A Mirror Neuron System for Syntax Acquisition
We investigate the use of a connectionist model of a mirror neuron cortical network for a context free syntax acquisition task. A finite state representation of the context free grammar is learned by an implicit knowledge system (IKS) modelled by a connectionist network. A mirror neuron system (MNS) whose evolutionary pedigree suggests adaptation for goal-directed sequential processing is used to track embedded recursions in a learned finite state model of the grammar. The mirror system modifies the output of the IKS depending on the depth of embedding. Reciprocally the IKS updates the MNS as natural ‘goals’ occur within a sequence during sentence production. This solves the computationally hard problem of inferring contexts from sequential input.
Unable to display preview. Download preview PDF.
- 1.Christopher M. Bishop. Neural Networks for Pattern Recognition. Oxford University Press, Great Clarendon Street, Oxford, OX2 6DP, United Kigdom, 1995.Google Scholar
- 3.Matthew A. Howard, F. McGlone, K. D. Singh, and N. Roberts. Brain activations to passive observation of relevant actions-an fMRI study. In Maxim Stamenov and Vittorio Gallese, editors, Mirror Neurons and the Evolution of Brain and Language, Advances In Consciousness Research. John Benjamins, John Benjamins, Netherlands, 2001 (In Press).Google Scholar
- 8.Edmund T. Rolls and Alessandro Trêves. Neural Networks and Brain Function, chapter 6, pages 95–135. Oxford University Press, Great Clarendon Street, Oxford, 0X2 6DP, United Kigdom, 1998.Google Scholar
- 9.Stefan Wermter. Hybrid Connectionist Natural Language Processing. Chapman & Hall, 2-6 Boundary Row, London., 1995.Google Scholar
- 10.Steve Womble and Stefan Wermter. Mirror neurons and feedback learning. In Maxim Stamenov and Vittorio Gallese, editors, Mirror Neurons and the Evolution of Brain and Language, Advances In Consciousness Research. John Benjamins, Netherlands, 2001 (In Press).Google Scholar