Experimental design and basic results.

A. Study timeline. A battery of 6 tasks was performed in a fixed order. The PIT-task always performed first to minimise interference between training and transfer. Average timings are indicated, with timings most relevant for the PIT-task data presented here, in purple. Tasks marked in black have been published / are in preparation elsewhere (Swart et al., 2017; Froböse et al., 2018; Cook et al., 2019; Rostami Kandroodi et al., 2021) MMSR = Mood & Medical Symptom ratings. B-D. PIT-task design and main task effects (averaged across drug conditions) for the three task phases. B. Instrumental training. Design (upper panel): Participants had to choose whether to click (‘Go’) or not click (‘NoGo’) with the mouse inside the blue frame. Trials were grouped into Approach and Withdrawal blocks. In the Approach context, a Go response resulted in ‘collecting’ the mushroom, while NoGo meant not collecting it. In the Withdrawal context, a Go would ‘discard’ the mushroom, while NoGo would keep it. Each mushroom only appeared in either Approach or Withdrawal contexts. For each mushroom, one action (Go or NoGo) was correct, and rewarded with 75% probability. The other action was punished with 75% probability. Participants were instructed that the outcomes received counted towards a bonus payment. Each block contained 6 unique mushrooms. Results (lower panel): mean ± standard error of the mean. Participants learnt to make the correct response for each of the required actions and Action Contexts. Note that for approach contexts, people initially displayed a Go bias, which led to above chance performance for Go-approach cues and below-chance performance for NoGo-Approach cues. However, at the end of training all cue/context combinations reached the same plateau, which approximately probability-matched the reward contingency. C: Pavlovian conditioning. Design (upper panel): In the conditioning trials, five fractals were presented repeatedly, followed deterministically by monetary outcomes of five value levels: high (100) or low (10) reward, nothing (0), low (-10) or high (-100) punishment. Intermixed with the conditioning trials, 18 query trials were presented, where participants had to select the most rewarding of random selection of two of the stimuli. The line plot shows average probability of selecting the higher CS across participants and drug sessions, as a function of valence level difference. Participants rapidly learned to select the better CS. Prior to and post conditioning, subjects were asked to rate how much they liked each fractal on a visual analogue scale (VAS). Results (lower panel): VAS significantly increased for appetitive conditioned stimuli, while they were reduced for aversive stimuli. Circles represent individual subjects. D. Pavlovian-Instrumental Transfer. Design (upper panel): Participants performed the instrumental training task in nominal extinction (they were instructed that their actions still counted towards their payment). On every trial, one of the Pavlovian CSs tiled the background. Results (lower panel): Proportion of Go responses independent of required response. In the Approach context (green), presence of a positive Pavlovian CSs increased Go (active approach), while a negative Pavlovian CS reduced Go (relative to a neutral CS). In contrast, in the Withdrawal context, the negative Pavlovian CS enhanced Go (active withdrawal), while the positive CS reduced Go. For panels C and D, results are collapsed within valence (i.e. high and low reward CS and high and low punishment CS were averaged). See supplemental materials results 1 for each of the valence levels separately.

Effects of Methylphenidate on Pavlovian-Instrumental Transfer.

A. Individual data points plus associated density distributions, for the difference in likelihood of a Go response under methylphenidate (MPH) minus placebo (PLA). Data are plotted as a function of Action and Valence. There was no main effect of methylphenidate. B&C. Methylphenidate affects action-specific PIT (i.e., the action-context specific impact of Pavlovian valence on invigoration), as a function of baseline working memory capacity. Plots show the regression line and 95% confidence intervals, and in B also individual data points. B. Effects of methylphenidate on action-specific PIT (Action Context x Valence), as a function of Pavlovian Valence. C. Breaking down the 4-way interaction demonstrates the full reversal of the effect of drug on action-specific PIT as a function of working memory (listening span) performance. In short, under placebo, action-specific PIT is present in people with low, but not high, working memory span, while under methylphenidate action-specific PIT is present in, people with high, but not low, working memory span.

Breaking down the 4-day interaction into component factors; PLA: placebo; MPH: methylphenidate; WM: working memory.

Demographic and background characteristics of participants included in the analysis. Questionnaires included the Beck Depression Inventory (BDI; (Beck et al., 1996), Behavioral Inhibition Scale/Behavioral Activation Scale (BIS/BAS; (Carver and White, 1994), Spielberger Trait Anxiety Inventory (STAI; (Spielberger et al., 1983), Multidimensional Scale of Perceived Social Support (MDSPSS; (Zimet et al., 1988), Social and Aggressive Dominance Questionnaire (SADQ; (Kalma et al., 1993) and Barratt Simplified Measure of Social Status (BSMSS; (Barratt, 2006)).

A. Pavlovian Conditioning effects and PIT effects as a function of all five levels of CS reinforcement. Liking ratings scale linearly with the levels of reinforcement, i.e. the CS associated with a large reward was liked more than the CS associated with a small reward, and vice versa for punishments. B. In contrast, PIT effects are a function of the sign (aversive, neutral, appetitive), but not the magnitude (10 vs. 100), of the CS valence. In other words, PIT effects were equally strong for large and small reinforcers of each valence. C. Effects of methylphenidate on PIT as a function of all 5 CS valence levels, which again are a function of sign but not valence.