The dominance of associative theorizing in implicit attitude research: propositional and behavioral alternatives.
Abstract: In the present article we re-examine one of the most deeply entrenched assumptions in modern attitude research, namely, that implicit social cognition is a product of associations between mental representations. More precisely, we argue that the analysis of implicit social cognition in psychology is curtailed by the widespread adoption of the associative assumption. We open with a brief overview of the attitude literature, with a particular emphasis on the fundamental structure, measurement, and conceptual differences that have emerged between implicit and explicit attitudes in recent times. Thereafter we address the influence of the associative assumption in shaping our methodologies, research questions, and theories regarding implicit and explicit attitudes. In the third and final section, we offer two alternative and perhaps complementary nonassociative models for understanding implicit cognition. While the first model situates its explanation at the mental (propositional) level of analysis and the second at the functional, each potentially allows for novel theoretical and empirical predictions and insight into attitudes above and beyond the boundaries of traditional associationism.

Key words: attitudes, implicit cognition, associative, propositional, functional
Article Type: Report
Subject: Attitude (Psychology) (Research)
Attitude (Psychology) (Psychological aspects)
Psychological research (History)
Psychological research (Methods)
Authors: Hughes, Sean
Barnes-Holmes, Dermot
Houwer, Jan De
Pub Date: 06/22/2011
Publication: Name: The Psychological Record Publisher: The Psychological Record Audience: Academic Format: Magazine/Journal Subject: Psychology and mental health Copyright: COPYRIGHT 2011 The Psychological Record ISSN: 0033-2933
Issue: Date: Summer, 2011 Source Volume: 61 Source Issue: 3
Topic: Event Code: 310 Science & research
Accession Number: 265302083
Full Text: A Brief History of Attitudes

The study of attitudes and their essential architecture--although more than 80 years old--continues to command considerable attention within social psychology (see Albarracin, Johnson, & Zanna, 2005; Crano & Prislin, 2006, 2008; Payne & Gawronski, 2010). At present, cognitive researchers broadly refer to attitudes as the integration of cognitive and affective evaluations experienced toward some object that can vary in strength (e.g., Crano & Prislin, 2006; Olson & Kendrick, 2008). According to this definition, an evaluation imputes an association between the attitude object and positive or negative valence that determines whether the individual responds in a favorable or unfavorable manner. Attitudes can be formed toward a diverse range of stimuli, from the self (e.g., Koole, Dijksterhuis, & Van Knippenberg, 2001; Pelham, Mirenberg, & Jones, 2002) to other individuals (McConnell, Rydell, Strain, & Mackie, 2008), social groups (e.g., Dovidio, Kawakami, & Gaertner, 2002), behaviors (Sherman, Rose, Koch, Presson, & Chassin, 2003), abstract psychological concepts (e.g., nationality; Devos & Banaji, 2005), and events (Esses, Dovidio, & Hodson, 2002). (1)

The Modern Renaissance in Attitude Measurement

Throughout the history of social psychology, researchers have attempted to explore individual variation in how people's attitudes are acquired, maintained, and changed across a multitude of domains. Traditionally this research localized investigation within direct measurement procedures that require participants to report on their deliberate and controlled evaluative judgments (termed explicit attitudes (2); Greenwald & Banaji, 1995; Nosek, 2007). Common examples of direct procedures include semantic differential scales, Likert scales, interviews, and focus groups. Each of these procedures capitalize on overt verbal reports to infer the individual's evaluation of the object of interest. For instance, an answer to the question "How do you feel about Black people?" would presumably provide an index of the respondent's explicit racial evaluation of Black people in general. Critically, two conditions are prerequisite if direct procedures are to provide a valid measure of explicit attitudes. First, people must possess reliable introspective access to the psychological attribute being measured. Yet research shows time and again that individuals are often limited in their ability to report their inner mental activity, even in situations where they are motivated to do so (e.g., Nisbett & Wilson, 1977; Wilson, 2009; Wilson & Schooler, 1991). Second, the outcomes generated by direct procedures should provide a sensitive index of the individual's evaluation invariant of deliberate attempts to manipulate that evaluation. Although self-report methodologies are still widely used today, researchers have long appreciated that these procedures are often contaminated by extraneous factors, such as demand characteristics and/or impression management (e.g., Cronbach, 1990; Holtgraves, 2004; Paulhus, 1989; although see Uziel, 2010). Put simply, individuals may strategically edit their privately held evaluation prior to reporting it, to better concord with prevailing social norms or the presumed expectations of the researcher. This editing or censorship process is typically beyond the researcher's control when direct procedures are deployed to target evaluations that the individual is unwilling or unable to report.

Therefore, the inaccuracy of introspection combined with motivation to strategically self-present severely threatens the validity of procedures that rely on verbal reports. Simply put, "asking people to report on their attitudes will almost always result in an answer--but it often remains unclear what exactly that answer means" (Schwarz, 2008, p. 49). With this realization, attitude research was presented with a Gordian knot--the need to capture evaluations uncorrupted by self-presentation biases or introspective inaccessibility.

Recent attempts to slice through these problems have been greatly facilitated by the evolving sophistication in attitude measurement technologies. Over the past two decades, the empirical impetus within experimental social psychology has shifted toward the development of a new heterogeneous class of indirect procedures that purportedly capture the individual's immediate, automatic, and nondeclarative evaluations (termed implicit attitudes; De Houwer, 2006a; De Houwer, Teige-Mocigemba, Spruyt, & Moors, 2009a, 2009b; Fazio & Olson, 2003; Gawronski, 2009). An arsenal of indirect methodologies has recently been advanced, including semantic and evaluative priming (e.g., Fazio, Jackson, Dunton, & Williams, 1995; Wittenbrink, Judd, & Park, 1997), the Go/No-Go Association Test (GNAT; Nosek & Banaji, 2001), the Extrinsic Affective Simon Test (EAST; De Houwer, 2003), the Affect Mis attribution Procedure (AMP; Payne, Cheng, Govorun, & Stewart, 2005), the approach-avoid task (Rinck & Becker, 2007), and the currently most well established response latency procedure, the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998).

Although methodologically diverse, each of these indirect procedures aims to provide an estimate of the individual's automatic evaluations. Typically, speeded categorization responses or subjective judgments of ambiguous stimuli are registered, but in principle one can also register other types of responses, such as activation of the sympathetic nervous system (e.g., Dotsch & Wigboldus, 2008), facial muscles (Vanman, Saltz, Nathan, & Warren, 2004), brain activity as measured by electroencephalography (EEG; He, Johnson, Dovidio, & McCarthy, 2009), and functional magnetic resonance imaging (fMRI; Stanley, Phelps, & Banaji, 2008; see De Houwer & Moors, 2010, for a review). In effect, indirect procedures do not require individuals to self-assess their attitudes; rather, attitudes are inferred from behavioral outcomes on experimental procedures and/or via the recording of physiological reactions. Such procedures are used on the assumption that a psychological attribute systematically moderates performance on these tasks, and the magnitude of that moderation provides a proxy of the underlying attitude.

Whereas these new indirect technologies have been ubiquitously adopted within social psychology (e.g., Banaji, 2001; Olson & Fazio, 2006), they have also been greeted with enthusiasm in health psychology (e.g., Teachman, Gapinski, Brownell, Rawlins, & Jeyaram, 2003; Von Hippel, Brener, & von Hippel, 2008), clinical psychology (Egloff & Schmukle, 2002), psychology and law (Greenwald & Krieger, 2006), forensic psychology (Brown, Gray & Snowden, 2009), and elsewhere throughout psychological science. There seem to be at least two central reasons why indirect procedures have been welcomed so broadly. First, as mentioned previously, by requiring the individual to self-assess his or her liking of an object, direct procedures are often exposed to a host of social desirability concerns. In contrast, indirect procedures aim to minimize the individual's ability to discriminate the relationship between his or her performance on these tasks and the attitude under. investigation by virtue of response speed and accuracy criteria or requiring judgments about ambiguous stimuli. In doing so, the degree of deliberate control exerted over responding is withheld from the individual and the potentially spurious impact of self-presentation is attenuated or eliminated. Second, in addition to solving the problem of response editing, indirect procedures apparently answer the issue of introspective inaccessibility, given that such methodologies do not require people to report on, or perhaps even be aware of, their attitudes (see De Houwer, 2006a, for a more in-depth discussion).

Relationship Between Direct and Indirect Procedures

The recent methodological revolution in attitude assessment also served to introduce an important empirical question: To what extent do direct and indirect procedures systematically capture and predict different types of behavior? In only a few years, an impressive amount of data has accumulated indicating that self-report methodologies consistently account for people's intentional and controlled behaviors (e.g., Dovidio, Kawakami, Johnson, Johnson, & Howard, 1997; Dovidio, Kawakami, Smoak, & Gaertner, 2009), whereas scores obtained on indirect metrics typically track spontaneous, immediate, and perhaps more automatic responses and judgments. Consider the research of McConnell and Leibold (2001). They found that when a participant's IAT performance revealed stronger negative evaluations toward Black individuals (relative to White individuals), their explicit social interactions with a Black experimenter were significantly more negative. In this case, automatic categorization of stimuli as either good or bad predicted behavioral consequences such that stronger implicit prejudice toward Black individuals was related to less time spent smiling and socially interacting with Black experimenters, as well as a greater number of speech errors or hesitations. Interestingly, participants' explicit reports did not reliably predict their discriminatory behaviors. Likewise, Galdi, Arcuri, and Gawronski (2008) provided evidence that automatic evaluations predicted the future voting choices of participants when they verbally reported that they were still undecided. In contrast, the future voting choices of already decided participants were found to be guided by their explicit beliefs about choice options rather than automatic evaluations. Finally, Friese, Hofmann, and Wanke (2008) reported that performance on an IAT predicted impulsive but not self-regulatory eating-and drinking-related behaviors. When participants had sufficient resources to monitor their behavior, outcomes on the self-report procedure were a strong predictor of consumption behavior, whereas indirect outcomes were not. Conversely, when participants' self-regulatory resources were depleted, the IAT demonstrated predictive power while the direct procedure was unrelated to behavioral consequences.

As we shall see, this well-established dissociation between the behaviors that direct and indirect procedures capture stimulated cognitive researchers to search for mental processes that could account for such findings.

Conceptualizing Attitudes: A Closer Look

The advent of indirect attitude measurement stimulated a hitherto unparalleled explosion of empirical research, not to mention a paradigmatic shift in the methodologies used to assess attitudes. Yet one of the most important contributions afforded by this measurement revolution was the renewed controversy concerning how attitudes should be conceptualized. The finding that direct and indirect procedures reliably predict different types of behaviors fueled the belief that distinct or separate psychological processes are captured by the two different procedures. Equally, disassociations between implicit and explicit attitudes nourished the assumption that these two types of attitudes are also distinguished from one another at the mental representational level.

Implicit and Explicit Attitudes: Mental Representation and Process Accounts

Early cognitive theorizing advocated a dispositional conceptualization of implicit attitudes as mental representations that are highly stable across time and context (e.g., Greenwald & Banaji, 1995; Smith & DeCoster, 1999; Wilson, Lindsey, & Schooler, 2000). In the eyes of dispositional theorists, implicit and explicit attitudes represent two dissociated, noninteracting types of evaluation simultaneously held toward the same object. Unlike explicit attitudes that develop in response to recent information, automatic evaluations were thought to reflect mental associations formed through early socialization experiences (e.g., De Hart, Pelham, & Tennen, 2006; Rudman, 2004). Once formed, these associations are highly robust and resistant to change, as well as stable across both context and time. Hence, the impact of contextual influences was assumed to obfuscate assessment of a person's "true" and enduring evaluative disposition as well as limit the capacity to predict subsequent behavior.

Support for this assumption was derived from the finding that explicit attitudes are often more responsive to manipulation than their automatic counterparts (e.g., Gawronski & Strack, 2004; Gregg, Seibt, & Banaji, 2006; Grumm, Nestler, & von Collani, 2009; Nosek, Banaji, & Greenwald, 2002). More recently, however, both explicit and implicit evaluations have been found to be malleable to a variety of social and contextual influences (Barden, Maddux, Petty, & Brewer, 2004; Ferguson & Bargh, 2004; Rydell & Gawronski, 2009; Wittenbrink, Judd, & Park, 2001). For instance, Dasgupta and Greenwald (2001) reported that exposing White participants to positive Black American exemplars and negative White American exemplars produced weaker implicit race effects (i.e., less of a pro-White/anti-Black bias as measured by the IAT) relative to participants in a control group. Moreover, several researchers have experimentally induced and manipulated novel implicit and explicit evaluations toward fictitious social groups in the laboratory for which no previously stored attitude could be responsible (e.g., De Houwer, 2006b; Hughes & Barnes-Holmes, 2011; Olson & Fazio, 2001, 2002; Ranganath & Nosek, 2008). Such findings support the sensitivity of both explicit and implicit attitudes to current contextual influences and challenge the view that implicit attitudes reflect highly stable trait-like associations.

In line with these findings, a number of researchers have shifted toward an active constructivist approach to attitudes (e.g., Gawronski & Bodenhausen, 2006; Petty & Brinol, 2006; Schwarz, 2007). Put simply, the very idea that attitudes are static, enduring "things" elicited from memory is rejected in favor of momentarily constructed associations that are formed as needed. In place of the dispositional approach--which views context effects as erroneous "noise" that obscures the person's "true" evaluative disposition--constructionist theorists view context sensitivity as playing a key role in attitude activation. Specifically, attitudes reflect contextually constructed evaluations that are responsive to the cognitive, motivational, and contextual influences operating for individuals at a particular point in time. Hence, this approach explains the malleability findings discussed above in terms of variations in contextual, cognitive, and motivational factors. Although implicit and explicit attitudes are distinguished by separate representations, result from different processes, and operate in different situations, constructionist models do allow for the interplay between those processes (see Gawronski & Bodenhausen, 2006).

Both dispositional and constructionist models discussed thus far treat implicit and explicit attitudes as consisting of structurally distinct mental representations. In contrast, Fazio and colleagues argue that while implicit and explicit attitudes are driven by two distinct levels of processing, they are not structurally separate from one another (e.g., Fazio & Towles-Schwen, 1999; Olson & Fazio, 2009). This single-attitude model treats both as a single form of mental representation (i.e., object-evaluation associations). However disassociations between the two types of attitudes (e.g., weak correlations, different predictive relationships with other variables) are explained in terms of the different mental processes that operate in indirect and direct measurement situations. In other words, attitudes are conceptualized along a continuum from relatively automatic to nonautomatic depending on whether motivation and opportunity to engage in effortful processing are employed.

Dual-process models of attitudes have emerged as a powerful, if not the dominant, paradigm currently active in the attitude literature (e.g., Gawronski & Bodenhausen, 2006; Ryden & McConnell, 2006; Smith & Defroster, 2000; Strack & Deutsch, 2004; Wilson et al., 2000). The key feature shared by these various accounts is an appeal to distinct mental representations and/or mental processes to explain the sometimes disassociated behaviors captured by direct and indirect procedures. Although the specific definitions applied to the elements of each cognitive model vary across researchers, they are typically referred to as associative and propositionai mental processes.

A core difference between associations and propositions resides in the manner in which knowledge is assumed to be mentally represented and the conditions under which they operate and guide behavior (see Mitchell, De Houwer, & Lovibond, 2009, for a thorough review). On the one hand, explicit attitudes are explained largely as a product of propositionai processes. These syllogistic rule-based inferences are thought to operate in a slow, deliberate, and conscious manner. Associative processes, in contrast, are argued to underlie the formation and change of implicit attitudes and are guided by perceptual similarity between stimuli as well as their spatio-temporal contiguity. Accordingly, associations are assumed to operate in a fast, automatic, and unconscious way. This distinction between explicit and implicit in terms of two separate (but potentially interacting) processes in turn guides the interpretation of the scores obtained on both direct and indirect procedures. Dual-process theorists consider outcomes on indirect metrics as generally providing a proxy for automatically activated associations in memory, while direct. procedures offer a qualitatively distinct estimate of the subjective validity of propositional evaluations of the attitude object (e.g., Gawronski & Bodenhausen, 2006; Rydell, McConnell, Mackie, & Strain, 2006).

In short, the unifying theme linking all of the foregoing models is the pre-analytic belief that implicit attitudes should be understood largely in terms of the formation, activation, and change of associations between mental representations. Yet, despite this widespread appeal to associations in the attitude literature, researchers rarely define exactly what they mean by an "association." To shed light on this issue, we now consider the associative assumption in more detail and its relationship to the study of implicit attitudes.

The Associative Assumption

Throughout the history of psychology, the explanatory concept of mental associations has enjoyed widespread appeal. Indeed, the belief that learning, for example, can be understood in terms of mental associations between stimuli shines bright in the works of early pioneers in the field, such as Thorndike (1931), Pavlov (1928), and Hull (1943; see Shanks, 2010, for a recent discussion). These associations constitute an unconscious, "automatic" excitation (or inhibition) of links between mental representations that are formed in response to the pairing of physical stimuli. Associations are such that encountering one stimulus is assumed to trigger activation in the representation of the second stimulus. However, while links are automatically and passively formed, certain environmental conditions are necessary, including spatial and temporal contiguity and/or perceptual similarity between stimuli (Shanks, 2007). Furthermore, these mental links carry no internal representational content or properties that specify the nature of the relationship between stimuli; they simply transmit contingent activation from one mental representation to another (see Figure 1).

[FIGURE 1 OMITTED]

The learning process of mental link formation is thought to apply equally to all stimuli across the various sensory modalities (Brunstrom, 2007) in a broad range of species (e.g., Hall, 2009). Not surprisingly, a number of research domains are still highly influenced by this assumption, including human learning (e.g., Shanks, 2010) and cognition (e.g., Greenwald et al., 2002). In the latter case, many cognitive researchers endorse a "theory-uncommitted" view of associations that does not specify any particular explanation of their structure and/or relationship (see Greenwald, Nosek, Banaji, & Klauer, 2005; Rothermund, Wentura, & De Houwer, 2005). In this case researchers assume that stimulus-environment interactions are coded as links between mental representations of those interactions in memory (e.g., Greenwald et al., 1998). Less often, researchers adopt "theory-committed" approaches to associative mental representations, some of which draw on semantic networks and spreading activation (e.g., Collins & Loftus, 1975; Gawronski & Bodenhausen, 2006). These researchers are explicit about how associations are thought to be represented and the situations under which they operate and interact with other mental processes as well as influence behavior.

Challenging the Associative Assumption

Although we do not reject the plausibility of an a priori associative assumption, the critical point is that any pre-analytic assumption is simply that--an assumption. When approached not as an ontological assumption but rather as an immutable truism, the associative basis of cognition is taken as a given, and thus research focuses on how associative networks operate and possibly interact with other psychological processes. Critically, the core associative hypothesis is rarely, if ever, targeted for empirical analysis itself. We would argue, however, that any assumption that forms the basis for theoretical claims or the development of methodologies must be subject to empirical verification (see De Houwer, Teige-Mocigemba, Spruyt, & Moors, 2009a). For instance, the claim that association formation is the psychological process responsible for implicit cognition is a theoretical one--an inference that extends beyond the observed behavioral performance. Equally, constructing procedures that aim to target the associative strength between mental representations presupposes that the psychological attribute of interest (i.e., implicit attitudes) is associative. Again, this theoretical claim needs to be confirmed empirically with appropriate evidence.

The Associative Assumption in Implicit Attitudes: Evidence From the Procedure, Effect, and Mental Process Levels

In the following section we provide an in-depth analysis of precisely how the associative assumption permeates the construction of indirect procedures, the interpretation of behavioral outcomes on such tasks, and the theoretical accounts developed to explain the processes underlying implicit attitudes. Recent work by De Houwer and colleagues may prove useful in this respect (e.g., De Houwer, 2006a, 2006b; De Houwer & Moors, 2007; De Houwer et al., 2009a). Specifically, they have offered a comprehensive explanatory framewrork that can be used to deconstruct (implicit) attitudes into three interrelated levels of analysis, namely, the procedure, process, and effect levels. We will now discuss the impact of associationism at each of these three levels in more detail.

Implicit Attitudes at the Procedural Level

A procedure simply constitutes a systemic structuring of a specific environmental situation aimed at generating a behavioral outcome. Procedures that tap "implicit attitudes" are that subset of all procedures that arrange the environment in such a way as to index psychological attributes (or relations between environmental events) that are automatic in nature. It is important to be clear that in examining the influence of the associative assumption at the procedural level, we do not aim to describe or even highlight every indirect procedure that is associative in nature (see De Houwer, 2009a, for an in-depth review of the functional and procedural properties of implicit attitude measures) (3). Rather, we limit our analysis to the two most widely deployed indirect assessment techniques in contemporary social psychology and evaluate whether and in what ways they reflect the associative assumption.

The first and currently most well-established protocol for indexing implicit cognition is the IAT (Greenwald & Banaji, 1995). This procedure was designed to provide an index of the strength of automatically activated associations. During one block of trials, the test of associative strength involves attaching the same response to two stimuli by presenting the associated label and target stimulus together (in close spatial proximity) on one side of the screen. (4) If a researcher is seeking to determine if Black people are associated with negative attributes, then names of Black people and negatively valenced words are paired (associated) on one side of the screen and the same response is required when either of these stimuli is presented in the middle of the screen. Relatively rapid responding when these stimuli are assigned to the same response is taken to indicate that they are associated in a mental system (i.e., exist in an associative network). Whereas physically assigning stimuli to the same or a different response in the procedure is an observable fact, the assumed mental association remains an inference and should not be granted the same ontological status as the procedural association. Unfortunately, this clear distinction between procedure and inferred process is often not maintained (e.g., Devine, Plant, Amodio, Harmon-Jones, & Vance, 2002).

A second associative methodology pervasively deployed in the measurement of implicit attitudes involves priming paradigms that target either the semantic or affective dimensions of evaluation (Fazio et al., 1995; Wittenbrink, 2007). In a typical evaluative priming task, each trial requires individuals to identify target stimuli as either positive or negative in valence. This evaluative judgment is assumed to be influenced by the valence of a prime stimulus that is presented before each of the target stimuli. In effect, the prime and target stimuli are associated in the procedure because they are presented in close temporal proximity. On half of the trials the valence of the prime and target stimuli are affectively compatible (e.g., Pleasant-Love); on other trials they are incompatible (e.g., Pleasant-Cancer). The interpretation of priming effects is based on the assumption that exposure to a (prime) stimulus activates related mental representations in memory, in turn reducing the time needed to identify the associated target stimulus. We reiterate, however, that an observable outcome resulting from a core property of the procedure (i.e., temporally contiguous presentations of stimuli) cannot be designated as ontologically equivalent to a cognitive process assumed to underlie that outcome (activation of mental representations).

Critically, the associative assumption is realized at the procedural level not only by the IAT and affective priming tasks but also by an accumulating number of new indirect protocols. Indeed, within the last 15 years a burgeoning industry of associative procedures has dominated social psychology research on implicit social cognition including the GNAT, the Single-Category IAT (Karpinski & Steinman, 2006), Single-Target IAT (Wigboldus, Holland, & van Knippenberg, 2004), the Implicit Association Procedure (Schnabel, Banse, & Asendorpf, 2006), the Single-Association Test (Blanton, Jaccard, Gonzales, & Christie, 2006), the Word Association Test (Stacy, Ames, & Grenard, 2007), the Sorting Paired Features Task (Bar-Anan, Nosek, & Vianello, 2009), and the brief IAT (Sriram & Greenwald, 2009). Though methodologically diverse, all these various procedures share a common design purpose: to provide an estimate of the strength of mental association between stimuli through pairing stimuli in space and/or time.

In short, the associative assumption has served as the conceptual bedrock upon which the vast majority of procedures designed to target implicit cognition have been built. Associative theorizing has dictated the direction of empirical research in this area by specifying the questions to ask and how those questions are to be answered (i.e., by constructing and employing associative procedures). In most cases, these tasks involve physically associating stimuli, in either time or space, and then interpreting the behavioral effect as a proxy of the extent to which the stimuli are associated as representations in memory.

Implicit Attitudes at the Effect Level

Stated technically, an implicit attitude effect is a behavioral outcome that is causally produced by a core property of the procedure in an automatic manner. Although behaviors or effects generated on indirect procedures can take many forms, they more often than not reflect performance on the speeded categorization or judgment tasks considered above. Critically, many cognitively oriented attitude researchers make the additional assumption that observable performances on indirect procedures are causally related to the individual's variation on the psychological attribute being measured (i.e., attitude) and thus that observable score acts as a proxy for mental representations or mental processes (see De Houwer, 2011). In other words, the behavioral outcomes derived from these procedures are in large part interpreted as providing an index of the strength of association between mental concepts in memory (e.g., Greenwald et al., 2002; although see Conrey, Sherman, Gawronski, Hugenberg, & Groom, 2005).

For illustrative purposes, take an IAT used to measure sexual attitudes. In this instance a pro-heterosexual IAT effect indicates that the individual responds more rapidly when required to categorize heterosexual words or images with positive stimuli and homosexual words and images with negative stimuli than vice versa. Contrastingly, a pro-homosexual IAT effect reflects an opposite response pattern (i.e., quicker categorization of Homosexual-Positive and Heterosexual-Negative). Attitude researchers often make the additional assumption that the IAT effect provides a measure of the individual's attitudes toward sexual orientation that is causally produced by automatically activated associations in memory (e.g., Inbar, Pizarro, Knobe, & Bloom, 2009). Note, however, that it does not necessarily follow that effects generated on a stimulus association procedure result from associative mental processes (see Fiedler, Messner, & Bluemke, 2006).

To extend this point, consider an evaluative priming task used to determine attitudes toward body weight. If performance on the task is more fluent (i.e., quicker and more accurate) when pictures of thin individuals are presented as primes with positive, rather than negative, words as targets, the observed priming effect results from arranging the environment in a particular way (i.e., temporally contiguous stimulus presentations). Concluding that such an effect is the result of associations in memory (between representations of thin and positive) is an additional assumption. In other words, the behavioral effects obtained with an associative procedure do not necessarily provide unequivocal evidence for the operation of associative (mental) processes. As we will see in the next section, the validity of the associative assumption is a theoretical issue determined by its explanatory, heuristic, and predictive power, all of which are empirical issues.

Implicit Attitudes at the Mental Process Level

The final level in De Houwer et al.'s (2009a) explanatory framework refers to the psychological process that is assumed to generate the effect obtained on a particular procedure. Within the implicit attitude literature, cognitive researchers often operate from the position that when a person completes a measurement procedure, a mental process within that person mediates the observed outcome, and this outcome can be used to infer the individual's variation in the attitude being measured. Hence, the role of the researcher is to construct and test theories that explain how mental representations and mental processes guide the formation, contextual-ization, and change of attitudes as well as account for behavior. For nearly 15 years now, the vast majority of theorizing about how implicit attitudes are acquired, represented, and changed has appealed to some form of associative mental process or representation. Indeed, the associative assumption has been formalized in a range of contemporary models that view the behavioral outcomes generated by indirect procedures as causally produced by the associative activation and interplay of mental representations in memory (e.g., Albarracin, Glasman, & Wallace, 2004; Fazio, 2007; Gawronski & Boclenhausen, 2006; Greenwald et al., 2002; Lieberman, Gaunt, Gilbert, & Trope, 2002; Petty, Brihol, & DeMarree, 2007; Rydell & McConnell, 2006; Smith & DeCoster, 2000; Strack & Deutsch, 2004). To illustrate this point more clearly, we will detail two process models of attitudes that have received considerable attention in the attitude literature.

The MODE model-Although it predates the measurement revolution that radically altered the research landscape, and was originally advanced to explain the relation between attitudes and behavior (Fazio, 1990), the Motivation and Opportunity as Determinants (MODE) model remains a prominent account of attitudes (Fazio, 2007; Olson & Fazio, 2009). This single-representation model treats all attitudes (i.e., implicit and explicit alike) as highly stable associations between an object and a summary evaluation of that object that are represented in memory. These associative knowledge structures are assumed to vary in strength along a continuum from weak to strong, such that the strength of object-evaluation associations dictates the attitude's accessibility from memory. For instance, if associations are strong, then an evaluative response will be elicited automatically without the person's awareness, intention, or control.

One of the central predictions of the MODE model is that the degree to which automatically activated associations guide "downstream" overt behavior depends on two factors; the cognitive resources available to the person and their motivation and opportunity to deploy them. On the one hand, sufficient motivation to respond coupled with an opportunity to do so will reduce the influence of automatic associations and produce behaviors that reflect deliberative processes. In this instance, the obtained behavioral effect will reflect deliberative propositional reasoning about specific attributes of the attitude object or an attempt to censor the response in line with normative standards. On the other hand, if the motivation and opportunity to engage in deliberative processing is low, then subsequent behaviors will be influenced by automatic associative processes. In short, the behavioral effects obtained on direct and indirect procedures are assumed to reflect a single type of mental representation (object-evaluation associations) that operates under two conditions (automatic vs. deliberative).

The Associative-Propositional Evaluation (APE) model. A second, influential dual-attitude model is Gawronski and Bodenhausen's (2006, 2007a) Associative-Propositional Evaluation (APE) model. The core assumption of the model is that implicit and explicit attitudes are rooted in two conceptually distinct kinds of mental processes (i.e., associative linking vs. propositional reasoning) that function under two very different types of conditions (i.e., automatic vs. controlled). Associative processes are thought to provide the basis for implicit attitudes and are defined as the activation of links between mental representations in memory independent of the assignment of truth-values. These associations do not reflect subjectively endorsed evaluations but instead reflect the automatic activation of links between mental representations formed through (a) perceptual similarity and (b) spatiotemporal contiguity. Once formed, these associations are stored in a network that is selectively activated in response to particular contextual influences (e.g., the momentary set of input stimuli). On the other hand, propositional processes consist of higher order, syllogistic inferences. Whereas associations are activated irrespective of whether the individual considers them to be true or false, propositions function to validate activated associations against other momentarily available propositions and are thus dependent on truth-values.

Gawronski and Bodenhausen (2006) argue that while these two mental processes are conceptually independent from one another, they are also bidirectionally interactive. On the one hand, associative processes can influence propositional processes. For example, when the affective reaction elicited by the activated association is consistent with the evaluative judgment implied by momentarily available propositions, then those reactions will be considered valid and provide the basis of explicit evaluation. Contrastingly, if the affective reaction elicited by automatically activated associations is not consistent with available propositional information, such affective reactions are rejected. Explicit judgements are thus formed on the basis of cognitive consistency with other momentarily considered propositions. In contrast, propositional validation processes may also influence automatic associations. This occurs when syllogistic inferences about the validity of currently accessible propositional information activate the relevant associations stored in memory. Put simply, the behavioral effects generated on indirect procedures typically reflect the activation of unqualified associations in memory, whereas performance on direct measures provides a proxy for the propositional validation of activated associations.

Conclusion. Understanding implicit attitudes in terms of associative mental processes is not a unique feature of the aforementioned models but rather a common thread running through much of modern attitude theorizing. Although mental process models vary in the type of representations that are assumed to be associated, conditions in which those representations operate and interact with other mental processes, as well as influence behavior, they each share an appeal to an associative mechanism as the core process responsible for implicit attitudes.

Summary

The foregoing analysis highlights that many researchers are guided by an associative approach when measuring, interpreting, and theorizing about implicit social cognition. Thus, when researchers say that implicit attitudes are associative, they can mean several things. First, it could imply that the procedure presents stimuli in close temporal and/or spatial contiguity, under certain conditions, to assess whether this leads to a behavioral effect. Here the associative nature of implicit attitudes refers to what the researcher does. Second, saying that the behavioral outcomes generated on these procedures are associative means that the obtained score is a product of physically associating stimuli in a certain manner. When defined as such, the associative nature of implicit attitudes refers to what the researcher finds. Finally, implicit attitudes can be understood in terms of the theoretical process of automatically activated associations in memory that produce the implicit effect observed on the indirect procedure. In this respect, the associative approach to implicit attitudes reflects what the researcher assumes.

Possible Alternatives to the Associative Approach

Although the associative tradition has provided a rich conceptual framework for the empirical study of implicit attitudes, we argue that in the interest of a flexible and progressive science, alternative approaches should be considered and explored. If attitude research is always conducted within the remit of this mctatheoretical framework, possibly useful and productive alternatives may be missed. As we have outlined, contemporary procedures designed to tap implicit attitudes almost exclusively do so by associating stimuli in close temporal and/or spatial proximity. Given the prevalence of these procedures, attitude researchers may become reluctant to accommodate nonassociative alternatives as providing a measure of implicit attitudes. This resistance may occur because the attribute captured by those procedures is not associative and therefore deemed not implicit. Furthermore, by defining implicit attitudes exclusively in terms of one mental process (i.e., automatic activation of linked concepts in memory), the viability of alternative (nonassociative) cognitive and functional theories is thoroughly undermined. In other words, it could be argued that the a priori associative assumption has become so woven into the fabric of attitude research that it is treated as an immutable truism that no longer requires research investigation. Consequently, the opportunity for novel theoretical expansion that deviates from this metatheoretical position is hampered. Indeed, the constraining effect of such an approach is evident in the fact that only now, after more than 20 years, the role of associations in implicit attitudes is beginning to be challenged.

The final section in this article will mount such a challenge in the form of two new, nonassociative, and potentially compatible conceptual frame-works in the study of attitudes. Although the first model situates its explanation at the cognitive (propositional) level of analysis and the second at the functional, each potentially allows for novel theoretical and empirical predictions about automatic social cognition above and beyond the boundaries of associationism.

The Propositional Account of Implicit and Explicit Attitudes

It is interesting to note that work in the field of human learning has paralleled to some extent the recent theoretical controversies observed in the attitude literature. Traditionally, human learning has been understood as the product of two independent and potentially competing systems (e.g., Evans, 2008; Kahneman, 2003; Sloman, 1996). One system is framed as a higher order conscious reasoning system that results in conscious propositional knowledge. The other learning system is thought to be associative and functions on the basis of the unconscious, automatic link-formation mechanism discussed previously. However, a number of researchers have sparked considerable debate by reassessing this dual-process conceptualization of learning in light of contemporary research. They contend that associative learning should be conceptualized not as a consequence of the automatic formation of excitatory or inhibitory links between stimuli and mental representations but, rather, as a product of propositional reasoning processes (e.g., De Houwer, 2007, 2009a; Lovibond & Shanks, 2002; Mitchell et al., 2009).

Insofar as this single-process model of learning has some explanatory or predictive value, it follows that such a conceptual analysis could also inform our theories concerning implicit and explicit attitudes. In other words, if one argues that a single psychological process (propositions) accounts for human learning, then it should follow that both implicit and explicit attitudes are a product of propositional learning. Given that it is generally assumed that propositions can determine explicit attitudes (i.e., nonautomatic evaluations of objects; see Gawronski & Bodenhausen, 2007a, 2007b), the challenge for a propositional model of attitudes is to explain how implicit attitudes (i.e., automatic evaluations of objects) can come about. Perhaps surprisingly, little is needed to formulate a basic propositional account of implicit attitudes. It suffices to assume that propositions can be activated automatically from memory. For example, once the proposition "I like Bob" has been formed based on the information that Bob often helps elderly people cross the road, this proposition can be stored in memory. The memory representation is propositional in nature in that it not only contains information about a link between Bob and positive valence but also specifies the nature of the link, namely, that it is me (and not necessarily someone else) who likes Bob. Once the proposition is stored in memory, it can be activated automatically (e.g., very quickly, without having the goal to retrieve that knowledge, or without being conscious of the retrieved information). There is no a priori reason why propositional knowledge could not be activated automatically from memory and thus lead to automatic evaluative reactions. It might also be the case that the automatic activation of nonevaluative propositions can lead to implicit attitudes. For instance, the proposition "Bob helps elderly people cross the street" could lead to an automatic positive response to Bob even when the proposition "I like Bob" has not been formed.

An important new prediction that can be derived from such a propositional model of attitudes is that (implicit) attitudes should be affected by information about the nature of the relation between two stimuli. The propositions "Bob helps the elderly" and "Bob robs the elderly" are associatively identical (Bob is associated with the elderly) but should form a propositional perspective leading to two very different attitudes. The propositional model of implicit attitudes implies that these differences should become apparent also in implicit measures of attitudes. Note, however, that this prediction holds only if information about the nature of the relation is retrieved from memory. In principle, it is possible that a proposition such as "Bob helps the elderly" is retrieved only partially. For instance, the presentation of Bob could lead to the activation of the concept "elderly" without retrieval of the nature of the relation between Bob and "elderly." Such a partial retrieval could occur when there is a long delay between the moments of storage and retrieval of the proposition or when very little time is available for retrieving the propositional information. In such cases, it would appear as if implicit attitudes are driven by associations (i.e., a mere link between two representations). The propositional account predicts that the probability that implicit attitudes seem to be impervious to information about the nature of the relation between stimuli will be moderated by those variables that determine the likelihood that information about the nature of the relation has been stored in or retrieved from memory.

Another set of predictions is derived from the fact that propositions about relations can result not only from the actual pairing of stimuli (i.e., spatiotemporal contiguity in the presence of stimuli) but also from other sources of information. For instance, propositions can be formed on the basis of instructions, inferences, and intervention (see De Houwer, 2009b). The propositional account therefore predicts that implicit attitudes will be influenced not only by the actual pairing of stimuli but also by other sources of information that are relevant for the formation and truth evaluation of propositions. Initial evidence indeed suggests that implicit attitudes can result from instructions (e.g., De Houwer, 2006a, 2006b). In sum, a propositional account of implicit attitudes not only is possible but also leads to a host of new predictions.

A Functional Approach to the Study of Attitudes

As we have seen, attitude research has traditionally focused on the causal role that mental representations, processes, and memory traces play in explaining the relation between external stimuli and overt responses. Interestingly, a number of authors have recently offered a purely functional approach to the study of attitudes that makes no such appeal to mediating mental mechanisms. This novel account, stemming from a philosophical tradition known as functional contextualism, is inherently concerned with understanding and influencing the environmental regularities responsible for producing behavioral performances, including those obtained on direct and indirect procedures (see Hayes, Barnes-Holmes & Roche, 2001). As such, this approach differs from the cognitive tradition in two fundamental ways. First, functional researchers treat behavior as an "act-in-context" (e.g., Biglan & Hayes, 1996; Fox, 2006). This context can be public or private (e.g., remembering, imagining) and can vary from the most immediate behavioral instance to temporally delayed and remote behavioral sequences. Moreover, behavior does not have to be mechanistically caused by preceding stimuli. Instead, functional relations between environment and responses can extend across both time and context. Second, and perhaps more importantly for the current article, behavior is not explained in terms of unobservable hypothetical constructs or mediating mental processes within the organism. Instead, behavior is understood in terms of behavioral principles, which refer to abstract patterns of generic interactions between the individual and their environment.

Consequently, the functional approach is sometimes subject to the misconception that because it situates its analysis at the environment-behavior level (rather than at the mental representational level), it is incapable of dealing with psychological phenomena that are typically located in a mental domain and, ipso facto, disinterested in cognitive phenomena. These beliefs have no doubt been fueled by the minimal interaction between cognitive and functional researchers in the past. However, in what follows, we aim to demonstrate that this approach does indeed provide a fertile conceptual and experimental platform for understanding and influencing cognitive phenomena--in this case, attitudes. Moreover, the fact that cognitive and functional researchers approach the study of attitudes at distinct and independent levels of analysis presents the interesting possibility that these intellectual traditions may in some respects be mutually-supportive (see De Houwer, 2011).

Relational Frame Theory

Despite the historical tension between the cognitive and functional traditions, the latter has sought to provide a viable explanation for the subject matter of the former. Indeed, for well over IS years now, a technical and conceptual account of human language and cognition has been articulated within the behavior-analytic tradition termed Relational Frame Theory (RFT; Hayes et aL, 2001). Broadly speaking, this theory differentiates between two different ways in which humans can respond to relations between stimuli. On the one hand, both humans and nonhuman animals can learn to discriminate the relation between stimuli on the basis of their formal or physical properties (e.g., selecting between the largest of two stimuli). In this case, relational responding comes under the control of the nonarbitrary properties of the to-be-related stimuli. On the other hand, a substantive body of research shows that verbal humans can relate stimuli not only on the basis of their formal physical properties but also according to arbitrary contextual and social cues (see Hayes et al., 2001; Rehfeldt & Barnes-Holmes, 2009; Sidman, 1994). This type of relational activity is termed arbitrarily applicable relational responding and, according to RFT, provides the basic units of human language and cognition.

Arbitrarily applicable relational responding is defined according to three core properties: mutual entailment, combinatorial entailment, and the transformation of stimulus functions (Hayes et al., 2001). To illustrate these properties: Imagine that an individual is directly taught, through either trial-and-error learning or direct instruction, that stimulus A is the same as stimulus B and stimulus B is the same as stimulus C. In this scenario, mutual entailment refers to the functional and bi-directional relation between two stimuli, namely, that if A is the same as B, then humans will derive (in the absence of training) that B is the same as A. Combinatorial entailment refers to the functional relations between two or more stimuli. In the current example, if A is the same as B and B the same as C, then verbally competent humans will derive that A is the same as C and C is the same as A. That is, humans can relate events in the absence of a direct learning history. A final property of relational responding, especially relevant to attitude researchers, is the finding that the behavioral functions of one stimulus may be transformed when it participates in a derived relation with other stimuli. For illustrative purposes, imagine that stimulus A (e.g., the word vomit) has a negative valence, and then subsequently it comes to be related as equivalent to a number of other stimuli (B, C, D). Given appropriate contextual cues, these other stimuli will acquire the negative evaluative functions of A.

Humans can learn to relate stimuli in multiple different ways, and RFT refers to these generic patterns as relational frames. In the above example the three stimuli are said to participate in a relational frame of coordination or equivalence. However other forms of relational frames are equally viable, including similar, opposite, hierarchy, comparison, temporal relations, cause/ effect, and deictic relations. In each case, the particular relational frame and the functions that may be transformed in accordance with that relational frame are determined by the contextual cues operating for the organism in that situation.

The ability to respond in accordance with multiple types of relational frames allows for more complex relational activity, referred to as relational networks. Such networks provide a functional account of increasingly complex verbal and cognitive activity. For example, the sentence "All my friends are White, but I am not racist" may be interpreted as responding in accordance with a coherent network of relational frames. The first part of the sentence involves responding in accordance with deictic and hierarchic relations, in which self (a deictic relational response) is hierarchically related to friends and friends, in turn, is hierarchically related to White. The second half of the sentence involves another deictic relation, but in this case self participates in a frame of distinction with the stimulus racist, which has negative evaluative functions. Finally, the word but functions as a contextual cue that establishes a frame of distinction between the two halves of the sentence (i.e., all White friends does not equal racism). In principle, all verbal phenomena, including story-telling, analogical reasoning, and complex instructional control, may be interpreted using this strictly functional account (see Hayes et al., 2001, for a detailed treatment).

Over the past two decades this relational account of human language and cognition has gained momentum and stimulated a growing body of empirical research (e.g., O'Hora et al., 2008; Weinstein, Wilson, Drake & Kellum, 2008), applications (e.g., Hayes, 2004; Hayes, Luoma, Bond, Masuda, & Lillis, 2006), and conceptual controversy (see Gross & Fox, 2009, for an overview). Recently, a number of researchers have argued that derived stimulus relations and the transformation of stimulus functions may also provide a basis for the study of attitudes (e.g., Grey & Barnes, 1996; Watt, Keenan, Barnes, & Cairns, 1991). In other words, attitudes are conceptualized as involving the acquisition of positive or negative evaluative functions based on arbitrarily applicable relational responding. The critical point here is that patterns of relational framing, as well as the transformation of stimulus functions, are all defined as overarching operant behaviors that are made possible through an extensive history of appropriate exemplars typically provided through interaction with the verbal community. When such a definition is applied, attitudes are explained not in terms of mental constructs or mediating mental processes but rather as the behavioral consequences of observable environment-behavior interactions. Put simply, attitudes are not hypothetical constructs that influence behavior and require a subsequent explanation. Attitudes are the behavior in question and can be understood in terms of behavioral principles.

The Relational Elaboration and Coherence (REC) model. Thus far we have offered a relational interpretation of attitudes in the broadest sense. However, it is incumbent upon RFT to articulate a model that can accommodate the behaviors captured by both direct and indirect procedures in the literature. One recent account emerging from RFT offered in this respect is the Relational Elaboration and Coherence (REC) model (Barnes-Holmes, Barnes-Holmes, Stewart, & Boles, 2010). At the core of the approach is the notion that relational responses, like all behaviors, unfold across time. Thus, when a stimulus is encountered, a relational response will occur relatively quickly and then may be followed by additional relational responses. These additional relational responses may occur toward the stimulus itself or toward the initial response to the stimulus. With sufficient time, these additional relational responses will likely form a coherent relational network. Imagine, for example, that a White participant is shown a picture of a Black man holding a gun. The first relational response to occur might involve a negative evaluation based on a verbal history in which Black men are portrayed (in the media) as dangerous and violent. However, additional relational responding may involve a different evaluation, such as "He may be a police officer" and/or "reacting on the basis of race is wrong," and so on. In effect, relational responding may be relatively brief and immediate or involve relational networks that extend beyond the initial response.

From the perspective of the REC model, brief and immediate relational responses provide the basis for what researchers commonly term implicit attitudes. Conversely, the extended relational responding that is needed to produce a response that coheres with one or more other relational responses in the person's behavioral repertoire provides the basis for so-called explicit attitudes. In short, the behavioral effects captured by direct and indirect procedures are thought to reflect the operation of the same behavioral process (i.e., arbitrarily applicable relational responding). However, two broad patterns of relational responding can be obtained--either brief and immediate or extended and elaborated--depending on the properties of the measurement situation (e.g., time-pressured).

The REC model also provides an explanation for the dissociation that may emerge between direct and indirect procedures, in terms of relational coherence. A relational network is said to cohere when all of the individual elements relate to each other in a manner that is consistent with the reinforcement history typically provided by the verbal community for such relational responding. According to RFT, the verbal community constantly reinforces coherence (and punishes incoherence) within relational networks, to the extent that relational coherence itself becomes a type of conditioned reinforcer for most language users. Imagine, for example, that you read the statement, "Tom is taller than Mark and Mark is taller than Peter, but Peter is taller than Tom." In this case, it is likely that you would recognize the incoherent nature of this simple relational network and question its veracity. (5)

The same search for relational coherence applies to our own verbal behavior. For example, responding to a picture of a Black man as "dangerous" (with no additional information) may not cohere readily with other, subsequent relational responses that follow that initial response, such as "I treat everyone equally, regardless of race." In this case, the individual has produced an incoherent relational network, and thus additional relational responding may follow in an effort to resolve the incoherence. This may involve responding to the initial response as "wrong" or mistaken, and thus divergence between implicit and explicit attitudes would be observed. In other words, it is assumed that individuals may "reject" their immediate and brief relational responses (or automatic evaluations) if they do not cohere with their more elaborate and extended relational responding. In some contexts, however, additional relational activity may reduce or remove the incoherence within a network. For instance, when the functions of the original stimulus are transformed, this may remove incoherence. In the example above, the individual may conclude that the Black man in the picture does, in fact, look rather dangerous, which would thus cohere with the original brief and immediate relational response to the picture. In short, brief and immediate evaluative responses may or may not cohere with subsequent relational responding--when they cohere, direct and indirect procedures will typically converge, but when they do not, the procedures will typically diverge.

The REC model thus attempts to explain the divergence in behavioral effects generated on implicit and explicit attitude procedures by appealing to the same process of arbitrarily applicable relational responding, but focusing on the extent to which such responding is brief and immediate or extended and coherent. (6) It should be noted that when implicit cognition is viewed as relational, rather than strictly associative, an alternative, nonassociative measure of implicit attitudes quickly becomes possible. A new methodology has recently been advanced in this respect, termed the Implicit Relational Assessment Procedure (IRAP; Barnes-Holmes et al., 2006).

A functional measure of implicit social cognition. The IRAP is a computerized response latency procedure designed to target stimulus relations rather than mental associations in memory. Specifically, the task involves presenting relational terms (e.g., similar, opposite, more than, less than) so that the properties of the relations among the relevant stimuli can be assessed. Similar to other response-latency methodologies, the IRAP involves asking participants to respond quickly and accurately in ways that are either consistent or inconsistent with their preexperimen-tally established verbal relations. The rationale behind the IRAP is that responding should be faster on consistent (e.g., Love Similar to Pleasant) relative to inconsistent trials (e.g., Love Opposite Pleasant) because brief and immediate relational responding will coordinate more often with consistent overt responding. The response time differential between consistent and inconsistent trials (defined as the IRAP effect) is assumed to provide a nonrelative index of the strength of the verbal or relational responses being assessed.

To illustrate this more clearly, consider an IRAP designed to index brief and immediate relational responding toward Black and White people. On each IRAP trial, one of two label stimuli--"Black Person" or "White Person"--is presented at the top of the computer screen with either a positive (e.g., "Safe," "Peaceful," "Intelligent") or negative (e.g., "Violent," "Dangerous," "Stupid") target stimulus presented in the center of the screen, and participants are required to choose between one of two response options (e.g., "Similar" and "Opposite") presented at the bottom left and right of the screen. During a block of consistent trials, participants have to respond in a manner assumed to reflect prevailing verbal contingencies for White people (e.g., choosing Similar given White Person and Safe) clears the screen for 400 ms and presents the next trial. If an inconsistent response is emitted (e.g., choosing Opposite given White Person and Safe), a red "X" appears immediately under the target stimulus. To remove the red "X" and continue to the 400-ms intertrial interval, participants are required to emit the consistent response. In contrast, during inconsistent blocks participants are required to make an inconsistent response in order to progress from one trial to the next (a consistent response produces the red "X").

The IRAP typically consists of a minimum of two practice blocks and a fixed set of six test blocks. Each block presents the same number of trials, comprising what are defined as four different trial-types. The trial-types are created by presenting each label stimulus with each of two sets of target words (see Figure 2 for a schematic representation of the IRAP). Given the previous example, a block of consistent trials thus requires the following pattern of responses: White Person-Positive-Similar; White Person-Negative-Opposite; Black Person-Positive-Opposite; Black Person-Negative-Similar. A block of inconsistent trials requires the opposite response pattern. The feedback contingencies are reversed across successive blocks of the IRAP, and thus participants are exposed to an alternating sequence of consistent and inconsistent blocks. For each block of IRAP trials, participants are typically required to reach a standard of 80% correct responses and a median response time of less than 2,000 ms. Failure to maintain these criteria across successive test blocks results in the removal of data (see Barnes-Holmes et al., 2010, for a more detailed overview of the procedure).

[FIGURE 2 OMITTED]

The IRAP differs from existing associative methodologies that target implicit attitudes in that neither spatial nor temporal contiguity is manipulated across the task--the presentation of the label and target stimuli remains unchanged throughout. However, the pattern of responding required by participants does change (responding Similar to White and Safe in one block but responding Opposite in another block), and thus the outcome of the procedure is not readily attributable to the spatial and/or temporal association of stimuli within the procedure itself (see below for discussion of implications).

The IRAP effect has now been replicated across a growing number of domains, ranging from implicit social stereotyping (Barnes-Holmes, Murphy, Barnes-Holmes, & Stewart, 2010; Power, Barnes-Holmes, Barnes-Holmes, & Stewart, 2009), implicit attitudes toward work and leisure (Chan, Barnes-Holmes, Barnes-Holmes, & Stewart, 2009), implicit ageism (Cullen, Barnes-Holmes, Barnes-Holmes, & Stewart, 2009), implicit self-esteem (Vahey, Barnes-Holmes, Barnes-Holmes, & Stewart, 2009), deviant implicit attitudes in child sex offenders (Dawson, Barnes-Holmes, Gresswell, Hart, & Gore, 2009), and experimentally induced evaluations (Hughes & Barnes-Holmes, 2011).

In situations where "socially sensitive" relational responses are targeted, the IRAP captures and predicts behaviors not accounted for by traditional self-report procedures. Moreover, recent work has provided compelling evidence that participants cannot manipulate the magnitude or direction of the IRAP effect even when given direct instructions to do so (McKenna, Barnes-Holmes, Barnes-Holmes, & Stewart, 2007). Finally, the IRAP demonstrates comparative levels of predictive validity to well-established procedures such as the IAT (Barnes-Holmes, Murtagh, Barnes-Holmes, & Stewart, 2010; Barnes-Holmes, Waldron, Barnes-Holmes, & Stewart, 2009; Roddy, Stewart, & Barnes-Holmes, 2010). Roddy et al. (2010), for example, investigated weight bias and found that both the IRAP and IAT correlated with each other and accounted for higher levels of body-size bias than revealed by self-report procedures.

Critically, in each of the foregoing studies participants had to respond directly to stimulus relations (or propositions), rather than to simple stimulus pairings or associations. Nevertheless, the IRAP produced behavioral effects that are typically defined as implicit attitude effects on associative procedures. These findings could be seen as counterintuitive from an associative perspective--the IRAP targets propositions and yet it yields effects that are typically attributed to associations. In time, it may be possible to develop an associative account of the IRAP effect. Nevertheless, the fact remains that thinking in relational rather than associative terms about implicit cognition yielded a new methodology and data that highlight possible limitations to a strictly associative approach.

The REC model provides one potential (nonassociative) explanation for the above-mentioned IRAP effects. According to this account, each IRAP trial involves asking participants to respond to the relationship between two stimuli. Each trial will thus cause the participant to emit a brief and immediate relational response prior to pressing the appropriate computer key. The probability of this response will be determined by the participant's prior learning history combined with current contextual variables. By definition, the most probable response will be emitted first most often. Thus, on IRAP trials correctly designated as consistent with the participant's history, and current contextual variables, the required key press will coordinate with the emitted response--producing faster response latencies. In contrast, inconsistent IRAP trials require a key press that opposes the most immediate relational response emitted by the individual; therefore, it occurs less quickly.7Thus, across multiple trials, the average latency for inconsistent blocks will be longer than for consistent blocks. In other words, given that consistent blocks are assumed to require responding in accordance with historically established stimulus relations (and inconsistent blocks, precisely the opposite), responding is assumed to be more fluent in the former relative to the latter. In short, the magnitude and direction of the IRAP effect (the difference between the response fluency in consistent and inconsistent trial blocks) reflect the strength or probability of the targeted relations. It should be noted that this interpretation of the IRAP effect precludes any appeal to mediating mental constructs and instead formulates an explanation in terms of behavioral events that may occur either publicly or privately.

Conclusion

To summarize, RFT (and, more precisely, the REC model) provides attitude researchers with a functional framework for understanding and analyzing the behavioral effects obtained on direct and indirect procedures. The IRAP is offered as one viable methodology for analyzing the brief and immediate relational responses that researchers typically describe as automatic or implicit. Nevertheless, we appreciate that considerable empirical work will be required to fully articulate this relational model as well as determine the validity and reliability of the IRAP effect as a measure of implicit attitudes. Work recently initiated within our laboratories has undertaken an experimental analysis of the behavioral histories and current contextual variables that form, maintain, or change implicit attitudes. We believe that such an experimental treatment of the environmental regularities that influence behavior should serve to provide an important test of the REC model and could also refine alternative theoretical accounts of implicit social cognition.

Epilogue

Over the course of 15 years the study of implicit attitudes has occupied center stage in many social psychology domains and flourished into a substantial and distinct research topic throughout psychological science. The current chapter suggests that the core concept at the heart of this research is that of mental associations. Without doubt this conceptual framework has afforded considerable utility in helping us to understand implicit attitudes at the procedural, effect, and process levels of analysis. Nevertheless, we challenge the dominance of the associative tradition in three respects. First and foremost, a sophisticated theoretical and methodological treatment of implicit attitudes cannot be fully realized within the boundaries of a single meta-the-oretical position. As long as the generic associative position is open to challenge (e.g., Mitchell et al., 2009), it is incumbent upon researchers to at least explore alternative perspectives. Doing so may protect the field of attitude research against possible biases or limitations in theoretical and methodological development. Second, research has shown that implicit attitudes indexed with associative procedures can be established via propositional knowledge (e.g., De Houwer, 2006b). Although not definitive evidence against the associative assumption (see Gawronski & Bodenhausen, 2006), it does serve to highlight that alternative, nonassociative accounts may also be worth exploring. Third, increasing evidence indicates that when stimulus relations (or in cognitive terms, propositions) are presented in the IRAP, they yield behavioral effects that overlap considerably with those found with associative procedures. Once again, an associative interpretation may be possible here, but there still remains a clear need to explore relational and/or propositional accounts. In short, there are both strategic and empirical grounds on which to question the assumption that implicit social cognition is inherently associative.

The current chapter highlighted that a nonassociative approach to the study of attitudes is indeed viable and is only now beginning to be fully appreciated and explored. In this respect, two potential alternatives have been articulated that derive from either a cognitive (propositional) or functional (relational) background. While these two accounts provide conceptually independent and separate levels of analysis, they may be complementary and perhaps mutually supportive of one another. Insofar as stimulus relations and propositions provide different ways of talking about a similar psychological domain, research on stimulus relations may be relevant to research on propositions and vice versa. For example, the IRAP could provide propositional researchers with a platform to test and further refine propositional explanations/theories. Furthermore, this research could then feed back into further behavioral analyses of the role of stimulus relations in implicit cognition (see De Houwer, 2011).

In challenging the associative assumption, and offering two possible alternatives, we recognize that researchers will adopt what they feel is the most convincing metatheoretical framework and the most effective methodologies for addressing their research questions. Nevertheless, we believe that it is important to consider and explore possible alternatives to even the most entrenched and established views within the discipline, provided that there is some empirical basis for doing so. Given that the associative assumption has been challenged within the psychology of learning, it seems reasonable to explore nonassociative accounts of implicit attitudes. We offer the current chapter in this spirit.

(1.) Controversy surrounding definitions of the attitude construct are further complicated depending on whether the researcher conceptualizes attitudes as reflecting hypothetical concepts (e.g., Gawronski & Bodenhausen, 2006) or as actual "things" residing in memory (e.g., Fazio, 2007; Krosnick, Judd, & Wittenbrink, 2005).

(2.) Following De Houwer (2006a), we favor the conceptual distinction in defining a measurement procedure as either direct or indirect on the basis of its procedural properties and the outcome or effect of a procedure as either implicit or explicit on the basis of the properties of the psychological attribute being measured.

(3.) We define a procedure as associative when it involves physically pairing stimuli according to some form of spatial and/or temporal contingency

(4.) The IAT typically presents the label and target stimuli on the computer screen, but it is also possible to associate the two stimuli with the same response via a verbal instruction in the form of "if A or B, press left."

(5.) It is worth noting that there may be some degree of conceptual overlap between relational coherence and the concept of cognitive consistency (e.g., Festinger, 1957; Gawronski & Bodenhausen, 2006). The latter refers to the process of assessing the logical consistency between two or more propositions based on the assignment of truth values and the application of syllogistic rules and logical principles.

(6.) Strictly speaking the REC model is not a single process model, given that it allows for the involvement of behavioral processes other than relational framing (e.g., respondent conditioning and primary stimulus generalization). That said, the REC model broadly explains the difference between direct and indirect procedures in terms of the elaboration and coherence involved in the single process of relational framing.

(7.) A potential behavioral explanation for the shorter latencies observed on consistent IRAP blocks is that relational coherence, as noted previously, is established by the verbal community as a conditioned reinforcer. Thus, we assume that brief and immediate relational responding coheres or coordinates more frequently than not with subsequent relevant responding in the day-to-day verbal behaviors of most individuals. Or more informally, thought and action frequently correspond with each other on a moment-to-moment basis. Consequently, in an IRAP performance, coherence between the initial brief and immediate relational response and the subsequent overt key-press will be at a higher probability for consistent relative to inconsistent trials.

References

ALBARRACIN, D., GLASMAN, L. R., & WALLACE, H. M. (2004). Survival and change in judgments: A model of activation and comparison. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 36, pp. 251-315). San Diego: Academic Press.

ALBARRACIN, D., JOHNSON, B. T., & ZANNA, M. P. (2005). The handbook of attitudes. Mahwah, NJ: Erlbaum.

BANAJI, M. R. (2001). Implicit attitudes can be measured. In H. L. Roediger, J. S. Nairne, I. Neath, & A. Surprenant (Eds.), The nature of remembering: Essays in honor of Robert G. Crowder (pp. 117-150). Washington, DC: American Psychological Association.

BAR-ANAN, Y., NOSEK, B. A., & VIANELLO, M. (2009). The sorting paired features task: A measure of association strengths. Experimental Psychology, 56, 329-343.

BARDEN, J., MADDUX, W. W., PETTY, R. E., & BREWER, M. B. (2004). Contextual moderation of racial bias: The impact of social roles on controlled and automatically activated attitudes. Journal of Personality and Social Psychology, 87, 5-22.

BARNES-HOLMES D., BARNES-HOLMES Y., POWER P., HAYDEN E., MILNE R., & STEWART I. (2006). Do you really know what you believe? Developing the Implicit Relational Assessment Procedure (IRAP) as a direct measure of implicit beliefs. The Irish Psychologist, 32, 169-177.

BARNES-HOLMES, D., BARNES-HOLMES, Y, STEWART, I. & BOLES, S. (2010). A sketch of the implicit relational assessment procedure (IRAP) and the relational elaboration and coherence (REC) model. The Psychological Record, 60, 527-542.

BARNES-HOLMES, D., MURPHY, A., BARNES-HOLMES, Y, & STEWART, I. (2010). The Implicit Relational Assessment Procedure (IRAP): Exploring the impact of private versus public contexts and the response latency criterion on pro-White and anti-Black stereotyping among white Irish individuals. The Psychological Record, 60, 57-66.

BARNES-HOLMES, D., MURTAGH, L., BARNES-HOLMES, Y., & STEWART, I. (2010). Using the Implicit Association Test and the Implicit Relational Assessment Procedure to measure attitudes towards meat and vegetables in vegetarians and meat-eaters. The Psychological Record, 60, 287-306.

BARNES-HOLMES, D., WALDRON, D., BARNES-HOLMES, Y., & STEWART, I. (2009). Testing the validity of the Implicit Relational Assessment Procedure (IRAP) and the Implicit Association Test (IAT): Measuring attitudes towards Dublin and country life in Ireland. The Psychological Record, 59, 389-406.

BIGLAN, A., & HAYES, S. c. (1996). Should the behavioral sciences become more pragmatic? The case for functional contextualism in research on human behavior. Applied and Preventive Psychology: Current Scientific Perspectives, 5, 47-57.

BLANTON, H., JACCARD, J., GONZALES, P. M., & CHRISTIE, C. (2006). Decoding the Implicit Association Test: Implications for criterion prediction. Journal of Experimental Social Psychology, 42, 192-212.

BROWN, A. S., GRAY, N. S., & SNOWDEN, R. J. (2009). Implicit measurement of sexual associations in child sex abusers. Sexual Abuse: A Journal of Research and Treatment, 21(2), 166-180.

BRUNSTROM, J. M. (2007). Associative learning and the control of human dietary behavior. Appetite, 49(1), 268-271.

CHAN, G., BARNES-HOLMES, D., BARNES-HOLMES, Y., & STEWART, I. (2009). Implicit attitudes to work and leisure among North American and Irish individuals: A preliminary study. International Journal of Psychology and Psychological Therapy, 9, 317-334.

COLLINS, A. M., & LOFTUS, E. F. (1975). A spreading activation theory of semantic processing. Psychological Review, 82, 407-428.

CONREY, F. R., SHERMAN, J. W., GAWRONSKI, B., HUGENBERG, K., & GROOM, C. (2005). Separating multiple processes in implicit social cognition: The quad model of implicit task performance. Journal of Personality and Social Psychology, 89, 469-487.

CRANO, W. D., & PRISLIN, R. (2006). Attitudes and persuasion. Annual Review of Psychology, 57, 345-374.

CRANO, W., & PRISLIN, R. (2008). Attitudes and attitude change. New York: Psychology Press.

CRONBACH, L. J. (1990). Essentials of psychological testing (5th ed.). New York: Harper & Row.

CULLEN, C, BARNES-HOLMES, D,. BARNES-HOLMES, Y., & STEWART, I. (2009). The Implicit Relational Assessment Procedure (IRAP) and the malleability of ageist attitudes. The Psychological Record, 59, 591-620.

DASGUPTA, N., & GREENWALD, A. G. (2001). On the malleability of automatic attitudes: Combating automatic prejudice with images of admired and disliked individuals. journai of Personality and Social Psychology, 81, 800-814.

DAWSON, D. L., BARNES-HOLMES, D., GRESSWELL, D. M., HART, A. J. P., & GORE, N. J. (2009). Assessing the implicit beliefs of sexual offenders using the Implicit Relational Assessment Procedure: A first study. Sexual Abuse: A Journal of Research and Treatment, 21, 57-75.

DE HART, T., PELHAM, B.W., & TENNEN, H. (2006). What lies beneath: Parenting style and implicit self-esteem. Journal of Experimental Social Psychology, 42, 1-17.

DE HOUWER, J. (2003). The extrinsic affective Simon task. Experimental Psychology 50, 77-85.

DE HOUWER, J. (2006a). What are implicit measures and why are we using them. In R.W. Wiers & A. W. Stacy (Eds.), The handbook of implicit cognition and addiction (pp. 11-28). Thousand Oaks, CA: Sage Publishers.

DE HOUWER, J. (2006b). Using the Implicit Association Test does not rule out an impact of conscious propositional knowledge on evaluative conditioning. Learning and Motivation, 37, 176-187.

DE HOUWER, J. (2007). A conceptual and theoretical analysis of evaluative conditioning. The Spanish Journal of Psychology, 10, 230-241.

DE HOUWER, J. (2009a). Comparing measures of attitudes at the procedural and functional level. In R. Petty, R. H. Fazio, & P. Brinol (Eds.), Attitudes: Insights from the new implicit measures (pp. 361-390). Mahwah, NJ: Erlbaum.

DE HOUWER, J. (2009b). The propositional approach to associative learning as an alternative for association formation models. Learning and Behavior, 37, 1-20.

DE HOUWER, J. (2011). Why the cognitive approach in psychology would profit from a functional approach and vice versa. Perspectives on Psychological Science, 6, 202-209.

DE HOUWER, JM & MOORS, A. (2007). How to define and examine the implicitness of implicit measures. In B. Wittenbrink & N. Schwarz (Eds.), Implicit measures of attitudes: Procedures and controversies (pp. 179-194). New York: Guilford Press.

DE HOUWER, J., & MOORS, A. (2010). Implicit measures: Similarities and differences. In B. Gawronski, & B. K. Payne (Eds.), Handbook of implicit social cognition: Measurement, theory, and applications. New York: Guilford Press.

DE HOUWER, J., TEIGE-MOCIGEMBA, S., SPRIIYT, A., & MOORS, A. (2009a). Implicit measures: A normative analysis and review. Psychological Bulletin, 135, 347-368.

DE HOUWER, J., TEIGE-MOCIGEMBA, S., SPRUYT, A., & MOORS, A. (2009b). Theoretical claims necessitate basic research: Reply to Gawronski, Lebel, Peters, and Banse (2009) and Nosek and Greenwald (2009). Psychological Bulletin, 135, 377-379.

DEVINE, P. G., PLANT, E. A., AMODIO, D. M., HARMON-JONES, E., & VANCE, S. L. (2002). The regulation of explicit and implicit race bias: The role of motivations to respond without prejudice. Journal of Personality and Social Psychology, 82, 835-848.

DEVOS, TM & BANAJI, M. R. (2005). American = White? Journal of Personality and Social Psychology, 88, 447-466.

DOVIDIO, J. F., KAWAKAMI, K., & GAERTNER, S. L. (2002). Implicit and explicit prejudice and interracial interactions. Journal of Personality and Social Psychology, 82, 62-68.

DOVIDIO, J. F., KAWAKAMI, K., SMOAK, N., & GAERTNER, S. L. (2009). The roles of implicit and explicit processes in contemporary prejudice. In R. E. Petty, R. H. Fazio, & P. Brinol (Eds.), Attitudes: Insights from the new implicit measures (pp. 165-192). New York: Psychology Press.

DOVIDIO, J. F., KAWAKAMI, K., JOHNSON, C, JOHNSON, B., & HOWARD, A. (1997). On the nature of prejudice: Automatic and controlled processes. Journal of Experimental Social Psychology, 33, 510-540.

DOTSCH, R., & WIGBOLDUS, D. H. J. (2008). Virtual prejudice. Journal of Experimental Social Psychology, 44, 1194-1198.

EGLOFF, B., & SCHMUKLE, S. C. (2002). Predictive validity of an Implicit Association Test for assessing anxiety. Journal of Personality and Social Psychology, 83, 1441-1455.

ESSES, V. M., DOVIDIO, J. F., & HODSON, G. (2002). Public attitudes toward immigration in the United States and Canadian response to the September 11, 2001, 'Attack on America.' Analyses of Social Issues and Public Policy, 2, 69-85.

EVANS, J. S. (2008). Dual-processing accounts of reasoning, judgment, and social cognition. Annual Review of Psychology, 59, 255-278.

FAZIO, R. H. (1990). Multiple processes by which attitudes guide behavior: The MODE model as an integrative framework. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 75-109). San Diego, CA: Academic Press.

FAZIO, R. H. (2007). Attitudes as object-evaluation associations of varying strength. Social Cognition, 25, 603-637.

FAZIO, R. H., JACKSON, J. R., DUNTON, B. C, & WILLIAMS, C. J. (1995). Variability in automatic activation as an unobtrusive measure of racial attitudes: A bona fide pipeline? Journal of Personality and Social Psychology, 69, 1013-1027.

FAZIO, R., & OLSON, M. (2003). Implicit measures in social cognition research: Their meaning and use. Annual Review of Psychology, 54, 297-327.

FAZIO, R. H., & TOWLES-SCHWEN, T. (1999). The MODE Model of attitude-behavior processes. In S. Chaiken & Y. Trope (Eds.), Dual process theories in social psychology (pp. 97-116). New York: Guilford Press.

FERGUSON, M. J. & BARGH, J. A. (2004). Liking is for doing: The effects of goal pursuit on automatic evaluation. Journal of Personality and Social Psychology, 87, 557-572.

FESTINGER, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University Press.

FIEDLER, K., MESSNER, C, & BLUEMKE, M. (2006). Unresolved problems with the "I", the "A", and the "T": A logical and psychometric critique of the Implicit Association Test (I AT). European Review of Social Psychology, 17, 74-147.

FOX, E. (2006). Constructing a pragmatic science of learning and instruction with functional contextualism. Educational Technology Research & Development, 54(1), 5-36.

FRIESE, M., HOFMANN, W., & WANKE, M. (2008). When impulses take over: Moderated predictive validity of explicit and implicit attitude measures in predicting food choice and consumption behavior. British Journal of Social Psychology, 46, 397-419.

GALDI, S., ARCURI, L., & GAWRONSKI, B. (2008). Automatic mental associations predict future choices of undecided decision makers. Science, 321, 1100-1102.

GAWRONSKI, B. (2009). Ten frequently asked questions about implicit measures and their frequently supposed, but not entirely correct answers. Canadian Psychology, 50, 141-150.

GAWRONSKI, B., & BODENHAUSEN, G. V. (2006). Associative and propositional processes in evaluation: An integrative review of implicit and explicit attitude change. Psychological Bulletin, 132, 692-731.

GAWRONSKI, B., & BODENHAUSEN, G. V. (2007a). Unraveling the processes underlying evaluation: Attitudes from the perspective of the APE Model. Social Cogniton, 25, 687-717.

GAWRONSKI, B., & BODENHAUSEN, G. V. (2007b). What do we know about implicit attitude measures and what do we have to learn? In B. Wittenbrink & N. Schwarz (Eds.), Implicit measures of attitudes: Progress and controversies (pp. 265-286). New York: Guilford.

GAWRONSKI, B., & SDRACK, L. (2004). On the propositional nature of cognitive consistency: Dissonance changes explicit, but not implicit attitudes. Journal of Experimental Social Psychology, 40, 535-542.

GREENWALD, A. G., &BANAJI, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 10Z(1), 4-27.

GREENWALD, A. G., BANAJI, M. R., RUDMAN, L. A., FARNHAM, S. D., NOSEK, B. A., & MELLOTT, D. S. (2002). A unified theory of implicit attitudes, stereotypes, self-esteem, andseif-concept. Psychological Review, 109, 3-25.

GREENWALD, A. G., & KRIEGER, L. H. (2006). Implicit bias: Scientific foundations. California Law Review, 94, 945-967.

GREENWALD, A. G., MCGHEE, D. E., & SCHWARTZ, J. K. L. (1998). Measuring individual differences in implicit cognition: The Implicit Association Test. Journal of Personality and Social Psychology, 74, 1464-1480.

GREENWALD, A. G., NOSEK, B. A., BANAJI, M. R., & KLAUER, K. C. (2005). Validity of the salience asymmetry interpretation of the IAT: Comment on Rothermund and Wentura (2004). Journal of Experimental Psychology: General, 134(3), 420-425.

GREGG, A. P., SEIBT, B., & BANAJI, M. H. (2006). Easier done than undone: Asymmetries in the malleability of implicit preferences. Journal of Personality and Social Psychology, 90, 1-20.

GREY, I., & BARNES, D. (1996). Stimulus equivalence and attitudes. The Psychological Record, 46, 243-270.

GROSS, A., & FOX, E. J. (2009). Relational frame theory: An overview of the controversy. The Analysis of Verbal Behavior, 25, 87-98

GRUMM, M., NESTLER, S., & VON COLLANI, G. (2009). Changing explicit and implicit attitudes: The case of self-esteem. Journal of Experimental Social Psychology, 45(2), 327-335.

HALL, G. (2009). Learning in simple systems. Behavioral and Brain Sciences, 32, 210-211.

HAYES, S. C. (2004). Acceptance and Commitment Therapy, Relational Frame Theory, and the third wave of behavior therapy. Behavior Therapy 35, 639-665.

HAYES, S. C., BARNES-HOLMES, D., & ROCHE, B. (Eds.). (2001). Relational Frame Theory: A Post-Skinnerian account of human language and cognition. New York: Plenum Press.

HAYES, S. C, LUOMA, J. B., BOND, F. W., MASUDA, A., & LILLIS, J. (2006). Acceptance and Commitment therapy: Model, processes, and outcomes. Behavior Research and Therapy, 44, 1-25.

HE, Y., JOHNSON, M. K., DOVIDIO, J. F., & MCCARTHY, G. (2009). The relation between race-related implicit associations and scalp-recorded neural activity evoked by faces from different races. Social Neuroscience, 4, 426-442.

HOLTGRAVES, T. (2004). Social desirability and self-reports: Testing models of socially desirable responding. Personality and Social Psychology Bulletin 30, 161-172.

HUGHES, s., & BARNES-HOLMES, D. (2011). On the formation and persistence of implicit attitudes: New evidence from the Implicit Relational Assessment Procedure (IRAP). The Psychological Record.

HULL, C.L. (1943). Principles of behavior. New York: Appleton-Century.

INBAR, Y., PIZARRO, D. A., KNOBE, J., & BLOOM, P. (2009). Disgust sensitivity predicts intuitive disapproval of gays. Emotion, 9, 435-439.

KAHNEMAN, D. (2003). A perspective on judgment and choice: Mapping bounded rationality. American Psychologist, 58, 697-720.

KARPINSKI, A., & STEINMAN, R. B. (2006). The single category implicit association test as a measure of implicit social cognition. Journal of Personality and Social Psychology, 91, 16-32.

KOOLE, S. L., DIJKSTERHUIS, A., & VAN KNIPPENBERG, A. (2001). What's in a name: Implicit self-esteem and the automatic self. Journal of Personality and Social Psychology, 80, 669-685.

KROSNICK., J. A., JUDD, C. M., & WITTENBRINK, B. (2005). The measurement of attitudes. In D. Albarracin, B. T. Johnson, & M. P. Zanna (Eds.), The handbook of attitudes (pp. 21-76). Mahwah, NJ: Erlbaum.

LIEBERMAN, M. D., GAUNT, R., GILBERT, D. T, & TROPE, Y. (2002). Reflection and reflexion: A social cognitive neuroscience approach to attributional inference. Advances in Experimental Social Psychology, 34, 199-249.

LOVIBOND, P. F., & SHANKS, D. R., (2002). The role of awareness in Pavlovian conditioning: Empirical evidence and theoretical interpretations. Journal of Experimental Psychology: Animal Behavior Processes 28, 3-26.

MCCONNELL, A. R., & LEIBOLD, J. M. (2001). Relations among the Implicit Association Test, discriminatory behavior, and explicit measures of racial attitudes. Journal of Experimental Social Psychology, 37, 435-442.

MCCONNELL, A. R., RYDELL, R. J., STRAIN, L. M., & MACKIE, D. M. (2008). Forming implicit and explicit attitudes toward individuals: Social group association cues. Journal of Personality and Social Psychology, 94, 792-807.

MCKENNA, I. M., BARNES-HOLMES, D., BARNES-HOLMES, Y, & STEWART, I. (2007). Testing the fake-ability of the Implicit Relational Assessment Procedure (IRAP): The first study. International Journal of Psychology and Psychological Therapy, 7, 253-268.

MITCHELL, C. J., DE HOUWER, J., & LOVIBOND, P. F. (2009). The propositional nature of human associative learning. Behavioral and Brain Sciences, 32, 183-198

NISBETT, R. E., & WILSON, T. D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84, 231-259.

NOSEK, B. A. (2007) Implicit-explicit relations. Current Directions in Psychological Science, 16, 65-69.

NOSEK, B. A., & BANAJI, M. R. (2001). The go/no-go association task. Social Cognition, 19, 625-664.

NOSEK, B., BANAJI, M., & GREENWALD, A. (2002). Math=male, me=female, therefore math [not equal to] me. Journal of Personality and Social Psychology, 83, 44-59.

O'HORA, D., PELAEZ, M., BARNES-HOLMES, D., RAE, G., ROBINSON, K., & CHAUDHARY, T. (2008). Temporal relations and intelligence: Correlating relational performance with performance on the WAIS-III. The Psychological Record, 58, 569-584.

OLSON, M. A., & FAZIO, R. H. (2001). Implicit attitude formation through classical conditioning. Psychological Science, 12, 413-417.

OLSON, M. A., & FAZIO, R. H. (2002). Implicit acquisition and manifestation of classically conditioned attitudes. Social Cognition, 20, 89-104.

OLSON, M. A., & FAZIO, R. H. (2006). Reducing automatically activated racial prejudice through implicit evaluative conditioning. Personality and Social Psychology Bulletin, 32, 421-433.

OLSON, M. A., & FAZIO, R. H. (2009). Implicit and explicit measures of attitudes: The perspective of the MODE model. In R. E. Petty, R. H. Fazio, & P. Brinol (Eds.), Attitudes: Insights from the new implicit measures (pp. 19-63). New York: Psychology Press.

OLSON, M. A., & KENDRICK, R. V. (2008). Origins of attitudes. In W. Crano & R. Prislin (Eds.), Attitudes and attitude change (pp. 111-131). New York: Psychology Press.

PAULHUS, D. L. (1989). Socially desirable responding. In D. M. Buss & N. Cantor (Eds.), Personality psychology: Recent trends and emerging directions (pp. 201-209). New York: Springer-Verlag.

PAVLOV, I. P. (1928). Twenty-five years of objective study of the higher nervous activity (behavior) of animals [W.H. Gantt, Trans.]. New York: International.

PAYNE, B. K., CHENG, S. M., GOVORUN, O., & STEWART, B. D. (2005). An inkblot for attitudes: Affect misattribution as implicit measurement. Journal of Personality and Social Psychology, 89, 277-293.

PAYNE, B. K., & GAWRONSKI, B. (2010). A history of implicit social cognition: Where is it coming from? Where is it now? Where is it going? In B. Gawronski & B. K. Payne (Eds.), Handbook of implicit social cognition: Measurement, theory, and applications. New York, NY: Guilford Press.

PELHAM, B. W., MIRENBERG, M. C, & JONES, j. K. (2002). Why Susie sells seashells by the seashore: Implicit egotism and major life decisions. Journal of Personality and Social Psychology, 82, 469-487.

PETTY, R. E., & BRINOL, P. (2006). A meta-cognitive approach to "implicit" and "explicit" evaluations: Comment on Gawronski and Bodenhausen (2006). Psychological Bulletin, 132, 740-744.

POWER, P. M., BARNES-HOLMES, D., BARNES-HOLMES, Y., & STEWART, I. (2009). The Implicit Relational Assessment Procedure (IRAP) as a measure of implicit relative preferences: A first study. The Psychological Record, 59, 621-640.

RANGANATH, K. A., & NOSEK, B. A. (2008). Implicit attitude generalization occurs immediately, explicit attitude generalization takes time. Psychological Science, 19, 249-254.

REHFELDT, R. A., & BARNES-HOLMES, Y. (2009). Derived relational responding: Applications for learners with autism and other developmental disabilities. Oakland, CA: Context Press/New Harbinger.

RINCK, M., & BECKER, E. S. (2007). Approach and avoidance in fear of spiders. Journal of Behavior Therapy and Experimental Psychiatry, 38, 105-120.

ROTHERMUND, K., WENTURA, D., & DE HOUWER, J. (2005). Retrieval of incidental stimulus-response associations as a source of negative priming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 482-495.

RUDMAN, L. A. (2004). Sources of implicit attitudes. Current Directions in Psychological Science, 13(2), 79-82.

RYDELL, R. J., & GAWRONSKI, B. (2009). I like you, I like you not: Understanding the formation of context dependent automatic evaluations. Cognition and Emotion, 23, 1118-1152.

RYDELL, R. J., & MCCONNELL, A. R. (2006). Understanding implicit and explicit attitude change: A systems of reasoning analysis. Journal of Personality and Social Psychology, 91, 995-1008.

RYDELL, R. J., MCCONNELL, A. R., MACKIE, D. M., & STRAIN, L. M. (2006). Of two minds: Forming and changing valence inconsistent attitudes. Psychological Science, 17, 954-958.

SCHNABEL, K., BANSE, R., & ASENDORPL, J. B. (2006). Employing automatic approach and avoidance tendencies for the assessment of implicit personality self-concept: The Tmplicit Association Procedure (lAP). Experimental Psychology, 53, 69-76.

SCHWARZ, N. (2007). Attitude construction: Evaluation in context. Social Cognition, Z5(5), 638-656.

SCHWARZ, N. (2008). "Attitude measurement." Tn W. frrano & R. Prislin (Eds.), Attitudes and attitude change (pp. 41-60). London: Psychology Press.

SHANKS, D. R. (2007). Associationism and cognition: Human contingency learning at 25. Quarterly Journal of Experimental Psychology, 60, 291-309.

SHANKS, D. R. (2010). Learning: from association to cognition. Annual Review of Psychology, 1, 273-301.

SHERMAN, S. J., ROSE, J. S., KOCH, K., PRESSON, C. C., & CHASSIN, L. (2003). Implicit and explicit attitudes toward cigarette smoking: The effects of context and motivation. Journal of Social and Clinical Psychology, 22, 13-39.

SIDMAN, M. (1994). Equivalence relations and behavior A research story. Boston: Authors frooperative.

SLOMAN, S. A. (1996). The empirical case for two systems of reasoning. Psychological Bulletin, 119, 3-22.

SMITH, E. R., & DE COSTER, J. (1999). Associative and rule-based processing: A connectionist interpretation of dual process models. Tn S. frhaiken & Y. Trope (Eds.), Dual-process theories in social psychology (pp, 323-336). New York Guilford.

SMITH, E. R., & DE COSTER, J. (2000). Dual-process models in social and cognitive psychology: fronceptual integration and links to underlying memory systems. Personality and Social Psychology Review, 4,108-131.

SRIRAM, N., & GREENWALD, A. G. (2009). The Brief Tmplicit Association Test. Experimental Psychology, 56,283-294.

STACY, A. W., AMES, S. L., & GRENARD, J. L. (2007). Word associations tests of associative memory and implicit processes: Theoretical and assessment issues. Tn R.W. Wiers & A.W. Stacy (Eds.), The handbook of implicit cognition and addiction (pp. 75-90). Thousand Oaks, frA: Sage.

STANLEY, D., PHELPS, E., & 13ANAJI, M. (2008). The neural basis of implicit attitudes. Current Directions in Psychological Science, 17(2), 164-170.

STRACK, F., & DEUTSCH, R. (2004). Reflective and impulsive determinants of social behavior. Personality and Social Psychology Review, 8, 220-247.

DEACHMAN, B. A., GAPINSKI, K. D., BROWNELL, K. D., RAWLINS, M., & JEYARAM, S. (2003). Demonstrations of implicit anti-fat bias: The impact of providing causal information and evoking empathy. Health Psychology, 22, 68-78.

THORNDIKE, E. L. (1931). Human learning. New York: Century

UZIEL, L. (2010). Rethinking social desirability scales: From impression management to interpersonally orientated self-control. Perspectives of Psychological Science, 5, 243-262.

VAHEY, N. A., BARNES-HOLMES, D., BARNES-HOLMES, Y., & STEWART, I. (2009). A first test of Implicit Relational Assessment Procedure (IRAP) as a measure of self-esteem: Irish prisoner groups and university students. The Psychological Record, 59, 371-388.

VANMAN, E. J., SALTZ, J. L., NATHAN, L. R., & WARREN, J. A. (2004). Racial discrimination by low-prejudiced Whites. Psychological Science, 15, 711-719.

VON HIPPEL, W., BRENER, L., & VON HIPPEL, C. (2008). Implicit prejudice toward injecting drug users predicts intentions to change jobs among drug and alcohol nurses. Psychological Science, 19, 7-11.

WATT, A. W, KEENAN, M., BARNES, D., & CAIRNS, E. (1991). Social categorization and stimulus equivalence. The Psychological Record, 41, 33-50.

WEINSTEIN, J. H., WTLSON, K. G., DRAKE, C. E., & KELLUM, K. K. (2008). A Relational Frame Theory Contribution to Social Categorization. Behavior and Social Issues, 17(l), 40-65.

WIGBOLDUS, D., HOLLAND, R. W., & VAN KNIPPENBERG, A. (2004). Single-target implicit associations. Unpublished manuscript.

WILSON, T. D. (2009). Know thyself. Perspectives on Psychological Science, 4, 384-389.

WTLSON, T. D., LINDSEY, S., & SCHOOLER, T. Y. (2000). A model of dual attitudes. Psychological Review, 107, 101-126.

WILSON, T. D., & SCHOOLER, J. W. (1991). Thinking too much: Introspection can reduce the quality of preferences and decisions. Journal of Personality and Social Psychology, 60, 181-192.

WITTENBRINK, B. (2007). Measuring attitudes through priming. In B. Wittenbrink & N. Schwarz (Eds.), Implicit measures of attitudes (pp. 17-58). New York: Guilford Press.

WITTENBRINK, B., JUDD, C. M., & PARK, B. (1997). Evidence for racial prejudice at the implicit level and its relationship to questionnaire measures. Journal of Personality and Social Psychology, 72, 262-274.

WITTENBRINK, B., JUDD, c. M., & PARK, B. (2001). Spontaneous prejudice in context: Variability in automatically activated attitudes. Journal of Personality and Social Psychology, 81, 815-827.

This article is also published in Spanish as a chapter in M. Valverdre & M. Alvaerz (Eds.), Current perspectives in human learning. The work was completed while the first author was in receipt of a scholarship from the Irish Research Council for Science, Engineering and Technology (IRCSET).

Correspondence concerning this article should be addressed to Sean Hughes or Dermot Barnes-Holmes, National University of Ireland Maynooth, County Kildare, Ireland. E-mail: sean.hughes@nuim.ie or dermot.barnes-holmes@nuim.ie

Sean Hughes and Dermot Barnes-Holmes National University of Ireland Maynooth

Jan De Houwer Ghent University
Gale Copyright: Copyright 2011 Gale, Cengage Learning. All rights reserved.