However, we measured several

variables related to the rat’s behavior and motivational state at and prior to the time of cue onset (precue variables), and only one of these was consistently correlated with neural activity: the proximity to the lever at time of cue onset. Critically, even when the effects of all of these precue variables were accounted for, we still observed a strong correlation between neural activity and the onset latency and speed of locomotion (Figure 3). Thus, if there were some underlying factor that influenced both locomotor behavior and NAc neural activity to produce a spurious correlation between them, it would have to be unrelated to the rat’s locomotion and selleck kinase inhibitor orientation at cue onset, unrelated to the level of motor activity during the ITI, and drug discovery unrelated to the time elapsed since the previous reward or operant event. Because at least some of these variables should have been influenced by motivational or attentional state, we think it is unlikely that the neural correlates of locomotor vigor

that we observed are attributable to trial-by-trial changes in these factors. The cue-evoked firing of NAc neurons was substantially greater for the reward-predictive DS than for the neutral NS. This difference occurred prior to movement onset in the majority of trials and therefore did not reflect ongoing differences in behavior elicited by the cues. Instead, the firing difference is likely due to afferent inputs that encode the reward value predicted by the cue, such as from dopamine neurons (Day et al., 2007) and the amygdala (Paton et al., 2006; Schoenbaum et al., 1998); consistent with this idea, inactivation of either of these inputs eliminates NAc DS-evoked firing (Ambroggi et al., 2008; Cacciapaglia et al., 2011; Jones et al., 2010; Yun et al., 2004). Whatever its origin, our results demonstrate that the value signal is transformed by NAc neurons such that their value-influenced firing is closely related to, and potentially sets, the vigor of the subsequent action. These findings

appear at odds with observations that pharmacological manipulations or lesions of the and NAc only minimally affected movement latency and speed in reaction time tasks (Amalric and Koob, 1987; Brown and Bowman, 1995; Giertler et al., 2004) and that NAc cue-evoked firing did not covary with these measures of vigor (Goldstein et al., 2012). The most likely explanation is that flexible approach was required in the DS task but not in these other paradigms. Locomotor approach is flexible in the DS task because a new path must be computed on every trial, but it is inflexible in the reaction time tasks and in Goldstein et al. (2012) because the start and end locations are fixed across trials, so that animals can reliably obtain reward using stereotyped approach trajectories.