IMPLICIT AND EXPLICIT ATTITUDE CHANGE 713 successfully predict a person's endorsement of target propositions, evaluative judgments not only directly via processes of proposi- given his or her level of endorsement of related propositions.Thus, tional reasoning but also indirectly via changes in associative changes in the perceived likelihood of one proposition can influ- evaluations.Most important,as direct effects on associative eval- ence the perceived validity of all other propositions that are logi- uations can occur even under suboptimal reception conditions cally connected to that proposition.The most important contribu- (e.g.,Dijksterhuis,2004;M.A.Olson Fazio.2001,2002,2006; tion of probabilogical models,however,is that they provide a set see also Baeyens,Eelen,Van den Bergh,1990;Krosnick,Betz. of mathematical equations that define probabilistic relations be- Jussim,Lynn,1992),reception processes may be more impor- tween logically related propositions and their subjective likeli- tant for attitude changes resulting from a direct influence on hoods of truth.These equations not only predict a person's en- propositional reasoning but less important for attitude changes dorsement of a particular proposition on the basis of his or her resulting from a direct influence on associative evaluations. beliefs regarding the subjective likelihoods of logically related propositions;they can also be used to predict changes in the Stability of Attitude Changes endorsement of a given proposition resulting from changes in the perceived likelihood of logically related propositions. An important question in the context of explicit and implicit From the perspective of the APE model,probabilogical models attitude change concerns the relative stability of attitude changes. primarily address processes of propositional reasoning.In partic- From a general perspective,one could argue that attitude changes ular,these models are concerned with the process of achieving and should be more stable when they imply a change of the associative maintaining logical consistency within a given set of propositions structure rather than a change in pattern activation.Even though (and also with the attainment of hedonic consistency,or compat- we generally agree with this assumption,we argue that the stability ibility between beliefs and desires).More important,probabilogi- of attitude changes is empirically defined only as the temporal cal models can provide a useful extension to the APE model by consistency of associative evaluations or evaluative judgments including a probabilistic (rather than binary)interpretation of (Lord Lepper,1999;Schwarz Bohner,2001;Wilson subjective truth.Such a probabilistic interpretation may provide a Hodges,1992).Temporal consistency,however,substantially de- higher level of accuracy in the prediction of changes in evaluative pends on both associative structure and contextual factors (see judgments resulting from processes of propositional reasoning. Smith,1996).For example,changes in associative structure should Still,the APE model goes beyond the notion of propositional result in a high level of temporal consistency only if future events reasoning addressed by probabilogical models,such that it in- consistently activate the same associative pattern.However,if cludes associative evaluations as an independent form of evalua- future events are unlikely to activate the same associative pattern, tion.In addition,the APE model specifies conditions under which changes in associative structure could also result in a low level of a given factor may or may not influence associative evaluations temporal consistency.In a similar vein,changes in pattern activa- and how such changes in associative evaluations may or may not tion may show a relatively high level of temporal consistency lead to corresponding changes in evaluative judgments when future events consistently activate the same pattern of asso- CPM.A relatively recent model of persuasion is Albarracin's ciations in memory.However,changes in pattern activation may (2002)CPM.Many of the predictions implied by the CPM corre- show a low level of temporal consistency when future events spond to the predictions by the ELM(Petty Cacioppo,1986)and activate different patterns of associations.In other words,changes the HSM (Chaiken et al.,1989).Thus,we refrain from restating in associative evaluations may exhibit either a high or a low level these predictions for the CPM.However,the CPM goes beyond of temporal consistency regardless of whether these changes are these two models by including several assumptions regarding the due to differences in pattern activation or to changes in associative reception of a persuasive message,the retrieval of information structure.Moreover,whether changes in associative evaluations from memory,and the role of affective states in persuasion.From show a high or a low level of temporal consistency is not only a the perspective of the APE model,the most important extension matter of associative structure but also a matter of contextual offered by the CPM is represented by its assumptions regarding factors that influence which associative pattern is activated. reception processes.That is,the CPM proposes several sequential Similar considerations apply to evaluative judgments (for a stages in the processing of a persuasive message,which are as more detailed analysis,see Albarracin,Wallace.Glasman, follows:(a)interpretation of information,(b)identification of 2004).For instance,if people consistently base their evaluative information,(c)selection of information,(d)validation of infor- judgments on associative evaluations,propositional evaluations mation,and (e)use of information in judgment may exhibit either a high or a low level of temporal consistency. From the perspective of the APE model,these processes are depending on the temporal consistency of associative evaluations. primarily related to direct influences of a persuasive message on However.even if associative evaluations show a low level of propositional reasoning.Thus,the assumptions of the CPM regard- temporal consistency,propositional evaluations could still exhibit ing intervening variables at these stages provide a useful extension a high level of temporal consistency if people generally reject their to the APE model when it comes to understanding how processes associative evaluations as a valid basis for an evaluative judgment, of propositional reasoning are influenced by persuasive messages. consistently consider the same set of propositions,and use the However,the APE model goes beyond the CPM(a)by addition- same strategy to achieve consistency.Such imperviousness to ally including associative processes as an independent form of influence can be regarded as the hallmark of attitude strength. evaluation and (b)by including an indirect method of explicit People who are strongly committed to their evaluative judgments attitude change via changes in associative evaluations (e.g.,EC). are unlikely to expose themselves to counterattitudinal information As outlined in the context of the ELM (Petty Cacioppo,1986), and are likely to vigorously counterargue it if they encounter it we argue that both central and peripheral cues can influence (e.g..Wyer Frey.1983).Thus,even if the automatic evaluativesuccessfully predict a person’s endorsement of target propositions, given his or her level of endorsement of related propositions. Thus, changes in the perceived likelihood of one proposition can influ￾ence the perceived validity of all other propositions that are logi￾cally connected to that proposition. The most important contribu￾tion of probabilogical models, however, is that they provide a set of mathematical equations that define probabilistic relations be￾tween logically related propositions and their subjective likeli￾hoods of truth. These equations not only predict a person’s en￾dorsement of a particular proposition on the basis of his or her beliefs regarding the subjective likelihoods of logically related propositions; they can also be used to predict changes in the endorsement of a given proposition resulting from changes in the perceived likelihood of logically related propositions. From the perspective of the APE model, probabilogical models primarily address processes of propositional reasoning. In partic￾ular, these models are concerned with the process of achieving and maintaining logical consistency within a given set of propositions (and also with the attainment of hedonic consistency, or compat￾ibility between beliefs and desires). More important, probabilogi￾cal models can provide a useful extension to the APE model by including a probabilistic (rather than binary) interpretation of subjective truth. Such a probabilistic interpretation may provide a higher level of accuracy in the prediction of changes in evaluative judgments resulting from processes of propositional reasoning. Still, the APE model goes beyond the notion of propositional reasoning addressed by probabilogical models, such that it in￾cludes associative evaluations as an independent form of evalua￾tion. In addition, the APE model specifies conditions under which a given factor may or may not influence associative evaluations and how such changes in associative evaluations may or may not lead to corresponding changes in evaluative judgments. CPM. A relatively recent model of persuasion is Albarracı´n’s (2002) CPM. Many of the predictions implied by the CPM corre￾spond to the predictions by the ELM (Petty & Cacioppo, 1986) and the HSM (Chaiken et al., 1989). Thus, we refrain from restating these predictions for the CPM. However, the CPM goes beyond these two models by including several assumptions regarding the reception of a persuasive message, the retrieval of information from memory, and the role of affective states in persuasion. From the perspective of the APE model, the most important extension offered by the CPM is represented by its assumptions regarding reception processes. That is, the CPM proposes several sequential stages in the processing of a persuasive message, which are as follows: (a) interpretation of information, (b) identification of information, (c) selection of information, (d) validation of infor￾mation, and (e) use of information in judgment. From the perspective of the APE model, these processes are primarily related to direct influences of a persuasive message on propositional reasoning. Thus, the assumptions of the CPM regard￾ing intervening variables at these stages provide a useful extension to the APE model when it comes to understanding how processes of propositional reasoning are influenced by persuasive messages. However, the APE model goes beyond the CPM (a) by addition￾ally including associative processes as an independent form of evaluation and (b) by including an indirect method of explicit attitude change via changes in associative evaluations (e.g., EC). As outlined in the context of the ELM (Petty & Cacioppo, 1986), we argue that both central and peripheral cues can influence evaluative judgments not only directly via processes of proposi￾tional reasoning but also indirectly via changes in associative evaluations. Most important, as direct effects on associative eval￾uations can occur even under suboptimal reception conditions (e.g., Dijksterhuis, 2004; M. A. Olson & Fazio, 2001, 2002, 2006; see also Baeyens, Eelen, & Van den Bergh, 1990; Krosnick, Betz, Jussim, & Lynn, 1992), reception processes may be more impor￾tant for attitude changes resulting from a direct influence on propositional reasoning but less important for attitude changes resulting from a direct influence on associative evaluations. Stability of Attitude Changes An important question in the context of explicit and implicit attitude change concerns the relative stability of attitude changes. From a general perspective, one could argue that attitude changes should be more stable when they imply a change of the associative structure rather than a change in pattern activation. Even though we generally agree with this assumption, we argue that the stability of attitude changes is empirically defined only as the temporal consistency of associative evaluations or evaluative judgments (Lord & Lepper, 1999; Schwarz & Bohner, 2001; Wilson & Hodges, 1992). Temporal consistency, however, substantially de￾pends on both associative structure and contextual factors (see Smith, 1996). For example, changes in associative structure should result in a high level of temporal consistency only if future events consistently activate the same associative pattern. However, if future events are unlikely to activate the same associative pattern, changes in associative structure could also result in a low level of temporal consistency. In a similar vein, changes in pattern activa￾tion may show a relatively high level of temporal consistency when future events consistently activate the same pattern of asso￾ciations in memory. However, changes in pattern activation may show a low level of temporal consistency when future events activate different patterns of associations. In other words, changes in associative evaluations may exhibit either a high or a low level of temporal consistency regardless of whether these changes are due to differences in pattern activation or to changes in associative structure. Moreover, whether changes in associative evaluations show a high or a low level of temporal consistency is not only a matter of associative structure but also a matter of contextual factors that influence which associative pattern is activated. Similar considerations apply to evaluative judgments (for a more detailed analysis, see Albarracı´n, Wallace, & Glasman, 2004). For instance, if people consistently base their evaluative judgments on associative evaluations, propositional evaluations may exhibit either a high or a low level of temporal consistency, depending on the temporal consistency of associative evaluations. However, even if associative evaluations show a low level of temporal consistency, propositional evaluations could still exhibit a high level of temporal consistency if people generally reject their associative evaluations as a valid basis for an evaluative judgment, consistently consider the same set of propositions, and use the same strategy to achieve consistency. Such imperviousness to influence can be regarded as the hallmark of attitude strength. People who are strongly committed to their evaluative judgments are unlikely to expose themselves to counterattitudinal information and are likely to vigorously counterargue it if they encounter it (e.g., Wyer & Frey, 1983). Thus, even if the automatic evaluative IMPLICIT AND EXPLICIT ATTITUDE CHANGE 713