(A) Task design and experimental timeline. During ten conditioning sessions, odors (A, B, C, and D) predicted an appetitive sucrose solution, an aversive denatonium solution, no outcome, and either no outcome or an unavoidable air puff, respectively. During the first probe test, A and B were mixed in three different ratios: 85%B/15%A (light red), 50%B/50%A (gray), 15%B/85%A (light blue). After completion of 10 reversal training sessions, in which A and B contingencies were reversed, mice were re-trained in a second probe test. (B) Licking rates in no air puff (top, 7 mice) or air puff (bottom, 5 mice) groups across days, during odor and delay period, for sucrose (blue), denatonium (red), and no-outcome (black) trials. Dashed lines indicate reversals on day 11. (C) Licking behavior from a representative test session from a mouse without (left) and one with (right) exposure to air puffs. Color gradations between blue and red indicate odor mixtures as in (A). Gray bars indicate a period of odor presentation. Dashed lines indicate outcome delivery. (D) Licking rates during sucrose (blue square) and denatonium (red square) trials and during the eight probe trials for no air puff (black) and air puff (orange) groups, during odor and delay period. (E) Mean trial-by-trial licking rates during sucrose (blue square) and denatonium (red square) trials and during the eight probe trials for each ambiguous cue (light red, gray, light blue squares) for no air puff (black) and air puff (orange) groups, during odor and delay period. Line and error bars represent mean ± SEM.