 4 where 1 > 4--accomplishes both requirements and lends unequivocal support to this theory. An outcome of Contrast 2 versus 3 where 2 < 3 combined with nonsignificant results from either Contrast 1 versus 3 or Contrast 2 versus 4 in the same study would also be evidence in favor of Aronfreed's theory. None of the studies in the literature reported an outcome o f Contrast 1 versus 2 where 1 > 2, but Bernal and Berger (1976) reported a significant outcome o f Contrast 1 versus 4 where 1 > 4. Notably, this was one of only two studies that did not confound either the unconditioned stimulus to the model or the unconditioned response o f the model with instructions. Hygge and Ohman (1976b) also reported an outcome of Contrast 1 versus 4 where 1 > 4, but in their study presence/ absence of the unconditioned stimulus to the model was manipulated via instructions alone. Their results support Hygge's (1976a) theory, but are also qualified support for Aronfreed's theory, because information about the unconditioned stimulus to the model produced vicarious instigation. Four studies (Berger, 1962, Exp. II; Craig & Weinstein, 1965; DiLollo & Berger, 1965; Hygge & Ohman, 1976a) reported no significant difference from Contrast I versus 4, disconfirming Aronfreed's hypothesis that the unconditioned stimulus to the model is the effective stimulus. Two other studies that contained instructional confounds obtained results that suggest effects of the unconditioned stimulus to the model in the absence of effects due to the unconditioned response of the model: DiLollo and Berger (1965) reported an outcome of Contrast 2 versus 3 where 2 < 3 with a finding of no significant difference from Contrast 2 versus 4; and Kravetz (1970) found an outcome of Contrast 2 versus 3 where 2 < 3 and no significant difference from Contrast 1 versus 3. Unconditioned emotional response of the model. Several studies reported results of Contrast 3 versus 4 (Berger, 1962, Exp. II; Bernal & Berger, 1976; Craig & Lowery, 1969; DiLollo & Berger, 1965; Hygge, 1976b, 1978; Hygge & Dimberg, 1982; Hygge & Ohman, 1978; Vaughan & Lanzetta, 1980, 1981). Not surprisingly, all reported statistically significant differences between the two conditions on at least one dependent measure, with more instigation occurring for Condition 3 than for Condition 4 regardless of whether events were indicated, instructed, or any combination thereof. Such results provide only partial support for the two theories that stress a relatively more important role for either single element in the vicarious instigation situation, the unconditioned stimulus to the model (Aronfreed) or the unconditioned response of the model (Bandura). In addition, they do not unequivocally support Berger's contention that both events are necessary to the observer's inference about the model's unconditioned emotional response, because Contrast 3 versus 4 alone does not permit comparison of effects of the presence of both events with effects attributable to either event alone. Stronger evidence for Berger's position comes from studies that find no or minimal effects attributable to either the unconditioned stimulus to the model or unconditioned response of the model alone and significant instigation when both are present. Only two studies (Berger, 1962, Exp. II; DiLollo & Berger, 1965) reported results of experimental contrasts that met these requirements. Berger (1962, Exp. II) found no significant difference from Contrast 1 versus 4 (suggesting no effect of the unconditioned stimulus to the model alone), and from Contrasts 1 versus 3 and 2 versus 3 found outcomes where 1 < 3 and 2 < 3, suggesting relatively small effects of the unconditioned stimulus to the model and the unconditioned response of the model, respectively. These results, in combination with a significant difference from Contrast 3 versus 4 where the outcome was 3 > 4, imply greater instigation to the joint occurrence of the unconditioned stimulus to the model and the unconditioned response of the model than to the occurrence of either event alone. DiLollo and Berger (1965) reported findings identical to Berger's (1962, Exp. II), with additional evidence against effects of the unconditioned response of the model alone inferred from their finding of nonsignificance from Contrast 2 versus 4, and a similar finding from Contrast 1 versus 2 that sug- A CRITICAL REVIEW OF VICARIOUS INSTIGATION gested that neither the unconditioned stimulus to the model nor the unconditioned response of the model alone produced instigation. Bernal and Berger (1976) reported an outcome of Contrast 1 versus 3 where 1 < 3 and an outcome of Contrast 3 versus 4 where 3 > 4, indicating greater instigation when both unconditioned stimulus to the model and unconditioned response of the model were present than when only an unconditioned stimulus to the model was present and both events were absent. However, they also reported an outcome of Contrast 1 versus 4 where 1 > 4, a finding that suggests significant effects accruing to the unconditioned stimulus to the model alone and detracts from this study's support for Berger's theory. Information as the unconditioned stimulus to the observer. As alluded to previously, the only study attempting to examine information as the sole necessary condition for vicarious instigation is Hygge and Ohman (1976b). In this study there were neither indicators of an unconditioned stimulus nor responses of a model. The only event was instructions about the unconditioned stimulus to the model. They reported an outcome of Contrast 1 versus 4 in which 1 > 4, arguing that the instructions contributed information and were sufficient to produce vicarious instigation. Summary of Evidence The foregoing analysis was drawn from a truncated sample of the vicarious instigation literature, because only 21 of 30 studies contrasted experimental conditions; the remainder did not examine necessary conditions. O f the remaining studies, five manipulated subjects' state in some way (e.g., arousal level) prior to having them observe an aversively stimulated model (e.g., Bandura & Rosenthal, 1966), whereas the others compared subjects who differed on some other dimension (e.g., sociopaths vs. nonsociopaths; Aniskiewicz, 1979; Sutker, 1970). From Table 3, it may be inferred that the available data do not provide unequivocal support for any single theory, although the sample of studies that made the contrasts necessary to test Berger's and Hygge's theories is insufficient for drawing any conclusions. Further, much of the existing evidence comes 15 from studies that confounded the putative instigating events with instructions about them. Where such confounding occurs, interpretation is difficult, especially when the unconditioned stimulus to the model, the unconditioned response of the model, and instructions covary. In such cases it is impossible to determine whether reported effects are attributable to the actual occurrence of an unconditioned stimulus of the model or unconditioned response of the model, instructions, or both. The single study that used indicated events without instructions and conducted multiple experimental contrasts (Bernal & Berger, 1976) found significant instigation to the unconditioned stimulus to the model alone, the unconditioned response of the model alone, and both events together. One study finding effects due to instructions about the unconditioned stimulus to the model (Hygge & Ohman, 1976b) supported Hygge's theory. Several studies contained instructions about the unconditioned stimulus to the model with no other indication of an unconditioned stimulus to the model, and an indicated unconditioned response of the model unconfounded by instructions; or confounded instructions and the unconditioned stimulus to the model, whereas the unconditioned response to the model was only instructed (no overt model response). However, all of those studies reported only outcomes of Contrast 3 versus 4 so they did not help determine which of the two contributed more to the obtained instigation effects. Conclusions No single theory of vicarious instigation is supported unequivocally by the available data. With regard to Bandura's and Aronfreed's theories, the single study that contained multiple contrasts and no instructional confounds (Bernal & Berger, 1976) obtained one finding in weak support of Bandura's theory and one finding unequivocally favoring Aronfreed's position. The single study that could make unconfounded comparisons to determine effects of instructions (Hygge & Ohman, 1976b) reported results of only one experimental contrast, which provides the only support in the literature for Hygge's (1976a) theory. Evidence related to Berger's theory comes only 16 GINA GREEN AND J. GRAYSON OSBORNE from studies containing multiple contrasts (i.e., Berger, 1962, E x p . II; B e r n a l & Berger, 1976; D i L o l l o & Berger, 1965), a n d t h o s e results a r e m i x e d . T h e effects r e p o r t e d in t h i s l i t e r a t u r e a r e d u e t o a n y o n e o f several c o n d i t i o n s : a n u n c o n d i t i o n e d s t i m u l u s to t h e m o d e l , a n u n c o n d i t i o n e d r e s p o n s e o f t h e m o d e l , o r ins t r u c t i o n s a b o u t either. It is n o t p o s s i b l e t o d e t e r m i n e w h i c h is m o r e i m p o r t a n t . G i v e n the small portion of the literature that successfully isolates t h e effective u n c o n d i t i o n e d s t i m u l u s in v i c a r i o u s i n s t i g a t i o n a n d t h e equivocal nature of the findings regarding t h a t effective u n c o n d i t i o n e d s t i m u l u s , o b s e r vational learning theory garners only weak support from the existing empirical evidence for t h e v i c a r i o u s i n s t i g a t i o n c o n s t r u c t . References Allport, E H. (1924). Socialpsychology. Boston: Houghton Mifflin. Allport, G. W. (1954). The historical background of modern social psychology. In G. Lindzey & E. Aronson (Eds.), The handbook of social psychology (Vol. 1, pp. 3-56). Reading, MA: Addison-Wesley. Aniskiewicz, A. S. (1979). Autonomic components of vicarious conditioning and psychopathy. Journal of Clinical Psychology, 35, 60-67. Aronfreed, J. (1969). The concept of internalization. In D. A. Goslin (Ed.), Handbook of socialization theory and research (pp. 263-323). Chicago: Rand McNally. Bandura, A. (1965). Vicarious processes: A case of notrial learning. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 2, pp. 1-55). New York: Academic Press. Bandura, A. (1969). Principles of behavior modification. New York: Holt, Rinehart & Winston. Bandura, A. (197 l a). Analysis of modeling processes. In A. Bandura (Ed.), Psychological modeling: Conflicting theories (pp. 1-62). New York: Aldine-Atherton. Bandura, A. ( 1971 b). Vicarious and self-reinforcement processes. In R. Glarer (Ed.), The nature of reinforcement (pp. 228-278). New York: Academic Press. Bandura, A. 0977). Social learning theory. Englewood Cliffs, N J: Prentice-Hall. Bandura, A., & Rosenthal, T. L. (1966). Vicarious classical conditioning as a function of arousal level. Journal of Personality and Social Psychology, 3, 354-362. Baron, R. A. (1970a). Effects of magnitude of model's apparent pain on observer reaction time. Psychonomic Science, 20, 229-230. Baron, R. A. (1970b). Magnitude of model's apparent pain and ability to aid the model as determinants of observer reaction time. PsychonomicScience, 21, 196197. Berger, S. M. (1962). Conditioning through vicarious instigation. Psychological Review, 69, 450-466. Berger, S. M., & Hadley, S. W. (1975). Some effects of a model's performance on an observer's electromyographic activity. American Journal of Ps3~'hology, 88, 263-276. Bernal, G., & Berger, S. M. (1976). Vicarious eyelid conditioning. Journal of Personality and Social Psychology, 34, 62-68. Brown, I. (1974). Effects of perceived similarity on vicarious emotional conditioning. Behavior Research and Therapy, 12. 156-173. Craig, K. (1968). Physiological arousal as a function of imagined, vicarious, and direct stress experiences. Journal of Abnormal Psychology, 73, 513-520. Craig, K. D., & Lowery, H. J. (1969). Heart-rate components of conditioned vicarious autonomic responses. Journal of Personality and Social Psychology, 11, 381387. Craig, K. D., & Weinstein, M. S. (1965). Conditioning vicarious affective arousal. Psychological Reports, 17, 955-963. DiLollo, V., & Berger, S. M. (1965). Effects of apparent pain in others on observer's reaction time. Journal of Personality and Social Psychology, 2, 573-575. Geer, J., & Jarmecky, L. (1973). The effect of being responsible for reducing another's pain on subjects' response and arousal. Journal of Personality and Social Psychology, 26, 232-237. Gewirtz, J. L. (1971). The roles of overt responding and extrinsic reinforcement in "self-" and "vicarious reinforcement" phenomena and in "observational learning" and imitation. In R. Glaser (Ed.), The nature of reinforcement (pp. 279-309). New York: Academic Press. Gormezano, I., & Kehoe, E. J. (1975). Classical conditioning: Some methodological-conceptual issues. In W. K. Estes (Ed.), Handbook of learning and cognitive processes (Vol. 2, pp. 143-179). Hillsdale, NJ: Erlbaum. Hygge, S. (1976a). Emotional and electrodermal reactions to the suffering of another: Vicarious instigation and vicarious classical conditioning. (Studia Psycbologica Upsaliensia 2). Uppsala, Sweden: Acta Universitatis Upsaliensis. Hygge, S. (1976b). Information about the model's unconditioned stimulus and response in vicarious classical conditioning. Journal of Personality and Social Psychology, 33. 764-771. Hygge, S. (1978). The observer's acquaintance with the model's stimulus in vicarious classical conditioning. Scandinavian Journal of Psychology, 19, 231-239. Hygge, S., & Dimberg, U. (1982). The differentiation between vicarious instigation and classical conditioning of electrodermal responses. Manuscript submitted for publication. Hygge, S., & Ohman, A. (1976a). Conditioning of dectrodermal responses through vicarious instigation and through perceived threat to a performer. Scandinavian Journal of Psychology, 17, 65-72. Hygge, S., & Ohman, A. (1976b). The relation of vicarious to direct instigation and conditioning of electrodermal responses. Scandinavian Journal of Psychology, 17, 217-222. Hygge, S., & Ohman, A. (1978). Modeling processes in the acquisition of fears: Vicarious electrodermal conditioning to fear-relevant stimuli. Journal of Personality and Social Psychology, 36, 271-279. Kanekar, S. (1976). Observational learning of attitudes: A CRITICAL REVIEW OF VICARIOUS INSTIGATION A behavioral analysis. European Journal of Social Psychology, 6, 5-24. Kravetz, D. E (1970). Heart rate as a minimal cue for the occurrence of conditioned vicarious autonomic responses. Psychonomic Science, 19, 90-91. Kravetz, D. E (1974). Heart rate as a minimal cue for the occurrence of vicarious classical conditioning. Journal of Personality and Social Psychology, 29, 125131. Krebs, D. L. (1970). Altruism: An examination of the concept and a review of the literature. Psychological Bulletin, 73, 258-302. McDougall, W. (1908). An introduction to social psychology. London: Methuen. Ogston, K. M., & Davidson, P. O. (1972). The effects of cognitive expectancies on vicarious conditioning. British Journal of Social and Clinical Psychology, 11, 126134. Ohman, A., & Hygge, S. (1977). When apple definitions become orange realities in banana theories: Epistemological and psychophysiological considerations in Lindahl's analysis of vicarious instigation research. Scandinavian Journal of Psychology, 18, 92-102. Schwartz, G. E., & Shapiro, D. (1973). Social psychophysiology. In W. E Prokasy & D. C. Raskin (Eds.), Electrodermal activity in psychological research (pp. 377-416). New York: Academic Press. Silver, A. I., & Greco, T. S. (1975). A comparison of the effects of vicariously instigated classical conditioning 17 and direct classical conditioning procedures. Pavlovian Journal of Biological Sciences, I0, 216-225. Social Science Citation Index. (1969-1983). Philadelphia: Institute for Scientific Information. Sutker, E B. (1970). Vicarious conditioning and sociopathy. Journal of Abnormal Psychology, 76, 380-386, Thelen, M. H., & Rennie,. D. L. (1972). The effect of vicarious reinforcement on imitation: A review of the literature. In B. A. Maher (Ed.), Progress in experimental personality research (pp. 83-108). New York: Academic Press. Vaughan, K. B., & Lanzetta, J. T. (1980). Vicarious instigation and conditioning of facial expressive and autonomic responses to a model's expressive display of pain. Journal of Personality and Social Psychology,, 38, 909-923. Vaughan, K. B., & Lanzetta, J. T. (1981). The effect of modification of expressive displays on vicarious emotional arousal. Journal of Experimental Social Psychology, 17, 16-30. Venn, J. R., & Short, J. G. (1973). Vicarious classical conditioning of emotional responses in nursery school children. Journal of Personality and Social Psychology, 28, 249-255. Yamaguchi, S., Harano, K., & Egawa, B. 0978). Effects of differentially modeled stimuli on vicarious autonomic arousal. Perceptual and Motor Skills, 46, 643-650. Received April 25, 1984 •