73 GAWRONSKI AND BODENHAUSEN affective atitude object aligns with this preferred cister No 1994:Dt Teiatiow and evalua s may be rgely nents depending align with the preferred outcome cople may eithe the valuation in both the e ELM (Pe ed propositions with the socially de oppo s of low cognitive elabor tion.attitudes are more likely t of po motivation may either increase or decrease the relation etween .more likely to be inu The al-heuristic cues is often (but not alwa ereas imp defense motivation in s the likelibe that a p centr stemati er,the that cog ither to n the strates o ach the s and evaluative in nd of high clab ration (e g Pier the di in statine that h is pers des (a)bothin without prejud ELM ntly to the n of a o th the Hsm tates that the relative degr nd and his e ed by the thre of a o lish their suffic d rea al rel s of EC pro in the App Another highly luative judgme s.the number of mo Thus. asion models is the probabil of belief orga n's actual and desired levels of iudgmental conf ews.sce Wyer.2004:Wyer&Alban cin,2005).These mod- ence.Again.these proce s should primanly influence processe els are concemed with the degree to which logical principles canaffective response to the attitude object aligns with this preferred outcome, the search for additional relevant propositions may be truncated (Baumeister & Newman, 1994; Ditto & Lopez, 1992), and evaluative judgments may be based largely on affirmation of the automatic affect. If the automatic affective response does not align with the preferred outcome, people may either change the strategy to achieve consistency within a given set of propositions or actively search for additional propositions that make the set of considered propositions consistent with the socially desired out￾come. Depending on whether these strategies confirm or discon￾firm the validity of associative evaluations, increased impression motivation may either increase or decrease the relation between automatic affective reactions and evaluative judgments. The process underlying defense motivation theoretically corre￾sponds to impression motivation, except that whereas impression motivation results in the search for socially desirable evaluative outcomes, defense motivation increases the likelihood that a per￾sonally desired proposition is set as a preferred outcome for propositional reasoning. As with impression motivation, this can lead either to a change in the strategy to achieve consistency within a given set of propositions or to a deliberate search for additional propositions that make the set of considered propositions consis￾tent with the desired outcome. Accordingly, increased defense motivation may either increase or decrease the relation between automatic affective reactions and evaluative judgments, depending on whether these strategies confirm or disconfirm the validity of associative evaluations. Despite the functional similarities between impression and defense motivation, the distinction between the two is still important given that they may differ with regard to their particular outcomes, because what is socially desirable is not always what is personally desirable. An illustrative example for this is research on prejudice and stereotyping, which has shown that (a) both internal (personal) and external (social) motivation to respond without prejudice contribute to the overt expression of prejudiced beliefs and (b) the two kinds of motivation can con￾tribute independently to the rejection of associative evaluations as a valid basis for propositional evaluations (Plant & Devine, 1998). With regard to motivational influences on cognitive elaboration, the HSM states that the relative degree of cognitive elaboration depends on the momentary difference between a person’s current level of confidence and his or her desired level of confidence. In addition, the model states (a) that people will spend only as much cognitive effort as is required to satisfy their goal-related needs, as defined by the three types of motivation we have discussed (least effort principle), and (b) that people will spend whatever cognitive effort is required to attain a sufficient level of confidence to accomplish their goals (sufficiency principle). Thus, any motiva￾tional influence on cognitive elaboration should be mediated by people’s sufficiency thresholds, such that any motivationally rel￾evant factor (e.g., personal relevance, task importance, need for cognition) should affect the desired level of judgmental confi￾dence. These aspects of the HSM can be easily applied to the notion of propositional processing embodied in the APE model. As we have noted, cognitive elaboration should primarily influence the number and complexity of propositions that are considered for an evaluative judgment. Thus, the number of momentarily consid￾ered propositions should increase as a function of the discrepancy between a person’s actual and desired levels of judgmental confi￾dence. Again, these processes should primarily influence processes of propositional reasoning. Moreover, enhanced elaboration result￾ing from larger discrepancies may either increase or decrease the relation between automatic affective reactions and evaluative judg￾ments, depending on whether enhanced elaboration leads to a confirmation or disconfirmation of the validity of associative evaluations. Unimodel. A central assumption in both the ELM (Petty & Cacioppo, 1986) and the HSM (Chaiken et al., 1989) is that under conditions of low cognitive elaboration, attitudes are more likely to be influenced by peripheral–heuristic cues rather than by central– systematic features of a persuasive message. Under conditions of high elaboration, in contrast, attitudes are more likely to be influ￾enced by central–systematic features, whereas the influence of peripheral–heuristic cues is often (but not always) attenuated. These assumptions were recently challenged by Kruglanski and Thompson (1999), who argued that both peripheral–heuristic cues and central–systematic features are processed propositionally. Moreover, they argued that cognitive elaboration influences atti￾tude changes via the complexity of propositional reasoning pro￾cesses, such that highly complex cues or arguments influence attitudes only under conditions of high cognitive elaboration, not under low cognitive elaboration. Less complex cues or arguments, in contrast, should influence attitudes under conditions of both high and low elaboration, unless their subjective relevance is overridden under conditions of high elaboration (e.g., Pierro, Man￾netti, Kruglanski, & Sleeth-Keppler, 2004). The APE model agrees with Kruglanski and Thompson’s (1999) unimodel in stating that cognitive elaboration is primarily a deter￾minant of propositional reasoning. That is, it is assumed that increased cognitive elaboration is likely to increase the number and complexity of propositions considered for an evaluative judg￾ment. The APE model also agrees with the unimodel in assuming that many peripheral–heuristic cues may be processed proposi￾tionally (see discussion of the ELM). However, the APE model goes beyond the unimodel by arguing that both peripheral– heuristic cues and central–systematic features may change propo￾sitional evaluations in two qualitatively different ways—that is, either directly by influencing the set of considered propositions or indirectly by influencing associative evaluations. Moreover, when people reject their associative evaluations as a valid basis for an explicit evaluative judgment, indirect influences of peripheral– heuristic cues or central–systematic features may still affect asso￾ciative evaluations, even when this change is not reflected in propositional evaluations of an attitude object (e.g., Dasgupta & Greenwald, 2001; Karpinski & Hilton, 2001; M. A. Olson & Fazio, 2006). As outlined in the context of the ELM, such dissociations should be particularly likely when processes of propositional rea￾soning lead to a deliberate rejection of the evaluation implied in the persuasive message but this evaluation nevertheless influences associative evaluations via processes of EC. Moreover, dissocia￾tions between associative evaluations and evaluative judgments are likely to occur when the persuasive argument implies a nega￾tion of an already activated evaluation rather than an affirmation of a new evaluation. Probabilogical models. Another highly relevant class of per￾suasion models is the probabilogical models of belief organization and change (e.g., McGuire, 1981; Wyer & Goldberg, 1970; for reviews, see Wyer, 2004; Wyer & Albarracı´n, 2005). These mod￾els are concerned with the degree to which logical principles can 712 GAWRONSKI AND BODENHAUSEN