Using the past to anticipate the future in human foraging behavior.

Humans engage in many tasks that involve gathering multiple targets from their environment (e.g., picking berries from a patch). Such foraging tasks raise questions about how observers maximize target collection - e.g., how long should one spend at one berry patch before moving to the next patch. Classic optimal foraging theories propose a simple decision rule: People move on when current intake drops below the average rate. Previous studies of foraging often assume this average is fixed and predict no strong relationship between the contents of the immediately preceding patch or patches and the current patch. In contrast to this prediction, we found evidence of temporal effects in a laboratory analog of a berry-picking task. Observers stayed longer when previous patches were better. This result is the opposite of what would be predicted by a model in which the assessment of the average rate is biased in favor of recent patches. This result was found when patch quality varied systematically over the course of the experiment (Experiment 1). Smaller effects were seen when patch quality was randomized (Experiment 2). Together, these results suggest that optimal foraging theories must account for the recent history to explain current behavior.