How to learn from our mistakes

Abstract

A new approach to developing models of folk psychology is suggested, namely that different models exist for different folk psychological practices. This point is made through an example: the explanation and justification of morally heinous actions. Human folk psychology in this area is prone to a specific error of conflating an explanation for behaviour with a justification of it. An analysis of the error leads me to conclude that simulation is used to generate both explanations and justifications of heinous acts. It is needed in both these cases because most of us lack theoretical information about evil actors. I will argue that it is difficult to simulate such acts, and hence difficult to develop explanations for behaviour widely accepted as evil. This difficulty explains the judgements made against successful simulators by those who don't succeed, and so explains the common problem of conflating an explanation with a justification.

Acknowledgements

Thanks to James Harold, Brian Huss and Christopher Eliot for helpful comments on drafts of these papers. I would also like to thank the audience at CENSS 2002, the philosophy department at the University of Cincinnati and two anonymous reviewers for this journal.

Notes

See, for example, the debate over the cognitive penetrability of mental simulations. It has been argued that if we fail to predict a common behaviour, we must not be using our own cognitive mechanisms to make that prediction, and thus we must not be simulating (Stich and Nichols 1995). This is the case with many behaviour patterns uncovered by social psychologists such as the endowment effect (placing a higher monetary value on objects when we own them) and the fundamental attribution error (minor situational factors can have a dramatic effect on behaviour, as in the case of Milgram's infamous experiment). One response to this criticism is that the failed prediction may be caused by imputing faulty initial states (e.g. Gopnik and Wellman 1992).

As an anonymous reviewer for this journal pointed out, the eliminativists seem to hold folk psychology to too high a standard.

For example, Beaman and colleagues demonstrated that the more hurried a person is, the less likely she is to offer aid to a person obviously in need (even if the reason for hurrying is not a very good one). However, they also found that people who were given a lecture on this effect were less likely to be influenced by it even two weeks later (Beaman et al. 1978). It is still not clear how much education is needed to immunize people against such biases, and whether ongoing instruction is required.

See Andrews (2003) for a discussion of these issues.

Of course, there is good reason to think that no finite set of beliefs and desires can determine any particular action. There could always be an unknown defeater outside the set that would cause the actor to engage in some other act. For a discussion of this point, see Morton (2003). However, this metaphysical point is entirely consistent with the claim that humans do in fact understand belief/desire sets to be sufficient for predicting or explaining behaviour. As we know from Nisbett and Ross (1985) and Kunda (2002) our heuristics are not limited to those that guarantee truth.

My focus in this section is on the justification of behaviour rather than justification more broadly construed, so I am not interested in epistemic issues dealing with justified belief. However, I am interested in more than moral justification. While justifications of behaviour are value claims, they are not all ethical claims. A behaviour might be justified on aesthetic grounds, for example. The issues at hand is whether a behaviour is acceptable on some grounds rather than whether it is truth oriented or reliable.

Trait attribution and induction might at first appear to be describing the same method of predicting behaviour. Since we may come to attribute a certain trait to someone after observing her behaviour for a while, induction certainly has a role to play. However, one can come to attribute traits to people via other methods as well. For example, if I am told that a person that I have just met has a certain trait, I may use that information in order to predict what he will do, even though I have no independent evidence that the person actually behaves in a way consistent with that trait.

Our decisions are not based on a good understanding of probabilities; we often ignore base rates, we underestimate the importance of consensus information and the power of situations to influence behaviour, we ignore sample size, and so forth (Kunda 2002).

Of course it is possible that there are some other problems; for example, the person might have a theory of mind deficit. However, here I am only concerned with people whose cognitive capacities are considered normal.

One might also reject Sontag's explanation because they think the facts she presented are false. However, that wasn't the criticism. It seems that it was the mere attempt at providing an explanation that illuminated the cause of the attacks which was so offensive.