On-Line Processing of Grammatical Structure Using Reservoir Computing

Previous words in the sentence can influence the processing of the current word in the timescale of hundreds of milliseconds. The current research provides a possible explanation of how certain aspects of this on-line language processing can occur, based on the dynamics of recurrent cortical networks. We simulate prefrontal area BA47 as a recurrent network that receives on-line input of "grammatical" words during sentence processing, with plastic connections between cortex and striatum (homology with Reservoir Computing). The system is trained on sentence-meaning pairs, where meaning is coded as activation in the striatum corresponding to the roles that different "semantic words" play in the sentences. The model learns an extended set of grammatical constructions, and demonstrates the ability to generalize to novel constructions. This demonstrates that a RNN can decode grammatical structure from sentences in an on-line manner in order to generate a predictive representation of the meaning of the sentences.

[1]  Razvan Pascanu,et al.  A neurodynamical model for working memory , 2011, Neural Networks.

[2]  J. Elman.  Learning and development in neural networks: the importance of starting small , 1993, Cognition.

[3]  A. Friederici,et al.  Why the P600 is not just a P300: the role of the basal ganglia , 2003, Clinical Neurophysiology.

[4]  F. van der Velde,et al.  Neural blackboard architectures of combinatorial structures in cognition , 2006, Behavioral and Brain Sciences.

[5]  Peter Ford Dominey,et al.  A Neurolinguistic Model of Grammatical Construction Processing , 2006, Journal of Cognitive Neuroscience.

[6]  Peter Ford Dominey,et al.  Neural network processing of natural language: II. Towards a unified model of corticostriatal function in learning sentence comprehension and non-linguistic sequencing , 2009, Brain and Language.

[7]  Peter Ford Dominey,et al.  Cortico-striatal function in sentence comprehension: Insights from neurophysiology and modeling , 2009, Cortex.

[8]  Garrison W. Cottrell,et al.  2007 Special Issue: Learning grammatical structure with Echo State Networks , 2007 .

[9]  E Donchin,et al.  Syntactic parsing preferences and their on-line revisions: a spatio-temporal analysis of event-related brain potentials. , 2001, Brain research. Cognitive brain research.

[10]  Brian MacWhinney,et al.  Cross-linguistic research in aphasia: An overview , 1991, Brain and Language.

[11]  P. Hagoort.  The fractionation of spoken language understanding by measuring electrical and magnetic brain signals , 2008, Philosophical Transactions of the Royal Society B: Biological Sciences.

[12]  Herbert Jaeger,et al.  Reservoir computing approaches to recurrent neural network training , 2009, Comput. Sci. Rev..

[13]  A. Anwander,et al.  The brain differentiates human and non-human grammars: Functional localization and structural connectivity , 2006, Proceedings of the National Academy of Sciences of the United States of America.

[14]  Peter Ford Dominey,et al.  Structure Mapping And Semantic Integration in a Construction-Based Neurolinguistic Model of Sentence Processing , 2006, Cortex.

[15]  Peter Ford Dominey,et al.  Neurological basis of language and sequential cognition: Evidence from simulation, aphasia, and ERP studies , 2003, Brain and Language.