IMPLICIT AND EXPLICIT ATTITUDE CHANGE 71 :(US)be with th CS ala.Brinol.and Petty (2004)foun itive i des than di a direct e on d only for implicit bu not explicit attitud differ fro al cues with be t of a cripheral infl ces e (a) evalua such that that what typ nd on whether the ed a ropositiona the Future with the type he firs should lead to a direct effect on explicit attitudes,the aand pe :of d al.'s(1989 th effect o plicit attit tha d t auaivcneeing (or perip heral)cues and syst c (or central)features of the two an Eaoly Chail al implications of thei tot th notivational infl in t the ELM.That is,it is a a bu ation to forn ially desi rable attitudes ve judg tent of hat eople foc 106 .stating that th impa a f a p eak argu nin olds c desired judgmental n tumn, influence cognitiv umber of additional propositions typically lead o three types of motivation proposed in the HSM r number of dditional pr With regar such that they determine (a)whethe f the ive al impl uch.the strong argu ns that is c onsidered to be relevant for an evaluativ be applied te such that additi aluative iudement.Dep whether thes sitionally r sed cue te o e expertise as an in prop ase decrease the relation between associative s a US ewha ons and e urprising that th notivation should in f only two s the likelihood that a socially desired propositio issetas implicit attitud preferred outcome for an evalu tive judgment.If the automati 2002)four licit no aried as a functior argument quality.In this udy,implici Neither Brifol et al.(2002)nor Tormala et al.(2004)included a ard vegetal s were e po e after set of propositions. Rather, the attractiveness of the source is more likely to result in an EC effect, such that the positive valence of the source (US) becomes associated with the object (CS). In other words, whereas the first case (expertise) should lead to a direct influence on propositional evaluations, the second case (attractive￾ness) should lead to a direct influence on associative evaluations (see also Petty & Wegener, 1999). Hence, even though peripheral cues often differ from central cues with regard to their complexity and thus with regard to the amount of cognitive elaboration re￾quired for their processing, the distinction between central and peripheral influences is orthogonal to the distinction between associative and propositional evaluations, such that both central and peripheral cues may influence attitudes associatively or propo￾sitionally. It is important to note that what type of influence central or peripheral cues have on explicit or implicit attitudes depends on (a) whether a given cue, be it central or peripheral, is used as a propositional argument and (b) whether the same cue implies direct evaluative information that could function as a US. Whereas the first case should lead to a direct effect on explicit attitudes, the second case should lead to a direct effect on implicit attitudes. Moreover, a direct effect on explicit attitudes may additionally lead to an indirect effect on implicit attitudes, given that the persuasive message implies an affirmation (rather than a negation) of evaluative information; a direct effect on implicit attitudes, in turn, may additionally lead to an indirect effect on explicit atti￾tudes, given that recipients base their evaluative judgments on the propositional implications of their automatic affective reactions. Another important question in the context of the ELM is why argument strength usually influences explicit attitudes only under conditions of high elaboration but not low elaboration and why the impact of peripheral cues is often (but not always) attenuated under conditions of high elaboration. According to the APE model, cognitive elaboration primarily influences the number of proposi￾tions that are considered for an evaluative judgment. This assump￾tion resembles the notion of cognitive response (Greenwald, 1968; Petty et al., 1981), stating that the impact of a persuasive message depends on the cognitive responses the message elicits in the recipient. In particular, we argue that for weak arguments, a higher number of additional propositions typically leads to a rejection of the persuasive argument. For strong arguments, in contrast, a higher number of additional propositions often implies a validation of the argument, thus increasing the recipient’s confidence in the validity of the argument (Petty, Brin˜ol, & Tormala, 2002). As such, the impact of strong arguments on explicit attitudes often increases as a function of cognitive elaboration, whereas the im￾pact of weak arguments typically decreases. Similar considerations can be applied to peripheral cues, such that additionally considered propositions may challenge either the perceived validity of a propositionally processed cue (e.g., source expertise as an argu￾ment in propositional reasoning) or the perceived validity of an associative evaluation resulting from processes of EC (e.g., source attractiveness as a US). It is somewhat surprising that there is almost no research on the effects of persuasive messages on implicit attitudes. In fact, we are aware of only two studies that have used implicit attitude measures in a typical persuasion design. Brin˜ol, Horcajo, Becerra, Falces, and Sierra (2002) found that only implicit, not explicit, attitudes varied as a function of argument quality. In this study, implicit attitudes toward vegetables were more positive after exposure to a persuasive message when that message contained strong rather than weak arguments for the consumption of vegetables. Using a similar manipulation, Tormala, Brin˜ol, and Petty (2004) found that (under conditions of high elaboration) strong arguments led to more positive implicit attitudes than did weak arguments. Again, this effect emerged only for implicit but not explicit attitudes.9 Taken together, these results are consistent with our assumption that central and peripheral processing cannot be equated with influences on explicit versus implicit attitudes. Instead, central and peripheral cues may show very different patterns, such that they may influence (a) implicit but not explicit attitudes, (b) explicit but not implicit attitudes, or (c) both explicit and implicit attitudes. Moreover, which of the three potential outcomes will occur should depend on whether these cues are directed at associative or prop￾ositional processes and whether their direct influence on one kind of evaluation involves a mediating effect on the other kind of evaluation. Future research with the type of manipulations typi￾cally used in persuasion research may provide deeper insights into how exactly central and peripheral cues influence explicit and implicit attitudes. HSM. Chaiken et al.’s (1989) HSM largely corresponds to the ELM with regard to its assumptions about the influence of heu￾ristic (or peripheral) cues and systematic (or central) features under conditions of high and low cognitive elaboration (for a comparison of the two models, see Eagly & Chaiken, 1993). Thus, we refrain from restating these assumptions for the HSM. However, the HSM differs from the ELM by additionally positing several assumptions about motivational influences in persuasion that are not explicitly addressed in the ELM. That is, it is assumed that recipients may have a motivation to form accurate attitudes (accuracy motivation), a motivation to form socially desirable attitudes (impression mo￾tivation), or a motivation to form personally desired attitudes (defense motivation). These motivations are assumed to influence the content of persuasion processes, such that people focus on particular kinds of heuristic cues or systematic features for an attitudinal judgment. In addition, motivation is assumed to deter￾mine individual thresholds of desired judgmental confidence, which, in turn, influence the relative degree of cognitive elaboration. With regard to three types of motivation proposed in the HSM, we argue that these motivations primarily influence processes of propositional reasoning, such that they determine (a) whether people base their evaluative judgments on the propositional impli￾cations of their automatic affective reactions and (b) the set of propositions that is considered to be relevant for an evaluative judgment. Increased accuracy motivation, for instance, may in￾crease the number and the complexity of propositions considered for an evaluative judgment. Depending on whether these changes in the set of considered propositions confirm or disconfirm the validity of associative evaluations, increased accuracy motivation can either increase or decrease the relation between associative evaluations and evaluative judgments. In a similar vein, enhanced impression motivation should in￾crease the likelihood that a socially desired proposition is set as a preferred outcome for an evaluative judgment. If the automatic 9 Neither Brin˜ol et al. (2002) nor Tormala et al. (2004) included a manipulation of peripheral cues. IMPLICIT AND EXPLICIT ATTITUDE CHANGE 711