Figure 1. dmPFC lesions alter the explore/exploit balance.

(a) Bilateral excitotoxic lesions of dmPFC. Cresyl-violet-stained sections show the extent of a typical lesion. (b) The behavioural choices were classified as explore or exploit choices using a reinforcement learning model. Lesion mice made fewer explore choices in discrimination compared with sham-operated mice (lesion: n=11, sham: n=9; strategy: F(1,17)=5.14, P=0.04; lesion: F(1,17)=7.34, P=0.02; interaction: F(1,17)=9.06, P=0.008; two-way repeated measures analysis of variance (ANOVA), Bonferroni post hoc test). (c) In reversal, lesion mice made more exploit choices (strategy: F(1,17)=90.66, P<0.0001; lesion: F(1,17)=5.22, P=0.04; interaction: F(1,17)=8.27, P=0.01; two-way repeated measures ANOVA, Bonferroni post hoc test). (d) Lesion mice shifted their strategy less from discrimination to reversal compared with sham mice, where a large number indicates more exploit choices in reversal (t(17)=2.19, P=0.04; unpaired t-test). ‘Exploit index'=(exploit−explore)/total trials. *P<0.05, **P<0.01, ***P<0.001. Bars represent the mean±s.e.m. Behavioural data for modelling were previously published¹².