Reading What Machines “Think” - Brain Informatics - page 159

Reading What Machines “ Think ”

Fabio Massimo Zanzotto and Danilo Croce

University of Rome Tor Vergata
00133 Roma, Italy
{ zanzotto,croce } @info.uniroma2.it

Abstract. In this paper, we want to farther advance the parallelism be-
tween models of the brain and computing machines. We want to apply the
same idea underlying neuroimaging techniques to electronic computers.
Applying this parallelism, we can address these two questions: (1) how
far we can go with neuroimaging in understanding human mind? ( foun-
dational perspective ); (2) can we understand what computers “ think ”?
( applicative perspective ). Our experiments demonstrate that it is possible
to believe that both questions have positive answers.

1 Introduction

Studies of machines and of living organisms are strongly related. Biological princi-
ples have been often used to design machines. In cybernetics [1], artificial adaptive
machines, i.e., machines that can control their states, have been studied with re-
spect to the adaptivity of natural living organisms. This relation between studies of
living organisms and design of machines is even more strict when we observe com-
puting machines. It is not hard to imagine that the Von Neumann architecture [2]
and the neural-based computing architecture originally introduced by Turing [3]
have been inspired by concepts coming from the study of the mind and of the brain.
Viceversa, also theories of computer architectures inspired some studies of the mind
and the brain. Cognitive Psychology [4] uses computing machines as a metaphor for
defining models of the human mind. Cognitive Science (see [5]) is even more radical
as computing machines are exploited to test and validate models of the mind.
The strict relation between models of the mind and computing machines is
fascinating. Staying in this tradition, in this paper, we want to farther advance
this parallelism.
Brains are often studied using neuroimaging techniques to discover areas re-
lated to particular cognitive processes. Neuroimaging techniques are also used
to induce activation patterns for high-level cognitive processes related to specific
semantic categories [6]. These activation patterns can be used to determine what
cognitive process a brain is performing.
Computing machines nowadays are extremely complex. These machines per-
form complex tasks. These tasks seem to be “ cognitive processes ” 1 . For example,
manipulating symbols can be seen as a “ cognitive process ”.
1 We often use terms in wrong contexts, e.g., “ cognitive processes ” as related to ma-
chines. Yet, we need to make use of human centered terms for describing machines.
Whenever we misuse a term, we will indicate it with different characters as we did
for “ think ” in the title and for “ cognitive processes ”.