Wednesday, 26 March 2014

Implicit Bias and Epistemic Innocence

In this post I will suggest some reasons for thinking that at least some beliefs based on implicit bias are epistemically innocent. An implicit bias is a bias ‘of which we are not aware […] and can clash with our professed beliefs about members of social groups’, and which can ‘affect our judgments and decisions’ (Crouch 2012: 7). Empirical work has shown that such biases are held by ‘most people’, even those people who avow egalitarian positions, or are members of the targeted group (Steinpreis et al. 1999). 

As Lisa and I have said in previous posts, we understand the epistemic innocence of a cognition as that cognition's meeting two conditions. Here are the conditions a belief based on implicit bias would have to meet in order to be epistemically innocent:

1. Epistemic Benefit: The belief delivers some significant epistemic benefit to an agent at a time (e.g., it contributes to the acquisition, retention or good use of true beliefs of importance to that agent).

2. No Relevant Alternatives. Alternative beliefs that would deliver the same epistemic benefit are unavailable to the agent at that time.

Let us look first to the No Relevant Alternatives condition. My thought is that if my belief that p is guided by, or grounded on, my implicit bias pertaining to some group which my belief is about, there is a sense in which alternative beliefs are unavailable. This is because my bias might incline me to selectively attend to evidence, and to generally form beliefs in ways which are not truth-regulative as they would be without the bias. Further, because my bias is implicit, because it is not something of which I am aware, I cannot correct for it (though note that it does not follow that I could correct for it even if I become aware of it, so alternative cognitions may still be unavailable).

With respect to unawareness, Jennifer Saul claims that the literature supports the view that we are unaware of the implicit biases we display (Saul 2013: 43). Similarly, Jules Holroyd claims that our harbouring implicit biases is not something that we can know by ‘introspection, reflection, or self-report on one’s motives’ (Holroyd 2012: 275). However, Holroyd resists the claim that implicit biases may be solely present because of cultural factors, suggesting instead that ‘the extent to which we manifest biases may rather be a function of other cognitive states we have, and over which we plausibly have control’ (Holroyd 2012: 280). Whatever is right about the etiology of implicit biases, that is, whether they are due solely to our surroundings, or whether they are due also in part to other attitudes we have, I do not think this bears on their meeting the No Relevant Alternatives condition on epistemic innocence.

Let us move to the Epistemic Benefit condition. This condition is less obviously met by beliefs based on implicit bias, as they are not appropriately sensitive to evidence, and are often false. Though beliefs based on implicit biases may, at the very least, have indirect epistemic benefits. We might usefully think of implicit biases as schemas, which are 'mental frameworks of beliefs, feelings, and assumptions about people, groups, objects' which can 'help us made sense of the world' (Anderson 2010: 10-11). Further, '[t]hese schemas filter information, helping us to determine what should be paid attention to and what can be disregarded. They save us time' (Crouch 2012: 7).

Having the belief that a set of CVs is less good than a comparable (perhaps identical!) set of CVs (see Steinpreis et al. 1999) might be epistemically beneficial insofar as it makes decision making processes quicker. This is in line with Crouch’s claim that 

'[P]eople who are pressed for time and cognitively overloaded tend to rely on their schemas or stereotypes more automatically. People higher up in hierarchies tend to be people who juggle a lot of tasks and information, and to be pressed for time, and so are not likely to take the time to consider decisions in ways that might avoid their pre-existing schemas' (Crouch 2012: 8).

So it looks like at least some beliefs based on implicit bias might be epistemically innocent, in virtue of alternatives being in some sense unavailable, and their having at least indirect epistemic benefits. In my next post I'll write about the implications of this result for responsibility for implicit bias and how we ought to tackle it.

No comments:

Post a Comment

Comments are moderated.