Table 2.

Predicted and observed transition percentages for the reading and choice phases combined in Experiments 1 and 2 (for Experiment 2, percentages are given separately for easy/difficult problems).

	r = 1	r = 2	r = 3
OUTCOME-PROBABILITY TRANSITIONS
Predictions
Priority heuristic	50	50	42
Expectation model	57	57	57
Random search	14.29	14.29	14.29
Results
Experiment 1	36.2	37.5	35.4
Experiment 2	–	43.2/42.8	–
OTHER WITHIN-GAMBLE TRANSITIONS
Predictions
Priority heuristic	25	20	25
Expectation model	29	29	29
Random search	28.57	28.57	28.57
Results
Experiment 1	19.0	19.4	17.2
Experiment 2	–	18.8/16.4	–
WITHIN-REASON TRANSITIONS
Predictions
Priority heuristic	25	30	33
Expectation model	14	14	14
Random search	14.29	14.29	14.29
Results
Experiment 1	24.4	23.2	25.6
Experiment 2	–	18.9/21.7	–

See Appendix A for detailed description of the derivation. r = number of reasons inspected by the priority heuristic. Note that the observed transition percentages do not add up to 100 as participants also made transitions that were both between-reasons and between-gambles. Such transitions, which could, for instance, be due to noise, are not predicted by the models. For the derivations of the predictions under random search, it was assumed that transitions between all boxes were equally likely. This yielded 42.86% transitions that were both between-reasons and between-gambles.