We recorded LFPs and spiking activity from ventral striatum in rats (n = 4) running laps for food reward on a spatial decision task (Multiple-T, Figure 2A). As described previously (Johnson and Redish, 2007; Schmitzer-Torbert and Redish, 2004a; van der Meer and Redish, 2009), this task contained several decision points (turns T1–T4); at the final choice point T4, choosing one return rail, but not the other, triggered delivery of food reward. As shown before for these subjects (van der Meer and Redish, 2009), rats learned to choose the correct return rail within the first 10 laps of a session, but continued to exhibit more gradual reductions in lap time within sessions. For comparison with later results, a pseudocolor heat map of average running speed over the linearized track as a function of lap is shown in Figure 2B. The feeder sites F1 and F2 are clearly highlighted by low average speed (as the rats pause to eat), while similar speeds are reached on the T4-F1 and F2-S (start) segments. Rats ran an average of 75.4 ± 10.2 laps per session, of which 1.43 ± 0.51 were unrewarded errors (SEM over rats).
