Abstract Reward expectations based on internal knowledge of the external environment are a core component of adaptive behavior. However, internal knowledge may be inaccurate or incomplete due to errors in sensory measurements. Some features of the environment may also be encoded inaccurately to minimise representational costs associated with their processing. We investigate how reward expectations are affected by differences in internal representations by studying rodents’ behaviour and dopaminergic activity while they make time based decisions. Several possible representations allow a reinforcement learning agent to model animals’ choices during the task. However, only a small subset of highly compressed representations simultaneously reproduce, both, animals’ behaviour and dopaminergic activity. Strikingly, these representations predict an unusual distribution of response times that closely matches animals’ behaviour. These results can inform how constraints of representational efficiency may be expressed in encoding representations of dynamic cognitive variables used for reward based computations.

Dopamine responses reveal efficient coding of cognitive variables