(A) Timeline of the experimental sequence. (B) Operant conditioning maintained by chocolate-flavored or standard chow (SC) pellets. Mice trained with chocolate increased the number of reinforcers in 1 hour during FR5 daily sessions compared with mice trained with SC (data indicate the mean ± SEM; repeated-measures ANOVA; ***P < 0.001, pellet effect and pellets per session). (C–E) The 3 addiction-like criteria for the SC and chocolate-trained groups in the late period. (C) Persistence of response (Mann-Whitney U, *P < 0.05). (D) Motivation (Mann-Whitney U, ***P < 0.001). (E) Compulsion-like behavior. The dashed horizontal lines indicate the 75th percentile of the distribution of the chocolate-trained group. It was used as a threshold for considering a mouse positive for 1 criterion. (F) Percentage of addicted and nonaddicted mice trained with chocolate and SC pellets classified in the late period (χ² test, ***P < 0.001). n = 51 mice trained with chocolate pellets; n = 7 mice trained with SC. (G–J) Tests for the 4 phenotypic traits in the late period for the chocolate-trained group, divided into addicted (A) and nonaddicted (NA) mice. (G) Impulsivity (Student’s t test, ***P < 0.001). (H) Cognitive flexibility (Mann Whitney U, *P < 0.05 and **P < 0.01. (I) Appetitive cue reactivity. Increased active response after presentation of the cue light (Mann-Whitney U, *P < 0.05). (J) Aversive cue reactivity. The number of nonreinforced active responses after the shock test with the same discriminative stimulus (grid floor) for the shock test. Pressing the active lever had no consequences: no shock, no pellets, and no cue light (Student’s t test, *P < 0.05). (K) Pellet intake and (L) body weight for mice classified as addicted or nonaddicted among those trained with chocolate pellets. n = 38 addicted mice; n = 13 nonaddicted mice trained with chocolate pellets. Data are presented as individual values with the IQR. Statistical details are included in Supplemental Table 10.