Massachusetts Institute of Technology
At the Center for Biological and Computational Learning (CBCL), we study the theory of learning under physical, computational, and biological constraints. Using a multidisciplinary approach, we investigate when and how learning is possible to better understand the brain and to build better machines.
New Seminar Series
The landscape of AI research is shifting. Many problems that once defined academia are now dominated by scale, data, and incumbent advantage. Yet the most important questions remain open. What are the fundamental principles of intelligence, and how can they guide systems that learn efficiently, generalize robustly, and create real societal and economic value? This seminar argues that academia remains the place to identify and test principles of intelligence. Startups remain the place to turn those principles into systems that matter. We convene the MIT-area community to identify foundational principles that are shared by biological and artificial intelligence, and translate them into deployable, venture-scale technologies. We will connect empirical evidence from cognitive development, systems neuroscience, and modern AI systems with formal structure in learning and reasoning, including sample efficiency, sparse compositionality, invariances, memory, optimization dynamics, and generalization. By the end of the series, we will have identified and tested key conjectures about intelligence and built new collaborations between labs and startups.
From the Lab
In the rapidly evolving field of NeuroAI, a significant amount of energy is dedicated to 'alignment', the idea that representations from artificial intelligence should converge towards biological inte...
The following is an edited transcript of a conversation between John Gabrieli Buchet (Poggio Lab, MIT) and Blaise Agüera y Arcas (VP, Google) recorded as part of the Poggio Lab's [AI: Foundations -...
Principal Investigator
Tomaso A. Poggio is a founder of computational neuroscience. He pioneered models of visual perception, bridged neuroscience and machine learning, and helped establish regularization theory and learning theory in vision. His work now focuses on the mathematics of deep learning and visual recognition. He has founded, advised, or invested in multiple technology companies, including DeepMind and Mobileye, and mentored leaders such as Christof Koch, Amnon Shashua, and Demis Hassabis. Poggio is the Eugene McDermott Professor at MIT and former co-director of the Center for Brains, Minds, and Machines. As of 2024, he stands as the largest individual recipient of NSF AI funding over a 14-year period, responsible for over a quarter of all such funding awarded to MIT.
Research Focus
Foundational research into the mathematical principles of learning and the theoretical boundaries of predictive models.
Analyzing how complex neural networks generalize and the optimization dynamics in high-dimensional parameter spaces.
Exploring novel architectures and training methods that move beyond standard approaches to improve efficiency and robustness.
Investigating how biological systems process information to derive principles for human-like machine intelligence.
Our students and visiting researchers have expertise in physics, computer science, cognitive psychology, and complexity theory.
Researcher
Administrative Assistant to Dr. Tomaso Poggio
Researcher
Researcher
Researcher
Researcher
Researcher
Researcher
Researcher
Researcher
Researcher
Publications
Liao, Q., Ziyin, L., Gan, Y., Cheung, B., Harnett, M. & Poggio, T.
Preprint (February 25, 2026)
Poggio, T.
CBMM Publication
Poggio, T.
CBMM Publication
Dubach, R., Abdallah, M. S. & Poggio, T.
CBMM Publication
Danhofer, D. A., D’Ascenzo, D., Dubach, R. & Poggio, T.
CBMM Publication
Latest Updates
I spoke with Tomaso Poggio about future of AGI, AI’s impact on human society, the signs of sentience in AI, and AI predictions…
(PDF) by Tomaso Poggio, Gemini+ChatGPT. Intelligence and its Fundamental Principles.
