I’m Valentin, a computational neuroscientist and psychologist modeling latent processes of brain and behavior.
I work with Caroline Charpentier in the Social Learning and Decisions Lab (SLD Lab) at the University of Maryland. I am also an affiliated member of the Artificial Intelligence Interdisciplinary Institute at Maryland (AIM), an affiliate researcher of the Neuroscience and Cognitive Science (NACS) program, an elected Full Member of Sigma Xi, the Scientific Research Honor Society, a Fellow at Thinking about Thinking. My doctoral thesis was supervised by Jean-Claude Dreher and Marie Claire Villeval.
My research focuses on judgment and decision-making, information processing, social learning, and beliefs under uncertainty. I study these processes across behavioral, computational, and neural levels. I use experiments to characterize behavior, generative models to infer latent cognitive dynamics, and neuroimaging to examine how the corresponding variables and computations are represented and integrated in the brain.
One axis of my work concerns information-related decision-making: how uncertain information is evaluated, transformed into beliefs, and used to guide action. To act in the world, people must estimate its current state and what future states might be. My work has examined how uncertainty in people’s evaluation of news shapes subsequent information seeking. I have also examined how people combine their own evaluation of news — a first-order estimate about the world — with higher-order inferences about others’ beliefs to guide communication. I use computational models to recover the latent beliefs and inferences underlying these decisions and neuroimaging to investigate how they are represented and integrated across brain systems. I am now extending this framework to more complex information environments, including large-scale social environments and environments in which AI systems mediate what information people encounter, evaluate, and act upon.
A second axis concerns heterogeneity in cognition, particularly in social learning. People can reach the same choices in a given situation through different combinations of learning, inference, and belief. These differences may reflect distinct cognitive profiles that population averages obfuscate. I use computational models and neural data to identify this latent heterogeneity and to ask whether apparently similar behavior is supported by different computational and neural mechanisms. I am particularly interested in how this variation relates to cognitive and psychiatric constraints, including neurodevelopmental differences such as ASD and dimensions of psychopathology such as anxiety. I am developing computational phenotypes that capture this variation, with the longer-term goal of contributing to precision psychiatry. I am also developing methods for task design and parameter estimation that can recover these processes reliably across tasks and sessions.
On Substack, I write about cognition, AI, and epistemology - for instance, why cognition matters more than intelligence for comparing biological and artificial systems, and how AI systems create decision environments that reshape human evaluation and reasoning. Outside of research, I boulder, I do photography, and I spend a great deal of time thinking about the works of Walt Whitman, Saul Leiter, and The Birth of Tragedy.