Our work can be divided in 4 complementary but synergistic axes:
- First (SA1), we studied how the activity of 5-HT neurons correlates with specific events such as anticipation and reception of uncertain rewards or processing of expected vs unexpected stimuli. We used innovative imaging techniques to record neural activity in the dorsal raphe nucleus (DRN) where most 5-HT neurons innervating the forebrain originate. We showed that the activity of 5-HT neurons reflects more than affective properties of events and stimuli: it appears to encode a generic surprise signal independent of affective valence;
- Second (SA2), we manipulated activity of 5-HT neurons during different behaviors (locomotion, exploration, decision-making, learning) using techniques as optogenetics (excitation), chemogenetics (inhibition) and pharmacology (unravel role of 5-HT receptors). Our results suggest an opponent role for 5-HT2a and 5-HT2c receptors in decision making, and that 5-HT has different short and long time scale effects on learning and decision making. Short term effects tend to modify persistence of ongoing behaviors (e.g. actively waiting for predictable rewards) and speed at which animals learn to predict events based on cues. Activating 5-HT neurons immediately reduces locomotion whereas stimulation over weeks tends to increase baseline locomotion. In addition, we observed paradoxical effects of 5-HT activity in foraging. While a transient 5-HT photo-activation promotes persistence, a sustained pharmacological increase promotes flexibility. Adopting a new probabilistic foraging task, we characterized the specific computational contribution of 5-HT photostimulation, finding that persistence is driven by a reduction of negative impact of outcomes in the decision to leave a patch;
- Third (SA3), we developed experimental paradigms to address the impact of 5-HT on olfactory processing in the primary olfactory cortex, densely innervated by 5-HT. Olfaction is ethologically fundamental for rodents, and mice excel at predicting upcoming events based on odors. Thus, we found that 5-HT release in the olfactory cortex suppressed the spontaneous activity of olfactory neurons, a phenomenon which could translate into a reduced weight of predictable events in guiding behavior.
We did not observe the expected reduction of olfactory neural responses when odors were predictable (SA3a). Although, across the neuronal population, net activity change was zero, the representational space was rotated. Furthermore, neuronal representations appeared before the predicted odor, as a recall-induced reinstatement or a prior. Upon stimulation of 5-HT neurons (SA3b), there was no simple reduction of spontaneous activity of neurons in the olfactory cortex, as hypothesized. These results are currently being prepared for publication.
Framing our studies in a computational framework allowed us to track and quantify some underlying algorithmic principles that organize individual behaviors and formulate predictions to arbitrate between theories of 5-HT function. We aim in future work to translate our findings in mice to humans and computational models are useful to help apprehend the functions of 5-HT independently of the species;
- Finally (SA4), we combined functional magnetic resonance imaging (fMRI) and optogenetics in awake rodents. Since 5-HT neurons innervate most of the forebrain, its net effects on behavior depend on coordination of multiple target areas. However, conventional imaging methods in neuroscience are often limited to 2D fields of view at a submillimeter scale, restricting the analysis of 5-HT effects brain-wide. fMRI in awake mice entailed several technical challenges, progressively solved. We provide one of the first demonstrations of brain-wide blood oxygenation levels (BOLD signal, an indirect marker of neural activity) in mice performing simple cognitive tasks. This approach will be a complement to the more local dynamics investigated in the other research axes.
