Université de Strasbourg

Demian Battaglia

Biography

CNRS and Institute for Systems Neuroscience (INS), Aix-Marseille University, France & USIAS Fellow, Laboratory of Cognitive and Adaptative Neuroscience (LNCA), University of Strasbourg and CNRS

Demian Battaglia, USIAS Fellow 2020

Demian Battaglia obtained his PhD in 2005 at the International School for Advanced Studies (SISSA, Trieste, Italy), working at the interface between statistical physics and information sciences. He then reoriented his research interests toward theoretical neuroscience, under the direction of Dr. David Hansel and Dr. Nicolas Brunel at the Paris Descartes University (France). Since 2009, he has held the positions of independent staff scientist at the Max Planck Institute for Dynamics and Self-Organisation (MPIDS) in Göttingen (Germany) in the Nonlinear Dynamics department (led by Professor Theo Geisel); and of project leader within the Bernstein Center for Computational Neuroscience (BCCN). Bernstein Fellow in 2013 and Marie Curie fellow in 2014, he obtained a permanent position at the French National Centre for Scientific Research (CNRS) in 2015. Since then, he serves as faculty member at the Institute for Systems Neuroscience at Aix-Marseille University (led by Dr. Viktor Jirsa).

Dr. Battaglia’s current research explores the links between structure, dynamics and function in neural circuits at different scales, from neuronal cultures to brain-wide networks, using a combination of computational modelling, information theory and machine learning approaches. He is particularly interested in the role that oscillatory and network dynamics play in neuronal information processing.

During his USIAS Fellowship, Demian Battaglia will be hosted by Professor Jean-Christophe Cassel and Dr. Romain Goutagny in the Laboratory of Cognitive and Adaptative Neuroscience (LNCA). He will collaborate with Dr. Goutagny within the team for Neurobiology of Cognitive decline.

Project - Taming the complexity of oscillatory organisation in the hippocampus (and its larger scale networks)

01/12/2020 - 31/12/2022

Oscillations of neuronal activity play a crucial role in neural information processing, allowing the selective routing of information through brain regions and providing a reference frame to organise neural codes. Oscillations have been particularly well explored in the hippocampus, a brain structure involved in key cognitive functions, such as spatial navigation and memory. Over the years, a veritable “standard model” of hippocampal oscillations has emerged, in which a multitude of different “gamma” oscillatory rhythms have been labelled in a near-entomological fashion, according to their frequency, the phase at which they occur relative to slower ongoing “theta” oscillations and the anatomical localisation of their putative source. Furthermore, specific functions (encoding of cues, memory retrieval…) have been attributed to each of these different rhythms in a very Cartesian one-to-one correspondence.

Despite its popularity, this “standard model” may be flawed by implicit assumptions that do not respect the real nature of neuronal oscillatory dynamics. Indeed, a majority of studies have studied the properties of hippocampal oscillations by relying on averages taken over multiple trials and long times. However, instantaneous oscillatory events may strongly deviate from these well-behaved averages. In a preliminary analysis of mouse recordings performed by the team’s experimental partner (Dr. Romain Goutagny at the University of Strasbourg), the estimate is that fewer than 4 % of individual oscillatory episodes bear some similarity to the text-book average templates. On the contrary, the frequencies and phases of gamma oscillatory bursts instead constitute a structured continuum. Therefore, the standard model may well have been elaborated in terms of entities and average oscillatory behaviours, that simply do not exist in the “here and now” of actual cognitive processing, responding to current stimuli and context to produce real-time behaviour.

In this project, Dr. Battaglia proposes to forge a new language and analytic framework to characterise the complex oscillatory behaviour of the hippocampus, haphazardly standing between order and randomness. To do so, the team will adopt quantitative approaches (information and network theory, supervised and unsupervised machine learning) to extract both the “dictionary” and the “grammar” of the language of oscillatory dynamics, at the level of hippocampus (recorded via multichannel silicon probes), but also of its wider networks of interactions with other cortical regions (measured via implanted EEG arrays). They will decode this language to predict actual behaviour during navigation tasks, in which a mouse explores a maze to find the fixed location of a reward, in the same fashion we would have to do to find a shop in a neighbourhood we are gradually getting acquainted with. This oscillatory language will be modified not only by learning but also by pathology. The team will thus also consider recordings from mutant mouse models, developing traits of Alzheimer’s disease (in which spatial navigation and memory performance are affected), or Amyotrophic Lateral Sclerosis (in which altered cortical excitability properties are expected to perturb oscillatory network dynamics). Computational modelling at multiple nested scales, using veritable “virtual mouse” whole-brain models, will allow reverse-engineering of the possible cellular and circuit-level causes of the observed alterations, serving as powerful inference, hypothesis generation and conceptualisation tools.

Links


Relevant publications:

Investissements d'Avenir