Page 45 - Deep Learning

28 Introduction

we can also ask, as children do, what makes a car go? To answer this question,
we need to deconstruct the car into its parts (engine, wheels, etc.) and their
connections. This type of description does not answer questions about traffic
flow and questions about how the engine works are not answered by a descrip-
tion of the traffic flow. The answers are not mutually exclusive or competing
because they are answers to different questions.
Similarly, questions about how minds work and questions about how
groups work are equally valid and interesting, but answers to the former ques-
tion do not answer the latter or vice versa. The answers are not alternatives
because they refer to different levels of description. Researchers in the socio-
cultural approach have contributed interesting studies of the cognitive aspects
of groups and teams, and their emphasis on the fact that cognitive practices
are embodied in communities is useful. But the claim that we can replace
12
descriptions of mind with descriptions of social interactions and cultural sys-
tems is absurd. How does the mind work, such that a person can create and
participate in social and cultural systems?
In short, mind cannot be reduced to conscious experience, the brain, the
material environment or sociocultural factors. Scholars working within those
approaches have made, and continue to make, significant contributions to
cognitive psychology, but they err in adorning those contributions with the
unnecessary and unwarranted metaphysical claim that their research pro-
grams will answer all the questions of psychology. They escape rather than
tackle the central question: How does the mind work? The answer has to be
cast in terms of mental entities and processes. Nothing less will satisfy. In the
words of Zenon W. Pylyshyn: “There are many reasons for maintaining that
explanations of behavior must involve cognitive terms in a way that does not
serve merely as a heuristic or as something we do while waiting for the neuro-
physiological theories to progress.” The name for this stance, if it needs any
13
other name than common sense, is mentalism. One difference between men-
talism and other approaches is that mentalism runs the risk of being compre-
hensible to nonpsychologists; so be it. It also suggests a particular specification
of what a theory of human cognition should look like.

THE ARCHITECTURE OF COGNITION: THE BASICS
Mind is a system, but what kind of system? Scientists are familiar with
ecosystems, electrical systems, mechanical systems, weather systems and
many others. Different types of systems differ in what kinds of stuff they
are made of, and what kinds of processes and transformations that stuff