Fig. 6.

Illustration of the evidence constraints in Experiment 2. As in Experiment 1, evidence was correlated across participants: If a part failed on one flight, it failed for all participants in that generation on that flight (part failures are displayed in the figure as red circles). In contrast to Experiment 1, participants could not change their spaceship designs in between flights, and evidence about a particular spaceship part was only shown if the part was chosen in the sample design shown to the participant.