Our findings of neural correlates of our SPE signal in lateral PFC may relate to studies of human causal learning, which report activity in the region while subjects learn causal relationships between cues and consequences (Fletcher et al., 2001). Prediction errors have been proposed as a putative mechanism for guiding learning of such causal associations (Dickinson, 2001), although to our knowledge, the precise mathematical form of how such a putative causal learning error signal is implemented in the brain has not yet been specified (Fletcher et al., 2001) and previous imaging studies have not examined its trial-by-trial computational dynamics. Furthermore, recent recording studies in monkeys have associated neuronal activity in this area with sequential planning behavior (Mushiake et al., 2006) and with the monkey’s performance during dynamic competitive games (Barraclough et al., 2004; Lee et al., 2004). On these grounds the lateral PFC has been proposed to contribute toward the implementation of model-based RL, possibly in the form of Bayesian belief states (Samejima and Doya, 2007), an account consistent with our proposed role of this area in model-based reinforcement learning.
