Thread by @karelkieslich

In this new preprint we ask: What drives #anhedonia? Problems with experiencing pleasure, reward valuation, or reward learning? @jonroiser, @vincentvalton and I reviewed the evidence from mostly cross-sectional studies in depression and schizophrenia psyarxiv.com/cya6r 🧵1/

The term anhedonia was classically understood as “inability to experience pleasure”, but do patients with anhedonia actually experience less pleasure? 2/

The simplest way to find out is to ask. On questionnaires and in clinical interviews, anhedonic patients report diminished experience of pleasure. However… 3/

…responding to a questionnaire requires imagining a hypothetical past or future reward, and so responses may reflect what you recollect or imagine, rather than what you actually experienced, and may be biased. 4/

When anhedonic people are asked to rate their in-the-moment enjoyment from various rewarding events throughout the day, or rate the pleasantness of sweet tastes and pleasant odours in the laboratory, surprisingly they do not report less enjoyment than healthy controls! 5/

Similarly, when presented with pictures and tastes of sweet things in a brain scanner, the brains of anhedonic patients react similarly to those of non-anhedonic people. 6/

So, if patients with anhedonia don’t have problems experiencing pleasure, why do they report that they enjoy (and are interested in) things less? Could it be because they internally value them less? 7/

Unlike pleasure, how much one values something can actually be measured. Not directly (because value is subjective), but we can measure how frequently people choose one thing over another, or how much money or physical effort they are willing to expend to obtain something. 8/

In general, patients with anhedonia behave as if they valued rewards less. While non-anhedonic people are typically willing to expend more effort for larger rewards, in anhedonia this pattern is less marked. 9/

While non-anhedonic people usually develop a preference for stimuli which are more frequently rewarded, patients with anhedonia in depression (but, interestingly, not in schizophrenia) are less likely to develop such preference. 10/

And when anticipating a rewarding outcome, anhedonic patients show attenuated brain responses compared with non-anhedonic people. 11/

OK, so people with anhedonia seem to behave as if they value rewards less. But why? One intriguing hypothesis is that this is because they are somehow less able to learn about reward from experience. 12/

Often, you don’t know how rewarding something is beforehand: you need to try it, experience the reward, and update your expectations. This difference between expected and obtained reward is termed “reward prediction error”. 13/

Do anhedonic patients value rewards less because they are less able to compute reward prediction errors and use them to learn about rewards? The evidence is mixed… 14/

Some studies found that reward prediction error-related responses measured in the brain scanner are weaker in anhedonic patients, or that they learn from them to a lesser degree. But others didn’t find this. 15/

Problematically, it isn’t always easy to separate impaired reward valuation from impaired reward learning: if all rewards appear equally low in value to you, it will look like you can’t learn which one is better, even if your learning per se is intact. 16/

One way to separate reward valuation from learning is to use computational modelling. Studies that did this found that learning appears to be intact in anhedonia. 17/

So, we can’t convincingly say it’s impaired learning that explains lower reward valuation in anhedonia. So what explains it then? We don’t really know yet. 18/

To recap, anhedonic patients behave as if they valued rewards less. The reason behind this is likely nuanced, and may differ across patients. (But it doesn’t seem to be that they derive less pleasure from rewards, or that they are less able to learn about them.) 19/

Figuring this out will require approaches that can isolate the various cognitive processes that contribute to reward valuation, computationally modelling these, studying them in anhedonic patients across diagnoses, and moving beyond cross-sectional designs to assess causality. 🏁