1. Language and the sensorimotor system

My current research focusses on the interface between language and the sensorimotor system. We can use language to talk about what we see and do in the world: I am interested in the neural mechanisms that allow this to happen. Much of my work explores embodied models of language, whose central claim is that language is closely related to the sensorimotor system. I propose a particularly strong embodied model of language. My proposal is that the syntactic structure of a sentence reporting a concrete episode in the world can be interpreted as a direct description of the sensorimotor processes through which this episode was experienced. I introduced this hypothesis in a recent book, and in several recent publications.

• Sensorimotor Cognition and Natural Language Syntax (MIT Press, 2012)
• Full list of publications on language and sensorimotor cognition

I am currently working on a Marsden project to test the hypothesis, jointly with Lubica Benuskova.

2. Neural network models

Most of my work involves developing neural network models of sensory and motor neural mechanisms, and of language processing mechanisms in the brain. Here are some papers presenting models of language processing and language development:

• M Takac, L Benuskova and A Knott. Mapping sensorimotor sequences to word sequences: A connectionist model of language acquisition and sentence generation. Cognition 125:288-308 (2012).
• G Caza and A Knott. Pragmatic bootstrapping: A neural network model of vocabulary acquisition. Language Learning and Development 8(2):113-135 (2012).

Models of visual attention and object categorisation:

• H Walles, A Robins and A Knott: A neural network model of visual attention and group classification, and its performance in a visual search task. Proceedings of the 26th Australasian Joint Conference on Artificial Intelligence (2013).
• H Walles, A Knott and A Robins: A model of cardinality blindness in inferotemporal cortex. Biological Cybernetics 98(5):427-437 (2008).

Models of motor control:

• J. Lee-Hand and A. Knott: Learning and representation of causative motor actions: a neural network model and its use in an embodied theory of language. Proceedings of the 35th Annual Meeting of the Cognitive Science Society (CogSci) 2013.
• J. Lee-Hand, T. Neumegen and A. Knott: Representing reach-to-grasp trajectories using perturbed goal motor states. Proceedings of the Pacific Rim Conference on Artificial Intelligence (PRICAI) 2012 pp250-261.

Models of episode representations in working memory:

• M. Takac and A. Knott: A neural network model of working memory for episodes. Proceedings of the 35th Annual Meeting of the Cognitive Science Society (CogSci) 2013.

3. AI and society

I'm interested in the effects that Artificial Intelligence technologies will have on society. I have recently helped to organise a cross-disciplinary working group to discuss these questions: the Artificial Intelligence and Society Discussion Group..

4. Computational linguistics

By training I am a computational linguist. I have worked on computational models of discourse and dialogue structure, and of natural language generation. My most recent work in this area has been in human-computer dialogue systems: together with students and colleagues, I developed a system to teach certain aspects of conversational Mäori. This system is described in the following papers:

• P Vlugter, A Knott, J McDonald, C Hall: Dialogue-based CALL: a case study on teaching pronouns. Computer-Assisted Language Learning 22(2):115-131 (2009).
• A Knott and P Vlugter: Multi-agent human-machine dialogue: issues in dialogue management and referring expression semantics. Artificial Intelligence 172:69-102 (2008).

For more information about my computational linguistics research:

• Overview of my computational linguistics research projects
• Full list of publications on computational linguistics