Vries, M. de
Author: Michael de Vries
Graduation Year: 2009
Department: Theoretical Philosophy
Thesis: On Cognition and Computability
Computationalism is a popular position in cognitive science. Some researchers defend good-
old fashioned AI or connectionism, while others argue that cognition has to be computed in
relation to its environment (embodied cognition). Despite these relevant differences,
computationalism states that cognition can be simulated by computational procedures. In
other words, humans are some sort of computers/machines. Although this position is
popular, it has its critics. The aim of this master thesis is to explain the relation between
computability and cognition, and discuss some arguments against computationalism.
There are arguments against computationalism based on thought-experiments in
philosophy, but the discussion of such arguments is beyond the scope of this thesis. The
arguments discussed in this thesis try to refute computationalism by using logic and facts
from the natural sciences. For instance, some argue (such as Lucas and Penrose) that
Gödel's incompleteness theorems in mathematical logic imply that the human mind has
properties no machines can have. Bringsjord and Zenzen are recently the most outspoken
critics of computationalism and some of their arguments are discussed in this thesis. For
example, they try to show that computation is reversible while cognition is not, therefore
cognition cannot be computation. Two others arguments of Bringsjord and Zenzen, "the
narrative argument" and "the argument from infinitary reasoning", are also discussed.
This thesis has two main conclusions. Firstly, I argue that discussed arguments
against computationalism fail. I counter every discussed argument with a list objections.
Secondly, I give some hints for further research in cognitive science and artificial intelligence.
Computationalism is the most plausible position in cognitive science and different forms of it
(good-old fashioned AI, connectionism, artificial life, behavior-based robotics and so on)
should work on computational models to simulate cognition artificially.
|Last modified:||01 November 2013 2.20 p.m.|