Abstract
Importance To support treatment assignment, mechanistic biomarkers should be selectively sensitive to specific interventions. Here, we examine whether different components of reinforcement learning in humans satisfy this necessary precondition. We focus on pharmacological manipulations of dopamine and serotonin that form the backbone of first-line management of common mental illnesses such as depression and anxiety.
Objective To perform a meta-analysis of pharmacological manipulations of dopamine and serotonin and examine whether they show distinct causal effects on reinforcement learning components in healthy humans.
Data Sources Ovid MEDLINE/PubMed, Embase, and PsycInfo databases were searched for studies published between January 1, 1946 and January 19, 2023 (repeated April 9, 2024, and October 15, 2024) investigating dopaminergic or serotonergic effects on reward/punishment processes in healthy humans.
Study Selection Studies reporting randomized, placebo-controlled, dopaminergic or serotonergic manipulations on a behavioral outcome from a reward/punishment processing task in healthy humans were included.
Data Extraction and Synthesis Standardized mean difference (SMD) scores were calculated for the comparison between each drug (dopamine/serotonin) and placebo on a behavioral reward or punishment outcome and quantified in random-effects models for overall reward/punishment processes and four main subcategories. Study quality, moderators, heterogeneity, and publication bias were also assessed.
Main Outcome(s) and Measure(s) Performance on reward/punishment processing tasks.
Results In total, 68 dopamine and 39 serotonin studies in healthy volunteers were included (Ndopamine=2452, Nplacebo=2432; Nserotonin=1364, Nplacebo=1393 participants). Dopamine increased overall reward (SMD=0.21; 95%CI [0.12 0.30]) but not punishment function (SMD=-0.09; 95%CI [-0.27,0.10]). Serotonin did not meaningfully affect overall punishment (SMD=0.22; 95%CI [-0.04,0.49]) or reward (SMD=0.01; 95%CI [-0.33,0.35]). Importantly, dopaminergic and serotonergic manipulations had distinct and selective effects on subcomponents. Dopamine affected reward learning/sensitivity (SMD=0.25; 95%CI [0.10,0.40]), reward discounting (SMD=-0.08; 95%CI [-0.14,-0.01]) and reward vigor (SMD=0.32; 95%CI [0.11,0.54]). By contrast, serotonin shaped punishment learning/sensitivity (SMD=0.32; 95%CI [0.05,0.59]), reward discounting (SMD=-0.35; 95%CI [-0.67,-0.02]), and aversive Pavlovian processes (within-subject studies only; SMD=0.36; 95%CI [0.20,0.53]).
Conclusions and Relevance Pharmacological manipulations of both dopamine and serotonin have measurable effects on reinforcement learning in humans. The selective effects on different components suggests that reinforcement learning tasks could form the basis of selective, mechanistically interpretable biomarkers to support treatment assignment.
Key Points Question: Do pharmacological manipulations of dopamine and serotonin affect components of reinforcement learning in humans?
Findings: Upregulating dopamine increases reward learning/sensitivity and reward response vigor, and decreases reward discounting. Upregulation of serotonin leads to increased punishment learning/sensitivity and decreased reward discounting.
Meaning: Pharmacological manipulations of dopamine and serotonin have dissociable effects on different components of reinforcement learning. This forms a necessary basis for the development of selective markers for treatment assignment.
Competing Interest Statement
QJMH has received fees and options for consultancies for Aya Technologies and Alto Neuroscience. MB has received consulting fees from J&J, Engrail Therapeutics, CHDR, and travel expenses from Lundbeck. He was previously employed by P1vital Ltd.
