Our computational model revealed that people tended to make worse decisions when intoxicated and possibly also when in a bad mood not because they were insensitive to punishments but because they were more sensitive to rewards.

New preprint with @hollysully.bsky.social & @kevinmking.bsky.social:

Alcohol intoxication and negative mood similarly affect reward learning but not punishment learning in the Iowa Gambling Task

doi.org/10.31219/osf...

This sums up life lately

Bonus: This is the bar laboratory in which we conducted this study

Stress moderately increased choices for alcohol (only in sober participants), but the effect on the bias parameter was much stronger. This is interesting, because in our paradigm stress had to shift the relative value of alcohol enough to overcome a preference for the alternative.

Using drift diffusion modeling, we decomposed choices and response times into latent decision-making mechanisms - how carefully participants make decisions, how much they relied on their baseline drink preferences, and whether they developed a bias toward alcohol regardless of preference.

We tested this with a 2 (stress/no stress) x 2 (intoxicated /sober) design: Participants (N = 160) rated drinks and made repeated choices between alcoholic and non-alcoholic drinks before and after manipulations of stress and alcohol intoxication (BrAC=.06%).

Instead of summing over various consequences (e.g., having a hangover, drunk driving), Megan modeled random slopes nested in pps & consequences. She found that, irrelevant of its average valence (neg/pos), the more often a consequence is experienced the higher its expected likelihood and valence.