Titile：Between Brains and Machines

Speaker：Tomaso Poggio，Eugene McDermott Professor，MIT

Time： June 10,15:00-17:00 p.m.

Venue：Youcai Deng Lecture Hall (#101), Jin-Guang Life Science Building

Host：Dr. Yixin Zhu, Prof. Si Wu, Prof. Huan Luo

Abstract

Will science or engineering win the race for AI? Do we need to understand the brain in order to build intelligent machines, or not? Are there theoretical principles underlying those architectures, including the human brain, that perform so well in learning tasks? Though we do not have a full theory of deep learning, there are good reasons to believe in the existence of some fundamental principles of learning and intelligence. In the race for intelligence, understanding such principles and applying them to brains and machines is a compelling and urgent need for our society. Complementary to all of this is the burning need to find a connection between artificial and natural intelligence at the more basic mechanistic level of the biophysics of neurons. How is synaptic plasticity used in circuits of neurons and synapses that may be the brain counterpart of Stochastic Gradient Descent? If we could establish such a link, the synergies between neuroscience and AI would be so much more grounded and powerful. I claim that brain sciences are still missing the neural equivalent of backpropagation: a common neural circuit motif, present in several brain areas and powerful enough to allow for different forms of supervised learning. 

In this talk will jump from machine learning to brains and back. I will first discuss some of the first neural networks proposed as models of associative memory. I will then review a synaptic motif for supervised learning in cortex circuit and how to test the idea experimentally. Next I will speak about a specific agentic AI system that we are developing. I will conclude with some thoughts about connectomics as a tool to understand brains — focusing on the connectome of Drosophila.

Bio

Tomaso Poggio，Eugene McDermott Professor 

Core founding scientific advisor, MIT Quest for Intelligence 

McGovern Institute, CSAIL, Brain Sciences Department 

ex-Co-Director Center for Brains, Minds and Machines (CBMM) 

M.I.T.

2026-06-08