March 24, 2014, by Jules Holroyd
The ‘Bias’ in Implicit Bias
What is it about some implicit associations that should lead us to characterise them as ‘biases’?
The idea of ‘bias’ has been unpacked in other philosophical debates, notably, by feminist philosophers concerned with bias in scientific methodology. For example, a candidate understanding of bias is offered by Louise Antony (1993) as an interest or perspective that shapes aspects of the enquiry undertaken. This does not seem quite right for our purposes in thinking about implicit biases: an upshot of Antony’s understanding is that some biases might be vindicated as helpful and fruitful for enquiry. But we want to isolate the implicit associations that are problematic, and focus on those.
One way of thinking about how to characterise the ‘bias’ part of implicit bias is in terms of those associations which exert a distorting influence on judgement. There are two important contrasts here: first, the bias is in the resulting judgement, rather than a property of the stored associations alone: they are merely disposed to produce biased judgement. Second, the influence is on judgement, rather than on behaviour more broadly construed. The influence at issue can be characterised as distorting the judgement, in that it leads to a judgement which departs from the norms of rationality. This can be most clearly seen in the CV studies mentioned in earlier posts, whereby the gender or race of the name on the CV lead to differential evaluations of the quality of the CV. But clearly, the name at the top of a CV does not provide a reason for judging to be better or worse an otherwise identical CV.
It is less clear how this analysis explains the behavioural outputs, such as increased seating distance from stigmatised groups. One possibility would be to extend the definition to include not only distortions of judgment but also undesired or undesirable influences on action. The disjunct is needed in order to accommodate the possibility that some implicit biases might not be undesired (for example, someone with explicitly racist views may not disavow the influence of negative race associations on their behaviour). Another would be to suppose that these behaviours are preceded by (tacit) judgements, which are distorted. Take the finding that individuals arrange seating so as to sit further away from stigmatised group members against whom they show biases on an IAT. We could understand this as involving a judgement about the suitable place to place the seat, say. What speaks in favour of one or other way of proceeding?
The judgement route provides a nice contrast with explicit bias, if by that we understand a distorted attitude or judgement that is explicitly held (which again, could inform behaviour in various ways). In the case of implicit bias, then, what is at issue is a judgement that is distorted. This judgement is sometimes the output measured; in other cases it informs the behavioural output which is measured. But might this understanding be unable to capture all the cases of ‘implicit bias’ we would want it to? Might it encounter further difficulties?