FIGURE 3.

Laser-related learning. (A) Mouse 5982 went through three different forms of the task: first with standard laser suppression (random rewards), then with fibers disconnected and random rewards, then finally with fibers disconnected and normally rewarded laser trials. Vertical dotted lines indicate the transitions between these segments. Performance is represented as percent correct averaged across both stimuli in a 100-trial sliding window. Black, performance on control trials; cyan, performance on laser trials; gray, bias on laser trials. Performance on control trials remained consistent over time, but performance and bias on laser trials were sensitive to the specific task conditions. During standard suppression, there was a ∼15% impairment in performance on laser trials, but when the fibers were disconnected from the brain, there was an initial period during which laser had no effect on performance. In this condition, the salient visual laser cue indicates random rewards. After several thousand trials, indicated by the red arrows, this mouse developed a strong bias. One interpretation of this is that the mouse learned that its responses were irrelevant on laser trials, and adopted a bias strategy to minimize effort on those trials. In the final segment, we rewarded laser trials normally, which reinforced responses on laser trials. Over time, performance on laser trials became indistinguishable from control, and the bias diminished; green arrows indicate the approximate time of the major shift in performance. (B) After initial testing showed a moderate effect of the standard laser suppression, mouse 5916 was tested with normally rewarded laser trials during optogenetic suppression of auditory cortex. The onset of normal rewards on laser trials is indicated by the vertical dotted line. After this, the impairment on laser trials persisted for several thousand trials, but then diminished. By about 15k trials after the onset of normal rewards, performance on control and laser trials was indistinguishable. This suggests that this mouse was eventually able to learn to respond correctly even when auditory cortex was suppressed.