Observing the same individuals across different lakes, we can disentangle inter-individual and ecological variation in foraging behaviors. Social information use and giving-up times were stable across individuals, whereas other strategies were used more flexibly.

12/n

We find that patch-leaving tendencies decreased in areas with high social densities. Crucially, however, these effects were small, suggesting that personal information trumps social information for patch-leaving decisions.

11/n

To predict patch-leaving, we compared the performance of learning models with simpler heuristic models and find that ice-fishers use fixed giving-up-time rules with patch-leaving increasing with time and decreasing after resource discovery.

9/n

Social learning is beneficial in many situations, but often bears the risk of maladaptive herding if used unstrategically. Here, we find that foragers use social information conditionally, relying more on social information when not finding fish, and less when successful.

8/n

This behavior mirrors the actual resource distribution: Social density and recent patch discoveries predict the presence of fish, while ecological information did not.

7/n

Using computational models inspired by cognitive science and movement ecology, we find that both personal and social information determine where icefishers go: foragers select locations close to others and in areas in which they are successful. Ecological information plays only a minor role.

6/n

We equipped large groups of Finnish ice-fishers competing for resources with GPS devices and head-cams, capturing continuous data on foraging success, behavioral states and the social environment for more than 400 foraging trips, over 16,000 foraging locations across 10 lakes.

3/n