Cross-species Regulation of Action by Controllability: a Keystone of 5-HT signalling?

Informations projet

CRACK-5HT

N° de convention de subvention: 101078795

DOI

10.3030/101078795

Date de signature de la CE 20 Novembre 2023

Date de début 1 Decembre 2023

Date de fin 30 Novembre 2028

Financé au titre de

• European Research Council (ERC)

Coût total

€ 1 483 738,75

Contribution de l’UE

€ 1 483 738,75

Coordonné par

INSTITUT NATIONAL DE LA SANTE ET DE LA RECHERCHE MEDICALE
France

Periodic Reporting for period 1 - CRACK-5HT (Cross-species Regulation of Action by Controllability: a Keystone of 5-HT signalling?)

Période du rapport: 2023-12-01 au 2025-05-31

Résumé du contexte et des objectifs généraux du projet

CRACK-5HT investigates how the mammalian brain estimates controllability—the extent to which one’s actions causally influence the environment—and how this process is implemented in the serotonin (5-HT) system, with relevance to depression and antidepressant treatment.

In neuroscience, reinforcement learning and predictive coding emphasize utility and uncertainty as core principles of adaptive behaviour. CRACK-5HT posits a third: controllability. While perceived control is known to shape cognition, motivation, and stress responses, its computational and neurobiological bases remain poorly understood. Standard definitions, based on correlations between actions and outcomes, confound controllability with predictability. We adopt a causal definition grounded in control theory and formalized in computational models that compare predictions from an action-dependent “actor” and an action-independent “spectator.” Discrepancies between the two generate a controllability prediction error—a signal hypothesized to guide learning and strategy selection.

Serotonin is a strong candidate for broadcasting this signal. The dorsal raphe nucleus (DRN) projects broadly to brain regions involved in decision-making, and 5-HT is implicated in diverse forms of plasticity. Although traditionally linked to aversive processing and patience, recent work reveals more context-dependent functions. Learned helplessness studies tied uncontrollable stress to abnormal 5-HT dynamics and depression-like symptoms, yet the relationship between controllability and 5-HT remains unclear, largely due to inadequate paradigms and limited time-resolved recordings. CRACK-5HT addresses this gap by integrating theory-driven tasks, modelling, and rodent and human experiments.

Aim 1 — Encoding of controllability by 5-HT neurons in rodents
Aim 2 — Causal role of 5-HT in controllability-related behaviour in rodents
Aim 3 — Controllability in depression and antidepressant treatment in humans

By aligning rodent and human experiments through homologous tasks and unified computational models, CRACK-5HT seeks to reveal mechanisms of controllability estimation and their role in behaviour. This work could refine theories of 5-HT function, inform phenotyping in psychiatry, and inspire novel antidepressant screening approaches.

Travail effectué depuis le début du projet jusqu’à la fin de la période considérée dans le rapport et principaux résultats atteints jusqu’à présent

The first 18 months of the project have involved substantial preparatory work to secure ethics and regulatory approvals for rodent and human experiments. Ethics approval for freely moving rodent experiments was obtained on January 14th, 2025, and a second submission for head-fixed experiments is in preparation for submission in the first semester of 2026. The main human neuroimaging experiment (Aim 3) posed significant organizational and regulatory challenges. We first secured a network of general practitioners and private psychiatrists willing to redirect drug-naïve patients in their first depressive episode to our study. Our protocol was selected for the first wave of studies supported by the “Laboratoire participatif,” which facilitates collaborations between researchers and a network of 1,500 clinicians in the Lyon region. We then refined our recruitment strategy and data-processing pipeline to meet regulatory requirements while maintaining feasibility. The final proposal will be submitted by the end of August 2025.

In parallel, we developed the human version of the foraging task and conducted an online pilot study that yielded highly promising results (see next section). These findings were presented at an invited symposium at Neurofrance 2025, the flagship event of the French Society for Neuroscience. All cognitive tasks for the human study are now programmed and validated. A subset, including the updated explore-and-predict task (Ligneul et al., Nature Human Behaviour, 2022), is also used in projects led by close collaborators (see next section). We also developed a translational foraging task for rodents and humans, with an online human study producing promising preliminary results (see next section).

Setting up a state-of-the-art systems neuroscience lab at the Centre de Recherche en Neurosciences de Lyon (CRNL) required extensive work. We acquired and configured a multimodal system for manipulating and recording neuronal activity using optogenetics and photometry, and built and validated apparatuses for the freely moving rodent foraging task (Fig. 1). For high-throughput testing, we can already run four animals in parallel. Several transgenic mouse lines have been acquired, and colonies are being expanded.

Overall, the experiments planned in the three aims of CRACK-5HT have either started or will commence by year’s end. Alongside experimental preparations, my scientific focus has been on developing and refining computational models to track and update controllability estimates across contexts (see next section).

I have also worked to raise the visibility of my group (website, conferences), enabling the recruitment of an exceptional postdoctoral researcher (Renato Marciano Maciel) for rodent work and a highly promising PhD student (Julie Hoomans) who start in December 2025. A second PhD position in animal experimentation will be advertised in September 2025.

Progrès au-delà de l’état des connaissances et impact potentiel prévu (y compris l’impact socio-économique et les conséquences sociétales plus larges du projet jusqu’à présent)

To date, one paper has been published, providing an up-to-date overview of brain 5-HT signalling and cognition (Ligneul & Mainen, Current Biology, 2023).

An online pilot study with the human 6-port foraging task (Aim 3) showed that SSRI-treated participants (N=73) were significantly impaired in learning and prediction in controllable contexts, with no difference from controls (N=52) in uncontrollable contexts (Fig. 1). Aim 3 hypothesizes that this impairment occurs in a subset of depressed patients, and that SSRIs should mitigate it. Our results confirm the impairment but suggest antidepressants alone do not normalize behaviour. The upcoming pre–post neuroimaging study will be crucial for assessing how antidepressants influence learning and prediction in controllable environments.

International collaborations have also yielded important results. With Roshan Cools's lab, we are replicating findings on the effects of prior exposure to uncontrollable stressors and have examined the impact of controllability estimates on low-level cognitive control (results presented at RLDM 2025). With Hanneke Den Ouden's lab, we showed that controllability estimation biases (assessed by the explore-and-predict task) have excellent test–retest reliability, and we are investigating how acute SSRI treatment affects behaviour and model parameters. While not overlapping with experiments in my group, these collaborations highlight broad interest in the framework, models, and behavioural tasks developed within CRACK-5HT.

Since our seminal paper (Ligneul, Trends in Cognitive Science, 2021), controllability estimation models have been refined. We developed a Bayesian version enabling optimal inference of controllability, allowing investigation of bias origins in health and disease. We also refactored actor–spectator computations to estimate transfer entropy in changing environments, overcoming a long-standing limitation in applying information-theory metrics to real-world cognitive problems. A manuscript describing these advances is in preparation.

Finally, for animal experiments, we have validated the apparatus for the 6-port foraging task and plan to make its design open source, enabling other groups to exploit its versatility.

Figure 1

Figure 2