Fig. 2
Choice behavior and sniff responses. a On free choice trials, choices for the high-intensity version of both the sweet and savory US were above chance and did not differ. Error bars depict across-subjects s.e.m. *p < 0.05, t-test vs. chance. b Average trial-wise sniff traces, time-locked to sniff cue onset show a main effect of odor intensity on both sniff peak and sniff duration but no effect of odor identity and no interaction. Error bars depicted within-subjects s.e.m. c Black circles are the percentage of choices for the CS predicting the high-intensity US (post reversal), plotted for three free choice trials before and after value reversals. Average model-derived choice probabilities are plotted in gray lines. The task state schematics above the plot illustrate an example of one possible value reversal, but the data plotted are averaged across all value reversals. Error bars depict across-subjects s.e.m. d Black circles are the percentage of choices for the CS predicting the high-intensity US, plotted for three free choice trials before and after identity reversals. Average model-derived choice probabilities are plotted in gray lines. The task state schematic above the plot illustrates an example of one possible identity reversal, but the data plotted are averaged across all identity reversals and subjects. Error bars depict across-subjects s.e.m
