Philosophical Review

Intention, Plans, and Practical Reason by Michael E. Bratman

Review by: J. David Velleman
The Philosophical Review, Vol. 100, No. 2 (Apr., 1991), pp. 277-284
Published by: Duke University Press on behalf of Philosophical Review
Stable URL: http://www.jstor.org/stable/2185304 .
Accessed: 24/06/2014 22:26

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Duke University Press and Philosophical Review are collaborating with JSTOR to digitize, preserve and extend
access to The Philosophical Review.

http://www.jstor.org

This content downloaded from 195.78.109.162 on Tue, 24 Jun 2014 22:26:19 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS

The PhilosophicalReview, Vol. C, No. 2 (April 1991)

INTENTION, PLANS, AND PRACTICAL REASON. By MICHAELE.

BRATMAN.Cambridge, Mass., Harvard University Press, 1987. Pp. xii,
200.

This important contribution to the philosophy of action is the culmina-

tion-though not in any sense a compilation-of Michael Bratman's nu-
merous studies on the nature and function of intentional With the excep-
tion of one chapter, the book is composed of new material, which not only
synthesizes Bratman's previous work but also ventures well beyond it in
offering a unified, positive theory of intention, defended with an un-
common degree of clarity and honesty.

METHODOLOGICALCLAIMS

Some of Bratman's most significant contributions to the study of inten-

tion are methodological. The broadest of Bratman's methodological
theses is that intention should be characterized in terms of its function in
rational action. The way to study intention, Bratman claims, is not to sys-
tematize our linguistic intuitions about words like "intend" or "inten-
tional," nor to puzzle over hard cases like weakness of will, but rather to
articulate the norms and regularities in which intention is implicated.
Holding an intention makes one subject to various norms about how to
reason and how to act; and insofar as one is rational, those norms will be
reflected as regularities in one's deliberations and actions. Bratman sees
his task as that of articulating these characteristic patterns, ideal and ac-
tual, in the reasoning and behavior of intenders.2

'Foremost among these are: "Intentionand Means-EndReasoning,"ThePhilo-

sophicalReview90 (1981), pp. 252-265; "Castafieda'sTheory of Thought and Ac-
tion," in Agent, Language, and the Structureof the World:EssaysPresentedto Hector-Neri
Castafieda,withHis Replies,ed. James E. Tomberlin (Indianapolis,Ind.: Hackett,
1983); "TakingPlans Seriously,"SocialTheoryand Practice9 (1983), pp. 271-287;
"Two Faces of Intention," The Philosophical Review 93 (1984), pp. 375-405; and
"Davidson's Theory of Intention," in Actions and Events: Perspectiveson the Philosophy
of Donald Davidson, ed. Ernest LePore and Brian McLaughlin (Oxford, England:
Basil Blackwell,1985), pp. 14-28.
2Manyfunctionalistswould not regard the functional role of a mental state as
including norms. Bratman invokes the name of functionalismwithout clearly ar-
ticulatingwhat he takes it to denote.

277

This content downloaded from 195.78.109.162 on Tue, 24 Jun 2014 22:26:19 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS

This functionalist approach enables Bratman to cut through much of

the fog generated by the methods of linguistic intuitions and hard cases. It
enables him, for example, to give an especially sensible and sensitive ac-
count of the distinction between intending and merely expecting the side-
effects of one's actions. (This account, which occupies Chapter 10, should
be of interest to a wider audience than some other parts of the book.)
A further methodological claim of Bratman's is that the functional role
of intention is more readily apparent in plans for the future than in imme-
diate or present-directed intentions. Bratman therefore devotes most of
the book to a theory of planning. Here Bratman's methodology meets
with mixed success. On the one hand, his theory of planning is a valuable
contribution in its own right to our understanding of practical reasoning.
On the other hand, Bratman cannot completely fulfill his promise to
apply the lessons of future-directed plans to the case of present-directed
intentions. Many of the norms and regularities that define the role of
future-directed plans, in Bratman's theory, are designed to facilitate coor-
dination among temporally disparate episodes of deliberation and action,
so that the agent can solve a practical problem in advance and in stages
rather than at the moment when action is required. Since these norms and
regularities have no obvious application to immediate intentions, the ques-
tion arises whether future-directed plans and present-directed intentions
share enough of a common role to be treated alike by a functionalist philo-
sophy of mind.

THE FUNCTIONALIST MODEL

According to Bratman, the main role of intention is to commit an agent

to action, and this role is best understood as having two aspects-
reasoning-centered commitment and volitional commitment. The norms
and regularities constituting reasoning-centered commitment include:
first, a requirement (and corresponding tendency) not to reconsider or
revise an intention, except in light of new and relevant information;
second, a requirement (and tendency) to engage in practical reasoning, as
necessary, about the means or methods by which the intention is to be
executed; and third, a requirement (and tendency) to exclude from one's
future deliberations any options believed to be incompatible with exe-
cuting the intention.3 These regularities and norms enable an agent's
present deliberation to lay a stable groundwork on which future delibera-
tions can build, so that he can solve his practical problems in stages, as
time and intellectual resources permit.

3These regularitiesare summarizedon pp. 108-109.

278

This content downloaded from 195.78.109.162 on Tue, 24 Jun 2014 22:26:19 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS

The norms and regularities that constitute volitional commitment are

less clear. Bratman says that intention embodies a volitional commitment
in the sense that it controls action rather than merely influences it. Some-
times this function of controlling action is described in causal terms: "If I
intend to A now, my intention will normally lead me at least to try to A" (p.
108). At other times, however, this function is described in terms that
sound distinctly reasoning-centered. Thus, Bratman argues that the dif-
ference between the control exerted by an intention and the mere influ-
ence exerted by a desire is that "my desire . . . still needs to be weighed
against conflicting desires," whereas having formed the intention, "I do
not normally need yet again to tote up the pros and cons" (p. 16). At this
point Bratman sounds as if he's merely applying the norms of reconsid-
eration that belong to the reasoning-centered aspect of intention's func-
tional role.
Indeed, even the causal formulation of volitional commitment sounds as
if it might collapse into one of the reasoning-centered regularities. Desires
merely influence behavior, Bratman suggests, because they do no more
than contribute to a sum of competing forces, whereas intentions (in the
normal case) do not have to contend with opposition from conflicting in-
tentions and are therefore in control. But the reason for this difference is
surely that intentions, unlike desires, are governed by a norm and a regu-
larity that enforce strong consistency among them. Intentions usually face
no opposition, and are thus in control, because potential opposition is
usually eliminated by the requirement that each intended action be com-
patible with all other intended actions, in light of the agent's beliefs. I
therefore wonder whether Bratman's volitional commitment is any more
than the tendency of intentions to exert a motivational influence, together
with the agent's tendency to ensure that the influence of each intention
will be unopposed.

OTHER CONCLUSIONS

While developing this functionalist model of intention, Bratman offers

subtle and illuminating answers to many of the questions that have inter-
ested students of the subject. Bratman's conclusions include the following:

* Intention is "a distinctive attitude, not to be conflated with or reduced to

ordinary desires and beliefs" (p. 10).

* This distinctive attitude is involved both in intending to do something

and in doing something intentionally.

* The intention involved in doing something intentionally needn't be the

intention to do that thing. Under suitable conditions, one can do some-

279

This content downloaded from 195.78.109.162 on Tue, 24 Jun 2014 22:26:19 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS

thing intentionally by doing it in execution of an intention to do some-

thing else.

* Acting with the intention of doing something may involve the distinctive
attitude of intending to do that thing, but it may involve no more than
having the purpose or goal of doing the thing-which is not the same as
an intention.

* Intending to do something does not entail believing that one will do it,
although intending to do something while believing that one won'tis irra-
tional.

* An agent's rationality in performing an intentional action depends on

whether he obeyed norms of rationality in forming, revising, and re-
taining the intention on which he acts.

I cannot do justice here to Bratman's arguments for these conclusions. I

shall comment on them selectively, in a way that I hope will reveal some of
the substance and subtlety of Bratman's theory.

Bratman's Theoryof Agent Rationality

Bratman devotes two chapters to the project of developing a theory of

"agent rationality." In Bratman's eyes, an agent's rationality in performing
a particular action is not simply a function of the reasons available to the
agent at the time of acting. The action may have been dictated by a long-
standing plan; that plan may have been formed at a time when a different
set of reasons was available to the agent; and such a plan, once formed,
would have acquired a rational inertia, by virtue of the norm discouraging
undue reconsideration or revision of one's decisions. An agent's ratio-
nality in performing an action must therefore depend, according to
Bratman, on the agent's rationality in holding the immediate intention on
which he acts, which in turn must depend on whether he obeyed the
norms of rationality in forming, reconsidering (or not reconsidering), and
revising (or not revising) the intention.
Bratman's discussion of the norms governing the reconsideration and
revision of intentions is one of the best and most original parts of the
book. What I cannot understand, however, is Bratman's desire for further
principles specifying how the agent's record of obedience to such norms
determines his rationality in performing the resulting action. These fur-
ther principles do not provide any additional guidance about how to
reason or act; they merely specify which errors of practical reasoning
leave a taint of irrationality on the resulting action.

280

This content downloaded from 195.78.109.162 on Tue, 24 Jun 2014 22:26:19 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS

Thus, for example, Bratman argues that if an agent initially forms an

intention for bad reasons but subsequently reconsiders the intention and
reaffirms it for good reasons, he will be rational in performing the re-
sulting action, even if he was irrational to reconsider the intention (p. 95).
Yet this judgment of the agent's rationality doesn't modify any of the spe-
cific recommendations about how to deliberate; in particular, it doesn't
entail that the agent was correct to reconsider his intention, after all. Since
reconsidering his intention was irrational, he shouldn't have reconsidered
it. The theory of agent rationality merely says that if, having irrationally
reconsidered his intention, the agent then reaffirms it for good reasons,
he cannot be charged with irrationality in performing the intended action.
The question is why we should take such an interest in which errors are
ultimately charged to the agent's account.
Bratman says, "The theory of agent rationality is a part of the theory of
education," because the theory will guide "the development of basic habits
of thought and action, in our children and in ourselves" (p. 51). Yet all
that we need to guide us in developing habits of thought and action are
the specific norms about how to form, reconsider, and revise intentions.
We do not need a means of reaching an overall judgment of an agent's
rationality by keeping score of his successes and failures in obeying the
relevant norms throughout the history of his intention. I therefore
wonder what use we have for a theory of agent rationality.

The Relations among Intention, Belief, and Purpose

Bratman claims that having an intention and having a purpose or goal

are two distinct states of mind. His functionalist methodology requires
that he support this claim by citing some norms and regularities that
govern one of these states but not the other. Bratman cites strong consis-
tency and agglomerativity. Strong consistency is the requirement (and
corresponding tendency) for an agent to ensure that his intentions are
jointly satisfiable, by ruling out options that he believes to be incompatible
with what he already intends. Agglomerativity is a "rational pressure" (and
tendency) to combine an intention to A and an intention to B into an in-
tention to A and B (p. 134). Bratman claims that purposes or goals are not
subject to these two norms. He bases his claim on cases in which an agent
rationally pursues two goals known to be incompatible, in the hope of
attaining one of them. In such cases, Bratman argues, the agent's pur-
poses do not obey the norm of agglomeration, since he is in fact required
and inclined to avoid aiming at the joint attainment of his goals, on ac-
count of their known incompatibility; and his purposes do not obey the
norm of consistency, since his knowledge that the goals are incompatible

281

This content downloaded from 195.78.109.162 on Tue, 24 Jun 2014 22:26:19 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS

does not forbid or inhibit him from aiming at them simultaneously. The
agent aims at each goal in the hope of enhancing his chances of success by
"letting the world decide" which one he attains (p. 138).
Hugh McCann has argued that such cases cannot prove that having a
goal is an attitude distinct from intention.4 The cases may be no more than
allowable exceptions to the norms by which intention is characterized in
Bratman's theory. That is, rationality may require consistency and ag-
glomeration only when they are necessary, and they obviously aren't nec-
essary when the agent would do best to "let the world decide" which of two
goals he attains. Hence the agent's incompatible purposes may still satisfy
the applicable norms of consistency and agglomeration, thereby fulfilling
the functional role of intention, as Bratman defines it.
I agree with McCann that Bratman doesn't adequately support his dis-
tinction between intention and purpose, but I do not join McCann in re-
garding the distinction itself as mistaken. Rather, I suspect that the funda-
mental difference between intention and purpose is one that's obscured
for Bratman by his view on the relation between intention and belief. For I
suspect that the difference is that intending to do something, unlike
aiming to do it, entails believing that one will, and Bratman denies that
entailment.5
My reason for thinking that intention entails belief is that it is an atti-
tude from which we view an outcome as settled. Intention is the attitude
suited to close (or foreclose) deliberation, and deliberation is a process of
settling issues that are open from our perspective, in the sense that they
are up to us. If an issue isn't up to us, then there is no point in our deli-
berating about it; if it is up to us, then we are in a position to deliberate
about it, in order to reach a conclusion from which we view the issue as
settled, not only normatively but factually. This conclusion is simulta-
neously a practical and a theoretical attitude: it represents not only how
we would resolve the issue but also how we shall resolve it. It therefore
entails expecting or believing in the intended resolution.
Having a goal differs from this state precisely in lacking the aspect of
belief. Our goals are normally outcomes whose attainment is not suffi-
ciently up to us and hence not for us to settle by deliberation. We are
therefore not in a position to deliberate about whether to attain these out-
comes but only about whether to aim at their attainment. And the content

41n "On Behalf of the Simple View," presented at the Central Division of the
AmericanPhilosophicalAssociation,April 29, 1989.
5Myviews on the relationof intention to belief are laid out at greater length in
Chapter Four of PracticalReflection(Princeton, N.J.: Princeton UniversityPress,
1989).

282

This content downloaded from 195.78.109.162 on Tue, 24 Jun 2014 22:26:19 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS

of the resulting attitude is that we would attain them but not that we neces-
sarily shall.'
If this brief outline of the difference between intention and purpose has
any truth, then Bratman's attempt to distinguish between the two was
bound to fail, because of his view that intention does not entail belief-a
view that rests on cases that he admits to be less than decisive (p. 38). More
is at stake, however, than the interpretation of particular cases. For if in-
tention differs from purpose in being the attitude by which we settle, and
represent as settled, issues that are up to us, then one cannot properly
understand that difference until one understands the sense in which
issues are up to us and the sense in which we have the capacity to settle
them. In short, an understanding of intention requires an understanding
of our freedom or autonomy. And I think that Bratman's account of in-
tention falls short in some respects because he tries to study intention in
isolation from such questions about the fundamental nature of agency.
Bratman avoids these metaphysical-sounding questions by confining his
conception of agency to that embodied in the desire-based theory of
rationality. In Bratman's eyes, all that an account of intention ultimately
needs to explain is how intention enhances our capacity to get what we
want.7 Bratman's theory satisfies this requirement by showing how a state
committing us to action, in both the volitional and reasoning-centered
senses, helps us to overcome our limitations as pursuers of desired ends,
by enabling us to deliberate and act in coordinated stages.
I have no objection to the thesis that intentions, especially future-
directed intentions, improve our effectiveness in this way. What troubles
me is the sugggestion that the study of intention can confine itself to this
aspect of intention. I think that we need to explain not only how intention
contributes toward our being more effective pursuers of ends, but also
how it contributes toward our being autonomous persons. And I don't
think that an autonomous person is just a goal-pursuing creature whose
limitations have been offset in the ways that Bratman describes. If a robot
were programmed to coordinate temporally disparate bouts of computa-

6These two paragraphsare subjectto severalqualifications.The most important

of these is that I do not claim to be accountingfor our intuitionsabout how the
words "aim"and "intend"are applied in ordinarydiscourse.In my view, our ordi-
nary use of the word "intend"is ambiguousbetween the two attitudesthat I have
described. This ambiguity is discussed by Gilbert Harman in "Willingand In-
tending," in Philosophical Grounds of Rationality: Intentions, Categories, Ends, ed.
Richard E. Grandy and Richard Warner (Oxford, England: Oxford University
Press, 1986), pp. 363-380. See also Harman's Change in View:Principlesof Reasoning
(Cambridge,Mass.:The MIT Press, 1986), ChapterEight.
70r perhaps what we rationallywant. See pp. 21-22, 52.

283

This content downloaded from 195.78.109.162 on Tue, 24 Jun 2014 22:26:19 PM

All use subject to JSTOR Terms and Conditions
 BOOK REVIEWS

tion and behavior, it might then overcome some of its computational and
mechanical limitations, but it would not thereby become an autonomous
agent.
I do not mean to imply that autonomous agency, and the role that in-
tentions play in it, must be understood in other than consequentialist or
instrumental terms. No doubt, our being autonomous agents is, from the
evolutionary perspective, a solution to the problem of our survival. But it
is a more indirect solution than that of fitting out goal-seeking creatures
with the capacity to deliberate and act in stages. There is something more
to being a self-governing person, something more than being systematic
about pursuing goals, and a theory of intention ought to help us under-
stand what it is.

CONCLUSION

Despite these and other disagreements with Bratman, I think that he is

generally correct about the functional role of intention and about the
need to acknowledge that role in any theory of practical reasoning or ra-
tional action. In these respects, his book significantly advances our under-
standing of agency, and it does so in a spirit and style that make it a
pleasure to read.

J. DAVID VELLEMAN
Universityof Michigan

8Thanksto Paul Boghossianand Dan Farrellfor commentson an earlierdraft of

this review.

The PhilosophicalReview, Vol. C, No. 2 (April 1991)

THE RATIONALITY OF EMOTION. By RONALD DE SOUSA. Cambridge,

Mass., The MIT Press, 1987. Pp. xix, 373.

Emotions color the objects of perception and thought: the metaphor is

both hard to resist and hard to unpack. What does the added emotional
coloration consist in? And is it a representation of something that is, bar-
ring specific perceptual errors, "really there," as Meinong claimed?' Or is
it, as Hume thought, "a new creation?"2 Ronald de Sousa develops a thesis

'My source on Meinong is J. N. Findlay, "EmotionalPresentation,"Australasian

Journal of Philosophyand Psychology13 (1935), pp. 1 1- 121.
2InquiryConcerningthe Principles of Morals, Appendix i.

284

This content downloaded from 195.78.109.162 on Tue, 24 Jun 2014 22:26:19 PM

All use subject to JSTOR Terms and Conditions