Eating expectancies and reinforcement learning: a state-based test of affect regulation and expectancy models in the natural environment.

PURPOSE: Affect regulation and expectancy-based models suggest that improvement in affect following binge-eating (BE) episodes contributes to increased eating expectancies, which then promote BE maintenance.
METHODS: The current pilot study utilized ecological momentary assessment to examine the prospective independent and interactive effects of eating reinforcement experiences [operationalized as reductions in negative affect (NA) following BE episodes] and eating expectancies on subsequent BE behavior among 17 women with recurrent BE.
RESULTS: Greater reductions in momentary NA following a BE episode (i.e., greater reinforcement) predicted higher levels of eating expectancies on the following day. Further, current eating expectancies interacted with proximal reinforcement history to predict future BE episodes. Participants were more likely to report BE episodes on days that were characterized by higher eating expectancies and preceded immediately by a day during which they experienced greater reinforcement from BE.
CONCLUSION: These preliminary results are consistent with affect regulation and expectancy-based models of BE, suggesting a dynamic and potentially modifiable process of reward-based learning associated with BE behavior. LEVEL OF EVIDENCE: Level IV, multiple time series.