According to Gilbert Harman, lay attributions of character traits “tend to be wildly incorrect” (1999:329) and are “often deeply misguided” (1999:316). Because the attribution of highly inaccurate character traits leads to “much evil”, he recommends that “we should try to educate ourselves and others to stop doing it” (Harman, 1999:315).1 Harman follows the lead provided by prominent social psychologists Ross and Nisbett (1991), in explaining the alleged inaccuracy of lay attributions of character or personality traits2 by appeal to the effect of an allegedly pervasive error in person perception. This is the “Fundamental Attribution Error” (henceforth FAE). Harman’s case against a warrant for lay character trait attributions is grounded on the results of a number of experiments interpreted by him and others as directly demonstrating the existence of a pervasive FAE, as well as some studies that examine differences in individual behavior, such as Hartshorne and May (1928), which he interprets as failing to provide support for the existence of character traits (1999:316). Harman sees the FAE as the principal cause of a systematic distortion in “folk social psychology” (1999:16) analogous to the systematic distortions in folk physics unearthed by McCloskey (1983) and others. Harman appears to believe that a pervasive FAE is an established scientific result identified by social psychologists, akin to other entrenched results in science, such as the law of gravity and the theory of evolution. He refers to “the conclusions of social psychology” (2000:224), which he takes to be authoritative. His skepticism about character traits derives directly from his understanding and acceptance of these conclusions. If the FAE was a pervasive phenomenon, as Harman alleges, following Ross and Nisbett (1991), then it seems that our ordinary source of warrant for the attribution of character traits would be altogether undermined, and there would be no reason to believe that our ordinary explanatory appeals to the existence of character traits would ever be successful. Note that it would not follow from this that all possible sources of warrant for belief in the existence of character traits would be undermined.
Nor would it follow that lay inferences to the existence of aspects of personalities other than character traits are unwarranted. We will have more to say about how best to understand character traits and about the relation of character trait attribution to the FAE in Section 3 of the paper.5 Harman appears to believe that the case for the pervasive FAE is properly described in a particular book; Ross and Nisbett’s The Person and the Situation (1991). His case against there being evidence for the existence of character traits is specifically described, as an instance of summarizing Ross and Nisbett (1991) (Harman, 2000:223). (Clarke 350-68) Other philosophical proponents of the FAE also rely heavily on Ross and Nisbett (1991), Ross (1977) and a number of other works by Ross and his collaborators.6 Ross and Nisbett are leading figures in social psychology, and it might be thought that philosophers are entitled to rely on an appeal to their authority to warrant acceptance of the case for a pervasive FAE. This will not do, however. Ross and Nisbett’s The Person and the Situation (1991) appeared at a time when proponents of the FAE in social psychology were generally much more confident about the accuracy of their characterization, and about its acceptability, than they generally are today (Fein, 2001:16).
Disenchantment with the FAE affects even those who were formerly its staunchest advocates, including Lee Ross, who coined the term “fundamental attribution error” (Ross, 1977). In a response to criticism of the case for the FAE, Ross who now refers to “the purported bias” (2001:38) concedes that there is a “lack of conceptual clarity attending the term fundamental attribution error” (Ross, 2001:38). The FAE is better understood as a bias than an error, a bias that, according to its enthusiasts, has the propensity to lead to frequent error. In fact the FAE is often referred to as “correspondence bias”. According to Harman, it is “a bias towards explanations in terms of corresponding personality traits, the error of ignoring situational factors” (Harman, 2003:90). To assert that the FAE has been committed, then, is to assert a combination of three claims. First, that the person who commits the FAE explains or perceives the behavior of a third party to be the result of her possession of certain character traits. Second, that in actual fact, the behavior of the third party is not well explained by appeal to such character traits. Third, that such behavior is better explained by an appeal to the situations that the third party is in. This description of the FAE falls short of being a clear definition. I do not think that this lack of clarity can be helped, unfortunately. I am in agreement with Ross (2001) regarding the lack of conceptual clarity that attends the term FAE. Advocates of the case for a generalisable FAE, such as Ross and Nisbett (1991), locate themselves within the situations tradition in social psychology, a tradition that dates back to the 1920s and 1930s. Intuitionists hold that the situations we find ourselves in are more important determinants of our behavior than lay folk commonly believe, and that character traits are less important determinants of behavior than lay folk commonly believe. Situations can be said to come in degrees, depending on the extent to which its advocates emphasize the explanatory importance of situations and downplay the explanatory importance of character traits and other aspects of personality as components of the proper explanation of human behavior. Harman holds that character traits have no explanatory utility whatsoever, whereas he appears to believe that situations are highly explanatory of behavior. This makes him a very extreme situationist.8 Harman is not the only philosopher who has been influenced by situationist social psychology. Others include Flanagan (1991), Merritt (2000) and Doris (1998, 2002). Harman’s variant of situationism appears to be the most extreme of these. Whereas Harman advocates doing away with character trait claims altogether, Flanagan (1991) and Merritt (2000) hold that character concepts can be suitably revised to accommodate lessons gleaned from experimental social psychology. (Langan 549)
Doris’ position is closer in spirit to that of Harman than either Flanagan or Merritt. He tells us that “people typically lack character” (1998:506) and that character-based explanations “presuppose the existence of character structures that actual people do not very often possess” (2002:6) Doris is a true believer in the FAE (2002:92-106), however his case against character is more broadly based than Harman’s case, appealing to other forms of evidence, apart from evidence for the FAE. The focus of this paper is the FAE and the argument that Harman attempts to derive from it. I am not concerned to attack situationism as such. I do happen to think that situationism is a flawed philosophy, and I endorse its recent criticisms of it due to Sabini and Silver (2005) and Webber (2006). However, although the argument I present against Harman’s case against the reliability of character trait attributions serves to undermine one strand of Doris’ case against character, it does not otherwise engage with that case. The term “fundamental attribution error” was introduced by Ross in 1977. Ross (1977) re-interpreted the results of many classic experiments of social psychology, which had hitherto been interpreted in other ways, as exemplifying the influence of the FAE. Where others had said that the notorious Milgram experiment showed that peoples’ characters disposed them towards obedience to a degree that was not generally recognized, Ross (1977) argued that it was better interpreted as exemplifying people’s insensitivity to the power and subtlety of situations. Ross argued that the FAE pervaded the vast majority of lay judgments of personality, rendering these unreliable. He also argued that the FAE was the most important finding in social psychology. He was very persuasive. (Miller 365-92) Gilbert describes the initial reception, given to Ross’ (1977) case for the FAE as one of “unbridled enthusiasm” by social psychologists (Gilbert, 1998:130). In the 1980s Ross (1977) was the most cited article in social psychology (Ross and Nisbett, 1991:v) Philosophers such as Harman draw on arguments due to social psychologists to cast doubt on the reliability of attributions of character traits. The social psychologists that they rely on tend not to refer directly to character traits, but to a broader category of items inherent in persons, which they commonly describe as “dispositions.”12 Talk of dispositions in social psychology refers exclusively to dispositions inherent in persons, whether these are personality or character traits, attributions of attitude, temperament and so on. What it does not refer to are entities or events that are “outside the skin”. For the purpose of this paper, we follow this usage. As we saw earlier, Ross and Nisbett (1991), who champion the cause of the FAE, are situationists. They stress the importance of situations as determinants of behavior (1991:27-46) and they stress the extent to which “lay dispositionalists” fail to appreciate the importance of variations in situation (1991:125-139). Situationism should not be confused with behaviorism. Unlike behaviorists, situationists such as Ross and Nisbett (1991) do not deny, altogether that people have internal states and traits. What they deny is that the traits people do have are typically expressed in their behavior in ways that are consistent across variations in situation, which is to say that they deny that the traits people do have are typically expressed in their behavior with any significant degree of cross-situational consistency (CSC) (Ross and Nisbett, 1991:163). In denying that people typically have cross-situational consistently (CSC)13 expressed traits, Ross and Nisbett (1991) are, in effect, denying that people typically have character traits, because character traits are usually understood to require a substantial degree of CSC behavioral expression.14 CSC is a matter of degrees. If John is tidy in the office and tidy when he is camping, he exhibits some CSC behavior, even if he is not tidy at home or tidy in hotel rooms. Harman describes character traits as “relatively long-term stable dispositions to act in distinctive ways” (1999:317).15 His requirement for there being CSC behavioral expression of character traits is a requirement that there be a substantial degree of CSC expression, but it is less than a requirement of universal CSC expression. Someone who persistently displays courage across a broad range of different situations, but inexplicably fails to exhibit courage in the presence of mice, would still count as having a relatively long-term stable disposition to act in courageous way. So they would still count as a courageous person, on this definition. Empirical work suggests that the population is divided between those who do insist on universal CSC behavioral expression to warrant the attribution of character traits, and those who only insist on the substantial degree of CSC expression that Harman requires. The former, who believe that character is absolutely invariant across time and situation, are what Chiu et al. (1997) refers to as “entity theorists”. By contrast, “incremental theorists” attribute character more fluidly. (Ambady 201 -71)
We saw that the effect of the FAE appears to fade out over a short period of time, and that it may not affect considered dispositional attributions. And we saw that there is apparent evidence of accuracy in person perception that may be incompatible with the FAE being a pervasive error. We also saw that the FAE is best understood as the product of a combination of underlying effects, which may not combine in the right sort of ways to produce a systematic overall effect, and we saw that there are good reasons to doubt that experiments that demonstrate the occurrence of the FAE are representative of many of the circumstances that we ordinarily find ourselves in. an alternative explanation for the effects that ground the case for the FAE exists and it is not obviously a worse explanation than one that appeals to the effect of the FAE. So we are not even entitled to conclude that the FAE sometimes occurs. Harman’s case against the reliability of character trait attribution was based on the assertion that the existence of a pervasive FAE is a result experimentally established by social psychologists. Plainly it is no such thing. The case for the existence of a pervasive FAE is simply not supported by available evidence. Harman’s case against character is actually based on speculations about the existence and pervasiveness of the FAE that go far beyond the evidential basis established scientifically. (Ridge 158)
Clarke, Steve. South African Journal of Philosophy, 2006, Vol. 25 Issue 4, p350-368, 19p.
Ambady, N. Toward a Histology of Social Behavior: 2000, Judgmental Accuracy from Thin Slices of the Behavior Stream, in M. P. Zanna (ed.), Advances in Experimental Social Psychology (Vol. 32). New York: Academic, 201-271.
Ridge, Michael. Ethics: Oct2002, Vol. 113 Issue 1, p158, 3p.
Miller, Christian. Journal of Ethics: 2003, Vol. 7 Issue 4, p365-392, 28p.
Lagan, John. Theological Studies, 1980, Vol. 41 Issue 3, p549, 19p