Evaluative conditioning is best defined as an effect t - PDF document

Evaluative conditioning is best defined as an effect, that is, as a change in the valence of a stimulusthat results from pairing the stimulus with another stimulus. This definition has several advantagesthat are made explicit in this paper. One of the advantages is that it clarifies that evaluativeconditioning can be due to multiple processes. Therefore, the conditions under which evaluativeevaluative conditioning. This could explain why there are so many conflicting results about theconditions under which evaluative conditioning can be found. Future research should adopt aKeywords:evaluative conditioning,associative learning,attitudes,attitude formationEl condicionamiento evaluativo se define como un efecto, es decir, un cambio en la valencia deque se explicitan en este trabajo. Una de las ventajas es que aclara que el condicionamientoPalabras clave:condicionamiento evaluativo, aprendizaje asociativo, actitudes, formaci—n de AConceptual and Theoretical Analysis of Evaluative ConditioningGhent University, Belgium The Spanish Journal of PsychologyCopyright 2007 by The Spanish Journal of Psychology2007, Vol. 10, No. 2, 230-241ISSN 1138-7416 The preparation of this paper was made possible by Grant BOF/GOA2006/001 of Ghent University. Sociedad Espa–ola de Psicolog’a Comparada -Spanish Society of Comparative Psychology-)write this paper. I dedicate this paper to my mentor, Paul Eelen, who introduced me to evaluative conditioning, to the distinctprocedure, effect, and theory, and to good food and drinks, to name just a few things. I benefited greatly from many discussionAgnes Moors regarding the concepts ÒautomaticityÓ and Òfunctional characteristics.Ó Many thanks also to Tom Beckers and Agnes Mfor their comments on a first draft of this paper. Correspondence should be addressed to Jan De Houwer, Ghent University, Henri Dunantlaan 2, B-9000 Ghent, Belgium. Phone:0032 9 264 64 45. FAX: 0032 9 264 64 89. E-mail: Jan.DeHouwer@UGent.be 230 Allport, 1935; Martin & Levy, 1978). To give just a fewpreferences influence attention, memory, and judgments,understand, predict, and influence behavior, it is thus crucialneutral stimulus can be changed by pairing it with another,liked or disliked, stimulus. The first stimulus is often calledor US. Typically,negative US. Well known real-life examples of evaluativeCoca-Cola Company. In these ads, the Coke brand nameEvaluative conditioning has been examined in a largenumber of studies (see De Houwer, Thomas, & Baeyens,2001, for an extensive review, and De Houwer, Baeyens, &of the phenomenon is still very limited. There now is generalagreement about the fact that evaluative conditioning is aonly recently; see De Houwer, Baeyens, et al., 2005). It alsofound with various kinds of stimuli and procedures. At thebefore evaluative conditioning can emerge (see De Houweret al., 2001, for a review). There is little else that evaluativesome results suggest that the effect can be found even whenBaeyens, Eelen, & Van den Bergh, 1990; Dickinson &awareness (e.g., Pleyers, Corneille, Luminet, & Yzerbyt,2007). Likewise, whereas some argue that evaluativeconditioning is unaffected by extinction (i.e., presentationsBaeyens, Crombez, Van den Bergh, & Eelen, 1988; Diaz,Ruis, & Baeyens, 2005), others have reported data showingthat extinction does affect evaluative conditioning (e.g.,In the current paper, I present a conceptual andthis can be remedied in future research. I will first arguethat evaluative conditioning is best defined as an effect ratherthan as a specific procedure or theoretical process. ThisWhen defined as an effect, it becomes clear that differenteffects (also see De Houwer, Baeyens, et al., 2005). ThisAConceptual AnalysisThe Distinction between Procedure,Effect,and TheoryThe starting point of the present analysis is the distinctionbetween procedure, effect, and theory (see Bolles, 1979;procedure and a target stimulus that are either related (e.g., NURSE-DOCTOR) or unrelated (e.g., WALL-DOCTOR) and (b)target. An effectprocedure. More specifically, it is an observation that isto the target DOCTOR are fast when it is preceded by theprime NURSE can be labeled a priming effect if thistarget. The core feature ÒrelatednessÓ is abstract in that itAn observation becomes an effect only when there isevidence showing that the core element of the procedure iseffect, this implies comparing reaction times on trials withrelated prime-target pairs to reaction times on trials withequivalent prime-target pairs that are not related in the samemanner. In practice, using the term effect thus requires acomparison of observations in situations that differ onlyto be crucial for the effect. Finally, the term EVALUATIVE CONDITIONING processes that are assumed to be responsible for an effect.For example, priming effects are often attributed to activationof the prime to the representation of the target, thusfacilitating responses to the target. Atheory therefore impliesof a procedure (e.g., relatedness of prime an target), it alsoeffect, or theoretical process, it is important to always makesaying that NURSE primes DOCTOR can mean that theword NURSE is presented briefly before the word DOCTORspeeds up responses to DOCTOR because the two wordsare related (priming as an effect), or that NURSE speedsup responses to the related word DOCTOR because(priming as a theoretical process). To avoid confusion, oneApplying the Distinction between Procedure,Effect,and Theory to Evaluative ConditioningconditioningÓ can be used to refer to a procedure, an effect,Òhave-a- Coke-and-a-smileÓ ads. To say that this is anprocedure effect procedure itself. It refers to the effect of the ads, not to theads as such. More generally, evaluative conditioning as aneffect refers to an actual change in the liking of stimuli thatis due to the fact that stimuli were paired in a certain manner.evaluative conditioning as an automatic, bottom-up, andof associations between representations in memory. SayingEvaluative Conditioning as a ProcedureIn that case, however, the term cannot be used to refer toVan den Bergh, & Crombez, 1990). Hence, such a use ofdifferences with other types of proceduresAsecond option is to use a set of core proceduralelements as the defining criterion. As noted above, core position of the stimulus and the relevant position of the response match than when they mismatch. What few researchers realize Simon tasks do not necessarily involve spatial information. For instance, De Houwer and Eelen (1998) introduced an affective Sidiffered (e.g., say GOOD to CANCER because it is a noun). Structurally, both spatial and affective Simon tasks involve the maniof the match between an irrelevant stimulus feature and a relevant response feature (see De Houwer, 2003). They thus share the elements). The question is what core procedural elementsalso used to define evaluative conditioning effects. Giventhat evaluative conditioning effects can be described ashave in common and how they differ from other types ofexamined whether this leads to changes in valence. Differentevaluative conditioning procedures can differ with regardPavlovian conditioning procedures. As in other Pavlovian(e.g., salivation, skin conductance). Third, evaluativeprocedures differ from those other procedures in that stimuliEvaluative Conditioning as an EffectIn terms of an effect, evaluative conditioning can bepairing of stimuli. Like other effects, there is an observationdefining evaluative conditioning as an effect provides aconditioning as an effect.Defining evaluative conditioning as an effect clarifiesthe similarities and differences with other effectsprocedures, defining evaluative conditioning effects in termseffects. First, all changes in liking that are due to the pairingof stimuli can be regarded as evaluative conditioning effects.conditions under which these stimuli are presented, and theconditioning effects constitute a subclass of all Pavlovianconditioning effects. Like all other Pavlovian conditioningeffects, evaluative conditioning effects concern changes instimulus with another stimulus. What is unique to evaluativeconditioning effects is that they concern changes in onein the liking of the stimuli. Third, evaluative conditioningeffects are a subclass of all observed changes in liking. Astimuli. There are several other possible reasons for whyon the mere exposure effect has demonstrated that theLiking can change also as the result of a procedure in whicha behavior is paired with an outcome. Assume, for instance,present. The behavior Òpressing a leverÓ could become moreopportunity to receive a reward. These changes in liking arenot evaluative conditioning effects because the changes into these changes as operant evaluative conditioning effectsto mere exposure and evaluative conditioning effects, thereeffects (see Beckers, De Houwer, & Eelen, 2002, for anThere are clear criteria for determining the presence ofevaluative conditioning effects. conditioning as an effect implies that a change in liking canbe classified as an evaluative conditioning effect once it isdemonstrated that it is due to the pairing of stimuli. This candiffer from the experimental condition only with regard toinstance, some CSs are paired with a positive US whereastwo sets of CSs differs after presenting the pairings, thenthis difference can be called an evaluative conditioning effectprovided that the two sets of CSs are entirely equivalentthe two sets of stimuli cannot differ in how often they haveEVALUATIVE CONDITIONING been presented or in how similar they are to the differentUSs (see Field & Davey, 1999, for a discussion of a possiblestimuli. If the first group of participants shows larger changesin liking can be called an evaluative conditioning effectprovided that the groups differ only in how the stimuli wereOne complication should be noted, however. Althoughlaboratory environment, outside of the laboratory, suchwith someone who genuinely dislikes spiders, it is difficultof a brand, it is difficult to know whether this is due to theas an evaluative conditioning effect is merely a hypotheticalDefining evaluative conditioning as an effect helps toorganize research. The fact that a procedure can be classifiedit will actually generate an evaluative conditioning effect (i.e.,stimuli). Whether it does can depend on at least three aspectsmeasured. Virtually all research on evaluative conditioningof the procedure on evaluative conditioning effects. Eachpaired in order to result in a change in liking. These studiesthat is necessary for obtaining changes in liking. They focusrevaluation), whether a CS and US are paired directly orindirectly (i.e., sensory preconditioning, higher-order(see Rachman, 1977), interactions between differentfor a review). All these factors deal with the abstractprocedural properties of how stimuli are paired. TheseThe second set of studies focuses on the generality ofevaluative conditioning effects. They examine whether, givenconditioning effects have been found with a range of stimuliet al., 2001, for a review). Effects have also been observed(as an effect) appears to be a general phenomenon.The third set of studies focuses not on the nature of theenabling conditions that could modulate the effect of thecore procedure. Again, abstract means that the enabling conditioning effects. Such effects can be submitted to the same experimental analysis as other evaluative conditioning effects. This raisesone could argue that studies examining variations in how liking is measured belong to the first set of studies on the nature of the coreprocedure. This is especially true when abstract features of the measure are manipulated (e.g., verbal vs. nonverbal, direct vsng. However,when the aim is merely to examine whether evaluative conditioning effects generalize over different types of specific measures, the Houwer, 2006a, for a detailed discussion). The conditionsNote that the classification of research that is offeredhere differs from the classification that De Houwer et al.literature. They made a distinction between, on the one hand,evaluative conditioning. The section on generality includedstudies with different kinds of stimuli (e.g., visual, tactile)and different procedural parameters (e.g., order of CS andto USs). The section on functional characteristics summarizedstudies on extinction, contingency, awareness, occasionsetting, amongst others. It is not clear, however, whatcharacteristics. Although they did not provide an expliciteffect can also be seen as research on the conditions underwhich evaluative conditioning can be found. Thebetween different types of conditions, it also leads to thepresent? As will be discussed later on this paper, suchDefining evaluative conditioning as an effect allows oneto entertain multiple theoretical accounts. advantage is probably the most important one. Whenevaluative conditioning is defined as an effect, the observedprocedure, being the pairing of stimuli. There is noresult from these newly formed associations. However, thispairings into an output of changes in liking. All theoreticaloptions are open. This also implies that evaluativeconditioning as an effect does not depend on the validity ofon the validity or existence of evaluative conditioning asan effect. As we will see in the next section, problems canEvaluative Conditioning as a Theoretical Processmore restrictive use of the term than evaluative conditioningas an effect. It entails that not all evaluative conditioningeffects (i.e., changes in liking that are due to the pairing ofÒTrueÓ evaluative conditioning has occurred only when thechanges in liking were due to a certain process. As noted& Bodenhausen, 2006, p. 697; Walther, Nagengast, &Trasselli, 2005, p. 191). In this section, I will argue thatcarries at least two risks. First, it renders it difficult toHow to determine the presence of evaluative conditioningas a theoretical process? as a theoretical process renders it difficult to determinesuch as the automatic formation of associations in memory.Psychological processes are theoretical constructs that cannotmemory. It is simply a theoretical assumption that couldhelp understand and organize existing empirical facts (i.e.,to forget that associations in memory are merely theoreticalneural links between neurons in the brain. This similarity,however, is misleading. One can observe dendritesdeveloping between neurons, but this is completely differentEVALUATIVE CONDITIONING activate another representation. The similarity is purelydirectly, it could be inferred indirectly. To the extent that aconditioning effects can be observed in the absence ofan evaluative conditioning effect in the absence ofthat ÒrealÓ evaluative conditioning has occurred. Althoughaccount can also explain that effects can occur given certainconditions, then the presence of the effect under thosecontingencies). Whereas the former requires only thepresence of adequate controls, the latter can prove difficultcontingency unaware; see Shanks & St. Johns, 1994). Third,the approach works only for those effects that are observedclassify effects that are observed under standard conditionsbecause the effects under those conditions could be due totheoretical process, it becomes difficult to determine whetherevaluative conditioning. This is because it is impossible todifficulties and limitations with assessing the presence of aprocess indirectly.Defining evaluative conditioning as a process hinderstheoretical research. such as the automatic formation of associations in memory,conditioning effects are likely to slow down the developmentassociation formation implies that changes in valence thatcases of evaluative conditioning. Therefore, if research wouldare automatically formed in memory, this would also raisesuch. The possibility that automatic association formation(e.g., Lovibond, 2003). Moreover, if even renownedtheoretical principles such as NewtonÕs law of gravity haveconditioning in humans.Ó This conclusion was based onof the contingencies between the stimuli. Importantly, thethose stimuli. The evidence did, however, question theof associations in memory. Because this theoretical approachterms of this one particular theoretical view. That is, manyconditioning as conditioned changes in behavior that aredue to the automatic formation of associations in memory.effects in humans. As a result of BrewerÕs chapter, manyconditioning. This was a very unfortunate evolution becauseaccount of classical conditioning. In a similar way, evaluative EVALUATIVE CONDITIONING ATheoretical Analysiswould best be defined as an effect rather than as a specificof defining evaluative conditioning as an effect is that itunderlie evaluative conditioning effects. In this section, Ievaluative conditioning effects. Next, I will point out thatA Multiple-Process Perspectiveassumption that evaluative conditioning effects are due tobetween representations. Take the example of the Òhave-with the positive affect that is evoked by these smilingpeople. When people see the Coke brand after beingsmiling people or positive affect, leading to positivefeelings. Different associative models differ in theirbehavior (e.g., direct translation or comparison of differenteffects are due to the automatic formation and updating ofassociations in memory.have always been prominent in conditioning research. However,there is no a priori reason why evaluative conditioning effectsconditioning effects are by definition associative in nature (i.e.,in memory. For instance, De Houwer and colleagues (DeHouwer, Baeyens, et al., 2005) pointed out that people mightof those stimuli. Assume that you receive an electric shockevery time you see a picture of a triangle but never after seeinga picture of a circle. Afterwards you are asked to indicate howcircle. When asked why, you can point to the fact that theconditioning effect) but it is not produced by automaticassociative processes. Rather, it is a genuine change in likingBoth types of effect thus depend on the use of propositionalis used for different reasons (i.e., to arrive at a genuineevaluation of the stimuli versus to comply with the expectationsof the experimenter; see Meersmans, De Houwer, Baeyens,Randell, & Eelen, 2005). Whereas associations in memoryonly in a controlled, non-automatic manner (e.g., Strack & There are probably still other processes that could underlie evaluative conditioning effects. For instance, Davey (1994) argued thatpairing stimuli makes salient the features that these stimuli have in common. When pairing a neutral stimulus with a liked stim(and vice versa for a neutral stimulus paired with a disliked stimulus). According to this explanation, evaluative conditioning is due to A Possible Explanation for Conflicting Results in theLiteratureAt the beginning of this paper, I noted that there areevaluative conditioning effects can be observed. Theseextinction, others showed that evaluative conditioningeffects are unaffected by CS-only presentations after the2006, for recent conflicting evidence). Likewise, whereasawareness (see Dickinson & Brown, 2007, and Walter &to assume that different evaluative conditioning effectscan be due to different processes. For instance, thecontingencies. Also, such knowledge is likely to reflecteffects that are due to the use of conscious propositionalcontingency (e.g., Baeyens & De Houwer, 1995;Gawronski & Bodenhausen, 2006; Walther et al., 2005).Therefore, in studies where evaluative conditioning effectseffects might well occur independently of contingencyawareness and be unaffected by extinctionImplications for Future Researchimportantly, it can be predicted that the impact of certainconditioning effects that do not depend on contingencyawareness should also be resistant to extinction. The reverseTo the best of my knowledge, these predictions have not yetbeen tested in the literature. The reason probably is thatBased on this view, the purpose of research becomes toelucidate the conditions under which evaluative conditioningconditioning effects. From the viewpoint that two or moreprocesses can produce evaluative conditioning effects,effects and when certain processes are important. The aimis thus to find the conditions under which certain conditionsclusters of conditions that tend to co-occur.Until now, I have limited the discussion to the enablingcondition of contingency awareness and the core proceduralcondition of extinction. More generally, one can predict thatnon-automatic evaluative conditioning effects should reflectthe statistical contingency between the CS and US whereasautomatic evaluative conditioning effects should reflect. This prediction is derived fromnon-automatic and thus can produce evaluative conditioningeffects only when participants are aware of the contingencies, I do not exclude the possibility that evaluative conditioning effects are based exclusively on the operation of propositional p(see De Houwer, Vandorpe, & Beckers, 2005). In that case, evidence for evaluative conditioning effects that are resistant to exand independent of contingency awareness might indeed result from methodological problems. However, at present, it is too early toin principle to define it in terms of enabling conditions. For instance, one could argue that only automatic evaluative conditi but also thatthis effect occurred under a certain set of enabling conditions (e.g., that participants were not aware of the contingencies). Adefinition interms of enabling conditions is more complex than a definition in terms of core procedure only, but in principle, it should be verify such a definition empirically. Also, it does not completely restrict ideas about possible underlying processes. It does EVALUATIVE CONDITIONING see Moors & De Houwer, 2006b). Second, under optimalbetween the CS and US (e.g., Shanks, 1995). Third,evaluative conditioning effects even when participants areto engage in effortful processing, and do not have controlenabling condition can determine the role of another enablingor resources to use propositions, the observed effects wouldnot depend on contingency awareness. This follows fromthe assumption that different enabling conditions are related(but see Moors & De Houwer, 2006a, 2006b).conditioning research beyond the current boundaries, it willnot make research easier. There are several potential pitfallseffects. For instance, hundreds of pages of text have beenextinction (e.g., Lipp et al., 2003). The meta-conditionalbetween the impact of those conditions. Fortunately, a lotMoreover, the emphasis is not on how to establish that adifferent conditions is related. For instance, if one canconsistently observe that extinction occurs when evaluativeeffect of extinction or the role of contingency awareness islink between the effect of different conditions, this couldactually be taken as evidence for the validity of the criteriaAsecond potential pitfall of the meta-conditional approachon multiple, often ill specified theoretical assumptions. TheDeutsch, 2004). Therefore, one way to exclude an impact ofdeveloped implicit attitude measures such as affective primingeffects and the Implicit Association Test (IAT) effects areAnderson, & Lovibond, 2003; but see De Houwer, 2006a,2006b). Research has shown, however, that propositionalmeasure, namely IATeffects (De Houwer, 2006b; De Houwer,Beckers, & Moors, in press). This demonstrates that the IATknowledge can influence behavior in an automatic manner.might well be invalid. Such assumptions thus need to be testedempirically.Humans and other organisms tend to want, do, and buylike. To understand and control human behavior, it ishuman behavior. Unfortunately, we still do not know muchabout this important phenomenon. To make matters worse,satisfactory manner.In the present paper, I have argued that progress in ourin liking), an effect (i.e., an actual change in liking as theterms of a particular process. Not only is it difficult to Most importantly, it detracts attention from the possibilityconditioning effects, that is, for a change in liking that isdefine evaluative conditioning in terms of an effect and toallow for the possibility that such effects can be due to atpropositional knowledge about stimulus contingencies. Theconditioning effects were in some cases due to one processand in other cases due to the other process. This multiplethese conditions are relevant. Although one should be awareAllport, G.W. (1935). Attitudes. In C.M. Murchison (Ed.),Handbook of social psychology(pp. 796-834). Winchester,Baeyens, F., Crombez, G., Van den Bergh, O., & Eelen, P. (1988).Advances in Behaviour Research andTherapy,10,Baeyens, F., & De Houwer, J. (1995). Evaluative conditioning is aqualitatively distinct form of classical conditioning: Areply toBehaviour Research and Therapy,33,Baeyens, F., Eelen, P., & Van den Bergh, O. (1990). Contingencyawareness in evaluative conditioning: Acase for unawareaffective-evaluative learning. Cognition and Emotion,4,Baeyens, F., Eelen, P., Van den Bergh, O., & Crombez, G. (1990).Flavor-flavor and color-flavor conditioning in humans. and Motivation,21,Beckers, T., De Houwer, J., & Eelen, P. (2002). Automaticintegration of non-perceptual action effect features: The caseof the associative affective Simon effect. PsychologicalResearch,66ed.). New York: Holt,Rinehart, & Winston.Bornstein, R.F. (1989). Exposure and affect. Overview and meta-Psychological Bulletin,106Brewer, W.F. (1974). There is no convincing evidence ofconditioning in adult humans. In W.B. Weimer & D.S. PalermoCognition and the symbolic processes Journal of Experimental Psychology,83Davey, G.C.L. (1994). Defining the important questions to askabout evaluative conditioning: Areply to Martin and LeveyBehaviour Research and Therapy,32,De Houwer, J. (2003). Astructural analysis of indirect measuresThe psychologyof evaluation:Affective processes in cognition and emotionDe Houwer, J. (2006a). What are implicit measures and why arewe using them? In R.W. Wiers & A.W. Stacy (Eds.), handbook of implicit cognition and addiction (pp. 11-28).De Houwer, J. (2006b). Using the implicit association test doesLearning and Motivation,37De Houwer, J., Baeyens, F., & Field, A.P. (2005). AssociativeCognition and Emotion,19De Houwer, J., Beckers, T., & Moors, A. (in press). Novel attitudescan be faked on the Implicit Association Test. Journal ofExperimental Social PsychologyDe Houwer, J., & Eelen, P. (1998). An affective variant of theCognition and Emotion,12De Houwer, J., Thomas, S., & Baeyens, F. (2001). Associativelearning of likes and dislikes: Areview of 25 years of researchPsychological Bulletin,127De Houwer, J., Vandorpe, S., & Beckers, T. (2005). On the roleIn A. Wills (Ed.), New directions in human associative learningDiaz, E., Ruis, G., & Baeyens, F. (2005). Resistance to extinctionCognition and Emotion,19Dickinson, A., & Brown, K.J. (2007). Flavor-evaluative conditioningis unaffected by contingency knowledge during training withcolor-flavor compounds. Learning & Behavior,35Eelen, P. (1980). Klassieke conditioning: Klassiek en toch modern (1980). Klassieke conditioning: Klassiek en toch modernIn Liber Amicorum Prof. J.R. Nuttin, Gedrag,dynamischerelatie en betekeniswereld [Behaviour,dynamic relation,andField, A.P., & Davey, G.C.L. (1999). Reevaluating evaluativeconditioning: Anonassociative explanation of conditioningeffects in the visual evaluative conditioning paradigm. Journalof Experimental Psychology:Animal Behaviour Processes,25,211-224.Gawronski, B., & Bodenhausen, G.V. (2006). Associative andpropositional processes in evaluation: An integrative reviewPsychological Bulletin, Hermans, D., Baeyens, F., & Eelen, P. (2003). On the acquisitionand activation of evaluative information in memory: The studyof evaluative learning and affective priming combined. In J.The psychology of evaluation:Affective processes in cognition and emotion Levey, A.B., & Martin, I. (1975). Classical conditioning of humanÔevaluativeÕresponses. Behaviour Research and Therapy,4,Lipp, O.V., Oughton, N., & LeLievre, J. (2003). Evaluative learningLearning and Motivation,34Lipp, O.V., & Purkis, H.M. (2006). The effects of assessment typeJournal of Experimental Psychology:Animal Behavior Processes,32Lovibond, P.F. (2003). Causal beliefs and conditioned responses:Journal of Experimental Psychology:Learning,Memory,and Cognition,29Martin, I., & Levey, A.B. (1978). Evaluative conditioning. in Behaviour Research and Therapy,1Meersmans, T., De Houwer, J., Baeyens, F., Randell, T., & Eelen,P. (2005). Beyond evaluative conditioning: Searching forCognition and Emotion,19Mitchell, C.J., Anderson, N.E., & Lovibond, P.F. (2003). Measuringevaluative conditioning using the Implicit Association Test.Learning and Motivation,34Moors, A., & De Houwer, J. (2006a). Automaticity: AconceptualPsychological Bulletin,132Moors, A., & De Houwer, J. (2006b). Problems with dividing thePsychological Inquiry,17Pleyers, G., Corneille, O., Luminet, O., & Yzerbyt, V. (2007).Aware and (dis)liking: Item- based analyses reveal that valenceacquisition via evaluative conditioning emerges only whenJournal of ExperimentalPsychology:Learning,Memory & Cognition,33acquisition: Areview of the evidence from retrospective andBehaviour Research and Therapy,40,Rachman, S. (1977). The conditioning theory of fear-acquisition:Acritical examination. Behaviour Research and Therapy,15Quarterly Journalof Experimental Psychology,48AShanks, D.R., & St. John, M.F. (1994). Characteristics of dissociableBehavioural and Brain Sciences,17,Strack, F., & Deutsch, R. (2004). Reflective and impulsivedeterminants of social behavior. Personality and SocialPsychology Review,8Walther, E., & Nagengast, B. (2006). Evaluative conditioning andthe awareness issue: Assessing contingency awareness withthe four-picture recognition test. Journal of ExperimentalPsychology:Animal Behavior Processes,32Walther, E., Nagengast, B., & Traselli, C. (2005). EvaluativeCognition and Emotion,19Received July, 17, 2007Accepted August, 16, 2007EVALUATIVE CONDITIONING