The Indirect Effect of Serotonin on Reward Sensitivity and Mood

Post by Shireen Parimoo

What's the science?

Major depression is characterized by persistent low mood, as well as a heightened sensitivity to negative information and reduced attention to positive information. Cognitive and behavioral treatments often target these negative thought patterns by emphasizing positive thoughts and experiences, whereas pharmacological interventions like selective serotonin reuptake inhibitors (SSRIs) alleviate symptoms by altering neurotransmitter levels. It typically takes several weeks for antidepressants to become effective. Recent research indicates that antidepressants might influence the processing of positive and negative information during this time, but it is not clear how that impacts mood. This week in Nature Communications, Michely and colleagues used pharmacological intervention, cognitive testing, and computational modeling to investigate the effect of SSRIs on the interaction between reward processing and mood in healthy adults.

How did they do it?

Sixty-four healthy adults participated in four laboratory sessions over the course of a week. During the baseline pre-drug session, participants completed various affective questionnaires and a reward learning task. At the end of the baseline session, they took the medication (20 mg citalopram or placebo) before returning for the first testing session three hours later. Half of the participants were assigned to the citalopram group and the other half were assigned to the placebo group, and they took their respective medication daily. They completed the reward learning task on each subsequent testing session and they completed the questionnaires again during the final session. 

The reward learning task consisted of two learning blocks and a test block. During the learning blocks, participants chose between two images with different reward probabilities and received feedback with a monetary reward or no reward. They were also repeatedly prompted to indicate their current mood, via a happiness rating, in between trials. The test block was the same as the learning block except that no reward feedback was provided. In the test block, participants were asked to choose between images with the same reward probability but learned about in different learning blocks, which allowed the authors to evaluate preference for individual images. The authors added a mood manipulation with a wheel-of-fortune draw between learning blocks. In the draw, participants received an unexpected monetary reward or loss, which generated large prediction errors and had a substantial impact on the mood. This enabled the authors to examine whether distinct mood states, during learning, affected their preference for specific images (“mood bias”) in the subsequent test block. The authors then used reinforcement learning models to understand how mood impacts the computational mechanisms of learning from prediction errors, associated with positive and negative outcomes. They identified a dynamic learning model that showed how mood, at time of learning, biased the perception of outcomes, such that sensitivity to reward was upregulated in a positive mood state, and downregulated in a negative mood state. Lastly, they examined whether the model’s learning rate and reward sensitivity bias parameters for positive and negative mood differed between the two treatment groups.

What did they find?

Participants successfully learned about the images’ associated reward probability, but basic learning mechanisms were not different for participants in the citalopram and placebo groups across testing sessions. Additionally, participants were happier following a wheel-of-fortune win than after a loss, but this did not differ between the two treatment groups. Thus, SSRIs did not differentially affect reward learning or mood directly over the one-week treatment. Critically, after winning the wheel-of-fortune draw, participants preferred the images from the second learning block (which they had learned when they were happier), whereas after losing the wheel-of-fortune, they disliked the images from the second learning block (when they were in a low mood). Importantly, the positive effect of mood on learning was boosted by the SSRI treatment. This was shown in the dynamic learning model in which the computational parameter governing reward sensitivity in a positive mood was enhanced in the SSRI group. Moreover, this positive mood bias further extended into the test block, as participants in the citalopram group reported being happier following a wheel-of-fortune win and less unhappy following a loss, as compared to the placebo group. Overall, these results show that SSRIs selectively enhanced a positive mood bias, which had a subsequent delayed effect on participants’ mood ratings.

What's the impact?

SSRIs are the most commonly prescribed class of antidepressants, but thus far, the cognitive and computational mechanisms underlying their treatment effectiveness remain unclear. In this study, the authors identified a potential cognitive mechanism by which antidepressants alleviate the symptoms of major depression. Specifically, SSRIs increase a bidirectional interaction between mood and reward perception, where rewards led to improved mood, which in turn enhances subsequent sensitivity to future reward. This ‘vicious cycle of positivity’ can result in a positive biased perception of experience and thus to further mood improvement over time. Future studies in patients with clinical depression may test whether this mechanistic account of antidepressant drug action can help to build better treatment prediction models.

Michely et al. A mechanistic account of serotonin’s impact on mood. Nature Communications (2020). Access the original scientific publication here.