Unexpected events are often considered as leading to the deployment of attention, in the form of orienting or executive control. Further, the areas correlated with the SPE signal are thought to be involved in aspects of attention (the PPC with orienting/salience (Yantis et al., 2002) and frontal regions with executive control (Corbetta et al., 2000; MacDonald et al., 2000)) . Thus, it is natural to question whether this correlation is parasitic on some more general forms of attention. Our experimental design does not allow us to rule this out definitively; and indeed some models of associative learning use a signal akin to our SPE to control the assignment of salience for learning to particular stimuli (Pearce and Hall, 1980), according to some accounts via a cholinergic pathway associated with the PPC (Bucci et al., 1998). In this context, the model-based approach used here would provide a computational account of the means by which such attention is allocated on a trial by trial basis, and of how those allocations change as a function of learning and experience. Our findings would then
