We emphasize that we recorded from only Bcl-2 inhibitor small numbers of IL units, and we used behavioral measures that only indirectly accessed underlying performance strategies; other features of IL activity that track behavior trial-to-trial, directly or through its interactions with other regions, may have been covertly present. It is nonetheless striking that a strong correlation did hold between the dominant

IL ensemble activity pattern and habitual features of behavior measured at the level of sessions, which were at particular levels of learning and behavioral plasticity. Notably, the times at which the task-bracketing activity pattern was observed in IL cortex were nearly identical to the times at which optogenetic IL perturbation (of all layers) could disrupt the maze habits: during overtraining, as shown here, as well as after overtraining and after postdevaluation training when a second habit had become established GSK1210151A research buy (Smith et al., 2012). These times, in turn, were highly correlated with the periods in which the numbers of deliberative head

movements declined. Together, these results suggest that the task-bracketing pattern in the IL cortex could reflect the training-related development of a potent and active IL influence over the sculpting of habits as well as an influence over their execution. The lack of trial-level correlation with behavior suggests a contribution to habits at the level of states that bias behavior toward outcome insensitivity (or low deliberation). This view might help

account, for example, for the fact that the ILs bracketing pattern remained on PP day 1, when we had previously reported that IL perturbation does not affect behavior (Smith et al., 2012); the pattern, although present, was joined by marked increases in spiking variability and magnitude reflecting perhaps a mixed habit/nonhabit state. If the IL cortex were to have such a state-level influence, how would it interact with the DLS to promote habits, given that direct connections between them have not been detected? Potential indirect connectivity could include fiber projections via the ventral striatum or the amygdala and the substantia nigra or through by way of projections to other cortical areas and then to the DLS (Hurley et al., 1991). However, as favored here, the IL cortex and the DLS might work partly in parallel, promoting habits through distinct circuit mechanisms, with the IL cortex providing, by way of its many limbic connections, routes by which it could disrupt flexibility and mnemonic processes or invigorate learned behavior. An unexpected finding of this study is that the task-bracketing pattern that did form in the IL cortex was evident only in the superficial layers. Superficial cortical layers are especially important for transcortical processing, and deeper layers for cortical projections to subcortical regions including the striatum (Anderson et al.