Figure 2. Model parameters.

(A) Posterior distributions of the parameters’ group-level means for each group (left and right panels). Parameters ${\alpha }_{no-pain}$ and ${\alpha }_{pain}$ are learning rates for avoided and received pain outcomes, respectively; parameter $\beta$ is the inverse-temperature parameter. The middle panels are joint density plots of ${\stackrel{-}{\alpha }}_{pain}$ and ${\stackrel{-}{\alpha }}_{no-pain}$ (dots are samples from the Markov chain Monte Carlo [MCMC]), showing that ${\stackrel{-}{\alpha }}_{pain}$ is reliably greater than ${\stackrel{-}{\alpha }}_{no-pain}$ in the placebo group only. (B) The difference between the posterior distributions for each drug group vs. the placebo group, showing that ${\stackrel{-}{\alpha }}_{no-pain}$ is greater and $\stackrel{-}{\beta }$ is smaller in both drug groups compared to the placebo group. Red lines indicate 95% highest density intervals (HDIs).

Figure 2—figure supplement 1. The 95% highest density intervals (HDIs) of the posterior distributions of each participant’s learning rate for pain ( ${\displaystyle {\alpha }_{pain}}$) and no-pain (${\displaystyle {\alpha }_{no-pain}}$) outcomes.

Participants are sorted according to the difference between their two learning rates (${\alpha }_{pain}$ − ${\alpha }_{no-pain}$). Note that ${\alpha }_{pain}$ and ${\alpha }_{no-pain}$ were positively correlated in the levodopa group (r = 0.43, p = 0.03), but were not correlated in the placebo (r = −0.16, p = 0.4) and naltrexone (r = 0.31, p = 0.10) group.

Figure 2—figure supplement 2. Modeling results from an independent sample of untreated participants from a previous study (N = 23), replicating the asymmetric learning rates (${\stackrel{-}{\alpha }}_{pain}$ > ${\stackrel{-}{\alpha }}_{no-pain}$) found in our placebo group.