openness to the causal influence of, or resemblance to its
object—correspond to one of the range of strategies that
our various representation-forming and representation-
consuming systems have evolved to solve the biologically
fundamental problem of providing autonomous organisms
with guidance for action.
On the guidance theory action is fundamentally
intentional: it is first and last a directed engagement with
the world. Our basic claim is that representations come
into existence and derive their content from their role
supporting the basic intentionality of action. The fact that
subjects take action with respect to things is what confers
content on representations; it is how representations reach
outside the organism and touch things in the world. The
guidance theory presumes, then, that the intentionality of
representation can be grounded in the intentionality of
action.
A Formal Account of the Guidance Theory
Let us say that a token provides guidance to a subject by
making its features available to the subject’s motor
systems and rational control processes for use in making
discriminating choices between possible actions or
possible ways of executing actions. Below we introduce
the foundations of the guidance theory in terms of a set of
propositions, which together characterize the most central
features of the theory.
(1) An entity is anything that can be represented: a
property, a concrete particular, an aspect of a thing, a state
of affairs, a number, etc.
(2) A subject is any representation-consuming
cognitive engine. To be a representation consumer, it
must be capable of interacting in the world in a rational,
goal directed way due at least partly to guidance it
receives from tokens within its cognitive systems.
(3) A circumstance is a circumstance of the subject. A
circumstance consists in the subject’s internal states,
including the subject’s bodily changes, registrations,
representations, expectations, priorities, values, options
for action, homeostatic self-evaluations, procedural
knowledge, motor schemas and also the subject’s
immediate environment.
(4) A subject standardly uses tokens (of a type) to
provide guidance with respect to an entity E in a given
(type of) circumstance C if, and only if, the subject has an
enduring conscious preference or conditioned reflex to
use the tokens (i.e., members of the type) to provide
guidance with respect to E when in circumstance C.
(5) An action can be a motor process or a cognitive
process. This yields two clauses in the definition of
action:
(5.1) In the case of a motor process, a motor process
is an action if, and only if, it is activated under
control of perceptual/cognitive feedback processes
capable of effectively modulating or bringing about
changes in the organism or in the world
(5.2) In the case of cognitive processes, a cognitive
process is an action if, and only if, it is a mental
process under intentional control whose results
contribute to circumstances (as defined above) used
to direct motor processes. A cognitive process is
under intentional control if the working of that
cognitive process is subject to modification by
processes of attention, short-term memory, valuation,
assent and dissent, practiced learning, and
consciously administered self-criticism and praise.
As mentioned already, the fact that subjects take action
with respect to things is what confers content on
representations; it is how representations reach outside the
organism and touch things in the world. The central
importance of the intentionality of action means that it is
vital to correctly understand—without regress—what it is
for an action to be taken with respect to something.
(6) An action is taken with respect to an entity E if, and
only if,
(6.1) The action is a motor program, E is the focus of
the intended change or efforts at control in the world;
or
(6.2) The action is a motor program and an
assumption of information about E is a motivating
reason that the given action, rather than some
alternative non-E involving action, was undertaken;
or
(6.3) The action is a cognitive process undertaken to
discover or confirm facts, to modify values, or to
decide between alternative actions, and an
assumption of information about E is necessary if the
process as a whole is to provide guidance for the
subject’s motor actions.
This definition uses three further terms—motivating
reason, focus, and assumption of information—that
present the potential for regress and require further
discussion.
Motivating Reason
For an account of motivating reason, we hold only that
any analysis must be such that it would be applicable to
goal-directed behavior of entities that do not have
representations at all. For example, it must be of a piece
with how we would identify the motivating reasons for
why a plant turns toward the sunlight. The plant’s
behavior is goal-directed behavior even if it is not action
in the sense defined above, and the motivating reason for
the behavior is to maximize the amount of sunlight
available for photosynthesis. Because the plant does not
have representations, a correct account of motivating
reason cannot appeal to representational content.
We also distinguish motivating reasons from
applications of causal force. A child may go to bed early
on Christmas Eve to encourage Santa Claus to bring
presents, and this may be the child’s motivating reason,
even though Santa Claus is not capable of applying causal
force on the child’s mind. A hungry wolf may look for
prey and its motivating reason may be a future state of
satiety, even if the cause of its behavior is a present
internal state. Any account of motivating reasons must
allow for motivating reasons that are non-representational
facts and entities, even for agents that possess
representations.