Convergence and Crossover: The Permutation Problem Revisited

Some work that I did with Emmet Spier early on during my PhD is now available as a Cognitive Science Research Paper from the Informatics Department at the University of Sussex. A copy of the paper can be downloaded here.

—

Convergence and Converge: The Permutation Problem Revisited

Tom Froese and Emmet Spier

Standard crossover operators are often omitted from simple genetic algorithms (GAs) used for optimizing artificial neural networks because of the traditional belief that they generally disrupt the distributed functionality of the evolving solutions. The notion that crossover will be especially disruptive when a genetic representation is used which has a many-to-one mapping between genotype and phenotype has become known as the ‘permutation problem’. In contrast, this paper argues that these problems do not normally appear in practical use of simple GAs because populations converge quickly and then continue to move through search space in this converged manner until a fitness optimum is found. After convergence all individuals are genetically similar, and moreover, distinct genetic permutations of the same phenotypic solution are unlikely to co-exist in the population. Genetic convergence thus minimizes the possibility for disruption caused by crossover. We have termed this the ‘convergence argument’. This claim is investigated experimentally on standard benchmark problems and the results provide empirical support.

Visit to SCAI Lab in Skoevde, Sweden

From the 10th-31st of May 2008 I will be in Sweden to work with Prof. Tom Ziemke at the Skoevde Cognition and Artificial Intelligence (SCAI) Lab on the topic of enactive AI.

My travel to Skoevde is covered thanks to an euCognition network action, and my stay here is kindly supported by the University of Skoevde.

Simulation of Adaptive Behavior 2008

The paper which Ezequiel and me submitted to the 10th International Conference on the Simulation of Adaptive Behavior (SAB’08) has been accepted as an oral presentation. The title and abstract of our contribution are as follows:

Stability of coordination requires mutuality of interaction in a model of embodied agents

Tom Froese & Ezequiel A. Di Paolo

Abstract. We used an evolutionary robotics methodology to generate pairs of simulated agents capable of reliably establishing and maintaining a coordination pattern under noisy conditions. The agents were only evolved for this ability and not for their capacity to detect social contingency. However, when they were made to interact with a previously recorded, successful behavioral sequence, the coordination pattern could not be maintained. An analysis of the system’s underlying dynamics revealed (i) that stability of the coordination pattern requires mutuality of interaction, and (ii) that the interaction process is not only constituted by but also constitutive of individual behavior. We suggest that such termination of interaction is a general property of a certain class of interactively coupled dynamical systems, and conclude that psychological explanations of an individual’s sensitivity to social contingency need to take into account the role of the interaction process.

Live stream from Test_Lab at V2_

My talk on the Enactive Torch which I gave at Test_Lab: Multimodal in V2_ is available online here:

Live stream

Test_Lab at V2_

I have been invited to give a talk on sensory substitution and our recent work on the Enactive Torch during the next Test_Lab on the 21st of February at V2_ in Rotterdam. I think that this will also be an excellent opportunity to introduce a broader audience to the framework of enactive cognitive science, as well as learn something about the latest developments in the arts!

Tucson 2008

The proposal me and my colleague Adam Spiers have submitted to this year’s Toward a Science of Consciousness conference in Tucson, Arizona has been accepted as a talk. The Enactive Torch is going to the States!

The title of our contribution is:

“The Enactive Torch: Promoting first-person phenomenology in the study of enactive perception”

In order to read the extended abstract of our talk please click here.

Alergic seminar

Modeling the dynamics of social cognition

Tom Froese
Alergic seminar
16 Jan. 2008

We used an evolutionary robotics methodology to generate simulated agents capable of reliably establishing and maintaining a coordination pattern under noisy conditions. The agents were only evolved for this ability and not for their capacity to detect social contingency. However, when they were made to interact with a previously recorded,successful behavioral sequence, the coordination pattern could not be maintained. An analysis of the underlying dynamics revealed (i) that stability of the coordination pattern requires mutuality of interaction, and (ii) that the interaction process is autonomous in the sense that it is not only constituted by but also constitutive of individual behavior. We hypothesize that in many cases an explanation of the breakdown of coordination does not require the postulation of an individual’s sensitivity to social contingency; it is likely a general property of a certain class of interactively coupled dynamical systems.

euCognition in Venice

I presented some recent work on a simulation model of social cognition at euCognition’s 4th Six-Monthly Meeting entitled “An enactive approach to social cognition: Detection of social contingency or stability of interaction dynamics?”. The talk won the best student presentation prize and I presented it again during the main event!

Slides for the talk can be downloaded from the euCognition website here.

Many thanks to everyone at euCognition for making this such a vibrant and memorable event!

Pump-prime research funding

I originally posted the comment below on the Enactive Torch website, but I thought I should mention it here as well:

Thanks are due to the Informatics Department at the University of Sussex for providing some initial funding to pump-prime research development with the Enactive Torch. The accepted funding proposal covers the design of a PCB, building 4 Enactive Torches, as well as expenses for conducting systematic experiments.

Cognitive Science Research Paper 593

A slightly updated, extended and better formatted version of the “Toward a Phenomenological Pragmatics of Enactive Perception” publication has been released as Cognitive Science Research Paper 593 by the University of Sussex.

This improved version of the paper can be downloaded from here.