Neurons in dopamine-rich areas of the rat medial midbrain predominantly encode the outcome-related rather than behavioural switching properties of conditioned stimuli

Midbrain dopamine neurons are phasically activated by a variety of sensory stimuli. It has been hypothesized that these activations contribute to reward prediction or behavioural switching. To test the latter hypothesis we recorded from 131 single neurons in the ventral tegmental area and retrorubral field of thirsty rats responding during a modified go/no-go task. One-quarter (n = 33) of these neurons responded to conditioned stimuli in the task, which varied according to the outcome with which they were associated (saccharin or quinine solution) and according to whether they triggered a switch in the ongoing sequence of the animal's behaviour ('behavioural switching'). Almost half the neurons (45%) responded differentially to saccharin- vs. quinine-conditioned stimuli; the activity of a minority (15%) correlated with an aspect of behavioural switching (mostly exhibiting changes from baseline activity in the absence of a behavioural switch) and one-third (33%) encoded various outcome-switch combinations. The strongest response was excitation to the saccharin-conditioned stimulus. Additionally, a proportion (38%) of neurons responded during outcome delivery, typically exhibiting inhibition during saccharin consumption. The neurons sampled did not fall into distinct clusters on the basis of their electrophysiological characteristics. However, most neurons that responded to the outcome-related properties of conditioned stimuli had long action potentials (> 1.2 ms), a reported characteristic of dopamine neurons. Moreover, responses to saccharin-conditioned stimuli were functionally akin to dopamine responses found in the macaque and rat nucleus accumbens responses observed within the same task. In conclusion, our data are more consistent with the reward-prediction than the behavioural switching hypothesis.