Fig. 2: Stimulation of dopamine neurons at 20 Hz functions as a teaching signal to drive sensory-specific learning.

a, Unilateral example of bilateral ChR2 expression in dopamine neurons (scale bar, 1 mm). b, Left, minimum and maximum viral expression across rats. Right, placement of fiber tips across rats. The schematic above c and d shows the design of our blocking task using one counterbalancing example, which consisted of acquisition, blocking and a probe test. c, Rats (n = 9) first learned that two visual cues lead to two distinct rewards (acquisition). Then, two new auditory stimuli were introduced in compound with the visual cues and led to the same rewards (blocking). During blocking, we stimulated dopamine neurons (1 s, 20 Hz) at reward delivery for one of the compounds (‘stimulation’ cue). Rats acquired the food-port response during the cues and maintained high levels of responding after the introduction of the auditory cues and dopamine stimulation. d, We next subjected rats to a probe test and found that they responded more to the stimulation cue relative to the control cue. The schematic to the left of e shows the devaluation procedure, which was used to devalue the reward paired with the stimulation cue using lithium chloride (LiCl). e, Rats showed a significant reduction in responding to the stimulation cue following devaluation, confirming that dopamine stimulation unblocked the association between the cue and a sensory-specific representation of the reward. The schematic above f and g shows the design used for ICSS. Rats were provided access to a lever that produced 50-Hz stimulation of dopamine neurons (active) or nothing (inactive). f, Rats pressed more on the active lever. g, This ICSS measure was positively correlated with the unblocking effect in (d) (Pearson’s correlation: r = 0.56, P = 0.045). This experiment was replicated in Supplementary Fig. 4 with the rats used for the PIT experiment in Fig. 1. Data across sessions were analyzed with a repeated-measures ANOVA and follow-up simple-effect analyses, where appropriate. A one-tailed t test was used to determine the significance between two means with directional hypotheses. **P < 0.05. Error bars indicate the s.e.m. Scale bar, 1 mm.