Vector Symbolic Architectures answer Jackendoff’s challenges for cognitive neuroscience

Source: arXiv Authors: Ross W. Gayler Date: 2004-12-13 Primary category: cs.NE All categories: cs.NE, cs.AI

Abstract

Jackendoff (2002) posed four challenges that linguistic combinatoriality and rules of language present to theories of brain function. The essence of these problems is the question of how to neurally instantiate the rapid construction and transformation of the compositional structures that are typically taken to be the domain of symbolic processing. He contended that typical connectionist approaches fail to meet these challenges and that the dialogue between linguistic theory and cognitive neuroscience will be relatively unproductive until the importance of these problems is widely recognised and the challenges answered by some technical innovation in connectionist modelling. This paper claims that a little-known family of connectionist models (Vector Symbolic Architectures) are able to meet Jackendoff’s challenges.

Why it matters here

Classic VSA paper. It is a good bridge from symbolic compositionality problems to high-dimensional vector operations such as binding, bundling, and permutation.