Explore this article's sources with AI
Follow PeptideMethods on Google
Semaglutide and Decision-Making in MDD
A small randomized trial analysis found changes in modeled choice behavior, but it did not show that semaglutide treats depression.
Why we wrote this. The randomized design is meaningful, but the new paper reports an exploratory task analysis rather than a depression-treatment result.
In this article (5 sections)
A secondary analysis of a small randomized trial found that oral semaglutide changed three modeled features of decision-making in adults with major depressive disorder (MDD): sensitivity to expected reward, sensitivity to uncertainty and the tendency to repeat the previous choice. The experiment used a computer task after 16 weeks of treatment[1]. It did not show that semaglutide treats depression, improves everyday decisions or produces a direct effect on the brain independent of metabolic change.
The trial and the decision task
NCT04466345 was a completed phase 2 trial at one outpatient center in Toronto. Participants were adults aged 18 to 60 with MDD, a body mass index of at least 25 and below-average performance on a test of executive function. They were assigned to 16 weeks of oral semaglutide or matching placebo alongside their existing treatment. Psychotropic regimens had to remain unchanged from four weeks before randomization through the trial[2]. The registry identifies an executive-function composite as the primary outcome, not the explore/exploit task reported in this paper[3].
Participants completed a 300-trial, four-option computer task at baseline and week 16. The reward assigned to each option drifted over time, so a participant could choose the option believed to offer the highest reward or sample an uncertain alternative. A reinforcement-learning model estimated three separate choice parameters from trial-by-trial behavior. These parameters are mathematical descriptions of performance on this task, not clinical diagnoses or direct measurements of neural activity.
The authors state that the trial's prespecified primary analysis did not find an improvement in its objective executive-function measure[2]. ClinicalTrials.gov lists tests of attention, memory and quality of life among the secondary outcomes, but it does not list the restless-bandit parameters reported here[3]. The current analysis can still generate useful evidence, but its exploratory status should travel with every result.
What changed in the semaglutide group
The model estimated a positive average treatment effect for exploitation, meaning greater sensitivity to learned expected reward (0.0242, 95% confidence interval 0.0143 to 0.0341; P < .001). Directed exploration, or sensitivity to uncertainty, also increased (0.0619, 95% confidence interval 0.0132 to 0.1106; P = .013). Choice perseveration increased as well (0.3017, 95% confidence interval 0.0633 to 0.5402; P = .013)[2]. Exploitation and directed exploration are separate model parameters, so both can rise at the same time.
An exploratory analysis of normalized regret asked whether choices accumulated less lost reward than the best available option. Estimates favored semaglutide at task rounds 100, 200 and 300, but none remained statistically significant after adjustment for the three checkpoint comparisons. The adjusted P values were .204, .096 and .204, respectively[2]. That result is more cautious than saying participants became better decision-makers.
Weight loss did not settle the mechanism
The semaglutide group had a larger reduction in body mass index. In exploratory mediation models, measured BMI change did not statistically explain the treatment associations with exploitation or directed exploration. The indirect estimate for perseveration approached the conventional threshold but remained inconclusive[2]. A nonsignificant mediation test does not prove that weight or wider metabolic changes played no role, especially in a small sample.
The paper proposes links between metabolic signals, reward processing and choice behavior, but this experiment did not measure a specific GLP-1 pathway in the brain. The task also cannot establish whether any modeled change would help or harm decisions outside the laboratory. Increased perseveration, for example, means greater dependence on the previous action after accounting for modeled value and uncertainty. Its clinical meaning remains uncertain.
Why the result needs a narrow reading
Only 54 of 72 randomized participants completed the endpoint evaluation: 28 of 35 assigned to semaglutide and 26 of 37 assigned to placebo. The authors describe the analysis as secondary and exploratory. The sample was restricted to people with overweight or obesity and pre-treatment cognitive dysfunction, which limits generalization to the broader MDD population. Participants used mixed psychiatric medication regimens, and the study was not large enough to separate effects by drug class[2].
Blinding may also have been compromised because gastrointestinal effects and noticeable weight loss could reveal treatment assignment. Randomization reduces baseline confounding, but functional unblinding can still influence behavior or expectations. Moderator analyses by age, sex, BMI and depression severity were adjusted for false discovery rate, yet the authors label them hypothesis-generating because the sample was modest.
What this trial does not establish
This paper does not support using semaglutide for depression. The registered trial studied adjunctive treatment for cognitive dysfunction, and the new readout concerns a specific experimental task. It did not test remission from MDD, suicide risk, relapse prevention or daily functional decisions. The study received support from the Physicians' Services Incorporated Foundation, which the paper says had no role in the study or publication decision[2]. Larger preregistered studies would need to reproduce the task findings and connect them to outcomes that matter to patients.
Medical disclaimer: This article is for educational and journalistic purposes only and does not constitute medical advice. Peptides discussed may be classified as prescription medicines or research chemicals depending on your jurisdiction. Always consult a qualified healthcare professional before using any peptide product. PeptideMethods.com does not sell, distribute, or facilitate the sale of any peptide product.
Frequently asked
Did semaglutide improve depression in this study?
This paper did not report semaglutide as a treatment for depressive symptoms. It reported a secondary analysis of modeled behavior on a computer decision task.
What do explore and exploit mean in the task?
Exploitation means choosing based on the highest learned expected reward. Directed exploration means sampling an option because its value is more uncertain. They are separate modeled tendencies, not opposite ends of one score.
Were the findings independent of weight loss?
Measured BMI change did not statistically mediate the exploitation or directed-exploration findings, but the small exploratory analysis cannot rule out weight-related or other metabolic pathways.
Can the results be generalized to everyone with MDD?
No. The single-center sample included adults with overweight or obesity and evidence of cognitive dysfunction, and only 54 participants completed the week 16 evaluation.
Sources
- [1]Zhang Z et al. Semaglutide modulates explore/exploit decision-making in major depressive disorder: A randomized clinical trial. Psychological Medicine. 2026. PMID 42904347.Tier 1 · primary↩
- [2]Zhang Z et al. Full open-access article. Psychological Medicine. Cambridge University Press. 2026.Tier 1 · primary↩
- [3]NCT04466345. Semaglutide for the Treatment of Cognitive Dysfunction in Major Depressive Disorder. ClinicalTrials.gov.Tier 1 · primary↩
No revisions yet. First published .