Dr Matthew Egbert

Senior Lecturer, School of Computer Science, University of Auckland
Director of Graduate Studies, School of Computer Science, University of Auckland
Principal Investigator, Centre for Computational Evolution, University of Auckland
Chair of Education and Outreach, Board of Directors, International Society for Artificial Life
Associate Editor of Adaptive Behavior

email: mde@matthewegbert.com
tel: +64 9 923 7027

Research Topics

Artificial Minds

Much of AI research focuses upon creating machines that can solve problems, but there is more to natural minds (human and animal minds) than just ability to solve problems. We pick the problems we want to solve, we experience the world through our senses, we have desires, experience pain, etc. What do classical approaches in AI (computationalist cognitivism, connectionism / machine learning) miss in their focus upon problem solving? How can we understand subjective elements of mind (such as those listed above) sufficiently well so as to create artificial instances of them?

My ultimate goal here is not to actually create artificial minds, but to improve our understanding of our own minds and what it means to be a thinking being that is capable of experience. My approach to these topics is inspired by sensorimotor contingency theory, enaction, cybernetics and "embodied" approaches to cognition which focus on the important role that bodies and environments play (in addition to brains!) in cognition. I build computational models to explore and evaluate new ideas or ways of thinking about mind.

(The Metabolic Organisation of) Artificial Life

Biological research often focuses upon evolutionary dynamics. Genetic mutations, selection pressures, etc. While evolution is a fascinating and important aspect of biology, there are other essential, universal and fascinating properties of biology that are also worthy of in-depth study. I am particularly interested in the 'metabolic organization' of life, whereby all living systems are perpetually falling apart and yet managing to persist thanks to processes of self-(re)construction.

Using computational models of highly simplified self-constructing systems I am exploring the advantages conferred by having this 'metabolic organization' instead of the passively stable organization that e.g., modern technology employs. One interesting advantage is that systems that are perpetually falling apart can 'sense' how well they are currently doing at persisting and then take action if the current trend is a bad one. In so doing, living systems can respond to their own emergent and open-ended set of existential needs, possibly allowing us to develop a scientific understanding of agency and purposefulness, i.e. how a system can have its own needs, and how can respond those needs.

Descriptions of some of my current projects can be found by following the research link above.