Revisiting the importance of model fitting for model-based fMRI: It does matter in computational psychiatry
Fig 5
The effects of within-group differences in true learning rates on beta values (regression coefficients) and true learning rate.
The between-group effect size of learning rate (gray), RPE of GLM1 (red), and negative value of GLM2 (blue) are plotted as a function of the standard deviation (s.d.) of true learning rate. The s.d. of learning rate is calculated for each group and then averaged across groups. Note that the s.d. is calculated for the actual parameter value after the truncation that restricts the range of learning rate to [0, 1].