Jason Bridges

University of Chicago


Phil 31410, A03—lecture 2 notes

Agents and actions

(Davidson, “Agency”)


As I mentioned last time, “Agency” is a very difficult article.  Extremely dense—much ground covered succinctly, almost elliptically.  And it’s not clear that it all holds together.

If our emphasis in this course were somewhat more traditional, so far as courses in the philosophy of action go, almost all of the myriad arguments of this paper would be worth examining, and measured against competing arguments in the literature.  But given what will actually be our focus in this course, we need look in depth at only some of the central claims.


I.  The constitutive question of agency

The first issue is the question of agency, and correlatively of action.

We human beings take ourselves to be, at least some of the time, participants in the course of worldly events, not just bystanders or victims.  Sometimes things befall us: the wind knocks us over, we get downsized, we win the lottery.  But sometimes we set things in motion.  We intervene in the course of events.  We throw a brick that breaks a window, we set a fire that burns down a forest, we carry a child to safety.

We have, it seems to us, a power—a power to participate in the course of events, to take an active role in these events.  Call this power agency.  Call those beings that have this power, amongst whom we, perhaps optimistically, number ourselves, agents.  And call the events in which agents exercise their agency actions.


This is all very inchoately expressed.  And that, at least at first, is how it should be.  The concept of agency, the idea of a power to participate, is a very fundamental one.  It is one of the most basic building blocks of our “conceptual scheme”.  So we should not expect to be able to immediately encapsulate this idea in other terms.

But that does not foreclose the possibility that, under scrutiny, some further elaboration of the concept might reveal itself.  At the very least, we should hope as philosophers to be able to link the idea of agency, and so of action, to other concepts in an illuminating way.  And perhaps we might in the end even be able to analyze that concept completely in other terms.  We don’t know until we try.

This constitutive question is the one that Davidson announces at the beginning as the topic of his paper.  Ultimately, he ends up spending more time on matters only indirectly related to this question.  But he does say something about it directly, and that’s what I want to talk about first.


We might put the question Davidson addresses as follows:

The constitutive question of agency

What relationship must hold between an event and a person for that event to be an action of that person?

An action, again, is an event in which a person exercises her agency.  So the question is how a person must be related to an event in order for it to be the case that she exercises her agency in it.

We can easily give a non-informative answer to this question.  The relationship that must hold between an event and a person for the former to be the action of the latter is that the former be the action of the latter.  Equivalently, for the latter to exercise her agency in the former.

But obviously in asking this question we want more.  We want a philosophical account of agency, of what it is to exercise agency, of what exercising agency consists in.  That’s what I mean in calling this the constitutive question.  It asks for an account of what constitutes an exercise of agency.


We might begin by trying to link acting with doing.  What makes something an action is that it is something a person does, as opposed to something that happens to her.  I punch you: that’s something I do.  So it’s an action of mine.  You get by punched by me.  That’s not something you do.  So it’s not an action of yours.

So here’s our first suggestion.

          Answers to the constitutive question:

Where e is an event and p is a person, e is an action of p’s = e is a doing of something by p

(Word on significance of “=”)

Thus e is an action of mine if it is, for example, a punching of you by me.


The problem with this suggestion is that the idea of a doing turns out to be too broad.  While it may be true that all actions are doings, not all doings seem to be actions.

Suppose I sneeze.  Then that is something I do.  At the very least, that’s how we ordinarily regard a sneeze.  Suppose you come into your living room and I say, “Sorry.  I startled the cat and he jumped out the window.”  You say, “What did you do?”  I say, “I sneezed.”

Or you find out that a date has gone badly, and you wonder, “What did I do?  What did I do?”  Then you realize you kept loudly sneezing.

But a sneezing isn’t an instance of the exercise of agency.  At any rate, sneezing doesn’t seem to be the kind of thing we had in mind when we were thinking of this power to participate, to take an active role, in worldly goings on.  Thinking in terms of the contrast between cases in which an agent acts, exercises that power, and cases in which things just befall her, a sneeze seems to fall into the latter category.


So the idea of an action can’t be identified with the idea of something a person does.  We need further material.

But now consider that not all sneezings seem alike in this regard.  Suppose I know that the person I’m out on a date with hates sneezes, and I’ve decided I want to nip this relationship in the bud.  Since I’m one of those people who can sneeze deliberately, I set myself to constantly sneezing to turn her off.  In this case, it does seem like these sneezings are actions, exercises of my agency, in the relevant sense.


What about this case is different from ordinary cases of sneezing, such that we are inclined to think of it as an exercise of agency?

That it was intentional, deliberate, done on purpose.

The difference between a sneezing that is a genuine action and a sneezing that isn’t, it seems plausible to say, is that the former sneezing was done intentionally.


So here’s our second proposal:

e is an action of p’s = e is an intentional doing of something by p

So if e is an intentional sneezing by me, then it’s an action and moreover my action: I am its agent.


This seems on reflection a promising suggestion.  But there are various worries one might raise about it.



1. Austin on differences between intentional, deliberate and on-purpose doings

The first is illustrated by Austin’s piece.  It might at first seem that talk of a person’s doing something intentionally is equivalent to talk of a person’s doing something on purpose, or deliberately.  But as Austin shows decisively, through among other things construction of ingenious examples, this is not the case.  There are myriad differences of meaning among these locutions.

This raises the question of whether these differences make a difference to our current question: whether, that is, the idea of agency should is more appropriately linked, a la, Davidson to intentional doing, or to deliberate doing, or to doing on purpose.  Or perhaps it can’t be linked definitively to any.

Perhaps the differences Austin tracks don’t bear on our general question, and any would do just as well.  But then given, as Austin points out, that one can do something intentionally but not on purpose, or deliberately but not intentionally, which concept we opt for will yield a different answer in these cases to whether the doing in question is an action, an exercise of agency.


There is a related potential issue here, which Austin does not really develop in this piece but does elsewhere: namely, that our ordinary talk of action, purpose, intention and so forth is so complex and variegated that it is bound to distort the phenomena in question by seeking anything like a general criterion of actionhood.

Davidson, unlike many philosophers who broach analyses of this or that concept, is alive to this possibility.  And he’s aware that it’s quite unlikely that we’ll find an account of action for which there are simply no problem cases, areas of indeterminacy or the like.  But he thinks the search for a general account, even if it will never quite fit all the phenomena can still be illuminating.


I will not discuss these issues further, which are really questions about the possibility of giving philosophical accounts of general phenomena.  But we shall have occasion later on to talk in much more depth about intentions and purposes in particular, and their differences.


2. Account doesn’t go deep enough

Suppose we accept that actions are the doing of things intentionally.  Even, this doesn’t seem a huge advance.  For the question may arise: what is it to do something intentionally?


One source of this question is that there is a thicket of complicated mental phenomena in this vicinity, and we might reasonably hope to sort our the interrelations.  For example, is doing something intentionally the same as doing it with some particular intention?  Must someone intend to do x before doing x in order to do x intentionally?  What, precisely, is it to have an intention to do something?

Moreover, it seems quite plausible to think, for reasons that we shall shortly consider, that actions done intentionally are done in light of or on the basis of beliefs, desires and perhaps other mental states.  What is the precise relationship?  Can we explain the idea of intentional doing in terms of these other mental concepts?


Another source of this question might be the sense that to explain agency in terms of any mentalistic concepts is unsatisfactorily.  For what we want in philosophy of mind, of which I suggested last time that the philosophy of action is a component, is to resolve the mind/body problem, to explain how the distinctive phenomena of mental life have a place in the natural world.  To give a mentalistic account of agency may not seem to make headway on this matter. (Velleman will disagree.  Sees such account as reducing agency/body problem to mind/body problem

We shall return to this question as well.


Davidson himself is aware that one might wish to go further, or deeper, in answering the constitutive question.  But he does not try to do so in this essay.  (Next time we shall look at a kind of further answer he gives in other work).


Rather, he focuses on a third worry about his proposal, and some further issues that it carries in its train.


II. Being both an intentional and an unintentional doing

3. Some unintentional doings count as actions.

Suppose I set the VCR to tape The Ring.  I believe the tape I am using is blank.  But in fact, it has an episode of Sex and the City my wife hasn’t seen.  So I tape over the episode.

Taping over this episode is certainly something I do.  Moreover, it seems to be an action.  It doesn’t seem to fall into the category of a doing that isn’t an action, like a sneeze.  It’s an exercise of agency, an intervention by me in the course of affairs. 

(Certainly my wife will be more likely to yell at me if I do this than if I sneeze.  Might say I should have known, should have checked.  Not so with sneeze.)


But on the other hand, I did not intend to tape over the episode.  My doing so was unintentional.

So we have an instance of an unintentional action.  And it seems to follow that Davidson is wrong to hold that all actions are intentional doings.

Davidson’s response is not to deny the premise of this argument, but rather to deny that the conclusion follows.


Davidson accepts:

Some unintentional doings of things are actions.

But what he wants to claim is that this is perfectly consistent with the claim, a consequence of his answer to the constitutive question, that:

          All actions are intentional doings of things.


How can both of these be correct?  Because, says Davidson, my doing of something intentionally can be identical to my doing of something else unintentionally.

In this example, I taped The Ring intentionally, and in doing so, unintentionally taped over an episode of Sex and the City.

On Davidson’s view, there is here a single event.  It is both a taping of The Ring and a taping over of an episode of Sex and the City.  The taping was intentional.  The taping over was unintentional.

To put it in the language of events that we have been adhering to:

There is an event e that is both an intentional taping of the Ring by me and an unintentional taping over of an episode of Sex and the City by me.


Thus the fact that my action is an unintentional doing of something—namely…--doesn’t make it a counterexample to the claim that every action is an intentional doing of something.

What can’t be the case, of course, is that an action is both an intentional and an unintentional doing of x. That would imply that the action is both an intentional doing of x and not an intentional doing of x.  And that is simply a contradiction.  It can’t both be the case that p and not-p.

But a single action can be both an intentional doing of x and an unintentional doing of y.


How is this possible?  Well, part of what’s at work here is the simple idea that a given event can have multiple properties.  In this respect, they are like objects.

Take a simple sentence of the form: O is p.

This sentence is about an object, O, and says of it that it is p.  It predicates something of O.  To put it another way, it ascribes a certain property to O.  So properties correspond to the predicate place of sentences (not that we can’t nominalize talk of them).


Now events are different from objects.  The soldiers who fought world war II are different from world war II itself.  And there are various things that can be said to isolate these categories, for example the manner of their existence in time.  Objects exist through time.  Events take place over time.  Some philosophers argue that whereas objects have spatial parts, events have temporal parts.  So just as a given soldier is composed of spatial parts—two arms, a torso, etc.—a war is composed of temporal parts, an early skirmish, an interlude of relative peace, a final battle, etc.

But events are like objects in that they correspond to the subject place of a sentence.  They have properties, things are predicated of them.

E has p.

So like objects, they can be ascribed properties.  “World war II was brutal”—this ascribes the property of being brutal to that event.


Then, just as a given object, say, me, can be a friend of x’s and not a friend of y’s, so a given event can be an intentional doing of one thing, but an unintentional doing of another.


This point, that events have multiple properties, is all we need to understand his response to this objection.  That an action can be an unintentional doing of something is consistent with every action being an intentional doing of something, because having the property of being an intentional doing of x does not preclude an event from having the further property of being an unintentional doing of y.


The point is simple enough, but we there is a complication that we have to be careful about.  The complication arises when we move from speaking of an event as an intentional doing of such-and-such to just speaking of it being intentional, period.

Take our example.  Davidson’s suggestion that there’s one single event that is both my taping of.. and my taping over of …..

But now suppose we ask: was that event intentional?  And the answer seems to be that it is and it isn’t.


To elaborate:

My taping of The Ring was intentional.

This sentence seems true, given the case as described.  Now it is Davidson’s thought that the phrase, “my taping of the Ring” picks out an event that has other properties besides being a taping of the Ring and so can be described in other ways.  For example, it is also an event of my taping over Sex and the City, and can thus be picked out by the description, “my taping over of Sex and the City”.


But then it seems to follows:

My taping over of Sex and the City was intentional.


Why should this follow?  Because my taping over of Sex and the City is my taping of the Ring.  They’re one and the same event.  So if my taping of the Ring has a certain property, so does my taping over of Sex and the City.

But in fact, my taping over of Sex and the City was not intentional.  So this is not a valid inference.



The author of Prior Analytics founded the Lyceum.


Now the author of Prior Analytics is identical to, the same person, as Plato’s most famous pupil

So it follows:

          Plato’s most famous pupil founded the Lyceum.


This is indeed a valid inference.  Since the author… is Plato’s most famous pupil what is true of the author… is true of ….


How we should we respond to the failure of the inference in the first case?  Well, the first thing we can do is register that this form of inference fails when the predicate is “is intentional”.  We can describe the situation as follows.  First of all, when two descriptions pick out the same item, as these two do and these two do, they are said to be coextensive.  So “The author…” is coextensive with… and….


Ordinarily, substitution of one coextensive description for another preserves truth value.  If the original statement was true, then the statement arrived at by substituting a coextensive description for a description in the original statement is true as well.  And if the original was false, the new one is false.


But sometimes the substitution of coextensive descriptions does not preserve truth-value.  Such is the case when the statement is of the form.

______ is (was) intentional.


Now here’s a confusing piece of terminology.  When substituting a coextensive description within a statement is not guaranteed to preserve  truth value, the statement is said to be intensional.   So statements of the form above are intensional.

It’s confusing that these words sound exactly the same.  But the meaning are completely different.  “Intentional” applies to, among other things, actions.  “Intensional” in the sense I just defined applies to linguistic items, including utterances like this one that talk about actions.  But many other kinds of statement are intensional as well, notably ascriptions of beliefs, desires and other mental states.

When Davidson talks about the intentionality of attributions of intention, he’s pointing out that they have the feature just described.  And that property is not special to attributions of intention.


So far, this is just a way of labeling the phenomenon we’ve noticed, namely that this kind of argument fails.  All kinds of questions arise about how to treat, account for, intensional statements in a semantic theory.

I won’t get into this issue at all here.  One thing I will say, though, is that we shouldn’t let the existence of intensional statements like this lead us to abandon the principle that if x and y are numerically identical, if x is the very same item as y, then x and y have all the same properties.  That principle, known as Leibniz’s Law is beyond reproach.  If x and y are the very same object, or event, then it simply must be the case that any property x has y has.  x is y; they’re one and the same, so whatever property is had by one must be had by the other.  In fact, that’s not the right way to put it: there is no “other”; there’s just the one object, which has the properties that it has.  One should suspect that denying this will lead one into tremendous difficulties, and it does, but we can’t here get into them.

What we must conclude instead, I think, is that being intentional is not a property of actions conceived as events.  Being an intentional doing of such-and-such, say of being an intentional taping of the ring, can be a property of an event.  But there’s no such property as being intentional, period.

Sometimes people say instead something like: being intentional is a property of an event as described in a certain way.  In fact, even Davidson himself sometimes resorts to speaking this way (p. 60).  But I suggest avoiding such locutions.  If read literally, they’re flat out false.  There’s no such thing as an event as described in a particular way, in contrast to that same event as described in a different way.  No more so than is there such a thing as object as described in a certain way.  A given object, such as me or this mug, can be picked out by any number of descriptions.  That doesn’t mean that for every object there’s an endless number of these other things—the mug as described thusly, the mug as described this way, etc.  There’s only the mug, and the various ways of describing it.


III. Doing one thing by doing another

Another application of this simple idea that actions are events, and that a given event can have any number of different properties, hence can be picked out by any number of different descriptions is in Davidson’s treatment of talk of doing one thing by doing another.


Some things we do we do by doing something else.  To an example of Davidson’s, the queen kills the king by pouring poison in his ear.  The relationship is not symmetrical.  It’s not the case that the queen pours poison in the king’s ear by killing him.

Why is the one the case and not the other?  Obviously, because of the nature of the casual relations involved.  The poison’s entering the King’s ear caused his death.  It’s not the case that the King’s death caused the poison’s entering his ear.  So the Queen killed the King by pouring the poison into his ear, not the other way around.

This chain can be extended.  Suppose that the King’s death led to the destruction of the kingdom.  Then the Queen destroyed the kingdom by killing the King by pouring poison into his ear.  This will be true, by the way, whether or not the Queen intended to destroy the kingdom.  She could have destroyed it without intending to.

And the chain can be extended in the other direction.  For how did the Queen pour the poison into the King’s ear?  She might have done it in a variety of ways.  But let’s say she did it by moving her hand in a certain way.  So she destroyed the kingdom by killing the king by….

The relation expressed by “by” here, by the way, is transitive.  So given what was just said, we can say that the Queen destroyed the kingdom by moving her hand.


So we have a series of events, each a cause of the succeeding:

Queen’s moving of her hand à poison’s pouring into King’s ear à King’s death à destruction of kingdom.


Because of this, we can say that Queen has done a variety of things, and done some by doing others:

The Queen destroyed the kingdom by killing the King by pouring poison into his ear by moving her hand.


Now here’s Davidson’s crucial move: he maintains that these are not descriptions of four separate actions by the Queen.  Rather, what we have here is the material for four separate descriptions of the same, single action.  The Queen’s moving of her hand is identical to, the very same event as, her pouring poison into the king’s ear, and identical to her….


The queen’s moving of her hand

          [equal’s sign here and below]

the queen’s pouring of poison into the king’s ear

the queen’s killing of the king

the queen’s destroying of the kingdom


The thought here is that actions are often described in terms of their consequences.  A killing, for example, is such in virtue of its causing a death.  To characterize an event as a killing is, among other things, to characterize it has the cause of something’s death.  A rescuing is such in virtue of its causing the cessation of a dangerous or otherwise problematic situation for someone or something.  A wowing, as in a wowing of the crowd, is such in virtue of its causing amazement or admiration on the part of the audience.  And so on.


So given that the queen’s action causes all of these events, it admits of all these descriptions.  But there’s only one action, one event here.

To make sense of this, one must resist the temptation to ask, “But what is this event that can be described in all these ways?”  Well, it is the queen’s moving of her hand, and killing of the king, and so on.  All of these are correct ways of picking out the event.

My mug can be described as Bridges’ mug, the metal object on the desk in the front of…, the thing purchased off of Amazon.com at precisely….  Here one feels no temptation to say, “But what is this thing that is….”  The same, Davidson wants to say, holds of actions and other events.


Moreover, one can’t confuse event that is action with event that is effect.  My boiling of the water caused the water’s boiling, and hence is not it.