The neural correlates of belief bias: activation in inferior frontal cortex reflects response rate differences.

Abstract

The belief bias effect in reasoning (Evans et al., 1983) is the tendency for logical problems with believable conclusions (e.g., some addictive things are not cigarettes) to elicit more positive responses than those with unbelievable conclusions (some cigarettes are not addictive things). The effect of believability interacts with conclusion validity (see the lower rows of Table ​Table11 for example data), leading many researchers to conclude that reasoning accuracy is greater for problems with unbelievable conclusions (e.g., Oakhill and Johnson-Laird, 1985; Newstead et al., 1992; Quayle and Ball, 2000). Dube et al. (2010, 2011) [see also Heit and Rotello (2014)] demonstrated that the typical ANOVA analysis of these behavioral data was inappropriate, and showed that a signal detection based interpretation of the data reached a different conclusion, namely that the effect of conclusion believability was to shift subjects’ response bias to be more liberal. Trippas et al. (2013) also concluded that conclusion believability consistently affected response bias, but that reasoning accuracy was additionally affected by believability under certain conditions (i.e., higher cognitive ability, complex syllogisms, unlimited decision time).