Author: Ben Johannes
Graduation Year: 2010
Graduation Year: 2010
Department: Theoretische filosofie
In the seventies and eighties of the 20th century, it was popular among cognitive scientists to conceptualize the mind as a computer. The basic idea was that brain is similar to hardware, and mind is similar to software. That software was thought to be of a symbolic-syntactic rule based nature. This classical conceptualization of mind as computer contains several metaphorical aspects, jointly called the ‘classical’ computer metaphor.
It is often thought that the computer metaphor is something from the past, and that nowadays it has been displaced by other viewpoints, like neural network models. I investigate whether this is indeed the case. Moreover, if so, whether thinking about the mind in computer concepts leads to misconceptions about the mind. Briefly, I touch upon the question whether a distinction can be made between a computer metaphor of mind and a computer model of mind.
I start by discussing two classical accounts from the 20th century, namely Aaron Sloman’s The Computer Revolution in Philosophy (1978), and John Searle’s ‘Mind, brains, and programs’ (1980). Then I concentrate on two 21st century theories, viz. Craik and Lockhart’s (2002) Levels of Processing theory of memory and Gallistel and King’s (2009) Computational Theory of mind.
The first version of Levels of Processing theory was conceived in the seventies of the 20th century and this original version is rooted in the classical computer metaphor. After it was falsified by empirical findings, Craik and Lockhart retained the central (computer) terms in which the theory was cast but changed their interpretations. In my thesis, I argue that, as a consequence, their Levels of Processing theory contains vacuous, superficial terms, circular reasoning, has severe empirical difficulties and falls prey to the so-called homunculus—the idea that there is a central, conscious observer in the brain.
The Computational Theory of Gallistel and King has no clear roots in the classical period. However, their theory implies that computers and brains must have similar computational mechanisms. Gallistel and King explicitly show how and why their theory of symbolic-syntactic, rule based mechanisms in the brain is better suited for implementing common functionality, like navigating, than are neural network models. In that sense, they revive the computer conceptualization of the mind, but in a modern, detailed and more model-like form. Nevertheless, I argue that they go too far in rejecting neural network models and that they still make some of the reasoning errors that plagued the classical computer metaphor.
I conclude that the computer metaphor is still around, albeit in a modified form.
|Laatst gewijzigd:||01 november 2013 14:03|