Neither Fish nor Fowl: Implicit Attitudes as Patchy Endorsements1

Implicit attitudes are mental states that appear sometimes to cause agents to act in ways that conflict with their considered beliefs. Implicit attitudes are usually held to be mere associations between representations. Recently, however, some philosophers have suggested that they are, or are very like, ordinary beliefs: they are apt to feature in properly inferential processing. This claim is important, in part because there is good reason to think that the vocabulary in which we make moral assessments of ourselves and of others is keyed to folk psychological concepts, like ‘belief’, and not to concepts that feature only in scientific psychology: if implicit attitudes are beliefs there is a prima facie case for thinking that they can serve as the basis for particular kinds of moral assessment. In this paper I argue that while implicit attitudes have propositional structure, their sensitivity and responsiveness to other mental representations is too patchy and fragmented for them to properly be considered beliefs. Instead, they are a sui generis kind of mental state, a state I dub patchy endorsements. There is a great deal of evidence that many people have implicit attitudes that conflict with their considered beliefs, and that these implicit attitudes sometimes affect their explicit judgments and their actions.2 For instance, negative implicit attitudes toward black people seem to play a role in explaining differences in political convictions and in a range of judgments—opposition to Barack Obama’s healthcare reforms (Knowles, Lowery & Schaumberg 2010), the judgment that blacks bear more responsibility than whites for the riots that followed the acquittal of the police officers accused of assaulting Rodney King (Fazio et al. 1995), and a preference for white job applicants over equally qualified black applicants (Dovidio and Gaertner 2000). Negative implicit attitudes also predict a range of much more subtle behaviors that it seems appropriate to describe as racist: for instance making less eye contact with a black person than a white person (Dovido et al. 1997). Overt racism has declined in the United States and other Western countries very significantly over the past 40 years, yet dramatic racial inequalities persist; it is likely that negative implicit attitudes toward minorities play some role in explaining this fact (Pearson, Dovidio & Gaertner 2009). Given the range of evidence suggesting that implicit attitudes play a role in explaining discriminatory behavior, understanding their nature is a pressing concern. A better understanding of what implicit attitudes are may enable us to avoid their C © 2014 Wiley Periodicals, Inc. 800 Neither Fish nor Fowl: Implicit Attitudes as Patchy Endorsements 801 inculcation, in ourselves and in others, and to prevent their expression when they are inculcated. Further, understanding their nature is central to a proper assessment of the agents who harbor them. As Ron Mallon (forthcoming) has recently emphasized, our attributions of responsibility and other kinds of moral assessment are keyed to folk psychological concepts; to the extent to which implicit attitudes are different kinds of beasts, we may find ourselves at a loss when it comes to moral assessment. For instance, discovering that someone harbors negative beliefs about black people, even implicit ones, may warrant a higher degree of moral condemnation than a finding that they harbor negative associations alone. Perhaps the first might justify the epithet “racist”, when the second does not.3 Several philosophers have recently suggested, on the basis of evidence that implicit attitudes have propositional structure, that they are beliefs (Smith 2005, 2012; Egan 2011; Mandelbaum 2013; Mandelbaum unpublished) In this paper, I shall argue that implicit attitudes are not beliefs, not even unconscious beliefs.4 Though they are more than mere associations, they are not sensitive enough to evidence and to the semantic contents of other attitudes to qualify as beliefs. Rather, they are what I will call patchy endorsements. In section 1 of the paper, I describe the ways in which implicit attitudes are standardly probed by psychologists, in order to lay out the background for the arguments to come. In section 2, I sketch the current debate concerning the nature of implicit attitudes. This debate has focused on how much propositional structure they exhibit, with most psychologists contending that they are essentially associations, and therefore have little structure. The associative view has difficulty accommodating a range of evidence—forcefully driven home in recent work by Eric Mandelbaum—that implicit attitudes are involved in seemingly inferential processes: explaining this evidence requires us to attribute propositional structure to implicit attitudes. But evidence of propositional structure is not sufficient to establish that implicit attitudes are beliefs. The evidence that implicit attitudes have propositional structure consists in evidence that they respond to the semantic contents of other states, and beliefs are states that respond in this kind of way, but beliefs do this systematically. If implicit attitudes are states that respond to semantic contents in a patchy and fragmented way, they are neither associations nor beliefs. In section 3 I argue that implicit attitudes—specifically the implicit attitudes that cause discriminatory behavior—are neither associations nor beliefs. They are not associations for the reasons set out by Mandelbaum: they feature in contentdriven transitions. But they are not beliefs because their responsiveness to content is fragmented, and often the responsiveness they exhibit is of the wrong sort to count as genuine inference. The upshot, I claim, is that implicit attitudes are not beliefs. Nor are they associations. What, exactly, are they? Something for which folk psychology lacks a word: patchy endorsements. If this conclusion is correct, we ought to be hesitant in applying our vocabulary of moral assessment to ourselves and to one another on the basis of finding that we harbor implicit attitudes: some kinds of moral assessment may need to await the development of a theory better suited to the kinds of psychological states that implicit attitudes are.5