Their runs to the devalued side, when so instructed, fell to the same 50% level that control rats had reached during the probe session (Figures 1F and 1G). Moreover, the rats drank the devalued reward on average fewer than half the times when they did run to it (Figure 1H). I-BET151 Instead, they ran the “wrong way” to the nondevalued goal in response to the instruction cues directing them to the devalued side (Figure 1I). Despite remaining unrewarded, the wrong-way runs increased in frequency over days (Figure 1I) and grew equivalent in speed to correct runs to the same goal and to predevaluation behavior, suggesting that they became insensitive to outcome value and

became habitual (Smith et al., 2012). The occurrence of deliberative head movements also suggested that these wrong-way runs represented a new habit. The head movements, in which the rats looked to the nonchosen run side before running the other way at the choice point (Figure 1J), decreased in frequency as performance improved during

training and overtraining (Figure 1K). This result is in accord with previous suggestions that they reflect purposefulness in decision making (Muenzinger, 1938, Redish et al., 2008 and Tolman, 1948). http://www.selleckchem.com/products/XL184.html In the sessions after devaluation, the deliberative movements during wrong-way runs were initially high, but then they fell again (see Figure 3B). Run speeds similarly rose during overtraining Linifanib (ABT-869) and, after devaluation,

were eventually higher for both wrong-way runs and correct runs to the nondevalued goal, and lower for runs to the devalued goal (Figures 1L and 1M). Based on these behavioral indices of habit formation, blockade, and replacement, we analyzed the spike activity patterns of IL and DLS neurons relative to the rats’ performance across both the early training and overtraining periods and also the postdevaluation period. We recorded activity in the IL cortex and DLS simultaneously for up to 4 months with chronically implanted multiple-tetrode assemblies as rats learned the tasks (n = 7, OT rats in Figure 1). Tetrodes were not moved or were lowered only in small (ca. 40 μm) steps to maintain the quality of recordings. For the DLS recordings, we focused on putative striatal projection neurons (n = 1,479 total and n = 858 task-related units; Supplemental Experimental Procedures available online). For the IL cortical recordings, we analyzed 1,694 units, of which 1,013 were task-related units. Because of the near-vertical orientation of the medially situated IL cortex, we were able to monitor activity recorded from tetrodes placed in relatively more superficial (ILs) or deep (ILd) depths of the neocortex (Figures 2A and S1). We found a marked contrast between the changes in ensemble activity in the DLS and IL cortex that occurred as learning proceeded.