710 GAWRONSKI AND BODENHAUSEN an od evaluation ("Diet sof drinks are not healthier than Relations to Theories of Persuasion Zwaan 2003).More ver.becau enegating a given propositio Over the last decades.social psychological research on attitud een guid pers As a result.persuasive argu Smith-McLallen.205).Henceto judgments but not associative implying a negated positive evaluation lead to more favorabl 196).the heuristic s 1989: 1970).and the Beyond indirect effects on associative evaluations,it is impo 002 106 of this d may creat cues uation via processes of EC.Because evaluative judg ents of ar aitud object are usually bas on the proposi age.such as the expertise of the the likability f the ated fi (Case 1:see Figure 3).Thus,persuasive arguments implying an attitudes are p y to be influenced by of implicitand Figur 9g f the of h on of an old evaluation.such that the re m ore likely tobe d by the quality of the ame argument r periph ence attitudes the s is und Diet soft drinks promote when recipients con s we have alread lluded to in the ontext of the causes o usly reject the proposition that diet soft drink thus should influenc .Ac Considerin the influence Pro of periphera e one could be pted to equa n eva to the set of p sitions t are considere to b Smith DeCo 2000m.1n trast to thi for an e nts e jud imilar vein. cues are d in a pro lead to a r sociative ev ing a particula ter as an additional arer t in the se ents lead to a pre ctive if it owever.may be different when the co uter is adver APE mode the crucia peop of the in press:Gilbert.199 a spe ge ent (e effect on as iative evaluations mediated by pro frequently sitional rea soning.If.however.a persuasive argument c ontains a negation osearch, and (e) some methodological issues related to the applica￾tion of the APE model. Relations to Theories of Persuasion Over the last decades, social psychological research on attitude change has been guided largely by theories of persuasion (for reviews, see Crano & Prislin, 2006; B. T. Johnson, Maio, & Smith-McLallen, 2005). Hence, to provide a genuine integration of research on attitude change, it seems necessary to discuss both the commonalities and the differences between the APE model and persuasion models. For this purpose, we discuss the relation of the APE model to the elaboration likelihood model (ELM; Petty & Cacioppo, 1986), the heuristic systematic model (HSM; Chaiken et al., 1989; Todorov, Chaiken, & Henderson, 2002), the unimodel (Kruglanski & Thompson, 1999), probabilogical models of atti￾tude change (McGuire, 1981; Wyer & Goldberg, 1970), and the cognition in persuasion model (CPM; Albarracı´n, 2002). ELM. The core assumption of Petty and Cacioppo’s (1986) ELM is that attitudes can be influenced either by central or by peripheral cues present in a persuasive message (for a review, see Petty, Cacioppo, Strathman, & Priester, 2005). Whereas central cues refer to the quality of the arguments presented in a persuasive message, peripheral cues refer to other characteristics of the mes￾sage, such as the expertise of the source, the likability of the source, or consensus information. A well-replicated finding in research on persuasion is that under conditions of low cognitive elaboration, attitudes are more likely to be influenced by peripheral cues rather than by the quality of the arguments (cf. Kruglanski & Thompson, 1999). Under conditions of high elaboration, in con￾trast, attitudes are more likely to be influenced by the quality of the arguments, whereas the influence of peripheral cues is often (but not always) attenuated. As we have already alluded to in the context of the causes of attitude change, persuasive arguments (or central cues) are—by definition—propositional statements and thus should influence attitudes primarily by processes of propositional reasoning. Ac￾cording to the APE model, persuasive arguments contribute new propositions to the set of propositions that are considered to be relevant for an evaluative judgment, which may then influence evaluative judgments about an attitude object. Whether these pro￾cesses will, in turn, affect associative evaluations depends on the particular nature of the new propositions. If the persuasive argu￾ments lead to a retroactive rejection of an already activated asso￾ciative evaluation, these arguments should leave associative eval￾uations unaffected (Case 3; see Figure 5). If, however, persuasive arguments lead to a proactive construction of a new associative evaluation, these arguments may indirectly influence associative evaluations mediated by processes of propositional reasoning (Case 4; see Figure 6). According to the APE model, the crucial difference between these two cases is whether the persuasive argument implies an affirmation of a new evaluation or a negation of an old evaluation (Deutsch et al., in press; Gilbert, 1991). That is, if a persuasive argument contains an affirmation of a new evaluation (e.g., “Diet soft drinks promote convulsions”), propo￾sitional processing of the argument leads to corresponding changes in the activated pattern of associations, thus resulting in an indirect effect on associative evaluations mediated by propositional rea￾soning. If, however, a persuasive argument contains a negation of an old evaluation (e.g., “Diet soft drinks are not healthier than standard soft drinks”), propositional processing increases the ac￾tivation level of the original (nonnegated) evaluation (see Kaup & Zwaan, 2003). Moreover, because negating a given proposition requires a reversal of its truth value, negations usually reverse only propositional judgments but not the activation of associations in memory (Deutsch et al., in press).8 As a result, persuasive argu￾ments implying negations (i.e., a retroactive rejection of an already activated associative evaluation) should influence only evaluative judgments but not associative evaluations. In fact, associative evaluations may even show ironic effects, such that arguments implying a negated positive evaluation lead to more favorable implicit attitudes and arguments implying a negated negative eval￾uation lead to more negative implicit attitudes (see Grant, Mala￾viya, & Sternthal, 2004). Beyond indirect effects on associative evaluations, it is impor￾tant to note that persuasive arguments implying an affirmation of new evaluative information can have an additional direct effect on associative evaluations. That is, arguments of this kind may create an associative link between the attitude object and the new eval￾uation via processes of EC. Because evaluative judgments of an attitude object are usually based on the propositional implications of associative evaluations, such changes in associative structure may additionally lead to an indirect effect on evaluative judgments (Case 1; see Figure 3). Thus, persuasive arguments implying an affirmation of a new evaluation are likely to result in a complex pattern of direct and indirect effects on both implicit and explicit attitudes (Case 8; see Figure 10). Thus, arguments of this kind should generally be more effective than arguments implying a negation of an old evaluation, such that the same argument may still influence attitudes even if one of the multiple ways is under￾mined (e.g., persistent effects of EC resulting from the message “Diet soft drinks promote convulsions” even when recipients con￾sciously reject the proposition that diet soft drinks promote convulsions). Considering the influence of peripheral cues, one could be tempted to equate central influences with changes in evaluative judgments and peripheral influences with changes in associative evaluations (e.g., Smith & DeCoster, 2000). In contrast to this assumption, however, we have already proposed that central-route influences may or may not influence associative evaluations. In a similar vein, peripheral cues are sometimes processed in a propo￾sitional manner (Kruglanski & Thompson, 1999). For example, people may consider the expertise of a computer engineer recom￾mending a particular computer as an additional argument in the set of relevant propositions. Recipients may assume that the computer must be good if it is recommended by an expert. This effect, however, may be different when the computer is advertised with a highly attractive supermodel. In this case, people will probably not consider the attractiveness of the source as a valid argument in the 8 An exception to this case is when the semantic content of the negated proposition is already stored independently in associative memory. This may be the case when the negated proposition has a specific referent (e.g., “no war” automatically activates “peace”; see Mayo, Schul, & Burnstein, 2004) or the negated proposition is used frequently in language (e.g., frequent use of “no problem” automatically activates positivity rather than negativity; see Deutsch et al., in press). 710 GAWRONSKI AND BODENHAUSEN