Fig. 1: Summary of participant performance.

A Average stay durations (over trials) for each task condition are plotted in a colour scale reflecting total reward obtained over the course of the experiment (lighter colours indicate higher reward, same colouring across panels). Participants stayed for longer than they should in all conditions, but recovered the rank ordering correctly (as a group). B Reward was strongly, but not entirely, determined by average stay durations. In particular, it was reduced by overstaying, which most participants exhibited, but also by under-staying, making the relationship nonlinear. C Response latency was a second significant determinant of total reward, and was not related to total stay durations (see main text). D Participants’ exit thresholds, which show differential overstaying by condition, with the least overstaying in the Low-Slow condition and the most in the Low-Fast. E Exit threshold deviations (from by-participant grand means) for each condition. Having a higher exit threshold deviation is generally better (less suboptimal). A minority of participants were in the vicinity of optimal performance, or even under-stayed, and for these participants, this higher-is-better interpretation will not always be correct. Participants can be seen to perform most suboptimally in the Low-Fast condition, followed by the High-Fast, High-Slow, and lastly, Low-Slow condition. F Participants with larger differences between their highest and lowest exit threshold, i.e., who had less consistent exit thresholds, had higher average exit thresholds and performed better on the task.