NeOpTo

NeOpto is an interdisciplinary team that brings together researchers from different disciplines – physiology, psychophysics, physics, computational neuroscience – to understand how perception and action are shaped by the dynamic computations that occur within topographically organised networks.

To reach this goal, we combine:

1 – Photonic imaging and electrophysiological recordings at multiple scales in-vivo, in particular in non-human primates (NHP). 

2 – Behavioral (psychophysical and oculomotor) experiments in NHPs and humans, focusing on predictive/Probabilistic models, dynamic perception and perception/action loops.  

3 – Computational frameworks constrained by experimental observations, particularly neural encoding/decoding models as well as AI-inspired (but biologically plausible)  hierarchical/predictive learning models.

4 – Innovative methods for data analyses for extracting perceptual/behavioral signals from complex, dynamical, and “noise-dominated” neural activity.    

5 – Pushing new technologies to record neural activity across scales from microscopic (2 photon microscopy using calcium or voltage sensors) to mesoscopic (wide-field optical imaging), with our partners from the departments of Optics  (Fresnel Institute Marseille) and Astrophysics (Laboratory of Astrophysics Marseille). 

We are members of CoNeCT and the NeuroTechCenter. Students interested in joining our groups can contact us and check the opportunities at Marseille Neuroschool (NeuroMarseille). 

Our projects are grouped around three main actions:

Encoding: Multi-scale processing and representation of dynamic stimuli within topographic maps

We wish to understand how neuronal dynamics within maps participate in the processing and representation of naturalistic stimuli. To do so, we compare the neuronal dynamics at mesoscopic (using Voltage-sensitive dye imaging -VSDI) and microscopic scales (using laminar probes, 2 photon microscopy of GeCIs and soon also GeVIs) of the early visual cortex (V1, V2, V4, MT…)  in NHP (macaques and marmosets). We combine computational and data-driven approaches to guide the development of dynamic hierarchical generative models of the visual system functions across different contexts. 

Decoding: From dynamic interactions between cortical areas to behaviour

Going beyond the specific cortical area, we position ourselves in a more global neuro-behavioural dynamical system. In particular, we wish to understand how intra-areal dynamic interactions influence (and are influenced by) downstream areas, and how, together, they control arm and eye movements. We make simultaneous recordings in multiple visual (V1, V4, MT…) and motor areas (in collaboration with the InVIBe and COMCO teams) in NHP reaching for objects in motion. In parallel, we study how predictive information influences, and closes the loop between various visual and motor behaviour (smooth pursuit, saccades, reaching and grasping movements).  Our ultimate goal is to achieve a more unified dynamical and probabilistic view of sensorimotor processes. 

Neurodevelopmental and psychiatric disorders, clinical and theoretical aspects 

We research visual pathologies with a focus on amblyopia, especially in relation to its cortical origin.  In addition, we study visuo-perceptual and visuo-motor dimensions of psychiatric disorders (schizophrenia and autism). We focus on how probabilistic inference is affected in these disorders, particularly in visual motion and 3D processing, bistable perception, hallucinations, and visuo-motor control. We use computational models to improve our understanding of the underlying neural mechanisms  for neuro-typical and atypical inferential/predictive processes.