The Effect of Hamming Distances in a Computational Model of Selection by Consequences Open Access

Popa, Andrei (2009)

Permanent URL: https://etd.library.emory.edu/concern/etds/8k71nj12p?locale=en
Published

Abstract

Abstract
The Effects of Hamming Distances in a Computational Model of Selection by
Consequences
Andrei Popa
McDowell (2004) instantiated the Darwinian principles of selection,
recombination, and mutation in a computational model of selection by consequences. The
model forces a population of behaviors to evolve under the selection pressure of the
environment, by applying low-level Darwinian principles; it has been tested under a
variety of conditions and the quantitative outcomes are remarkably similar to those
obtained in experiments with live organisms (McDowell et al., 2008). The computational
model animates a virtual organism with a repertoire of 100 behaviors, represented by
binary strings; this raises the specific issue of Hamming distances, the number of digits in
a binary string that must be changed in order to obtain another bit string of equal length
(Hamming, 1950). McDowell (2008) hypothesized that in environments that reinforce
two alternatives the Hamming distance may be computationally equivalent to a
changeover delay (COD). In experiments with live organisms that reinforce two
alternatives, an interesting phenomenon is sometimes observed: instead of responding to
the alternatives, the organism behaves "as if" switching itself is reinforced. One way to
prevent this phenomenon is the use of a changeover delay, a procedure that prevents the
organism from acquiring reinforcement if it switches too often (Findley, 1958). The computational model places the target classes next to each other, and, traditionally,
they are separated by a large Hamming cliff, which makes it more difficult for a behavior
to switch from one target class to the other. In order to investigate the effects of smaller
cliffs between the target classes, they were positioned at different locations along the
continuum; in addition, other parameters were systematically varied. Results
confirmed McDowell et al.'s Hamming-Distance-As-Changeover-Delay hypothesis and
also revealed a robust rule about the effects of Hamming distances within the model. The
steady state outcome is, therefore, a product of the reiteration of Darwinian rules, and not
an artifact of conveniently choosing an exceptional location for the target classes. This
study constitutes another argument for the robustness of the computational model of
selection by consequences as a valid account of the behavioral dynamics.

Table of Contents



Table of Contents
I. Introduction………………………………………………………………………….3
II. General Method…………………………………………………………………….22
III. General Discussion…………………………………………………...…………….44
IV. References……………..……………….…………………………………………..49
V. Tables…………………………………………………………………………….....53
VI. Figure Captions……..………………………………………………………………58
VII. Figures……….……………………………………………………………………..67

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
School
Department
Degree
Submission
Language
  • English
Research field
Keyword
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files