Technical Program

Paper Detail

Paper: PS-2B.72
Session: Poster Session 2B
Location: H Fl├Ąche 1.OG
Session Time: Sunday, September 15, 17:15 - 20:15
Presentation Time:Sunday, September 15, 17:15 - 20:15
Presentation: Poster
Publication: 2019 Conference on Cognitive Computational Neuroscience, 13-16 September 2019, Berlin, Germany
Paper Title: A Calculus for Brain Computation
Manuscript:  Click here to view manuscript
License: Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 Unported License.
DOI: https://doi.org/10.32470/CCN.2019.1381-0
Authors: Christos H. Papadimitriou, Columbia University, United States; Santosh S. Vempala, Georgia Tech, United States; Daniel Mitropolsky, Michael J. Collins, Columbia University, United States; Wolfgang Maass, Technische Universit├Ąt Graz, Austria; Larry F. Abbott, Columbia University, United States
Abstract: Do brains compute? How do brains learn? How are intelligence and language achieved in the human brain? In this pursuit, we develop a formal calculus and associated programming language for brain computation, based on the assembly hypothesis, first proposed by Hebb: the basic unit of memory and computation in the brain is an assembly, a sparse distribution over neurons. We show that assemblies can be realized efficiently and neuroplausibly by using random projection, inhibition, and plasticity. Repeated applications of this RP\&C primitive (random projection and cap) lead to (1) stable assembly creation through projection; (2) association and pattern completion; and finally (3) merge, where two assemblies form a higher-level assembly and, eventually, hierarchies. Further, these operations are composable, allowing the creation of stable computational circuits and structures. We argue that this functionality, in the presence of merge in particular, might underlie language and syntax in humans.