Fig. 3.

Retrodicting actual behavioural data in probabilistic learning by marmoset monkeys by the ‘winning’ (best-fitting) reinforcement learning model (right). Open squares, sham-operated group. Filled circles, amygdala 5-HT depleted group. Filled triangles, orbitofrontal 5-HT depleted group. The amygdala 5-HT depleted group exhibited impairments in responding to both ‘misleading’ and ‘truthful’ feedback, which were accurately modelled as a deficit in reinforcement sensitivity. Reproduced from Rygula et al. (2014)