inVibe

Questions & objectives

First, we aim at developing theoretically-grounded behavioral paradigms to dissect the representational hierarchy, from sensory processing (e.g., visual motion encoding) to motor output (e.g., eye movements) and cognitive functions (e.g., attention, decision-making, confidence) contributing to adaptive behavior. Second, we aim at identifying the underlying computations and link them to neuronal dynamics at multiple scales, ranging from local populations (e.g., in fronto-parietal cortex or basal ganglia) to functional maps (e.g., across occipito-parieto-frontal cortex) and recurrent networks (e.g., FEF-LIP, cortico-striatal loops).

Methods

We investigate neural computations underlying both visually-complex and naturalistic behaviors (e.g., visuomotor integration, foraging) through a comparative approach spanning humans, macaques, marmosets, rodents, and fish (Fig1 a, b). We developed specific high-end virtual reality setups to simulate ecological visuo-motor contexts that elicit naturalistic behaviors.

We combine our paradigms with physiological recordings, tracking behavior, on the one hand, with eye tracking, video monitoring and AI cutting-edge technology (e.g., DeepLabCut), and on the other hand, measuring neuronal population dynamics across spatiotemporal scales using techniques such as NeuroPixels probes, Utah arrays, and fMRI (Fig. c).

Theoretically, we aim to identify canonical computational principles governing these behaviors and their implementation in recurrent neural networks, including feedforward and feedback interactions between cortical and subcortical areas. We extend classical dynamical Bayesian inference frameworks to encompass recurrent network dynamics representing multiple interacting variables.

Composition

The team is composed of 5 permanent CNRS researchers (Guillaume Masson, Manuel Vidal, Martin Szinte, Guilhem Ibos, Fanny Cazettes), an associate researcher (Andrea Desantis, ONERA researcher). In December 2024, the team  hosts one post-doc fellow, 6 PhD students, 3 engineers and 1 Research assistant (see Team composition page)

Funded projects

Amidex – 2023-2027: Long Range Cortical Interactions during comparison of sensory and cognitive information
Led by Guilhem Ibos,, group: Camila Losada (Phd), Alexis Monnet-Aimard (Phd), Arno Feinstein (Phd), Lucio Condro (RA).

This project aims at comparing the implication of parietal, prefrontal and visual cortical areas during the active comparison of visual representation (what we are looking at) and cognitive representation of behaviorally relevant information (what we are looking for). We combine behavioral,  multi electrode, multi areas electrophysiology and modeling methods to decipher how goal directed behaviors emerge from the interaction of occipital parietal and prefrontal cortices.

ANR CRCNS/NSF – 2020-2025: Integrating sensory and prior information to control behavior  (Priosens)
Led by Guillaume Masson, group: Cleo Schoeffel (PhD), Guilhem Ibos (CR). US partner: Nicholas Priebe (Pr) Department of Neurosciences, University of Austin.

This project aims at understanding how visual and extra-retinal information about object motion are dynamically integrated to optimally control pursuit eye movements in both humans and monkeys. We investigate how complex, naturalistics moving images are encoded by neuronal populations in cortical areas V1 and MT and how they are decoded to drive both reflexive and voluntary pursuit in both humans and marmosets. We are also interested in unveiling how Prior information is shaped by perceptual learning and input contingencies. 

CNRS 80Prime- 2024-2028: NatSpeed: Visual motion perception, a naturalistic approach
Led by Guillaume Masson, group: Asma Bendahame (PhD). Paris-Cité partner: Jonathan Vacher, MAP5 Université Paris-Cité

Estimating the speed of moving natural objects implies to measure and selectively integrate the local motion energies across different spatiotemporal scales. To decipher the dynamics of motion integration and integration, we design a new class of motion stimuli. Random-phase textures, called Motion Clouds, allow to finally control the statistical structure (mean and variance) along several image dimensions. Combining psychophysical methods and computational models, we aim at grabbing the integration/segmentation rules within a new representational space: the Scale-Velocity space.

Simons Foundation (US) SCGB TTI award – 2023-2026: Neural computations and dynamics of flexible behavior.
Led by Fanny Cazettes,  group: Carole Marchese (IE), Saleha Siddiqui (IE), Antoine Courbi (IE)

This project investigates how the brain’s parallel processing creates a “reservoir” of potential solutions, enabling rapid cognitive flexibility. Using virtual reality and large-scale neural recordings in mice, our goal is to understand how this reservoir facilitates adaptation to changing conditions during foraging. This research aims to bridge the gap between neural circuit dynamics and normative theories of decision-making, ultimately providing insights into the mechanisms of flexible behavior.

ERC Starting Grant – 2025-2030: Understanding diversity in decision strategy: from neural circuits to behavior.
Led by Fanny Cazettes,  group: You?

This research explores the neural underpinnings of diverse decision-making strategies. Using cutting-edge techniques like two-photon holographic stimulation and electrophysiological recordings in mice performing a foraging task, we will dissect the circuits and mechanisms controlling strategy switching. This innovative approach will provide a mechanistic understanding of how behavioral diversity emerges, including deviations from typical behavior.

ANR JCJC – 2023-2027: Modeling human visual and oculomotor retinotopic maps.
Led by Martin Szinte,  group: Uriel Lascombes (PhD), Sina Kling (PhD)

Our brains deal and store information through spatial and motor retinotopic maps. The assessment of these maps was for long restricted to the animal model, different studies targeting distinct brain areas and distinct neuronal populations. We aim at determining through the whole computational neuroimaging method using high field (3T) and ultra-high field (7T) fMRI the retinotopic maps responsible for active vision.

DFG Pi-project – 2024-2028: Mechanisms of 3-Dimensional attention and visual constancy in primates.
Project in collaboration with Martin Szinte, Guillaume Masson, group: Baptiste Caziot (postdoc-PI), Dilara Erisen (PhD).

When we move our eyes, we experience visual stability despite constantly changing retinal inputs. Previous studies have focused on 2D visual scenes, neglecting depth information. This project will systematically investigate the orienting of attention in 3D, both through psychophysical experiments in humans and physiological experiments in non-human primates, with the goal of understanding how the brain maintains perceptual stability of 3D scenes. The results are expected to demonstrate that attention operates on a 3D representation of the environment and that these effects are reflected in the neural activity of the posterior parietal cortex.

Chaire « Santé » CMA CGM. AMIDEX Aix-Marseille Université – 2024-2027:. Translational Neuro-ophthalmology Lab Marseille –TRINOLAB–.
Project in collaboration with  Martin Szinte, group: Jan-Patrick Stellmann (PI).

The project aims at establishing a high-end clinical research platform in Marseille to study and provide expert care for rare optic neuropathies. It integrates clinicians, researchers, and laboratories to improve research, clinical care, and access to new therapies for patients in the region. The project aims to transfer knowledge from neuroscience and clinical trials into clinical practice, support the academization of paramedical staff, and improve patient information and communication. Overall, the TRINO lab will serve as a center of excellence for optic neuropathies, providing a unique opportunity to advance research and patient care in this field.

FRM postdoctoral fellowship – 2024-2027: Unraveling the Topography of Human Prefrontal Eye Fields: Anatomo-functional Organization and Connectivity explored with high-field MRI.
Project led by Guillaume Masson, Martin Szinte, group: Marco Bedini (postdoc)

The project aims to use multimodal MRI data to localize and characterize the different prefrontal retinotopic maps, assess their anatomical and functional variability, and investigate their contribution to various oculomotor tasks. It will build a probabilistic atlas capturing the overlap of these retinotopic maps across subjects, and systematically characterize their connectivity patterns using functional and diffusion MRI. The connectivity analyses will map the spatial organization of functional connectivity within eye fields and reconstruct the white matter tracts connecting them, to understand if the white matter also follows topographic principles of organization.