Strumia made the claim of the title in a controversial talk at CERN.* I will show that this claim is falsified by the psychological literature.
There are a large number of cognitive biases operative in our psychology: over 180 at the last count, and that is just the ones we know about so far. All of these biases share the characteristics of being largely invisible to us in their operation and extremely hard to eradicate. Data show for example that significant financial incentives do not cause reduction of the effects of some of these biases. And I have discussed myself at length (Short, 2017) the way biases can cause highly suboptimal decision-making, even when there are very serious financial consequences.
The types of bias I mean would be exemplified by Confirmation Bias, which occurs when people look for evidence which confirms hypotheses they already believe. I think we should also consider Gender Bias in this same arena, though the claim we should not represents an objection to my position. I will show below that my position has adequate resources to defeat that objection, but first I will show that intelligence offers no protection against implicit biases.
I will do that by mentioning three types of bias where it was not the case that more intelligent subjects exhibited less bias, and then making a broader point.
- Myside Bias — this is related to Confirmation Bias. It occurs when people evaluate and generate evidence or test hypotheses in a way that conforms to their prior opinions and attitudes. Stanovich, West and Toplak (2013, p. 259) found that the “magnitude of the myside bias shows very little relation to intelligence.”
- Dunning-Kruger Effect — this is best known as the claim that unskilled persons also lack insight into their relatively poor abilities in an area. However, similar bias effects operate at the other end of the spectrum. Schlösser et al. (2013, p. 85) report that their model “partially explained why top performers underestimate their performances.” (I am assuming a correlation here between high intelligence and an ability to be a top performer in the fields of endeavour examined by the authors.) But here we see that intelligent subjects are also not immune from a variant of the Dunning-Kruger Effect.
- The Gambler’s Fallacy — this is the tendency to think that fixed probabilities are altered by past events. For example, the odds of getting heads on throwing a fair coin are always 50%, irrespective of what has happened previously. If someone sees heads ten times in a row and then says either “it must be heads again next” or the opposite, they are exhibiting this apparently maladaptive heuristic. Xue et al. (2012) found that “individuals’ use of the [Gambler’s Fallacy] strategy was positively correlated with their general intelligence.’’
More generally, we may note that many experiments in social psychology are conducted on psychology undergraduates. This has been mooted in the past as a potential “ecological” objection, meaning that the results could be unrepresentative of the general population. Nevertheless, robust and widely replicated data exists to show the existence of 180+ cognitive biases. We may assume that undergraduates in psychology are a more intelligent subset than the population in general.
I will close by considering one potential objection to my account. This is that Gender Bias is not a cognitive bias and should not be considered in the group above where intelligence is not a protective factor. I will counter this objection in a number of ways.
- If Gender Bias is not a cognitive bias, what is it? It results in a systematic slanting of judgements away from what would be strictly rational, and that accords precisely with my working definition of a bias (Short, 2015).
- I do not need to assume a narrow and precise definition for Gender Bias. I am including within it all of what people refer to by the terms Sex Discrimination, Sexual Discrimination, Homophobia, Anti-LGBTQ+ prejudice etc. These discriminations often take place via stereotyping — assuming that everyone in group X has certain characteristics which may in fact be possessed by only some or indeed none of the members of group X. Stereotyping appears on the standard list of cognitive biases.
- Krieger (1995) explicitly considers racial bias within a cognitive bias framework and includes also discussion of Gender Bias.
I conclude that this objection fails, and that therefore the claim that intelligence protects against implicit bias is false.
*According to a letter published by the Office of the Chair, Department of Physics and Astronomy, University of California, Irvine on 01 October 2018.
- Krieger, L H 1995. The Content of Our Categories: A Cognitive Bias Approach to Discrimination and Equal Employment Opportunity. Stanford Law Review 47.6 pp. 1161–1248
- Schlösser, T, Dunning, D Johnson K L, Kruger J 2013 How unaware are the unskilled? Empirical tests of the “signal extraction” counterexplanation for the Dunning–Kruger effect in self-evaluation of performance Journal of Economic Psychology 39, December 2013, pp. 85–100, DOI: 10.1016/j.joep.2013.07.004
- Short, T L 2015 Simulation Theory: A psychological and philosophical consideration, Psychology Press, ISBN 9781317598145
- Short, T, L 2017 The Psychology of Successful Trading: Behavioural Strategies for Profitability, Routledge, ISBN 9781351601016
- Stanovich, K E, West, R F and Toplak, M E 2013 Myside Bias, Rational Thinking, and Intelligence, Current Directions in Psychological Science, 22. 4, pp. 259–264, DOI: 10.1177/0963721413480174
- Xue, G, He, Q, Lei, X, Chen, C, Liu, Y, Chen, C, Lu, Z-L, Dong, Q, Bechara, A The Gambler’s Fallacy Is Associated with Weak Affective Decision Making but Strong Cognitive Ability PLOS One, October 5, 2012 DOI: 10.1371/journal.pone.0047019