Table 2 Model comparison results, separately per drug condition

From: Dopamine regulates decision thresholds in human reinforcement learning in males

Model

η

Placebo

-elpd (SE)

Rank

L-dopa

-elpd (SE)

Rank

Haloperidol

-elpd (SE)

Rank

DDM0

-

296.5 (52.1)

3

375.9 (52.8)

3

480.2 (51.2)

3

RLDDM 1

Single

178.7 (47.7)

2

271.6 (50.0)

2

412.7 (49.5)

2

RLDDM 2

Dual

67.3 (48.5)

1

195.4 (51.2)

1

336.8 (50.7)

1

  1. We examined reinforcement learning drift-diffusion models (RLDDMs) with a single learning rate η (RLDDM 1) vs. separate learning rates for positive and negative prediction errors (RLDDM 2). We also included a null model without learning (DDM0). Model comparison was conducted using the estimated log pointwise predictive density (-elpd)65 where smaller values indicate a better fit.
Back to article page