Advertisement

< Back to Article

Model-Based Reasoning in Humans Becomes Automatic with Training

Fig 3

Computational modeling: Within-group comparison.

The weighting parameter w represents a measure of model-based (w = 1) relative to model-free (w = 0) control. At the group level, model parameters remained relatively stable across single-task trials, indicating that performance in the absence of load was modestly influenced by training. By contrast, we observed higher w values and higher learning rates with increased task exposure during dual-task trials. Vertical lines represent SEM. α = learning rate, β = inverse temperature, ε = lapse rate. (A) Mean best-fitting parameters when fitting data from the ‘high load group’ and days 1–3 of training for single-task trials. (B) Mean best-fitting parameters when fitting data from the ‘high load group’ and days 1–3 of training for dual-task trials.

Fig 3

doi: https://doi.org/10.1371/journal.pcbi.1004463.g003