Abstract 

Organisms regularly make decisions regarding foraging, habitat, mating, and danger avoidance. Individuals accumulate noisy and dynamic information and may share it with neighbors. We present a statistical inference model describing how rational Bayesian agents update their beliefs over time and commit to decision using criteria on the evidence gathered. Foraging decisions are idealized as a stay/go decision, so agents leave a food patch when they become certain enough the patch is low yielding. Asymmetries in information transfer exaggerate errors in patch quality estimates, triggering poorly timed departures. In a generic binary decision task performed by a large clique, the speed and accuracy of the majority group decision can be estimated by approximating the solution to the population density equation asymptotics for order statistics based on the flux through decision boundaries. Groups can improve their overall decision efficiency by adopting a diversity of decision criteria. We also find that correlations in accumulated evidence can serve to impoverish the accuracy of early deciders, so that pooling multiple decisions does not substantially increase collective decision accuracy.