Provided by Cognitive Sciences ePrint Archive
A Brief Introduction to the Guidance Theory of Representation
Gregg Rosenberg ([email protected])
Artificial Intelligence Center, University of Georgia
Athens, GA 30602 USA
Michael L. Anderson ([email protected])
Institute for Advanced Computer Studies, University of Maryland
College Park, MD 20742 USA
Abstract
Recent trends in the philosophy of mind and cognitive
science can be fruitfully characterized as part of the
ongoing attempt to come to grips with the very idea of
homo sapiens—an intelligent, evolved, biological agent—
and its signature contribution is the emergence of a
philosophical anthropology which, contra Descartes and
his thinking thing, instead puts doing at the center of
human being. Applying this agency-oriented line of
thinking to the problem of representation, this paper
introduces the guidance theory, according to which the
content and intentionality of representations can be
accounted for in terms of the way they provide guidance
for action. We offer a brief account of the motivation for
the theory, and a formal characterization.
Introduction and Background
Recent trends in the philosophy of mind and cognitive
science can be fruitfully characterized as part of the
ongoing attempt to come to grips with the very idea of
homo sapiens—an intelligent, evolved, biological agent—
and its signature contribution is the emergence of a
philosophical anthropology which, contra Descartes and
his thinking thing, instead puts doing at the center of
human being. Work that falls under this broad umbrella
includes accounts of human cognition which stress
embodiment and environmental situatedness (Anderson,
2003; forthcoming-a; Ballard et al., 1997; Clancey, 1997;
Clark, 1995; Varela, Thompson, & Rosch, 1991),
pragmatic and evolutionary accounts of human
knowledge and culture (Barkow, Cosmides, & Tooby,
1992; Guignon, 1983; Hacking, 1983; Munz, 1993;
O’Donovan-Anderson, 1997; Rescher, 1990) and action-
oriented accounts of perception (Aloimonos, 1992;
Ballard, 1991; Gibson, 1966; Milner & Goodale, 1995;
O’Regan & Noe, 2001), to name only a few categories,
and a few of the many works in each. The current essay
introduces the results of our effort to build a theory of
representation on the basis of the same kind of agency-
oriented approach. It is only an introduction, and many
difficult issues will have to be treated briefly, or not at all.
The interested reader is encouraged to consult the fuller
treatment given in (Rosenberg & Anderson, forthcoming).
A representation is something that stands in for, is in
some sense about, something else. How is one thing ever
about another? To answer this question is usually to
analyze this relation of aboutness—the intentionality of a
representation—in terms of some other, presumably more
basic relation. For instance, a typical causal theory of
representation might hold that a given representation R is
about E just in case it has a certain specified set of causal
relations to E, for instance, that perceiving an instance of
E will cause one to represent with R (Fodor, 1981; 1987).
Likewise an information-content approach might hold that
a given representation is about that object from which the
information it contains in fact derived (Dretske, 1981;
1986; 1988). Conceptual role theories, on the other hand,
try to analyze meaning in terms of the role played by the
concept in inferential and other conceptual/cognitive
processes: roughly speaking, the representation R is about
E just in case it is used to make warranted inferences
about E (Harman, 1982; 1987). Naturally, there are also
theories that try to combine these two approaches,
producing the so-called “two-factor” accounts (Block,
1986; Loar, 1981; Lycan, 1984). There is no need, nor is
this the place, to rehearse the standard critiques of these
various theories (but see Anderson, forthcoming-b).
However, by way of situating and introducing our own
account of representational content, let us say that we find
the various causal approaches too input focused, meaning
they give too much importance to the ways in which the
environment affects the organism to endow its states with
representational meaning, and while the conceptual role
theories seem to us a step in the right direction in that they
draw attention to the importance of cognitive actions
taken by the subject with its representations, none of the
theories outlined above give sufficient weight to the full
range of what a subject does with its representations.
In contrast, we ask first not what a representation is, but
what it does for the representing agent, and what the agent
does with it; what is a representation for? Our contention
is essentially that representations are what representations
do, and that what a representation does is provide
guidance for action. Whatever the details of its
instantiation or structure, whatever its physical,
informational, or inferential features (and these are quite
various across different representing systems), what
makes a given item representational is its role in
providing guidance to the cognitive agent for taking
actions with respect to the represented object. In our
view, each of those other special features a given
representing token might possess—e.g. co-variance with,