In artificial intelligence, the dominant paradigms for learning and reasoning are often designed without direct inspiration from the biological systems that embody natural intelligence. While these approaches have driven remarkable progress, their efficiency, interpretability, and adaptability limitations persist. For instance, conventional AI systems rely heavily on vast labeled datasets, lack transparency in decision-making, and struggle to generalize in dynamic environments.
Yet, the human brain excels at these tasks, operating with unmatched energy efficiency, continuous learning, and contextual reasoning. A growing body of research suggests that integrating neuroscientific principles into AI can address these shortcomings, offering novel frameworks for explainability, self-supervision, and cognitive modeling.
At the same time, deep learning is starting to play an important role in neuroscience applications where scientists face the daunting task of decoding massive, multiscale, and high-dimensional datasets. Deep learning models are the ideal fit in this setting, and they have already been employed to extract spatiotemporal patterns of neural dynamics, enabling early and precise decoding of brain states and favoring the principled formulation of brain connectivity and dynamics models.
This workshop will focus on advancing the integration of neuroscience and artificial intelligence (AI) through:
Explainable AI in Neuroscience: Explore methods to enhance AI interpretability using neuroscientific principles, such as artificial architectures inspired by the biological neuronal structure and function.
Neuro-Inspired Reasoning and Decision-Making: Develop AI models that mimic the brain's problem-solving mechanisms, leveraging distributed representations, adaptive learning, and context-sensitive architectures.
Cognitive Functions in AI: Design benchmarks to evaluate AI's ability to replicate human cognitive processes (e.g., creativity, language understanding, problem-solving).
Biological-inspired Autonomous Learning: Investigate how AI models can adopt adaptive, dynamical, and context-aware computing strategies that resemble those of natural neuronal circuits.
Across-Systems Multiscale Dynamics: Explore how biophysical constraints, biological connectivity models, and empirical brain dynamics can inform AI architectures, improving their ability to describe multiple scales simultaneously, generalize, and self-organize.
Mathematical Principles of Neural Computation: Survey mathematical models of neural computation investigating how physical principles can describe biological neural networks and can be embedded into artificial learning systems
Call for papers: join us to challenge conventional perspectives and shape the future of computer vision!
TL;DR: is an Associate Professor of Physiology at the University of Modena and Reggio Emilia (Unimore), and director of the Neuromorphic Intelligence Lab (NILab). His work spans experimental neurophysiology, advanced microscopy, and computational modeling of brain circuits. His research focuses on the structure and function of hippocampal microcircuits, combining high-resolution electrophysiology, optical imaging, and mathematical modeling. He leads projects aimed at developing digital twins of brain regions to better understand neural dynamics and support personalized pharmacology, particularly in treatment-resistant conditions like epilepsy. He also investigates synaptic plasticity, neuromodulation and the use of brain-inspired algorithms for neuromorphic computing.
TL;DR is a Full Professor in Medical Physics at the University of Rome “Tor Vergata” and Research Staff and Faculty at the Athinoula A. Martinos Center for Biomedical Imaging (Harvard Medical School). He has previously worked as a strategy consultant at McKinsey & Company, as a facilitator for the United Nations convention on Climate Change, with the Italian National Television (RAI) and as a project coordinator with AMREF. His research is interdisciplinary, with a focus on scientific and technological solutions for the deployment of advanced physical and mathematical techniques in order to extract quantitative information of investigative, diagnostic and prognostic value in a clinical context.
TL;DR is a postdoctoral researcher in computational neuroscience at Columbia University, New York. She collaborate closely with experimentalists to understand neural mechanisms and computational principles of cognitive processes, such as decision-making, and to characterize emotional states. She developed a representational geometry framework to uncover individual cognitive strategies in primate prefrontal cortex, probe decision-making computations in cerebellar Purkinje cells, and characterize internal states following stress events in the amygdala and hippocampus. She build interpretable, data-driven models using machine learning, dynamical systems modeling, and decoding analyses to understand how the geometry of neural activity shapes behavior.
TL;DR Federico is a PhD candidate in Artificial Intelligence & Machine Learning at Imperial College London. His research explores how humans learn and adapt their movements in real-world using embodied virtual reality (VR). By integrating immersive VR environments with brain and body measurements—such as EEG and kinematic tracking—he investigates the neural mechanisms underlying motor learning driven by error-based and reward-based feedback. Federico develops as well graph neural network models to capture the temporal and spatial structure of neural data, aiming to uncover how the brain supports flexible motor adaptation in natural behaviour.
14:30-15:00 Jonathan Mapelli
15:00-15:30 Nicola Toschi
15:30-16:15 Coffee Break
16:45-17:00 David Freire-Obregón
17:00-17:30 Federico Nardi
17:30-18:00 Valeria Fascianelli
Sapienza University of Rome
Sapienza University of Rome
University of Modena and Reggio Emilia
University of Pavia