Symbols in the Brain Workshop

Symbols in the Brain Workshop

March 31, 2019 | MIT - Cambridge, MA

CBMM and MIT Quest logos

Over the weekend of March 30th-31st, Prof. Tomaso Poggio, Director of the Center for Brains, Minds and Machines (CBMM), and Prof. Antonio Torralba, Director of MIT Quest for Intelligence, hosted an informal one-day workshop on the topic of Symbols in the Brain, at the Brain and Cognitive Sciences Complex, MIT. The workshop organizers invited a diverse group of researchers – including, computer scientist with expertise in simulating artificial neural networks; computer science researchers with background on the theory of recursive functions and circuits; and cortical physiologists recording intracellularly and extracellularly from mammalian cortex – to discuss the still elusive questions:

How do networks of neurons learn abstract concepts from the sensory world?

How do networks of neurons manipulate the associate symbols?

The workshop presentations and panel discussions were focused on two areas of research: 1.) physiology and anatomy studies on circuit motifs in primate cortex; and 2.) computational studies on networks that can be trained to implement specific symbolic computations.

Workshop organizers were motivated by the belief that - the question of the origins of symbols in the brain is a key step to enable a study of human intelligence.  Scientists will need to understand much more than symbolic computation per se - the goal of mathematical logic and computer science – they will also need to understand how evolution discovered symbols using networks of neurons.

The workshop speakers were welcomed to MIT with a special dinner and dinner talk by Prof. Christos Papadimitriou (Columbia U.) entitled “Logic(omix) and the birth of computation,” on Saturday, March 30, 2019. On Sunday morning, the workshop opened with a “Neuroscience: complex “pre-symbolic” computation” session and then progressed into the afternoon “Symbolic computations” session. The day consisted of research talks presented by Drs. Michale Fee (MIT), Christos Papadimitriou (Columbia U.), Mac Nickel (Facebook AI), Nancy Lynch (MIT), Shimon Ullman (Weizmann Inst., MIT),  Haim Sompolinsky (HUJI, Harvard U.), and Andrea Tacchetti (DeepMind). The workshop closed with a panel discussion lead by Prof. Poggio and which included panelists Profs. Ron Rivest (MIT), Josh Tenenbaum (MIT),  Gabriel Kreiman (Children’s’ Hospital Boston, Harvard Medical), Sam Gershman (Harvard U.), and Jim Dicarlo (MIT.)

Profs. Poggio and Torralba look forward to more intellectual conversations with colleagues on the origin of symbols in the brain. Both greatly anticipate Moon Shot projects, in the very near future, which will explore how humans develop the understanding of - and the ability to manipulate - symbols.

Agenda

8:45AM

Welcome
Tomaso Poggio

9:00AM-9:30AM

Session: Neuroscience: complex “pre-symbolic” computation

Symbols in the brain: Lessons from the songbird
Speaker: Michale Fee

9:30AM-12:00PM

Session: Symbolic computations

A calculus for brain computation
Speaker: Christos Papadimitriou, Columbia U.

Symbols, Associative Memory, and Relational Learning
Speaker: Max Nickel, Facebook

Algorithms in the Brain
Speaker: Nancy Lynch, MIT

12:00PM-1:15PM

Lunch
McGovern Seminar Room 46-3189

1:15PM-3:45PM

Session: Symbolic computations (continued)

On combining vision and cognition
Speaker: Shimon Ullman

Long-term memory of structured knowledge
Speaker: Haim Sompolinsky

Learning models of environments, and their inhabitants: an entities- and-relations approach
Speaker: Andrea Tacchetti

3:45PM-4:00PM

Break

4:00PM-6:00PM

Panel + General discussion + Wrap-Up
Members: Ron Rivest, Tomaso Poggio, Josh Tenenbaum, Gabriel Krei- man, Sam Gershman, Jim Dicarlo, ......