Table 2.
Comparison of mean score for each model, m, relative to the baseline for all participants in a particular condition. In parentheses is the percentage of participants for whom model m provides a better fit than baseline (i.e., is positive). The best fit model in each condition is indicated in bold.
| Experiment 1 | Experiment 2 | ||||
|---|---|---|---|---|---|
| Model (m) | cont. | prob. | no-cue | shuffled | consistent |
| Softmax | 18.4 (0.78) | 130.0 (1.00) | −11.2 (0.47) | 1.0 (0.59) | −26.9 (0.53) |
| ET | 41.1 (1.00) | 96.3 (1.00) | 30.5 (0.76) | 78.6 (0.82) | 11.1 (0.76) |
| Q-learning | 105.3 (1.00) | 137.5 (1.00) | 93.2 (0.94) | 166.7 (0.94) | 118.2 (0.94) |