The Role of Subject Awareness in Classical Conditioning: a Case of Opposing Ontologies and Conflicting Evidence

ABSTRACT - The role of subject awareness has been prominent in discussions of classical conditioning of human subjects for over three decades. Skeptics of classical conditioning contend, on the one hand, that the presence of subject awareness violates the conditioning model's ontological premises, and, on the other, that the presence of subject awareness implies the likelihood that putative conditioning effects are actually attributable to demand artifacts. This paper confronts both of these challenges.

Citation:

Terence A. Shimp (1991) ,"The Role of Subject Awareness in Classical Conditioning: a Case of Opposing Ontologies and Conflicting Evidence", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 158-163.

Advances in Consumer Research Volume 18, 1991      Pages 158-163

THE ROLE OF SUBJECT AWARENESS IN CLASSICAL CONDITIONING: A CASE OF OPPOSING ONTOLOGIES AND CONFLICTING EVIDENCE

Terence A. Shimp, University of South Carolina

ABSTRACT -

The role of subject awareness has been prominent in discussions of classical conditioning of human subjects for over three decades. Skeptics of classical conditioning contend, on the one hand, that the presence of subject awareness violates the conditioning model's ontological premises, and, on the other, that the presence of subject awareness implies the likelihood that putative conditioning effects are actually attributable to demand artifacts. This paper confronts both of these challenges.

INTRODUCTION

The past decade in consumer research has witnessed both increased activity and greater sophistication in the empirical study and theoretical/critical examination of classically conditioned learning (e.g., Allen and Janiszewski 1989; Allen and Madden 1985; Bierley, McSweeney, and Vannieukerk 1985; Gorn 1982; Gorn, Jacobs, and Mana 1987; Kahle, Beatty, and Kennedy 1987; Kellaris and Cox 1989; Macklin 1986; McSweeney and Bierley 1984; Nord and Peter 1980; Stuart, Shimp, and Engle 1987). However, empirical results claiming to demonstrate the classical conditioning of consumer attitudes have been challenged by skeptics on grounds they violate conditioning theory's ontological premises (Kahle, Beatty, and Kennedy 1987) or reflect little more than demand artifacts (Kellaris and Cox 1989).

The present paper seeks to illuminate the issues by (1) evaluating opposing ontologies on the role of subject awareness, and (2) reviewing relevant empirical work that provides commentary on demand artifacts and related issues. Before pursuing these objectives, it is worth noting that the role of subject awareness has long been controversial in the study of classical conditioning of human subjects (e.g., Insko and Oakes 1966; Page 1969; Staats and Staats 1957, 1958; Zanna, Kiesler, and Pilkonis 1970). The Staatses (1957, 1958) prompted the awareness-issue debate by asserting their results evidenced classical conditioning without awareness. Subsequent researchers (e.g., Insko and Oakes 1966; Page 1969) challenged this claim; most notably, Page (1969) asserted that there is no conditioning without awareness and implicated demand artifact as the villain (cf. Staats 1969). Insko and Oakes (1966) presented data showing that contingency awareness mediates the effect of conditioning but that demand artifacts is not a necessary precondition. Other studies (e.g., Zanna, Kiesler, and PiLkonis 1970; Frcka, Beyts, Levey, and Martin 1983; Baeyens, Crombez, Van den Bergh, and Eelen 1989) have demonstrated conditioning effects in the absence of subject awareness of the CS-US contingency. (See for further literature citations Kahle 1984; Kahle et al. 1987; and Allen and Janiszewski 1989.)

COMPETING ONTOLOGIES: PRISTINE AND COGNITIVE POSITIONS

As suggested above, the philosophical issue involved is whether subject awareness violates the ontological premises of the classical conditioning model. Two competing stances on this matter are termed here the Pristine and Cognitive positions.

The Pristine Position

This philosophical position extends from classical conditioning's historical roots in behaviorism. Because behaviorist ontology rejects mentalistic constructs, it follows that subject awareness of the CS-US contingency (itself a mentalistic phenomenon) cannot be accommodated by a theory of classical conditioning. Hence, according to this philosophical stance, to acknowledge contingency awareness is to disavow classical conditioning as a theoretical account for observed effects.

In an especially well-articulated position, Kahle et al. (1987) argue that conditioning theory, ontological premises require that conditioning occur without awareness. Kahle and colleagues favor complete separation of behaviorist and cognitive traditions and claim that cognitive classical conditioning is an inherently contradictory phrase. They go so far as to conclude that, since the evidence shows the widespread presence of subject awareness in conditioning experiments, "perhaps researchers should study new topics" (p. 413).

Cognitive Position

In contrast to the pristine position, modern Pavlovian theory warmly embraces cognitive concepts as elements of classical conditioning's ontology. This fully cognitive position rejects the view that classical conditioning is simply reflexive, passive, low-involvement learning, and views it, instead, as cognitive associative learning, that is, the learning of relations among events in the environment (cf. Dawson, Schell, Beers, and Kelly 1982; Furedy, Riley, and Fredrikson 1983, p. 126; Holland 1984; Holyoak, Koh, and Nisbett 1989; Rescorla 1988). Modern conditioning theorists contend that the presence of contingency awareness is indeed a necessary condition for classical conditioning: "the acquisition of autonomic CRs is not an automatic process, but rather requires conscious cognitive processing of the stimulus contingency" (Dawson and Schell 1987, p. 33).

Brewer (1974), whose brilliant essay claimed there is no convincing evidence of classical (or operant) conditioning in adult humans, is often cited as evidence that classical conditioning is nonexistent. In actuality, Brewer had simply anticipated the cognitive revolution in Pavlovian theory and was saying essentially what others since have recognizes: "all the results of the traditional conditioning literature are due to the operation of higher mental processes, as assumed in cognitive theory, and that there is not and never has been any convincing evidence for unconscious, automatic mechanisms in the conditioning of adult human beings" (p. 27). This argument did not deny conditioning, it rather required that cognitive notions such as awareness and expectations be incorporated within conditioning's purview if the theory were to represent an accurate account of human learning.

Neo-Pavlovians are clear in pointing out that awareness of the CS-US relationship is a necessary condition for classical conditioning. Indeed, survival (especially for animals) and goal achievement (especially for humans) require awareness of the many critically informative relations that exist between environmental events. For example, animals learn and probably are fully aware that smoke precedes fire, that the odor of a predator indicates the likelihood of a life-threatening attack, and that the sight of another animal's tracks implies the possibility of food. Brewer (1974) himself, in arguing that conditioning in human subjects involves cognitive processes, acknowledged that this leaves open the possibility, a possibility he was fully willing to accept, that cognitive operations are also present in the conditioning of lower animals.

Awareness of the temporal or spatial relation between environmental events is the sine qua non for conditioned learning in the natural world. There is no reason to expect it should be otherwise in laboratory experiments with human subjects. In fact, awareness is absolutely essential because consequences of behavioral responses in the laboratory are so relatively trivial in comparison to real-world consequences. That is, assignment of subjects to a conditioning group does not assure all will be conditioned, because some will not devote the level of attention necessary to become aware of the CS-US contingency (cf. Staats 1969; Gormezano in Petty and Cacioppo 1981, p. 46). It is only by paying attention and becoming aware of the CS-US contingency that an experiment stands a chance of conditioning a response. Maltzman (1987, p. 230) perhaps said it best:

Conditioning does not occur in a black box, an empty shell, or a buffer. We do not condition a disembodied response. We condition an organism, usually an inquiring, suspicious, curious, bored--all of these and more, or none--college sophomore. The conditioned response is a small sample of the complex behavioral, physiological, and neurochemical changes occurring at the moment. It is produced in a person with a variety of cortical sets, dominant foci, interests, attitudes, etc., present at the moment the CS and US are presented which influence and direct the nature of the CR.

Recapitulation

Modern theorists of classical conditioning (the Neo-Pavlovians cited above) have taken a fully cognitive position in formulating their view of how Pavlovian conditioning operates. These theorists readily accept subject awareness of CS-US relations as a fundamental property of conditioning. Hence, the opposing, pristine position that subject awareness violates conditioning's ontological premises is based on an obsolete view of classical conditioning theory. Robert Rescorla, a noted conditioning scholar, recently chastised this anachronistic view in a self-descriptive review article titled "Pavlovian Conditioning: It's Not What You Think It Is" (Rescorla 1988).

EMPIRICAL EVIDENCE ON THE ROLE OF SUBJECT AWARENESS

When discussing the awareness issue at the empirical level, it is important to make clear distinctions between contingency awareness and demand bias (cf. Allen and Janiszewski 1989; Insko and Oakes 1966; Petty and Cacioppo 1981; Stuart, Shimp, and Engle 1987). Specifically, it is necessary to delineate four constructs: contingency awareness, demand awareness, demand bias, and conditioned effect. The Figure presents the relations among these constructs and their corresponding necessity and sufficiency conditions.

Contingency awareness (CA) exists when experimental subjects know that CS and US have been related temporally in an experiment's sequencing of these two events. For example, subjects were judged to be contingency aware in Allen and Janiszewski's (1989) study when they knew that certain Norwegian words (the CS) were more likely than others to be followed by positive feedback (the US).

Demand awareness (DA) means that a subject has some idea about his or her expected role in a conditioning experiment. She or he thinks (conjectures, infers) that the experimenter expects him or her to emit a particular response, such as reflecting a favorable attitude on a response scale.

Demand bias (DB), also called demand artifacts, means that subjects, because they are demand aware, alter their natural (i.e., unbiased) response in a direction either intended to support or counter their assumption of what the experimenter expects them to do. The important point, though typically neglected (Allen and Janiszewski provide an exception), is that DB does not necessarily follow from DA. Because a subject has some idea of what the experimenter expects does not necessarily mean that s(he) will adjust her(his) behavior accordingly. Moreover, individual subjects may sometimes alter their behavior, but because different subjects likely alter their behavior in different directions, the effect of such random deviations is that individual data values may vary from true scores, but aggregate statistics are not necessarily systematically biased.

FIGURE

CONCEPTUAL FRAMEWORK FOR AWARENESS ISSUE

Conditioned effect (CE) is an individual- or aggregate-level measure of conditioned response. At the individual-subject level, the effect is a conditioned response which an experiment attempts to bring as close to asymptote as possible. When studying autonomic (e.g., GSR) or skeletal (e.g., eye-blink) responses, it is possible to test whether an individual subject has in fact been conditioned. This is done by comparing a baseline measure of response to the US (e.g., the heart rate when the US is present) with a post ne< sure of response to the CS alone. However, when studying attitudinal conditioning, which is the type of learning most relevant for consumer behavior, individual-level measurement is virtually impossible because unambiguous baseline measures are unavailable. Thus, with attitudinal conditioning, the CE is measured at an aggregate-level; i.e., the CE represents the statistical difference between conditioning and control groups' average attitudinal responses. CE will be used in this aggregate-effect sense throughout the remaining discussion. Returning to the Figure, the arguments can be summarized as follows:

1. Contingency awareness (CA) may be necessary for a conditioned effect (CE) to materialize. (Empirical evidence is inconsistent on this point, hence justifying the equivocal statement.)

2. But CA is not necessary nor sufficient for demand bias.

3. The relationship between CA and demand awareness (DA) is indeterminate; i.e., either type of awareness may alert the subject to the other, but whether it does or not is entirely contingent on the conditions in a particular experiment.

4. DA is necessary but not sufficient for demand bias (DB). Indeed, heroic assumptions must be made to assume that (aggregate-level) demand artifacts exists just because subjects may be aware of the experimenter's expectations.

5. Demand awareness is neither necessary nor sufficient for a conditioned effect (CE) to materialize.

6. Demand bias is sufficient for CE but is not necessary.

This rather protracted, though necessary, prelude makes it possible now to meaningfully discuss the various awareness issues and the challenges they pose for classical conditioning. The discussion is restricted to recent research evidence and is divided into two sections, contingency-awareness and demand-bias evidence.

Contingency-Awareness Evidence

The issue of contingency awareness can be summarized in this question: Is contingency awareness necessary for conditioning, or can conditioning occur without cognition? [Readers will note an obvious similarity with the more general issue of whether affect is formed independently of cognition or whether it is cognitively mediated (Anand, Holbrook, and Stephens 1988; Zajonc 1980; Zajonc and Marcus 1984).] As discussed below, the evidence is mixed.

Evidence Supporting Cognitive Mediation. Allen and Janiszewski (1989) performed two experiments employing a creative computerized word game to condition attitudes to five neutral Norwegian words. Experimental results (especially Experiment 1) evidenced attitudinal conditioning; however, the most important facet of this research is the showing in both experiments that contingency awareness is a necessary precondition for attitudinal conditioning. Allen and Janiszewski do not assert that their results disprove non-cognitive conditioning (i.e., conditioning without subject awareness), but they present an impressive argument that it is difficult to demonstrate conditioning without awareness.

Evidence Challenging Cognitive Mediation. Some influential recent European conditioning research has yielded results in conflict with Allen and Janiszewski's (1989) findings. Baeyens and his colleagues' (Baeyens, Crombez, Van den Bergh, and Eelen 1989; Baeyens, Eelen, and Van den Bergh 1989; Baeyens, Eelen, Van den Bergh, and Crombez 1989) experiments have demonstrated attitudinal conditioning without subject awareness. The significance of this finding is in the suggestion that affect for an initially neutral object can be formed automatically without the subject necessarily being aware of the CS-US contingency.

The disagreement between Allen and Janiszewski's findings and those from Europe is perplexing, but not necessarily inexplicable. Whereas Allen and Janiszewski conditioned attitudes toward neutral Norwegian words using verbal feedback as unconditioned stimuli, Baeyens and his colleagues conditioned attitudes toward human faces using other human faces (either liked or disliked) as unconditioned stimuli. Baeyens and colleagues contend that their experimental paradigm yields evaluative conditioning (i.e., affective learning), which is fundamentally different than the type of signal learning that is probably more typical of Allen and Janiszewski's paradigm as well as that employed by Stuart et al. (1987). Further research is certainly in order to confront this incongruity.

Demand-Bias Evidence

While acknowledging the presence of contingency awareness as necessary for conditioned effects, Allen and Janiszewski (1989) and Stuart et al. (1987) deny their results are due to demand bias. However, in an impressive recent challenge, Kellaris and Cox (1989) raise the possibility that the seminal consumer behavior demonstration of classical conditioning by Gorn (1982) may be attributable to demand bias. Briefly, Kellaris and Cox performed two true-experiments (one a slight modification of Gorn's procedures, and the other a close replication) that failed to replicate Gorn's findings (that subjects exposed to a pen paired with liked music were more likely to prefer that pen compared to subjects exposed to a pen paired with disliked music). However, a third study (a non-experiment that reenacted Gorn's procedure by merely describing rather than administering treatments to subjects) yielded results similar to Gorn's original findings.

Kellaris and Cox's (1989) overall results would appear to suggest that Gorn's findings (and, by implication, findings from other conditioning experiments) are purely artifactual. However, Kellaris and Cox provide no concrete evidence to show that specific features of Gorn's study would have rendered its findings artifact-laden. Moreover, Kellaris and Cox's experiments, although conscientious and close replications of Gorn's work, differ in the music used as US, which may account for why they were unable to replicate Gorn's findings. Finally, the fact that Kellaris and Cox's non-experiment supports Gorn's findings is not itself convincing evidence that Gorn's results were due to demand artifacts inasmuch as this procedure has been criticized on grounds that it tends to produce a special mental set that is unique to role-playing subjects (Kruglanski 1975).

In sum, although it is difficult to ever entirely discount a demand-artifacts challenge-especially in view of the fact that efforts to measure the presence of demand artifacts are themselves subject to demand artifacts (Gorn, Jacobs, and Mana 1987)--it would seem that the logic chain necessary to suggest that Gorn's results were artifact laden is based on considerable speculation. The fact is that contingency awareness is not tantamount to guessing the experimental hypothesis; furthermore, subjects who do correctly guess the hypothesis may or may not alter their behavior in the direction called for by the hypothesis.

CONCLUSION

Consumer researchers during the past decade displayed considerable interest in classical conditioning. To sustain this momentum will require (1) examining various heretofore untested issues (e.g., the role of CS-US similarity), (2) possibly pitting the conditioning model against alternative accounts of attitude formation/change in a comparative-testing sense (Sternthal, Tybout, and Calder 1987), and (3) safeguarding against alternative explanations that threaten the internal validity of experimental evidence. In the spirit of this last desideratum, the purpose of this paper has been to evaluate whether the empirical presence of subject awareness (of the contingent relation between conditioned and unconditioned stimuli) poses a pernicious blow to the classical conditioning paradigm.

This possibility was examined first by evaluating the philosophical argument which contends that mentalistic constructs are inherently incompatible with classical conditioning's ontological premises. This so-called pristine position was discounted on grounds that it is out of step with the modern, fully cognitive position on the nature of Pavlovian conditioning, a position that warmly embraces cognitive constructs.

The paper also examined the related issues of contingency awareness and demand bias. Regarding the latter, it was concluded that the evidence favoring a demand-bias challenge to claimed classical conditioning results (particularly Kellaris and Cox's (1989) challenge to Gorn's (1982) findings) is more speculative than definitive. With respect to the role of contingency awareness, the paper amplifies Allen and Janiszewski's (1989) position that conditioned learning is cognitively mediated by subject awareness of the temporal relation between CS and US. However, this position was taken with some degree of equivocation in view of the recent evidence from Europe by Baeyens and his colleagues, who have presented evidence of classical conditioning without contingency awareness. No definitive conclusion is possible at this point, but it would appear that the conditioning paradigm used by the European researchers may better evidence evaluative learning in comparison to the serial-learning procedure more characteristic of recent consumer behavior conditioning studies (especially Allen and Janiszewski 1989; Bierley et al. 1985; and Stuart et al. 1987). This provocative possibility demands future examination in view of Baeyens et al.'s claim that evaluative conditioning is possible without cognitive mediation, and, even more interestingly, that such conditioning is non-extinguishable.

REFERENCES

Allen, Chris T. and Chris A. Janiszewski (1989), "Assessing the Role of Contingency Awareness in Attitudinal Conditioning with Implications for Advertising Research," Journal of Marketing Research, 26 (February), 3043.

Allen, Chris T. and Thomas J. Madden (1985), "A Closer Look at Classical Conditioning," Journal of Consumer Research, 12 (December), 301-315.

Anand, Punam, Morris B. Holbrook, and Debra Stephens (1988), "The Formation of Affective Judgments: The Cognitive-Affective Model Versus the Independence Hypothesis," Journal of Consumer Research, 15 (December), 386-391.

Baeyens, Frank, Geert Crombez, Omer Van den Bergh, and Paul Eelen (1988), "Once in Contact Always in Contact: Evaluative Conditioning Is Resistant to Extinction," Advances in Behavioral Research and Therapy, 10, 179-199.

Baeyens, Frank, Paul Eelen, and Omer Van den Bergh (1989), "Contingency Awareness in Evaluative Conditioning: A Case for Unaware Affective-Evaluative Learning," Cognition and Emotion, 3.

Baeyens, Frank, Paul Eelen, Omer Van den Bergh, and Geert Crombez (1988), "Acquired Affective-Evaluative Value: Conservative But Not Unchangeable," Behavioral Research and Therapy, 27 (3), 279-287.

Bierley, Calvin, Frances K. McSweeney, and Renee Vannieuwkerk (1985), "Classical Conditioning of Preferences for Stimuli," Journal of Consumer Research, 12 (December), 316-323.

Brewer, William P. (1974), "There Is No Convincing Evidence for Operant Or Classical Conditioning in Adult Humans," in W. Weirner and D. Palermo (eds.), Cognition and the Symbolic Processes, Lawrence Erlbaum, Hillsdale, NJ., 142.

Dawson, Michael F. and Anne M. Schell (1987), "Human Autonomic and Skeletal Classical Conditioning: The Role of Conscious Cognitive Factors," in G. Davey (ed.), Cognitive Processes and Pavlovian Conditioning in Humans, Chichester: John Wiley & Sons Ltd., 27-55.

Dawson, Michael E., Annel M. Schell, James R. Beers, and Andrew Kelly (1982), "Allocation of Cognitive Processing Capacity During Human Autonomic Classical Conditioning," Journal of Experimental Psychology: General, m (September), 273-295.

Frcka, Gertrude, Johanna Beyts, A. B. Levey, and Irene-Martin (1983); "The Role of Awareness in Human Conditioning," Pavlovian Journal of Biological Sciences, 18 (April-June), 69-76.

Furedy, John J., Diane M. Riley, and Mats Fredrikson (1983), "Pavlovian Extinction, Phobias, and the Limits of the Cognitive Paradigm," Pavlovian Journal of Biological Science, 17 (July-September), 126-135.

Gorn, Gerald J. (1982), "The Effects of Music in Advertising on Choice Behavior: A Classical Conditioning Approach," Journal of Marketing, 46 (Winter), 94-101.

Gorn, Gerald J., W. J. Jacobs, and Michael J. Mana (1987), "Observations on Awareness and Conditioning," in Melanie Wallendorf and Paul P. Anderson (eds.), Advances in Consumer Research, 14, Provo, UT: Association for Consumer Research, 415416.

Holland, Peter C. (1984), "Origins of Behavior in Pavlovian Conditioning," G. H. Bower (ed.), The Psychology of Learning and Motivation, 18, 129-174.

Holyoak, Keith J., Kyunghee Koh, and Richard E. Nisbett (1989), "A Theory of Conditioning: Inductive Learning Within Rule-Based Default Hierarchies," Psychological Review, 96 (No. 2), 315-340.

Insko, Chester A. and William F. Oakes (1966), "Awareness and the 'Conditioning' of Attitudes," Journal of Personality and Social Psychology, 4, No. 3, 487-496.

Kahle, Lynn R. (1984), Attitudes and Social Adaptation: A Person-Situation Interaction Approach, Oxford, U.K.: Pergamon Press.

Kahle, Lynn R., Sharon E. Beatty, and Patricia Kennedy (1987), "Comment on Classically Conditioning Human Consumers," in Melanie Wallendorf and Paul F. Anderson (eds.), Advances in Consumer Research, 14, Provo, UT: Association for Consumer Research, 411-414.

Kellaris, James J. and Anthony D. Cox (1989), 'The Effects of Background Music in Advertising: A Reassessment," Journal of Consumer Research, 16 (June), 113-118.

Kruglanski, A. W. (1975), 'The Human Subject in the Psychology Experiment: Fact and Artifact," in L. Berkowitz (ed.), Advances in Experimental Social Psychology, 8, Orlando: Academic Press, 101 -147.

Macklin, M. Carole (1986), "Classical Conditioning Effects in Product/Character Pairings Presented to Children," in Richard J. Lutz (ed.), Advances in Consumer Research, 13, Provo, UT: Association for Consumer Research, 198-203.

Maltzman, Irving (1987), "A Neo-Pavlovian Interpretation of the OR and Classical Conditioning in Humans: With Comments on Alcoholism and the Poverty of Cognitive Psychology," in G. Davey (ed.), Cognitive Processes and Pavlovian Conditioning in Humans, Chichester: John Wiley & Sons Ltd., 211 -250.

McSweeney, Frances K. and Calvin Bierley (1984), "Recent Developments in Classical Conditioning," Journal of Consumer Research, 11 (September), 619-631.

Nord, Walter R. and J. Paul Peter (1980), "A Behavior Modification Perspective on Marketing," Journal of Marketing, 44 (Spring), 36-47.

Page, Monte M. (1969), "Social Psychology of a Classical Conditioning of Attitudes Experiment," Journal of Personality and Social Psychology, 11, No. 2, 177-186. l

Petty, Richard E. and John T. Cacioppo (1981), Attitudes and Persuasion: Classic and Contemporary Approaches, Dubuque, Iowa: Wm. C. Brown.

Rescorla, Robert A. (1988), "Pavlovian Conditioning: It's Not What You Think It Is," American Psychologist, 43 (March), 151-160.

Staats, Arthur W. (1969), "Experimental Demand Characteristics and the Classical Conditioning of Attitudes," Journal of Personality and Social Psychology, 11, No. 2, 187-192.

Staats, Carolyn K. and Arthur W. Staats (1957), "Meaning Established by Classical Conditioning," Journal of Experimental Psychology, 54, 74-80.

Staats, Arthur W. and Carolyn K. Staats (1958), "Attitudes Established by Classical Conditioning," Journal of Abnormal and Social Psychology, 57, 3740.

Sternthal, Brian, Alice M. Tybout, and Bobby J. Calder (1987), "Confirmatory Versus Comparative Approaches to Judging Theory Tests," Journal of Consumer Research, 14 (June), 114-125.

Stuart, Elnora W., Terence A. Shimp, and Randall W. Engle (1987), "Classical Conditioning of Consumer Attitudes: Four Experiments in an Advertising Context," Journal of Consumer Research, 14 (December), 334-349.

Zanna, Mark P., Charles A. Kiesler, and Paul A. Pilkonis (1970), "Positive and Negative Attitudinal Affect Established by Classical Conditioning," Journal of Personality and Social Psychology, 14, No. 4, 321-328.

Zajonc, Robert B. (1980), "Feeling and Thinking: Preferences Need No Inferences," American Psychologist, 35, -151 - 175.

Zajonc, Robert B. and Hazel Markus (1982), "Affective and Cognitive Factors in Preferences," Journal of Consumer Research, 9 (September), 123-131.

----------------------------------------