The Memory, Error and Learning (MEL) group

We use behavioral experiments, brain electrophysiology and non-invasive brain stimulation in humans to identify the neural computations of memory and learning

- News

- Research

Decoding working memory and replay from MEG signal

The spatiotemporal dynamic of information processing is visible using machine-learning decoding of high-temporal resolution neural signal from M/EEG. Using this approach, we identified a neural signature of content selection and characterize differentiated spatiotemporal constraints for subprocesses of working memory. In a context of learning, this approach allows us to test whether the brain is replaying silently what it just learned, as a replay mechanism to consolidate the new skill.

Read More

Behavioral dynamics of learning

We are investigating the behavioral short-scale dynamics of learning. During a learning session, some types of learning are evidenced during short rest periods and others during the practice itself. For example, statistical learning occurs during the practice itself while more complex rule learning occurs during short-breaks. Knowing when the brain learns is crucial for both the comprehension of memory formation and consolidation and for developing new training and neurorehabilitation strategies in healthy and patient populations.

Read More

Learning from errors

Learning through trial and error is required for daily living (“Experience is simply the name we give our mistakes” – Oscar Wilde), whether it be for a child learning to write or a stroke patient learning to walk again. Behavioral computational models provide a framework for understanding the impact of error on human learning. They pose that the brain continuously updates an error-sensitivity signal controlling how much is learned from past errors. Using machine-learning decoding of brain activity during learning tasks, we will identify the neural representation of error-sensitivity. This project will provide crucially missing information about the brain computations during learning.

Read More

Decoding working memory and replay from MEG signal

The spatiotemporal dynamic of information processing is visible using machine-learning decoding of high-temporal resolution neural signal from M/EEG. Using this approach, we identified a neural signature of content selection and characterize differentiated spatiotemporal constraints for subprocesses of working memory. In a context of learning, this approach allows us to test whether the brain is replaying silently what it just learned, as a replay mechanism to consolidate the new skill.

Read More

Behavioral dynamics of learning

We are investigating the behavioral short-scale dynamics of learning. During a learning session, some types of learning are evidenced during short rest periods and others during the practice itself. For example, statistical learning occurs during the practice itself while more complex rule learning occurs during short-breaks. Knowing when the brain learns is crucial for both the comprehension of memory formation and consolidation and for developing new training and neurorehabilitation strategies in healthy and patient populations.

Read More

Learning from errors

Learning through trial and error is required for daily living (“Experience is simply the name we give our mistakes” – Oscar Wilde), whether it be for a child learning to write or a stroke patient learning to walk again. Behavioral computational models provide a framework for understanding the impact of error on human learning. They pose that the brain continuously updates an error-sensitivity signal controlling how much is learned from past errors. Using machine-learning decoding of brain activity during learning tasks, we will identify the neural representation of error-sensitivity. This project will provide crucially missing information about the brain computations during learning.

Read More

- The team

My research aims at identifying neural computations crucial for learning in large-scale neural dynamics and at designing innovative ways to modulate and improve learning abilities. I am mainly using machine-learning techniques applied to magnetoencephalography (MEG) and psychophysics. I have also experience on brain stimulation and diffusion imaging.

Download my CV:

Szonja WEIGL

Coumarane Tirou

Alumni

Szonja WEIGL

Coumarane Tirou

Alumni

Join us

Job offer 

Type: Postdoc
Expected start: 01/02/2022
Duration: 24 months
Location: Lyon
Read More

We are continuously looking for motivated and talented students.

If you think that your CV could match the group activities, don’t hesitate to contact us.

Full Publication List

Differential brain mechanisms of selection and maintenance of information during working memory
Quentin Romain, King Jean-rémi, Sallard Etienne, Fishman Nathan, Thompson Ryan, Buch Ethan, Cohen Leonardo G.
Entrainment of local synchrony reveals a causal role for high-beta right frontal oscillations in human visual consciousness
Vernet Marine, Stengel Chloé, Quentin Romain, Amengual Julià L., Valero-cabré Antoni
mne-bids: organizing electrophysiological data into the bids format and facilitating their analysis
Appelhoff Stefan, Sanderson Matthew, Brooks Teon L., Vliet Marijn Van, Quentin Romain, Holdgraf Chris, Chaumon Maximilien, Mikulan Ezequiel, Tavabi Kambiz, Höchenberger Richard, Welke Dominik, Brunner Clemens, Rockhill Alexander P., Larson Eric, Gramfort Alexandre, Jas Mainak
From visual awareness to consciousness without sensory input: the role of spontaneous brain activity
Vernet Marine, Quentin Romain, Japee Shruti, Ungerleider Leslie G.
The longer the better? general skill but not probabilistic learning improves with the duration of short rest periods
Fanuel Lison, Plèche Claire, Vékony Teodóra, Quentin Romain, Janacsek Karolina, Nemeth Dezso
Biorxiv (2020)
Voluntary motor commands are preferentially released during restricted sensorimotor beta rhythm phases
Hussain Sara J., Vollmer Mary K., Iturrate Iñaki, Quentin Romain
Biorxiv (2021)
Consolidation of human skill linked to waking hippocampo-neocortical replay
Buch Ethan R., Claudino Leonardo, Quentin Romain, Bönstrup Marlene, Cohen Leonardo G.
Cell Reports (2021)
Statistical learning occurs during practice while high-order rule learning during rest period
Quentin Romain, Fanuel Lison, Kiss Mariann, Vernet Marine, Vékony Teodóra, Janacsek Karolina, Cohen Leonardo G., Nemeth Dezso
Basal ganglia activation localized in meg using a reward task
Sepe-forrest Linnea, Carver Frederick W., Quentin Romain, Holroyd Tom, Nugent Allison C.