Table 1.
Reinforcement learning parameter estimates.
| Parameter | Median (Q1,Q3) | Regression fit: age vs. age2 | ||
|---|---|---|---|---|
| Child | Adolescent | Adult | ||
| Lapse rate (ξ) | 0.075 (0.016, 0.250) | 0.067 (0.008, 0.229) | 0.026 (0.007, 0.139) | Age |
| Learning rate (α) | 0.223 (0.055, 0.467) | 0.364 (0.173, 0.517) | 0.482 (0.164, 0.573) | Age |
| Go bias (b) | 0.537 (0.143, 0.942) | 0.215 (− 0.007, 0.491) | 0.310 (0.006, 1.04) | Age2 |
| Pav bias (π) | 1.388 (0.502, 2.694) | 0.229 (0.167, 0.411) | 0.398 (0.113, 0.862) | Age2 |
| Reinforcement Sensitivity (ρ) | 3.159 (1.750, 4.247) | 4.846 (3.835, 8.383) | 4.512 (2.329, 8.758) | Age |
Median parameter estimates, as well as the first (Q1) and third (Q3) quartile, are shown separately for each categorical age group. Linear regressions were performed to test the relationship of each parameter estimates with age, which was included as a continuous variable. The addition of age-squared was compared against a model including age alone to identify the best-fitting model, which is listed in the column on the right.