In this work, Zhong et al. measure the activity of DRN serotonin and VTA dopamine neurons as mice learned to associate a cue with a reward. Using fiber photometry to measure the bulk calcium activity of each population of neurons in freely-moving animals, they show that dopamine neurons initially exhibit a strong reward response that gradually transfers to the cue, as expected for a reward prediction error, while serotonin neurons exhibit a constant response to the reward itself and a pre-reward ramp that emerges gradually over the course of learning. These observations are very important for reinforcement learning-based theories of serotonergic function, because they suggest that serotonin neurons are involved in signalling expected future reward, rather than surprise or a prediction error.

The idea that serotonin neurons encode reward expectation rather than punishment or surprise contrasts with ideas in another branch of the serotonin literature. Possibly in an effort to reconcile these ideas, the authors also show that various sources of stress, including head restraint used routinely in the branch of the literature in question, dramatically suppress the responses of both serotonin and dopamine neurons. Exactly how this suppression could produce the appearance of a surprise or punishment code remains unclear.

A series of optogenetic studies in rodents showed that activating serotonin neurons promotes passive waiting for delayed rewards. While these results have been interpreted as evidence that serotonin promotes patience or persistence, it could also be that dorsal raphe serotonin neurons simply decrease locomotor activity. In this work, Lottem et al. show that optogenetically activating serotonin neurons increases active reward seeking behaviour (nose poking), supporting an effect of serotonin on patience or persistence rather than locomotor inhibition.

In computational and systems neuroscience, a neural system usually refers to a group of neurons with a common function in terms of behaviour, perception, or cognition (for example, the visual system). In other areas of neuroscience, the term system might refer to parts of the brain that express a common molecular marker but with diverse effects on behaviour, etc. (for example, the endocannabinoid system). This work represents an early example of an ongoing shift in the serotonin literature away from the idea that the dorsal raphe can be seen as a neural system in the computational and systems neuroscience sense.

In this work, the authors show that serotonin neurons in different parts of the dorsal raphe nucleus (and in the dorsal vs. median raphe nuclei) have different biochemical, morphological, and electrophysiological features. While there is overlap in the properties of neurons from different raphe sub-regions, this work marks an important step towards a "glass half heterogenous" view of the raphe.

If there were a single part of the brain we could point to and call "the serotonin system" in a computational and systems neuroscience sense (see details related to Calizo et al., 2011), the dorsal raphe nucleus would be it. Here, Ren et al. make the case that not even the dorsal raphe fits the bill.

Much like the earlier work of Fernandez et al., Ren et al. show that dorsal raphe serotonin neurons projecting to the amygdala and frontal cortex are mostly not the same cells, are generally located in different parts of the raphe, and receive somewhat different inputs. Ren et al. take these results a step further by showing that these two groups of neurons display different responses to rewarding and aversive stimuli, and that optogeneticaly activating them has different behavioural effects.

The idea that serotonin neurons within the dorsal raphe are at least somewhat specialized is now impossible to ignore.