Research interests

I am a French student in computational sociology/philosophy at the Interdisciplinary Centre for Science and Technology Studies in Wuppertal. I have a training in fundamental physics (École Normale Supérieure de Cachan and Université Pierre et Marie Curie) and in history and philosophy of science (Université Paris-Cité).

I study human and social behavior using computational techniques, with a special interest in collective knowledge. My PhD focuses on modern science, as it illuminates the challenges raised by complex knowledge and the division of cognitive labor. I have broad interests that include epistemology, the philosophy and sociology of science, and cultural evolution. I remain interested in contemporary physics as well. I am also interested in Bayesian and causal inference, which I use for my PhD and other projects.

In a previous life, I have also been a journalist and the publication manager for the french online media Le Média.

Current projects

Research dynamics in high-energy physics (PhD, IZWT, Wuppertal)

Long story short, I am interested in the challenges and trade-offs involved in the production and cultivation of complex knowledge, such as:

• Challenges related to coordination. Specialization entails that different scientific communities hold distinct bodies of knowledge with little overlap. How can they understand each other and exchange knowledge? How can they sustain cooperation as they develop their own, sometimes orthogonal goals? My first paper in the context my PhD addresses this particular issue.
• Challenges related to adaptation. Complex knowledge can be difficult to adjust and re-purpose as circumstances change. How scientists navigate the balance between specialization and adaptation? Here is a work in progress addressing this question.
• Challenges related to scale. Complex knowledge and complex experiments require more brains and material resources over longer periods of time. What are the relationships between the many scales (social scale, institutional complexity, time-scale, etc.) involved in the edification of complex knowlege?

I look into what quantitative analyses of scientific literature can tell us about these issues, by developing statistical models grounded in theoretical literature from sociology, anthropology, social epistemology, and the philosophy of science. I focus on high-energy physics, a field where these challenges are very relevant, and, I would argue, at the root of a “complexity crisis”. The empirical material I study includes large amounts of data from the physics literature, including semantic and social data.

This project is highly interdiscplinary as it draws from various fields (“HPS”, the sociology of scientific knowledge, anthropology, social epistemology, …). This project is financed by the Research Training Group “Transformations of Science and Technology since 1800”.

Language Acquisition Across Cultures

I am also collaborating with the LSCP, as part of the Language Acquisition Across Cultures team led by Alejandrina Cristia. This team studies Language Acquisition by collecting and analyzing long-form recordings capturing the sound environment of infants (what they ear) and their own vocal production. This method generates large amounts of audio (typically thousands of hours of audio per corpus) which can be challenging to analyze. My work consists into the elaboration of statistical models and in the development of tools that enable performant systematic analyses across many corpora.

Recently, I have been investigating the effect on the quantity of “input” – that is, how much speech children are exposed to – on their speech output. This is challenging for a number of reasons. First, several time scales are involved in the correlations between input and output, as various effects are superposed: conversational effects, confounding contextual effects (such as child-care activites), and finally long-term developmental effects. Moreover, the algorithms we use to detect and classify speech in the data tend to confuse different speakers, which leads to spurious correlations between input and output. My goal is to develop statistical models that overcome these two challenges.

Energy transition

I have also started working on various pet projects about electric power systems and the energy transition. I have started designing a device for minimising the carbon footprint of the charge of electric batteries by optimising their schedule. My progress is documented here. For this project I use various components including an Arduino, combined with some modelling of the power system in order to forecast the optimal charge program for reducing carbon emissions.

For another project, my goal is to model electricity production and consumption in order to assess the resiliency of different scenarios, using various optimization and statistical modelling techniques. The project can be found on my GitHub.