The Resilience of Computationalism

Roughly speaking, computationalism is the view that cognition is computation.  Although some form of computationalism has been mainstream in psychology, neuroscience, and philosophy of mind for decades, many people remain sceptical of it.  Their reasons vary, but none of them amount to a refutation.

In the philosophical literature, the most prominent objections to computationalism are of the following form:  cognition involves feature P, but computation is insufficient for P; therefore, cognition is not computation.  Candidate P’s include mathematical insight, intentionality/understanding, consciousness, embodiment, and embeddedness.

The main problem with these objections is that even if they are correct that computation is insufficient for some of the P’s, computationalism can still be retained in a slightly weakened form:  cognition is computation plus X, where X is what is needed, in addition to computation, to account for P(s).  In fact, most computationalists already subscribe to this weaker version of computationalism.

But there is another series of objections, which have attracted far less attention by philosophers.  They have the following form:  cognitive processes are (implemented in) neural processes, neural processes have feature Q, and feature Q is incompatible with being (implementing) computations.  In my opinion, this is the only kind of consideration that can settle whether computationalism holds.

I have written a short paper, The Resilience of Computationalism,” in which I discuss objections to computationalism and why they don’t succeed as they stand.  Most of the paper is devoted to objections of the second kind, arguing that as they have been formulated to date, they are either confused or insufficiently precise to refute computationalism.  But I suggest that they can be improved upon by relying on a more precise account of computation.  I will present this paper at the 2008 PSA meeting in a couple of weeks.  If anyone cares to look at it, comments are extremely welcome, especially by November 5.

 

4 Comments

  1. Re: “Third, and most importantly, even if computation is insufficient for some cognitive phenomenon P, computationalism can be retained in its most plausible form.”

    It is possible that it is only computation “as we know it” that is insufficient: the brain’s methods of computation, with hundreds and even thousands of interconnections between neurons, is currently beyond our comprehension, but obviously is quite sufficient for all our observed cognitive phenomena.

    Adam Leonard (author of Man by Nature: The Hidden Programming Controlling Human Behavior)

  2. gualtiero

    Adam,

    Thanks for your comment (and thanks to the several people who send me comments privately). I agree, although I prefer a different way of putting the point. I think “computation” has a fairly definite meaning, related to computer science and computability theory. What the brain does may or may not be computation in this strict sense. You are using “computation” in a broader sense, which includes neural processes by definition.

Comments are closed.