Understanding Natural Intelligence


Natural intelligence (NI) is the opposite of artificial intelligence: it is all the systems of control present in biology.  Normally when we think of NI we think about how animal or human brains function, but there is more to natural intelligence than neuroscience.  Nature also demonstrates non-neural control in plants and protozoa, as well as distributed intelligence in colony species like ants, hyenas and humans.  Our behaviour co-evolves with the rest of our bodies, and in response to our changing environment.  Understanding natural intelligence requires understanding all of these influences on behaviour and their interactions.

One of the best methods for understanding how NI systems work is to try to replicate their behaviour in simulation.  Just as learning to paint forces you to understand the details of what you are seeing, building a working model forces you to understand the intricacies of what the target intelligent system is doing.  For example:

An AI model of an organism is a very-well-specified hypothesis about how that organism thinks and behaves.  Like any hypothesis, we assess an AI model by testing its predictions against the performance of the real system and by evaluating the plausibility of its assumptions. The predictions of a model are its behaviour, which we simply record after running simulations.  Its assumptions are its components; for example, the computations it makes, the information it has access to, the things it perceives and remembers.  We can use standard statistical tests to see how close we come to modelling behaviour in order to argue the validity of our assumptions.

The Artificial models of natural Intelligence (AmonI) group at Bath is dedicated to understanding natural intelligence.  Building AI models of NI systems requires designing intelligent systems.



These are selected publications generally contributing to understanding NI.  For full references and a complete list of publications, see my publications page. For focused lists concerning task learning, the evolution of primate social structure and the evolution of culture, see my understanding primate intelligence web page.

  • Agent-based models as scientific methodology: A case study analysing primate social behaviour, with Yasushi Ando and Hagen Lehmann.  In Philosophical Transactions of the Royal Society - B, Biology 362(1485):1685-1698, September 2007.  This paper talks about how ABM fits in as a part of scientific methodology, and in particular analyses macaque social structure in the DomWorld model of Charlotte Hemelrijk. Penultimate version in case you don't have access to PTRS-B.  The case analysed in this paper concerns Hemelrijk's DomWorld, that link includes the associated software.
  • Tony J. Prescott,  Joanna J. Bryson and Anil K. Seth, Introduction. Modelling Natural Action Selection in Philosophical Transactions of the Royal Society, B -- Biology.  This is actually quite a substantial article which covers the concept of action selection.
  • Primate errors in transitive `inference': A two-tier learning model, with Jonathan C. S. Leong.  In Animal Cognition 10(1):1-15, January 2007.   A model of transitive inference as the implicit learning of relationships between context-action pairs.  Associated software.
  • Embodiment vs. Memetics.  Discusses the importance of the discovery that human-like semantics can be learned simply from observing large corpora, with ramifications for the evolution of language.  In Mind & Society 7(1), June 2008 (online now).  Draft from August 2007 in case you don't subscribe to M & S.
  • Representational Requirements for Evolving Cultural Evolution, a 2007 target article in the web magazine interdisciplines' on-line conference, Adaptation and Representation.
  • Ivana Cace and I, Agent Based Modelling of Communication Costs: Why Information Can Be Free, in Emergence of Communication and Language on Springer, edited by Caroline Lyon, Chrystopher L. Nehaniv and Angelo Cangelosi. Shows that the tendency to communicate information can be adaptive even though it has immediate costs to the communicators and there are free riders / information hoarders around the place. This is a draft version from early March 2006.
  • Modular Representations of Cognitive Phenomena in AI, Psychology and Neuroscience (in HTML or PDF) in Visions of Mind, Darryl Davis ed. (2004)
  • ACT-R is almost a Model of Primate Task Learning:  Experiments in Modelling Transitive Inference, coauthored with Mark Wood and Jonathan Leong, from Cognitive Science 2004.
  • Language Isn't Quite That Special (HTML). Commentary on The cognitive functions of language by  Peter Carruthers, both in Behavioral & Brain Sciences (BBS) December 2002.
  • What Monkeys See and Don't Do: Agent Models of Safe Learning in Primates (in pdf), with Marc D. Hauser. A position paper. In the proceedings of the AAAI Spring Symposium on Safe Learning Agents.(2002).
  • Intelligent Control Requires More Structure than the Theory of Event Coding Provides (HTML). Commentary on The Theory of Event Coding: A Framework for Perception and Action Planning by Bernhard Hommel, Jochen Müsseler, Gisa Aschersleben and Wolfgang Prinz, both appeared in Behavioral & Brain Sciences (BBS).  (2001)
  • Modularity and Specialized Learning in the Organization of Behavior in pdf (or postscript). Written with Lynn Andrea Stein. From NCPW6, the final version is © Springer-Verlag.
  • The Study of Sequential and Hierarchical Organisation of Behaviour via Artificial Mechanisms of Action Selection.  MPhil Dissertation: University of Edinburgh, Faculty of Social Sciences (Department of Psychology) 2000
  • Cognition without Representational Rediscription coauthored with Will Lowe. This is a commentary on Dana H. Ballard, Mary M. Hayhoe, Polly K. Pook, and Rajesh P. N. Rao, Deictic Codes for the Embodiment of Cognition; both articles appeared in Behavioral & Brain Sciences (BBS) (1997)

  • Joanna  Bryson
    Last updated 29 February, 2008