InterJournal Complex Systems, 231
Status: Rejected
Manuscript Number: [231]
Submission Date: 981212
Does Evolution Make Sense? Finding Order in a Class of Evolved Continuous Time Recurrent Neural Network Central Pattern Generators
Author(s): John Gallagher

Subject(s): CX.66, CX.67, CX.31

Category: Brief Article

Abstract:

Is there any reason to believe that evolution "makes sense"? In other words, is there any reason to believe that an evolutionary process will produce products whose internal operation is understandable to the humans that study them? This is a very difficult question to answer because there are at least two factors that determine whether a particular evolved system is understandable. The first is that a particular system may or may not be understandable as a result of the quality of the analytical tools used in making the study. The second is that a successful evolved system is by no means required to be "understandable" if "understandability" is not somehow related to"survivability". Scientists might be in the habit of cutting with Occam's razor -- but evolution is certainly not so required. This paper presents some observations on what features of a specific artificial evolutionary process encouraged the formation of continuous time recurrent neural networks (CTRNNs) that are understandable using some tools and techniques of dynamical systems. Some speculations on how these observations might impact both engineers and working biologists are also discussed.

Retrieve Manuscript
Submit referee report/comment


Public Comments: