Thomas Deneux
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Thomas Deneux has a total of 79 co-authors. Here are the top co-authors by number of shared publications:
- Brice Bathellier
Centre National de la Recherche Scientifique
3 shared publications - Marie Absalon
Université Paris-Saclay, CNRS, Institut de Neurosciences Paris-Saclay
3 shared publications - Morgane Chevalier
Haute école pédagogique (HEP) Vaud, Lausanne, Switzerland
3 shared publications - Andreas Savas Tolias
Department of Ophthalmology, Byers Eye Institute, Stanford University School of Medicine
2 shared publications - Emmanouil Froudarakis
Institute of Molecular Biology & Biotechnology, Foundation of Research & Technology - Hellas, Heraklion, Greece
2 shared publications - Jakob H. Macke
Machine Learning in Science, Excellence Cluster ‘Machine Learning’, Tübingen University, Tübingen, Germany
2 shared publications
Latest Publications
- 9 October 2026
- Published by Springer Nature
- p. 117-126
From Logic Gates to Hidden Layers: a Hands-on Introduction to Neural Networks for Young Learners
- 9 October 2026
- Published by Springer Nature
- p. 117-126
From Logic Gates to Hidden Layers: a Hands-on Introduction to Neural Networks for Young Learners
- Open Access
Adolescent social profiles predict adult social behaviour, monoamine brain content and regional gene expression
The study of inter-individual variability in animal behaviors during cognitive processes is crucial to better grasp the emergence of personality traits in healthy animals. Social experiences occurring at adolescence can affect the expression of the behavioural variability as it is a critical period of brain and cognitive plasticity. Here, we asked how individual variability of social profiles is shaped at adolescence and whether the profiles are associated with regional brain metabolites and gene expressions in the adult brain. At adolescence we determined a ‘dominant’ profile expressing more dominance, a ‘submissive’ profile expressing more submission and an ‘explorer’ profile showing more novelty exploration than social interaction. Adolescent social behaviors are conserved at adulthood. Our data suggest that dominance tendencies already present at adolescence are likely to stabilize throughout life, thus shaping future social behaviors. Our results show that social profiles are not related to sex and life environment. However, being more dominant, interactive and socially motivated shape dopaminergic olfactory activity at adolescence. Also, showing the explorer or submissive profiles is associated with an opposite serotonin pattern in the cerebellum, but a similar kynurenine pattern in the prefrontal cortex and cerebellum, thus pointing toward major neurotransmitter systems in key regions for brain health. This work opens new perspectives on the importance of studying social profiles early in life and the individual neurobiological associated features that could potentially predict individual vulnerability. Graphical
Open AccessAdolescent social profiles predict adult social behaviour, monoamine brain content and regional gene expression
The study of inter-individual variability in animal behaviors during cognitive processes is crucial to better grasp the emergence of personality traits in healthy animals. Social experiences occurring at adolescence can affect the expression of the behavioural variability as it is a critical period of brain and cognitive plasticity. Here, we asked how individual variability of social profiles is shaped at adolescence and whether the profiles are associated with regional brain metabolites and gene expressions in the adult brain. At adolescence we determined a ‘dominant’ profile expressing more dominance, a ‘submissive’ profile expressing more submission and an ‘explorer’ profile showing more novelty exploration than social interaction. Adolescent social behaviors are conserved at adulthood. Our data suggest that dominance tendencies already present at adolescence are likely to stabilize throughout life, thus shaping future social behaviors. Our results show that social profiles are not related to sex and life environment. However, being more dominant, interactive and socially motivated shape dopaminergic olfactory activity at adolescence. Also, showing the explorer or submissive profiles is associated with an opposite serotonin pattern in the cerebellum, but a similar kynurenine pattern in the prefrontal cortex and cerebellum, thus pointing toward major neurotransmitter systems in key regions for brain health. This work opens new perspectives on the importance of studying social profiles early in life and the individual neurobiological associated features that could potentially predict individual vulnerability. Graphical
- 29 August 2025
- Published by Springer Nature
- p. 89-102
Fostering Metacognitive Behaviors Through AI-Driven Educational Robotics in Mathematical Problem Solving
- 29 August 2025
- Published by Springer Nature
- p. 89-102
Fostering Metacognitive Behaviors Through AI-Driven Educational Robotics in Mathematical Problem Solving
- Open Access
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (28) , 29096-29103
“AlphAI”: Teaching AI Algorithms to K12 by Training Learning Robots and Visualizing How It Works
Given the massive transformation of all areas of society by AI, it is becoming essential to integrate AI literacy into the various school curricula from an early age. However, teaching the basic concepts of AI and Machine Learning (e.g. training a model; artificial neural networks) at the K-12 level might seem too abstract, whereas teaching only how to use AI fails to really "open the black box". To overcome these difficulties, we have developed AlphAI, a software resource designed to make the understanding of AI algorithms accessible and attractive to the general public and children as young as 8 years old. This is achieved by making AI very concrete, first by manipulating the learning of educational robots that users train for different behaviors, such as circuit racing, using either supervised or reinforcement learning; second by visualizing in real time in a graphical interface the details of AI algorithms (neural networks, k-nearest neighbors, Q-learning, etc). In addition, the use of the software is not limited to beginners, since it allows to write one's own AI in Python to control the robots. In this paper, we present the basic principles of the software, its graphical interface, how to use it with various educational robots, and example activities with classes from Elementary school to University. AlphAI software and robotic kits are commercially available from Learning Robots.
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (28) , 29096-29103
Open Access“AlphAI”: Teaching AI Algorithms to K12 by Training Learning Robots and Visualizing How It Works
Given the massive transformation of all areas of society by AI, it is becoming essential to integrate AI literacy into the various school curricula from an early age. However, teaching the basic concepts of AI and Machine Learning (e.g. training a model; artificial neural networks) at the K-12 level might seem too abstract, whereas teaching only how to use AI fails to really "open the black box". To overcome these difficulties, we have developed AlphAI, a software resource designed to make the understanding of AI algorithms accessible and attractive to the general public and children as young as 8 years old. This is achieved by making AI very concrete, first by manipulating the learning of educational robots that users train for different behaviors, such as circuit racing, using either supervised or reinforcement learning; second by visualizing in real time in a graphical interface the details of AI algorithms (neural networks, k-nearest neighbors, Q-learning, etc). In addition, the use of the software is not limited to beginners, since it allows to write one's own AI in Python to control the robots. In this paper, we present the basic principles of the software, its graphical interface, how to use it with various educational robots, and example activities with classes from Elementary school to University. AlphAI software and robotic kits are commercially available from Learning Robots.
- 27 September 2024
- Published by Springer Nature
- p. 403-410
From Automaton to AI Robot: the Added Value for Learning
- 27 September 2024
- Published by Springer Nature
- p. 403-410
From Automaton to AI Robot: the Added Value for Learning
- 4 October 2023
- Published by Springer Nature
- p. 169-183
Effects of Introducing a Learning Robot on the Metacognitive Knowledge of Students Aged 8–11
- 4 October 2023
- Published by Springer Nature
- p. 169-183
Effects of Introducing a Learning Robot on the Metacognitive Knowledge of Students Aged 8–11
- 11 May 2022
- Published by MathDoc/Centre Mersenne in Comptes Rendus. Biologies
- Vol. 345 (1) , 75-89
Differential roles of auditory and visual cortex for sensory detection in mice
Sensory cortex encompasses the regions of the cerebral cortex that receive primary sensory inputs and is crucial for conscious sensory perception in humans. Yet, some forms of perception are possible without sensory cortex. For example in animal models, the association of a sound detection to a simple behavior resists to the inactivation of auditory cortex. In contrast, post-training inactivation experiments conducted in visual or somatosensory cortex led to much stronger effects. Here we show that muscimol inactivation of visual or auditory cortex in the same detection protocol transiently abolishes visual but not auditory detection. We also observe that cortex-dependency correlates with longer reaction times. This suggests that auditory cortex is more easily bypassed by other circuits for stimulus detection than other primary sensory areas, which may be due to timing differences between auditory and visual associations.
- 11 May 2022
- Published by MathDoc/Centre Mersenne in Comptes Rendus. Biologies
- Vol. 345 (1) , 75-89
Differential roles of auditory and visual cortex for sensory detection in mice
Sensory cortex encompasses the regions of the cerebral cortex that receive primary sensory inputs and is crucial for conscious sensory perception in humans. Yet, some forms of perception are possible without sensory cortex. For example in animal models, the association of a sound detection to a simple behavior resists to the inactivation of auditory cortex. In contrast, post-training inactivation experiments conducted in visual or somatosensory cortex led to much stronger effects. Here we show that muscimol inactivation of visual or auditory cortex in the same detection protocol transiently abolishes visual but not auditory detection. We also observe that cortex-dependency correlates with longer reaction times. This suggests that auditory cortex is more easily bypassed by other circuits for stimulus detection than other primary sensory areas, which may be due to timing differences between auditory and visual associations.
- Open Access
- 24 March 2021
- Published by eLife Sciences Publications, Ltd in eLife
- Vol. 10
Network dynamics underlying OFF responses in the auditory cortex
Across sensory systems, complex spatio-temporal patterns of neural activity arise following the onset (ON) and offset (OFF) of stimuli. While ON responses have been widely studied, the mechanisms generating OFF responses in cortical areas have so far not been fully elucidated. We examine here the hypothesis that OFF responses are single-cell signatures of recurrent interactions at the network level. To test this hypothesis, we performed population analyses of two-photon calcium recordings in the auditory cortex of awake mice listening to auditory stimuli, and compared them to linear single-cell and network models. While the single-cell model explained some prominent features of the data, it could not capture the structure across stimuli and trials. In contrast, the network model accounted for the low-dimensional organization of population responses and their global structure across stimuli, where distinct stimuli activated mostly orthogonal dimensions in the neural state-space.
- 24 March 2021
- Published by eLife Sciences Publications, Ltd in eLife
- Vol. 10
Open AccessNetwork dynamics underlying OFF responses in the auditory cortex
Across sensory systems, complex spatio-temporal patterns of neural activity arise following the onset (ON) and offset (OFF) of stimuli. While ON responses have been widely studied, the mechanisms generating OFF responses in cortical areas have so far not been fully elucidated. We examine here the hypothesis that OFF responses are single-cell signatures of recurrent interactions at the network level. To test this hypothesis, we performed population analyses of two-photon calcium recordings in the auditory cortex of awake mice listening to auditory stimuli, and compared them to linear single-cell and network models. While the single-cell model explained some prominent features of the data, it could not capture the structure across stimuli and trials. In contrast, the network model accounted for the low-dimensional organization of population responses and their global structure across stimuli, where distinct stimuli activated mostly orthogonal dimensions in the neural state-space.
- Open Access
- 23 May 2019
- Published by eLife Sciences Publications, Ltd in eLife
- Vol. 8
Context-dependent signaling of coincident auditory and visual events in primary visual cortex
Detecting rapid, coincident changes across sensory modalities is essential for recognition of sudden threats or events. Using two-photon calcium imaging in identified cell types in awake, head-fixed mice, we show that, among the basic features of a sound envelope, loud sound onsets are a dominant feature coded by the auditory cortex neurons projecting to primary visual cortex (V1). In V1, a small number of layer 1 interneurons gates this cross-modal information flow in a context-dependent manner. In dark conditions, auditory cortex inputs lead to suppression of the V1 population. However, when sound input coincides with a visual stimulus, visual responses are boosted in V1, most strongly after loud sound onsets. Thus, a dynamic, asymmetric circuit connecting AC and V1 contributes to the encoding of visual events that are coincident with sounds.
- 23 May 2019
- Published by eLife Sciences Publications, Ltd in eLife
- Vol. 8
Open AccessContext-dependent signaling of coincident auditory and visual events in primary visual cortex
Detecting rapid, coincident changes across sensory modalities is essential for recognition of sudden threats or events. Using two-photon calcium imaging in identified cell types in awake, head-fixed mice, we show that, among the basic features of a sound envelope, loud sound onsets are a dominant feature coded by the auditory cortex neurons projecting to primary visual cortex (V1). In V1, a small number of layer 1 interneurons gates this cross-modal information flow in a context-dependent manner. In dark conditions, auditory cortex inputs lead to suppression of the V1 population. However, when sound input coincides with a visual stimulus, visual responses are boosted in V1, most strongly after loud sound onsets. Thus, a dynamic, asymmetric circuit connecting AC and V1 contributes to the encoding of visual events that are coincident with sounds.
- Open Access
- 21 May 2018
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 14 (5) , e1006157
Community-based benchmarking improves spike rate inference from two-photon calcium imaging data
In recent years, two-photon calcium imaging has become a standard tool to probe the function of neural circuits and to study computations in neuronal populations. However, the acquired signal is only an indirect measurement of neural activity due to the comparatively slow dynamics of fluorescent calcium indicators. Different algorithms for estimating spike rates from noisy calcium measurements have been proposed in the past, but it is an open question how far performance can be improved. Here, we report the results of the spikefinder challenge, launched to catalyze the development of new spike rate inference algorithms through crowd-sourcing. We present ten of the submitted algorithms which show improved performance compared to previously evaluated methods. Interestingly, the top-performing algorithms are based on a wide range of principles from deep neural networks to generative models, yet provide highly correlated estimates of the neural activity. The competition shows that benchmark challenges can drive algorithmic developments in neuroscience. Two-photon calcium imaging is one of the major tools to study the activity of large populations of neurons in the brain. In this technique, a fluorescent calcium indicator changes its brightness when a neuron fires an action potential due to an associated increase in intracellular calcium. However, while a number of algorithms have been proposed for estimating spike rates from the measured signal, the problem is far from solved. We organized a public competition using a data set for which ground truth data was available. Participants were given a training set to develop new algorithms, and the performance of the algorithms was evaluated on a hidden test set. Here we report on the results of this competition and discuss the progress made towards better algorithms to infer spiking activity from imaging data.
- 21 May 2018
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 14 (5) , e1006157
Open AccessCommunity-based benchmarking improves spike rate inference from two-photon calcium imaging data
In recent years, two-photon calcium imaging has become a standard tool to probe the function of neural circuits and to study computations in neuronal populations. However, the acquired signal is only an indirect measurement of neural activity due to the comparatively slow dynamics of fluorescent calcium indicators. Different algorithms for estimating spike rates from noisy calcium measurements have been proposed in the past, but it is an open question how far performance can be improved. Here, we report the results of the spikefinder challenge, launched to catalyze the development of new spike rate inference algorithms through crowd-sourcing. We present ten of the submitted algorithms which show improved performance compared to previously evaluated methods. Interestingly, the top-performing algorithms are based on a wide range of principles from deep neural networks to generative models, yet provide highly correlated estimates of the neural activity. The competition shows that benchmark challenges can drive algorithmic developments in neuroscience. Two-photon calcium imaging is one of the major tools to study the activity of large populations of neurons in the brain. In this technique, a fluorescent calcium indicator changes its brightness when a neuron fires an action potential due to an associated increase in intracellular calcium. However, while a number of algorithms have been proposed for estimating spike rates from the measured signal, the problem is far from solved. We organized a public competition using a data set for which ground truth data was available. Participants were given a training set to develop new algorithms, and the performance of the algorithms was evaluated on a hidden test set. Here we report on the results of this competition and discuss the progress made towards better algorithms to infer spiking activity from imaging data.