Fitted Bayesian mixed models to estimate learning/trial effects in the **training** section. Separate models for effort, neutral, performance conditions (because they should be considered to have come from different "populations")—3 models.
- only include reward trials (i.e., no probe trials)
- controlled for pre_training/baseline choices
```r
# model specification
brm(choice2 ~ trial + pre_training + (trial | subj))
```
# Dot-motion task
![[1660203516.png]]
```r
Group | Parameter1 | Parameter2 | r | 95% CI | t | df | p
---------------------------------------------------------------------------------------
performance | post_training | b1_trial | 0.36 | [0.25, 0.46] | 6.17 | 253 | < .001***
neutral | post_training | b1_trial | 0.44 | [0.33, 0.53] | 7.65 | 250 | < .001***
effort | post_training | b1_trial | 0.32 | [0.20, 0.42] | 5.29 | 252 | < .001***
```
# Math task
![[1660203561.png]]
```r
Group | Parameter1 | Parameter2 | r | 95% CI | t | df | p
---------------------------------------------------------------------------------------
performance | post_training | b1_trial | 0.21 | [0.09, 0.33] | 3.43 | 253 | < .001***
neutral | post_training | b1_trial | 0.17 | [0.04, 0.28] | 2.66 | 250 | 0.008**
effort | post_training | b1_trial | 0.38 | [0.27, 0.48] | 6.45 | 252 | < .001*** 252 | < .001***
```