Logo

Axe Interdisciplinaire de Recherche de l’Université de Nice – Sophia Antipolis

Accueil > Evènements passés > Séminaires passés

Séminaires passés

mardi 14 octobre 2014

Counterfactual processing of values in private and social contexts

Jeudi 2 Juin 2016 - 16h-17h
GREDEG
Giorgio Coricelli (USC Dornsife)

I will discuss the theoretical implications of incorporating counterfactual processing of values into the process of choice, and into adaptive models of decision-making. Counterfactual evaluation means to compare the outcome of our choice with what we could have obtained with a different choice. In social settings the counterfactual outcome could be the outcome of a choice taken by another individual. We hypothesize that private and social counterfactuals share similar features – because both support learning – but social evaluations have distinguishing characteristics, such as keeping track of our social ranking. Results from a neuropsychological study (patients with lesions in the ventromedial prefrontal cortex) show a neuroanatomical dissociation between private and social counterfactuals, and fMRI data shows that the interplay between reward and social reasoning networks mediates the influence of social counterfactuals (social comparison) on the decision process.

Modélisation mathématique et simulations multi-échelles pour l’étude de la relâche vésiculaire dans les synapses neuronales.

Vendredi 20 Mai 2016 -11h à 12h
LJAD - Salle de Conférence
Claire Guerrier (ENS, Paris)

Nous avons étudié plusieurs structures neuronales à différentes
échelles allant des synapses aux réseaux neuronaux. Notre objectif était de développer et analyser des modèles mathématiques, afin de déterminer comment les propriétés des synapses au niveau moléculaire façonnent leur activité, et se propagent au niveau du réseau. Ce changement d’échelle peut être formulé et analysé à l’aide de plusieurs outils tels que les équations aux dérivées partielles, les processus stochastiques ou les simulations numériques. Dans un premier temps, nous avons calculé le temps moyen pour qu’une particule brownienne arrive à une petite ouverture définie comme le cylindre faisant la jonction entre deux sphères tangentes. La méthode repose sur une transformation conforme de Möbius appliquée à l’équation de Laplace. Nous avons également estimé, lorsque la particule se trouve dans un voisinage de l’ouverture, la probabilité d’atteindre l’ouverture avant de quitter le voisinage. De nouveau, cette probabilité est exprim
ée à l’aide d’une équation de Laplace, avec des conditions aux limites mixtes. En utilisant ces résultats, nous développons un modèle et des simulations stochastiques pour étudier la libération vésiculaire au niveau des synapses, en tenant compte de leur géométrie particulière. En nous appuyant sur ces résultats, nous avons développer un modèle pour le terminal pré-synaptique, qui couple un système d’équations d’action de masse à un ensemble d’équations de Markov, ce qui permet d’obtenir des résultats analytiques et de réaliser des simulations stochastiques rapides.

Why do we need “quantum-like” models for cognition ?

Mardi 10 Mai 2016 - 14h à 15h
LJAD - Salle de Conférence
Eric Guerci (GREDEG, Nice)

Quelques modèles d’EDP en neurosciences

Jeudi 10 Décembre 2015 - 14h à 15h
LJAD - Salle 2
Delphine Salort (Laboratoire de Biologie Computationnelle et Quantitative, UPMC)

Dans le cadre de cet exposé, nous présenterons quelques modèles classiques d’équations aux dérivées partielles permettant de modéliser un réseau composé de neurones en interactions. Dans chacun de ces modèles, nous étudierons l’impact de la force des interconnexions entre les neurones sur la dynamique qualitative et asymptotique des solutions. Nous montrerons en particulier que, dans certains cas, il est possible d’exhiber de façon très naturelle des solutions périodiques pourvu que les interconnexions entre les neurones soient suffisamment fortes et que dans le cas de faibles interconnexions, sous éventuellement certaines contraintes, le réseau tend asymptotiquement vers un état stationnaire.

Heteroclinic dynamics in neuronal information processing

Vendredi 12 Décembre 2014 - 11h à 12hAutomatic word wrap
INRIA - Salle Byron beigeAutomatic word wrap
Martin Krupa (INRIA Rocquencourt)

Homoclinic and heteroclinic dynamics are robust dynamic phenomena which can encode a large amount of information and have the flexibility to switch from one pattern to another. In a series of papers Rabinovich and co-workers have proposed a role for such dynamics in various functions of the brain, including memory formation and decision making. In this talk we review these ideas and present our recent work on the subject, which shows how a digital string of information can be encoded as a dynamical pattern of a neural network model. : Homoclinic and heteroclinic dynamics are robust dynamic phenomena which can encode a large amount of information and have the flexibility to switch from one pattern to another. In a series of papers Rabinovich and co-workers have proposed a role for such dynamics in various functions of the brain, including memory formation and decision making. In this talk we review these ideas and present our recent work on the subject, which shows how a digital string of information can be encoded as a dynamical pattern of a neural network model.

Computational properties of hippocampal neurons under physiological and pathological conditions

Mardi 2 Décembre 2014 - 11h à 12h
INRIA - Salle Euler bleu
Michele Migliore (Palermo, Italy ; Yale, USA)

The brain region most involved in higher brain functions is the hippocampus, which has a critical role in memory recollection or associative memory tasks. The underlying mechanisms depend on synaptic integration processes at the single neuron level that are extremely difficult to explore experimentally, and almost impossible to connect with psychological or behavioral findings. The pyramidal neurons in the hippocampal CA1 region are in a critical position in the hippocampa

Spike-based computing and learning in brains, machines, and visual systems in particular

Vendredi 17 Octobre 2014 - 14h30 à 15h30
Salle de conférence du LJAD
Timothée Masquelier

Using simulations, we have first shown that, thanks to the physiological learning mechanism referred to as spike timing-dependent plasticity (STDP), neurons can detect and learn repeating spike patterns, in an unsupervised manner, even when those patterns are embedded in noise [1], [2]. Importantly, the spike patterns do not need to repeat exactly : it also works when only a firing probability pattern repeats, providing this profile has narrow (10-20ms) temporal peaks [3]. Brain oscillations may help in getting the required temporal precision [4], in particular when dealing with slowly changing stimuli. All together, these studies show that some envisaged problems associated to spike timing codes, in particular noise-resistance, the need for a reference time, or the decoding issue, might not be as severe as once thought. These generic STDP-based mechanisms are probably at work in particular the visual system, where they can explain how selectivity to visual primitives emerges [5], [6]. Finally, these mechanisms are also appealing for neuromorphic engineering : they can be efficiently implemented in hardware, leading to reactive systems with self-learning abilities [7].

— 

Timothée Masquelier was trained as an engineer (Ecole Centrale Paris and MIT), before doing a PhD in computational neuroscience at Université Toulouse III, under the supervision of Simon Thorpe, working on spiking neuron network models of the visual system. He graduated in 2008, and next joined Gustavo Deco’s group at Universitat Pompeu Fabra, Barcelona, for a postdoc during which he studied the consequences of a biological learning rule known as spike timing-dependent plasticity in neuronal networks, and its implications for neural coding and information processing. In 2012, he was awarded a permanent research position at CNRS.

References

[1] T. Masquelier, R. Guyonneau, and S. J. Thorpe, “Spike timing dependent plasticity finds the start of repeating patterns in continuous spike trains.,” PLoS ONE, vol. 3, no. 1, p. e1377, Jan. 2008.

[2] T. Masquelier, R. Guyonneau, and S. J. Thorpe, “Competitive STDP-Based Spike Pattern Learning.,” Neural Comput, vol. 21, no. 5, pp. 1259–1276, May 2009.

[3] M. Gilson, T. Masquelier, and E. Hugues, “STDP allows fast rate-modulated coding with Poisson-like spike trains.,” PLoS computational biology, vol. 7, no. 10, p. e1002231, Oct. 2011.

[4] T. Masquelier, E. Hugues, G. Deco, and S. J. Thorpe, “Oscillations, phase-of-firing coding, and spike timing-dependent plasticity : an efficient learning scheme.,” The Journal of neuroscience, vol. 29, no. 43, pp. 13484–93, Oct. 2009.

[5] T. Masquelier and S. J. Thorpe, “Unsupervised learning of visual features through spike timing dependent plasticity.,” PLoS Comput Biol, vol. 3, no. 2, p. e31, Feb. 2007.

[6] T. Masquelier, “Relative spike time coding and STDP-based orientation selectivity in the early visual system in natural continuous and saccadic vision : a computational model.,” Journal of computational neuroscience, vol. 32, no. 3, pp. 425–41, Jun. 2012.

[7] C. Zamarreño-Ramos, L. A. Camuñas-Mesa, J. A. Pérez-Carrasco, T. Masquelier, T. Serrano-Gotarredona, and B. Linares-Barranco, “On Spike-Timing-Dependent-Plasticity, Memristive Devices, and Building a Self-Learning Visual Cortex,” Frontiers in neuroscience, vol. 5, no. March, p. 22, 2011.