University of Southern California Dissertations and Theses

The constitution of action

(USC Thesis Other)

The constitution of action

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content The Constitution of Action
Kenneth Silver
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PHILOSOPHY)
December 2017
Doctoral Committee:
Professor of Philosophy, Gabriel Uzquiano, Co-chair
Professor of Philosophy, John Hawthorne, Co-chair
Professor of Philosophy, Kadri Vihvelin
Professor of Philosophy, Gary Watson
Research Assistant Professor, Ron Artstein, External Member
2

Table of Contents
Dedication…………………………………………………………………………...3
Acknowledgements………………………………………………………………......4
Introduction…………………………………………………………………………7
Chapter One: How to Reject the Identity Thesis but Maintain the CTA………….......20
Chapter Two: The Act of Omitting………………………………………………....45
Chapter Three: The Vague Time of a Killing………………………..............................69
Concluding Remarks………………………………………………………………..91
References………………………………………………………………………….94

3

To my mother, Jane Musselman Silver.

4
Acknowledgements
A dissertation is a clear culmination of one kind, of a doctoral program. Yet in another sense,
it is an initiation into academia itself – in this case, into Philosophy. Given this, it is hard to pinpoint
the appropriate scope of my gratitude. I am thankful to both those who directly contributed to my
ideas or supported me as I wrote this work as well as to those who contributed to my philosophical
development in general. Indeed, I am thankful even for the friends who saw me as ‘deep’ in high
school (Ian Shannon, Kyle Kramer, Adam Jordan Weck, and Abby Liberty in particular), thereby
helping me conceive of myself as a philosopher.
I was hell-bent on becoming a Philosopher long before I had any idea of what that meant. But
I’m confident that I never would have made it far in that ambition if it were not for the sincere warmth
and encouragement of many of the faculty and graduate students at my undergraduate alma mater, the
University of North Carolina at Chapel Hill. Their collective friendliness and patience were as
important for me to learn and model as the particular philosophical views we discussed. Individually
amongst the faculty, I would like to thank John Roberts, Susan Wolf, Dorit Bar-On, Keith Simmons,
Geoffrey Sayre-McCord, Joshua Knobe, Jesse Prinz, Thomas Hofweber, Bill Lycan, Marc Lange, Ram
Neta, and Matt Kotzen. John, Susan, Dorit, and Keith deserve special thanks for truly mentoring me
both philosophically and professionally; however, many of these faculty members went far beyond
any pedagogical obligations towards me. They provided me with opportunities to grow both as a
philosopher and as a person, and I am grateful for it. Meanwhile, the graduate students at Carolina
welcomed me in their seminars and helped me to understand what would ultimately be the next phase
of my life. Among them, I would like to give a special thanks to Patrick Connelly, whom I will always
consider my friend.
I would like to thank the University of Southern California for its support of me during
graduate school (and as a postdoc this coming fall). It has invested in my education and development,
and I hope to have made good on the promise seen in me when I was accepted. I am grateful both to
the university as a whole as well as the philosophy department. In my tenure, the department has
grown tremendously, and I have been fortunate to benefit from this. Among the past and present
faculty of USC, I would like to thank Andrew Bacon, John Dreher, Kenny Easwaran, Stephen Finlay,
Jeremy Goodman, James Higginbotham (RIP), Robin Jeshion, Shieva Kleinschmidt, Janet Levin,
Frank Lewis, Sharon Lloyd, Andrei Marmor, Ed McCann, Jon Quong, Jake Ross, Jeff Russell, Mark
Schroeder, Scott Soames, James van Cleve, David Wallace, Ralph Wedgwood, George Wilson, and
Gideon Yaffe. I would also like to thank current and former members of the administrative support
5
staff, including Tomiko Higuchi, Cynthia Lugo, J.G.N. Nikolai, Corey Resnick, Natalie Schaad, and
Barrington Smith-Seetachitt.
Deserving of the most direct acknowledgement for this project is of course my dissertation
committee: Gabriel Uzquiano, John Hawthorne, Kadri Vihvelin, Gary Watson, and Ron Artstein.
Thank you all. Gabriel, I know that you are sympathetic to the view that I present, but I learned a lot
from your willingness to recognize the limitations of this position. More was gained through my
understanding what assumptions really could legitimize my opponent’s view. John, thank you for many
fun conversations out in the courtyard, and for helping me to attain a level of precision to my work
that these questions deserve. Kadri, events involving your guidance have whether/whether influence
over the event of my finishing this dissertation, which is just to say that I could not have done it
without you. You helped me find a love for causation and metaphysics more generally early on in the
program, and those tools will serve me as long as I am in the field. Gary, thank you for all of your
guidance. Long discussions about the philosophy of action with you are originally what led me to want
to contribute to the metaphysics of action. Finally, Ron, thank you for being my external committee
member and giving me your honest feedback. It was great working with you.

I was so thankful for the culture that facilitated the positive environment of my time in
philosophy in Los Angeles, and paramount to the culture was the make-up of the USC graduate
students, including: Greg Ackerman, Mike Ashfield, Ara Astourian, Matt Babb, Rima Basu, Simon
Blessenohl, Renee Jorgensen Bolinger, Josh Crabill, Justin Dallmann, Kory DeClark, Alex Dietz, Sean
Donahue, Erik Encarnacion, Marina Folescu, Paul Garofalo, Bixin Guo, Keith Hall, Michael Hatcher,
Ryan Hay, Frank Hong, Joe Horton, Nathan Robert Howard, Nicola Kemp, Tanya Kostochka,
Junhyo Lee, Woo Ram Lee, Ben Lennertz, Matt Lutz, Shyam Nair, Elli Neufeld, Michael Milona,
Caleb Perl, Christa Peterson, Kenny Pierce, Quyen Pham, Abelard Podgorski, Michael Pressman,
Indrek Reiland, Justin Snedegar, Vishnu Sridharan, Julia Staffel, Douglas Wadle, Jesse Wilson, and Jon
Wright.
From this list, I have withheld some. The metaphysical force is strong with USC, and I was
especially happy to work with and talk to Maegan Fairchild, Thomas Hall, Jennifer Head, Matt
Leonard, and Dan Pallies on all matters metaphysical. Similarly, I was very fortunate to have a terrific
group of people to work with in the philosophy of action, including: Steve Bero, August Gorman,
Alida Liberman, Sam Shpall, Aaron Veek, and Aness Webster. Many of the USC graduate students
were not only colleagues, but my close friends, so I would like to thank them for their friendship as I
was engaged in this process. Among my closest friends and within my cohort were Ryan Walsh, Julian
6
Stone-Kronberg, and Nicholas Laskowski. Thank you for helping me find my footing in philosophy,
in Los Angeles, and in life. Nick, not only are you one of my closest friends and philosophical
confidants, but I would have had no idea how to structure this acknowledgements section without
you.
Apart from those associated with USC, a number of philosophers helped me to both develop
professionally and work through ideas related to this dissertation, including: Sara Bernstein, Andrei
Buckareff, Joe Campbell, Randy Clarke, Greg Janssen, and Carolina Sartorio. More generally, I could
not have hoped for a more welcoming group of philosophers than the philosophers I have met
working within the philosophy of action and responsibility.
I had so much wonderful support from friends and family outside of philosophy. While in Los
Angeles, I was lucky to count among my close friends: Ben Thompson, Greg Faletto, Amanda
Prunesti, Kayla Kurisu, Clo Pazera, Myles Webster, Eric Kunau, and my indie improv troupe
‘Balanced Breakfast.’ Each of you would let me ramble on about philosophy if necessary, but it never
was. I would also like to thank my immediate family: my mother, Jane Silver; my father, Alan Silver;
and my step-mother, Patricia Silver. I will never know how hard it is for some people to tell their
parents that they want to be a philosopher, because each of you could not have been more supportive
and trusting of my decisions regarding this field. The question was never why would I do it, but what
was the best way for me to engage with this subject that I love.
Finally, I would like to thank my partner, Ilana Winter. No one has had to put up with as much
of my whining, and no one could have handled it more gracefully, than you have. I love you. You
know what matters, and you constantly help me to remember. Because of you, I am a better person
and writer, and I will always be grateful that you were with me through the inevitable growing pains
of this doctoral program. I am also deeply thankful for your family, including: Cathy, Craig, Bryan,
Jeff, Velveeta, Florence, and many others. They have become part of my family, and they have given
me so much emotional support in this process.

7
Introduction
Our actions are important to us. They express if not determine who we are and what we value.
They are how we get what we want, do what we ought, and change the world. However, although it
is clear that our actions play these roles, it is less clear what action is such that it can do these things.
We act for reasons, but what is the relationship between our reasons and our actions? What are our
actions, and how can actions be a part of the natural world? For which actions are we responsible, and
does this conflict with our actions’ being part of the natural world? Are we the only creatures that can
act, or can animals, can computers, can babies, can businesses? What we need is an analysis of the
concept of action. With an understanding of the concept of action, we can judge to what extent the
concept applies and what it means for explanation and responsibility.
In this dissertation, I will defend a prominent theory of action, the Causal Theory of Action
(or the ‘CTA’ for short), against a particular class of objections. As we will see, a classic version of the
CTA identifies all actions with body movements. While this is initially a very plausible idea, there are
nevertheless a number of objections to it, and these have been used as grounds to reject the causal
approach to action altogether. I will attempt to throw out the bathwater but keep the baby. I will show
how we can reject the thesis that all actions are body movements without giving up on the reasons
that led us to accept a causal theory. I will argue that the metaphysical and linguistic resources are
available for accepting an augmented version of the CTA that is even more plausible than the classic
version.
As set up, I will first introduce the Causal Theory of Action and the main motivations for
accepting it. We will see that in order to give a precise version of the CTA, Donald Davidson initially
identified all actions with movements of the body. Although there are many different kinds of
problems that have been raised for the CTA in general, several objections have been given to this idea
that all actions are body movements in particular. I will motivate Davidson’s view, introduce the
objections, and lay out the plan of the dissertation to take on these objections in turn.

Being Led to the Causal Theory of Action
In the philosophy of action, the Casual Theory of Action has dominated the literature for the
last half century. Central to the CTA is the idea that actions are events that are distinguished by having
a special causal history. Why is my intentionally raising my arm different from my arm’s rising as a
8
reflex or against my will?
1
The former is caused by certain mental states (that I wanted to raise my arm
and I knew how to do it), while the latter is not. Many actions look intrinsically identical to possible
events that are clearly not actions, so what distinguishes the actions as actions on this kind of view is
their having particular causes.
Although this general idea behind the CTA has a long lineage in philosophy, it was brought to
prominence in the twentieth century by Donald Davidson in his landmark essay, “Actions, Reasons,
and Causes.” There, Davidson was concerned with the connection between our reasons and our
actions. He wanted to know why it is that our reasons rationalize our actions when they do. That is,
in what sense do our reasons for our actions explain our actions? And what distinguishes the reasons
for which we act from the reasons we merely have for acting?
His answers to these questions came in two parts. He argued (i) that reasons explain actions
when they do by causing those actions, and (ii) that these reasons can be identified with certain mental
states such as beliefs and desires. So, for instance, although I may have a number of desires that I
believe would be satisfied by mowing my lawn, it is possible for me to mow my lawn for the purpose
of satisfying one particular desire (e.g. to spite my neighbor). In this case, what explains why I acted
for that reason in particular is that my desire to spite my neighbor caused me to mow my lawn while my
other desires did not.
Both of these claims were innovative in the historical context. The idea that reasons cause
actions has roots that go back to Aristotle, and it is clearly articulated and endorsed by Hobbes.
Nevertheless, it was a breakthrough at the time to persuasively defend the idea that reasons cause
actions at all. Before this point, philosophers had become convinced that reasons could not cause
actions, because reasons were thought to be logically connected to actions.
2
If I could describe my
action in terms of my reason for doing it, it was thought, then my reason could not be the cause of
that action. Davidson’s innovation was in part to push against this assumption. He pointed out that
actions are events, and events are frequently described in terms of their causes. Further, he
compellingly argued that a reason for performing an action could cause that action, and that reasons
for actions could be identified with beliefs and desires of the agent.
Showing this causal connection between reasons and actions goes a long way towards
answering two of the questions with which we began. We wanted to understand the sense in which

1
This is roughly the same question as Wittgenstein’s famous ‘problem of action’: “What is left over if I subtract the fact
that my arm goes up from the fact that I raise my arm?” (section 621).
2
This line of thought can be found in Hampshire (1959), Kenny (1963), and Melden (1961).
9
reasons explain actions when they do, and we wanted to know how actions fit into the natural world.
If reasons explain actions in virtue of causing them, then this answers the first question. Further,
although we may have thought that actions are connected to mental states or events in some way (and
those mental states/events are in some way related to items in the natural world), learning that actions
are caused by mental events leads us to an even more specific view of how to naturalize actions.
3

Nevertheless, what Davidson had established is still a long way from an analysis of action, and the
questions we have about action are still largely unanswered.
For one thing, learning that some actions are caused by reasons does not show us that it is
necessary for something’s being an action that it is caused by some reason. That is, more needs to be
said to argue that all actions are caused by reasons. Anscombe (1957) argues that it is appropriate to
expect that an explanation could be offered by the agent for any intentional action she performs.
Insofar as such an explanation involves adverting to one’s reasons, and those reasons caused the
agent’s actions, this will give us the resources to defend the claim that at least intentional actions are
caused by reasons. Further, insofar as Davidson himself argues that every action admits of an
intentional description,
4
Davidson can claim that every action is caused by one’s reasons for acting.
It is controversial whether all intentional actions must be done for (and so caused by) some
reasons, and it is even more controversial how to think about whether unintentional actions must be
caused by reasons. However, more important for our purposes here is that even if we agree that it is
a necessary condition for action that actions are caused by reasons, we would be short of a sufficient
condition for action. That is, learning that actions are caused by reasons is not enough to tell us which
things are actions and how they are so caused. Still, this at least provides us with the tools needed to
establish a more general version of the CTA. Knowing only that actions are always caused by certain
mental states, we could give the following schema, offered by Aguilar and Buckareff (2010:1):
(CTA) Any behavioral event A of an agent S is an action if and only if S’s A-ing is caused in
the right way and causally explained by some appropriate nonactional mental event(s) that
mediate or constitute S’s reasons for A-ing.

3
This question of how to fit actions into a picture of the natural world, and the conviction that a causal theory of action
was a crucial means of answering this question, is the central focus of John Bishop’s Natural Agency.
4
He says, “…a man is the agent of an act if what he does can be described under an aspect that makes it intentional”
(1971/2001:46). So, if I am writing, and this is an action, then although there may be many ways of describing what I am
doing according to which I am not doing that thing intentionally (such as moving my fingers at some exact speed), there
is at least one way of describing what I am doing according to which I do what I do intentionally – namely, ‘writing.’
According to Davidson, even if I perform an unintentional action such as insulting you (perhaps by writing instead of
listening to you), what I do counts as an action because there is a description under which I am doing that thing
intentionally.
10
This schema can be endorsed by any proponent of the CTA. Further, it makes it clear where we would
need to be more specific in order to arrive at a particular version of the CTA. A less general version
of the schema would require three things: First, we would need a view concerning which nonactional
mental event(s) mediate or constitute S’s reasons and cause S to A. Second, as we will see below, there
is a challenge with saying exactly what ‘the right way’ is in which actions are caused. And third, the
term ‘behavioral’ is hardly clearer than ‘action,’ so we will need a view concerning which events caused
in the right way by one’s mental states count as actions. In the next section, I will consider these
challenges in turn, and I will show how Davidson arrived at a particular version of the CTA, focusing
on exactly which events will be actions.

Davidson’s CTA
A general schema for the CTA is helpful in that it codifies or does some justice to the idea
that something is an action in virtue of
5
its causal history. Still, this general view is not enough to
demonstrate that any particular version or spelling out of it will be true. For all we know, there may
be no mental events that can properly understood as one’s reasons for acting or way of characterizing
those events that when caused in the right way are actions. To allay these concerns, I want to focus
on the particular account of the CTA given by Davidson himself.
We have already seen which mental events are, for Davidson, those events that cause one’s
actions and that constitute one’s reasons for acting. These will just be belief/desire pairs – typically a
desire for some end and a belief that acting in a particular way will achieve that end. There is a lot
more to say here to justify focusing on these mental states in particular. It is still an open question
whether these are the right mental events to focus on or whether or not there are others (such as
intentions) that should also be included. And there is also an open question concerning whether these
mental states just are or can appropriately constitute reasons. Nevertheless, let’s take on board this
component of Davidson’s view and move on to consider the other specifics of his account.
I noted above that that the events that are actions must be caused by certain mental states in
the right way. The reason to make this distinction stems from Davidson as well, as he recognized cases
of what is known as ‘causal deviance’ – in which an agent’s reason for acting causes an event that

5
Although, as stated, it does not make any kind of grounding claim that might be necessary to license the talk of events
being actions ‘in virtue of’ their causal history.
11
looks like an action, and yet we would intuitively not call it an action. Here is a prominent case given
by Davidson:
A climber might want to rid himself of the weight and danger of holding another man on a
rope, and he might know that by loosening his hold on the rope he could rid himself of the
weight and danger. This belief and want might so unnerve him as to cause him to loosen his
hold, and yet it might be the case that he never chose to loosen his hold, nor did he do it
intentionally. (1973/2001:79)
In this example, we see that an event is caused by a reason for action, but because that event is
somehow caused ‘deviantly,’ we do not think of it as an intentional action (or indeed an action at all).
There are a number of different versions of these cases involving causal deviance, and collectively they
offer a formidable challenge. A causal theorist must say how to appropriately characterize the
connection between reasons and actions in completely causal terms while avoiding all cases of
deviance.
For his part, Davidson originally despaired at the analytic philosopher’s ability to fully
characterize the causation of action so as to avoid these wayward causal chains, but he did not see this
as a fundamental threat to the causal approach to action (ibid.:80). Instead, he thought that empirical
science would eventually tell us what kind of causation was not deviant. Given this, we should just be
satisfied with saying that actions are those events caused in the right way. Historically, many opponents
have been enormously unsatisfied with this response. We may worry that there is no such ‘right way,’
and the lack of such a way indicates that no causal theory of action can ultimately succeed. Examining
this issue has occupied many working within the causalist framework, and we will not take it up here.
Instead, I want to move on to the last component necessary to flesh out Davidson’s version of the
CTA. We will spend more time in the explanation for this component of Davidson’s account, since it
is the central focus of this dissertation.
Actions cannot just be those events caused in the right way by these mental events, because
those mental events appear to cause many events beyond our actions. Even considering a case in
which beliefs and desires cause actions, we do not want to say that everything that they go on to cause
via those actions counts as the same or some further action.
6
If a desire for milk causes me to stand

6
We could in some way restrict this by recognizing that the schema for the CTA given above restricts the actions of an
agent to ‘behavioral events’ of that agent; however, this is not very illuminating, as now we will be unsure about what
constitutes a behavioral event. For example, although Dretske (1988) spends a lot of space considering which events are
the behaviors of something, he ends up suggesting that behaviors will be primarily caused by something internal to the
behaving entity, but that these matters can be extremely context-sensitive given the interests of the individual investigating.
12
up to get milk which causes you to get angry at me which causes you to yell at me, we would not
characterize your yelling at me as my action. So, we need some way to restrict the domain of events
caused by reasons to pick out only the actions.
Davidson provides just such a theory of restriction in “Agency.” There, he is concerned
directly with this question of which events in the course of a life are expressions of one’s agency. As
he says, “…what are his deeds and his doings in contrast to mere happenings in his history; what is
the mark that distinguishes his actions?” (43).
7
After considering and abandoning an approach to
answering this question by arguing that all instances of agency are done by the agent under some
intentional description,
8
Davidson instead answers the question in two steps. He first gives a theory
for which events are primitive actions, and then he goes on to give a theory for non-primitive actions.
He says,
If we interpret the idea of a bodily movement generously, a case can be made for saying that
all primitive actions are body movements. The generosity must be openhanded enough to
encompass such ‘movements’ as standing fast, and mental acts like deciding and computing.
(49)
As our standing still involves our body, and our mental acts seem to involve our brain ‘moving’ or
being in various states (and it too is a part of our body), this may not seem too generous. So here is a
particular theory about exactly which events caused by certain mental states are primitive actions:
primitive actions are body movements caused by reasons.
We may for the moment accept this idea for primitive actions, but what about non-primitive
actions, and what are we to think about the distinction between primitive and non-primitive actions?
The latter is an interesting question that has been taken up a lot in the literature on the metaphysics
of action; however, the debate over this issue is somewhat academic given Davidson’s answer to the
former question. Davidson argues that non-primitive actions are, just like primitive actions, identical
to those same bodily movements. Although it may seem like playing basketball or voting or calling a
friend are actions I can perform that seem to involve elements beyond me or not be a matter of merely
moving my body, Davidson disagrees. Instead, he argues at length that although we may have many

7
A subtler but still important distinction we might make is between things I do and actions I perform. The former might
include events like my digesting that are clearly not acts, whereas my actions seem to be things that I do that express my
agency. Here, too, the difference will be the mark of agency.
8
Davidson saw this approach to the problem as unilluminating because he considered the concept of agency more basic
than the concept of intention (49). Although in “Actions, Reasons, and Causes” Davidson claimed that the phrase ‘the
intention’ was syncategorematic and did not refer to any item whatsoever, in his later paper “Intending” (written after
“Agency”) Davidson offered a theory of intentions.
13
ways of describing actions where those descriptions do not mention bodily movements, what is picked
out by those descriptions are those same bodily movements identical to primitive actions. For instance,
my ‘playing basketball’ will pick out those body movements I perform as I play, just as ‘making you
happy’ may be another description that picks out those same body movements, if my playing
basketball makes you happy. We have a whole method for offering different descriptions for actions
that incorporate or leave out the effects of body movements, but, according to Davidson, all actions
just are body movements variously described. He says,
We must conclude, perhaps with a shock of surprise…mere movements of the body – these
are all the actions there are. We never do more than move our bodies: the rest is up to nature.
(59)
This oft-quoted passage is provocative, and, as we will see below, it can be well-motivated. It gives a
very clear and definitive answer to our question of which events are actions – a subset of our body
movements, caused in the right way. So, Davidson can be seen as accepting the following formal view:

Identity Thesis: For every action a of agent S that occurs in an interval of time t, a is identical to some
body movement
9
m of S at t.

With this view in hand, we have all of the pieces necessary to formulate a specific version of
the more general CTA given above for Davidson.
10
So far, we have taken on board the idea that all
actions are caused by certain mental states (beliefs and desires), and for Davidson actions will be
caused in the right way by those mental states. Putting this together with the Identity Thesis gives us
the following account:

Davidson’s CTA: Event a is an action of agent S iff a is a body movement of S caused in the right way
by certain belief/desire pairs of S that are S’s reasons for performing a.

This analysis purports to capture all and only actions. We can also see how the importance of causation
comes through on this account. This account identifies actions with body movements and

9
This may be a cluster or extended sequence of body movements that occur during the interval. We would consider the
sequence of movements together to itself be the movement of the agent at t.
10
It is worth noting that there does not seem to be a place in his writing where Davidson gave the CTA (or his particular
version of it) as a formal analysis of action. Nevertheless, I take this to be a fair reconstruction of his view.
14
distinguishes actions as those body movements with the right causal history. Given this, we can see at
least the foundations for how we can answer the questions with which we began.
Remember, we began with four basic questions: (1) How are actions connected to or explained
by our reasons for acting? (2) How are actions items in the natural world? (3) What is the scope of our
responsibility for our actions, and how is it possible for us to be responsible? (4) And which things
can act? This account gives the answer to (1) that we saw above: actions are explained by reasons by
being caused by those reasons. And while we have a partial answer to (2) if we can identify reasons
with mental states (and if we can find mental states in the natural world), this account makes answering
(2) even easier. We already take for granted that the movements of our body are events that occur in
the world. So, rather than think that actions are some sort of special mysterious events imbued with
purpose or intentionality, Davidson’s view identifies actions with these movements that we already
accept.
I do not think that any version of the CTA alone is sufficient for providing a complete answer
to (3). However, the CTA can show us which events are actions, and if we have a view according to
which we can only be responsible for actions, then the CTA coupled with this view will deliver a
constraint on those events for which we might be responsible.
11
Finally, to answer (4) the CTA shows
us that we need only determine which things are capable of having bodily movements caused in the
right way by the right kinds of mental events. For instance, while it may be a very difficult task to say
what needs to be true to characterize an artificial intelligence as having certain beliefs and desires or
in some way ‘moving,’ the CTA at least shows the bar that the AI must meet to count as acting. So,
Davidson’s CTA has many advantages if it can be accepted, but of course there are problems that can
be raised at every stage that led us to it.
Some philosophers won’t grant that reasons cause actions.
12
Others will deny that actions are
always caused by reasons. Many others will balk at the inability to characterize ‘the right way’ that
actions are caused.
13
And still others will disagree with Davidson’s identification of all actions with
body movements. Although all of these challenges deserve due diligence, and much of the literature
has been centered around several of them, it is with this last group of philosophers that I will be

11
Answering the further question of how it is that we are responsible for our actions would require a theory of
responsibility, for one, and it would also require answering the various challenges to our ever being responsible on a
causalist’s framework. I will not attempt to engage with and resolve these issues in this dissertation, but it is crucial to
recognize that we must first accept a theory of action to make precise what the challenges to responsibility might be.
12
See Wilson (1989), Schueler (2003), and Sehon (2005).
13
For example, see Sehon (1998) and O’Brien (2012) for two instances of individuals who think the problem of causal
deviance cannot be solved and that a causal theory of action should be abandoned for this reason.
15
concerned with in this dissertation. That is, I am concerned with challenges to Davidson’s Identity
Thesis. So, how plausible is the Identity Thesis, what are the objections to it, and can these objections
be met? Or, if the Identity Thesis should be rejected, can it be replaced with a view that allows us to
maintain a respectable causal theory of action? To answer these questions, let’s first take a step back
and examine the motivations for accepting the Identity Thesis.

Reasons For and Against the Identity Thesis
As we saw, the Identity Thesis naturalizes actions by identifying them with events we already
take to exist, and it also provides a limit on which effects of our reasons count among our actions.
These are theoretical advantages of identifying actions with body movements, but these are not
reasons that motivate us to accept this identifying. Instead, I think there are three sorts of
considerations that suggest identifying actions with body movements. First, we can recognize that
many of our actions seem to involve or entail that our body moves. I make breakfast; I wave to a
friend; I ride a bike. These are all clearly actions, and we can easily see that the object that is had in
common between them is my body, or different parts of my body. In each case, my body is moving
around in various ways, and it is hard to imagine myself capable of performing these actions without
moving my body. We might even accept the stronger claim that all actions at least involve my body in
some way, even if it is not moving. Apart from cases in which we purposely stand still or are thinking
(which Davidson already addressed), every case of action I can think of somehow involves my body
or my body’s moving in some way. This at least strongly suggests that our actions are somehow closely
connected to certain events of our body’s moving in various ways. From here, it’s not a big leap to
think that these actions might be identical to these body movements.
In addition to this, our bodies (or at least our limbs) often seem under our control in a way
that external items are not. I can reliably control my winking in a way that I cannot control your
winking. Even if I extort you to ensure that you wink when I wink (and so in some sense do control
your winking), your winking will still not feel like an action of mine, and I do not seem capable of
guiding the manner of your winking as it’s happening.
14
As we saw above, a causal theorist needs an

14
I won’t venture into the nature of the phenomenology that seems to accompany this sense of control (or what Ginet
(1990:13) called the ‘actish phenomenal quality’ that our actions possess), but there does seem to be this sense of what it
feels like to act that is present when we move our bodies that is absent when I coerce someone into moving theirs. I also
won’t discuss here this question of how to understand control over or guidance of our actions. There is a question about
how the causal theorist can understand control and guidance (Frankfurt 1978 argues that causal theories of action do not
16
answer to the question of which effects of one’s reasons count among one’s actions. One place to
look for an answer to this question is those events over which we have reliable control. So this also
suggests that our body movements are among our actions. If we can make the case that we cannot
reliably control events beyond the movements of our own body, then we should think that our body
movements are the only events that are candidates for being actions.
Third, certain actions just strike us as body movements, and the way we describe these actions
is directly by describing our body’s moving. For example, my action of ‘raising my arm’ seems to be
an action I take by moving my arm upwards. If my action of raising or even just moving my arm was
not very closely related if not identical to the event of the movement of my arm, then what would it
be? Once we have on board this idea that at least some actions are our body movements, the thought
may go, then it would be odd to think that other actions are something entirely different in nature
from those body movements. We should at least conclude that all other actions are the same kind of
thing as body movements if not themselves body movements.
This is not sufficient to show that all actions are body movements, but these considerations
do strongly recommend this view. Further, if we accept what Davidson says in “Agency” concerning
how particular events can be described in a myriad of ways (for instance, to include to a greater or
lesser extent the effects of those events), then it is not hard to imagine that all of the descriptions we
use to pick out our actions are merely different ways of describing our body movements. This again
is not quite enough to show that all actions are body movements, but, for all of these reasons, we can
see that the Identity Thesis is a motivated and parsimonious view. In general, what I do, the
contribution of my agency, seems to be a matter of how my body’s moving. Davidson’s view gets to
the heart of this idea that how our bodies move is the only means we have of exerting our will in the
world; it’s all that we can directly control.
Although Davidson’s view is well-motivated, and it comes with several theoretical benefits,
there are a number of clear objections to it. In this dissertation, I will address three different kinds of
cases that offer distinct objections to the Identity Thesis. First, we saw that it was surprising to
conclude that all actions are body movements, because it does seem that many actions are not body
movements. Sure, using Davidson’s discussion of descriptions, there is a way to claim that actions like
my voting or my greeting a friend are identical to the movements of my body as I vote and greet, but
prima facie voting and greeting are very different from how I move as I perform these actions. These

adequately capture the sense in which we guide our actions). For a recent and compelling attempt to give such an account
of control, see Shepherd (2014).
17
actions seem to require certain social contexts or state institutions to be the actions that they are, and
they may seem to have causes and effects or other properties that differ from the movements of my
body. For these and a variety of reasons, it can be hard to accept that my voting, or my greeting, or
my driving, or my gambling just are instances of my moving in a particular way at a particular time.
Second, we often omit to do many things and are held responsible for these omissions. For
instance, I may omit to pick up a friend from the airport and instead take a nap while he stands waiting
on the curb. Historically, proponents of the CTA have struggled to figure out how to handle these
cases of omission. Should we say that we are not responsible for omissions? Or that we are responsible
for omissions, but omissions are not expressions of our agency? Or should we say that omissions (as
our other actions) are also identical to mere body movements (to those movements we perform as we
omit)? Davidson and those following in his tradition have opted for this third route – for identifying
omissions with body movements – but this is a difficult case to make.
Third and finally, a great many of our actions seem to extend beyond the movements of our
bodies. We may, with our movements, set in motion some greater plan over which we have control.
And we may thus seem capable of extending our agency. If Davidson is correct, however, then our
actions are completed once our body is done moving, even if the desired result is not yet secured. This
has led to, among others, the widely discussed problem of the time of a killing. An agent may move
her finger to pull the trigger at one time, although the victim may not die until days or even years later.
In this case, Davidson is committed to saying that the agent killed the victim long before the victim
died, and that sounds unintuitive. So, the Identity Thesis faces pressure concerning the location of
action, that actions seem capable of extending out beyond our movements.
Each of these objections is framed as an objection to the Identity Thesis. They purport to
show that there are actions or expressions of agency that are not identical to body movements because
they are either different in kind or else cannot be colocated with body movements. Now, these
objections in themselves do not directly tell against the general version of the CTA. Although the
Identity Thesis may be incorrect, there may be some other specific metaphysics of action that could
be given to fulfill the mandate of the general CTA. However, raising these challenges to the Identity
Thesis can be a fairly effective means of indirectly challenging the CTA, and these challenges are in
good company.
While some objections to the CTA are rather direct – for instance, arguing that not all actions
are up for being explained by reasons – other prominent challenges to the CTA are indirect. The
causal deviance worry is a challenge to the CTA not because it directly claims that actions are not
18
caused by reasons, but because it claims that there is no complete way to satisfyingly characterize the
causation of all and only actions. If there is no way to characterize the causation of action, then this
suggests the view that actions should not be conceived of as caused by reasons at all. Similarly, rejecting
the Identity Thesis not only allows one to reject Davidson’s CTA, but it casts doubt on the entire
project of analyzing actions as events caused by reasons. These objections may be pushed because we
have a view on which actions are events not up for being caused by reasons (perhaps they are within
us and contemporaneous with reasons or too distal to count as caused by reasons), or because we
think that actions are not events at all (perhaps they are the causing of certain events, rather than
events themselves). Regardless of what motivates these objections, we can see that maintaining the
CTA more generally requires a promising spelling out of which events are actions.

Defending the CTA
We have seen that there are many good reasons for accepting the causal approach. It does
require defense at every step, but the advantages of accepting it are so great that it is worth defending.
The causal approach has the promise of answering some of the biggest questions we have about the
nature of action and responsibility. Some version of it is simply taken for granted so regularly that it
is frequently and appropriately referred to as the standard view.
If we are to maintain a causal theory of action, then we must figure out how to answer these
challenges to the Identity Thesis. Either these objections need to be decisively answered, or we need
to admit that the Identity Thesis is false but not far off, and there is some other view about the
metaphysics of action that allows us to provide a better version of the CTA. It is this latter course that
I will chart in this dissertation. There are ways to push back in the spirit of Davidson and defend the
Identity Thesis, but what I will show is that this is not necessary. Another view about the metaphysics
of action can be given, and it will provide an even more plausible way of dealing with the above cases
as well as pave the way to an augmented but more acceptable version of the CTA.
In short, I am going to argue that the first two objections are spot-on, but that the third
objection can be adequately answered. In the first chapter, I will introduce the arguments for thinking
that there are many actions that cannot be identified with movements of the body (even if we assume
that they are colocated). I do not take these arguments to be decisive – there are certain views that a
Davidsonian could appeal to in order to answer them. What I want to show is that the metaphysical
resources are available for rehabilitating the CTA even if actions cannot always be identified with body
19
movements. After considering how actions are related to body movements, I will suggest that it is
constructive to think that many of our actions are constituted by our body movements, rather than
identical to them. This not only fits actions into an existing literature on dependent but colocated
items, but it allows us to formulate an improved version of the CTA.
In the second chapter, I will consider the case of intentional omissions. These seem to be
intentional expressions of our agency for which we can be responsible, so there should be a place for
them on the CTA. However, the Davidsonian view that omissions are identical to body movements
of the agent will be even less compelling than in the case of regular actions. In response, I will explore
the biggest challenges with reifying omissions as events, and argue that a view of omissions as fine-
grained events has the resources to answer these challenges. Further, I will consider how they can also
be integrated into a revised version of the CTA. I will suggest that we can also consider omissions to
be constituted by the agent’s body movements as she omits, and I will answer several objections that
have been given explicitly against this view.
In the third and final chapter, I will consider the objection that many actions are beyond, and
so not colocated with, our body movements. Whereas I claimed that the first two objections were
correct, I will argue that this objection is mistaken. To answer the problem, I consider the definitions
of the verbs for the actions that appear to possibly be beyond the movements of the body. These
actions would only constitute counterexamples to the Identity Thesis if our use of the verbs that pick
them out referred to events beyond the agent. However, I will argue that the reference of these verbs
is linguistically vague. It is indeterminate whether our use of these action verbs refers to the actions
themselves or the process that leads from the agent to the result of the action. This vagueness will
explain our conflicting intuitions about the cases, and it will also stifle this objection to the claim that
actions are colocated with body movements.
What this will leave us with is a version of the Causal Theory of Action that respects the many
different actions that we perform but also provides a clear limit for which events count among those
actions.

20
Chapter One: How to Reject the Identity Thesis but Maintain the CTA
We saw in the introduction that Davidson had good reasons for maintaining what I called the
Identity Thesis – the claim that all actions are identical to body movements. If any version of the
Causal Theory of Action is going to succeed as an analysis of action, then it must tell us which events
are the ones that, when caused in the right way by the right things, are actions. The movements of our
bodies are a category of events that are within our control, and Davidson provided an explanation for
how all actions can be identified with them. Nevertheless, it is hard to accept that all we ever do is move
our bodies. In this chapter, I want to consider arguments for thinking that many actions cannot be
identified with body movements, and I want to show how we can still accept a version of the CTA
even if we reject Davidson’s Identity Thesis.
In the first section, we will see prima facie evidence of actions that do not seem to be body
movements, and we flesh out how Davidson handles cases such as these. In the second section, we
will investigate a series of arguments against the Identity Thesis in a particular case, and we will see
how similar arguments will be available for many actions (maybe even all of them). In the third section,
we will explore how Davidson could respond to the arguments against the Identity Thesis. The
resources are available for a Davidsonian to defend the Identity Thesis, given Davidson’s views on
events and modality, but I find the views required to be implausible. The important point here will be
that rejecting those views and the Identity Thesis should not on their own recommend rejecting some
version of the CTA.
In the fourth section, I pursue how actions can be events colocated with body movements if
they are not identical to them. Given the features of the relationship between actions and body
movements, there is a strong case to be made for thinking that actions not identical to body
movements are nevertheless constituted by them. Constitution is understood here as a relation distinct
from identity, and it is typically thought of as that relation that obtains between the statue and the clay.
I introduce the relation of constitution, show why we might think that some actions are constituted,
and I discuss the precedence for this view in the literature. Finally, in the sixth section, I show how
the relation of constitution allows us to revise the CTA, and how it has several advantages for the
proponent of a causal approach to action.

21
Disagreeing with Davidson
There are many intentional actions or instances of agency that prima facie do not seem to be
identical to mere movements of our body. Consider the acts of greeting, moving a pawn in chess, and
proctoring an exam. For a given instance of each of these acts, some body movements may be related
to each. When we greet someone, we may wave our arm at them; when we move a pawn, we may pick
it up with one hand and place it down somewhere else; when we proctor, we may walk around a
classroom, looking to make sure students aren’t cheating, and putting on a face so as to seem both
stern and encouraging. Nevertheless, in each case, it is prima facie unintuitive to say that these actions
just are these body movements.
15

I may greet someone by moving my arm a certain way, but I’m not just waving my arm at them;
I’m communicating that I recognize them as a person and want them to know that I feel a certain way
about that. Notice that I could have just waved my arm at them. Perhaps I was convinced I have
flattering arms and I wanted to wave it in their face. Or, more boringly, I could have simply been
waving my arms around at the time in just the way that I would have waved, not realizing that they
were there. Then I really would be just waving my arm. But it seems like I did something else – I
greeted them, albeit by waving.
The case becomes stronger if we consider more abstract actions imbedded within cultural
traditions or social institutions – actions that require the context of certain institutions to be the actions
they are. It may be true that to marry someone there is some sense in which I need only to move the
muscles in my mouth a certain way at a certain time, and the rest is up to nature. But it really feels like
this leaves out a sense of what I do – this extra (special) thing – of marrying someone. I could have
moved muscles to voice something that sounds like “I do” at any time, but I seem unable to marry
someone without the appropriate linguistic context and social institutions in place. Less romantic cases
include when a broker moves her hand and underwrites a loan, when a priest moves his hand to
worship, or when legislators raises their hands in order to vote. These are not just cases in which the
presence of certain circumstances or institutions seem to partially explain how these actions are

15
The intuitions here may be a little more complicated than I am suggesting. For example, those that have argued that the
statue is not identical to the lump of clay that makes it up may nevertheless claim that sentences like ‘The statue just is the
lump’ are true because there is an ‘is’ of constitution. (See Wiggins (2004) for this suggestion, but see Pickel (2010) for
arguments against there being an ‘is’ of constitution.) Similarly, I may want to admit that there is a sense in which the
greeting just is the arm-movement, at least insofar as the greeting is nothing over and above than the arm-movement. That
being said, I think it is even less intuitive to say that the greeting is the arm-movement than it is to say that the statue is
the lump.
22
possible; these are cases in which the very role that agents take themselves to play in these institutions
is defined by their ability to perform these distinct actions.
This is not proof that these actions are in fact distinct from mere body movements, but it does
suggest that we take them to be. I can certainly pretend that by flicking my wrist I do some further
action of casting a spell that I do not actually perform, but most actions do not seem to be like this.
We may have invented chess, but now that we have it, I take myself to be able to move my pawn when
playing, and this does not seem to be just a matter of moving my hand. However, if moving my pawn
is not just a matter of my moving my hand, then what it is, and how is it related to my hand movement?
More generally, how can we maintain a causal theory of action on which some actions are not body
movements?
Given these worries, there is a familiar Davidsonian response. It begins by recognizing that
most actions and body movements do not seem to be the same kind or type of thing. For example,
‘commuting to work’ in general does not seem to be of the same kind as ‘moving one’s legs in a
circular motion’ or ‘walking.’ That is, there is no set of body movements essential or conceptually tied
to commuting to work, and so commuting to work is a type of action which is not identical to any
type (or series of types) of body movement(s). In this way, actions can admit types that are distinct
from types of body movements. Nevertheless, we might think that a particular instance or token of
my performing some kind of action is identical to my moving my body in some kind of way. So, my
commuting to work on one day is identical to my moving my legs in a circular motion as I bike to
work that day. By distinguishing between types and tokens, we can attempt to allay this concern about
how different some actions seem to be from movements. Again, the strategy is to acknowledge that
action-types may indeed be very different from types of body movements in general, but this does not
mean that particular instances of each type could not co-occur. Perhaps the same token event could
instantiate multiple types.
This response is initially compelling, but we should delve deeper to consider how we should
really understand the type/token distinction. After all, what are types and how are tokens related to
them? One way we might think of the type/token distinction is as between a property and an
instantiation of that property. There may be particular instances in which some property is instantiated
or exemplified, and this would be a token of that property, but more generally we could discuss that
property as a type. So, for instance, we might think of ‘being a commute’ as a property that many
different events could have, and we can distinguish between this set of events of commuting and one
particular instance of a commute as an event that has the property of being a commute. Understood
23
in this way, we can see how adverting to the type/token distinction is supposed to help. Davidson can
admit that ‘being a commute’ and ‘being a leg motion’ pick out different properties,
16
but he can claim
that these two properties are instantiated by the same event. There is only one event when we act, and
it is a body movement.
This distinction makes space for the possibility that these actions are all identical to certain
body movements, but it does not show us that this view is correct. Instead, it shows us where exactly
opponents are disagreeing with Davidson. Everyone will likely admit that there are certain action-
properties instantiated by those events that are actions (and similarly for body movements); the
question now is whether we should think that act-types are always instantiated by events that also
instantiate movement-types. What I will do in the next section is consider compelling arguments for
thinking that these properties are not instantiated by the same events. In the following section, I will
show how Davidson can respond to these arguments. This dialectic will reveal a larger lesson: Whether
we are convinced that Davidson is right or wrong may come down to how we are thinking about
events. Further, even if we are convinced that Davidson is wrong about this issue, this should not
push us to think that the central idea of the Causal Theory of Action is wrong.

Arguing for Distinctness
Not only do actions in generally seem different from body movements, but particular instances
of actions seem to differ in various ways from (and so not be identical to) particular body movements.
An argument from these apparent differences to non-identity uses Leibniz’s Law, and this style of
argument is common in the literature on metaphysics of ordinary objects. There, we might infer the
distinctness of a statue and the lump that makes it up from certain modal differences between them.
17

Here we are concerned with events, but the arguments will proceed in a similar fashion.
Consider a particular act of greeting and the arm-movement
18
associated with it. I want to
suggest that there are certain counterfactuals that are true when involving the greeting but not the

16
It’s worth noting that Davidson doesn’t even have to admit this much. Few believe that different predicates necessarily
refer to different properties, so he could maintain that even the properties are identical, despite our intuitions to the
contrary. If I am right in interpreting types in the type/token distinction as being a matter of properties (and I am not sure
that there is another way to interpret the distinction), then by adverting to the distinction Davidson is conceding that this
approach will not work. Instead, we will see that he is entirely relying on his particular view of events.
17
For example, see Fine (1982), Johnston (1992), and Baker (1997).
18
I will talk in terms of the arm-movement, but I mean to refer to whatever body movement Davidson or others might
argue is identical to my greeting. This might include my moving my face in particular ways to smile as I wave, for instance.
Others might include in this movement the muscle-contractions as my arm rises, or even certain mental events involved
24
arm-movement and vice versa. First, let’s fill out the story a little bit more. Suppose that Martha greets
Michael by waving to him because she believes that it would be polite to greet him and she wants to
be polite. Suppose further that after being greeted, Michael waves his arm in greeting in return. Let’s
also assume that Michael is new to town and slightly timid, so he will only greet people if they initiate
the greeting.
Even before considering counterfactual differences between Martha’s greeting and her arm-
movement, it is tempting to think that there are direct property differences between them. Martha’s
greeting may be warm or friendly, but it does not seem that the event of her arm-movement is warm
or friendly. Or, these events may come apart in what properties are had by them essentially. For
example, it may be essential to that arm-movement’s being the movement that it is that it occur at
exactly the speed that it does. However, if the greeting shares the property of occurring at that same
speed, it seems to only accidentally be that speed. This second difference points to a theme to which
we will return below, the idea that many actions do not seem to be as fragile as the movements to
which they are related. That is, there are more possible ways that the token action could have been
slightly different while remaining the same event than there are for the token movement. We need not
evaluate this idea yet, but I will claim that it in part motivates our intuitions concerning the differences
in true counterfactuals between the actions and the movements.
Moving on to counterfactual differences, let’s consider the counterfactual ‘If Martha had not
greeted Michael, then Michael would not have waved to Martha.’ That counterfactual sounds
intuitively true. After all, in the nearest world in which Martha had not greeted Michael, she likely
would have been doing something else, such as being lost in thought. Since Michael is timid and will
not initiate a greeting, we can see that in the closest world in which Martha happens not to greet
Michael, Michael will not greet Martha.
Now consider a similar counterfactual with the arm-movement: ‘If Martha had not performed
that arm-movement, then Michael would not have waved to Martha.’ That counterfactual seems false.
The closest worlds in which Martha not perform that exact token arm-movement are worlds in which
Martha moves her arm to wave in greeting to Michael, but in slightly different way (or with the other
arm). Given this, and that we might expect exactly how Martha greets to not influence whether
Michael perceives Martha as greeting him, it’s natural to think that Michael’s return wave would have

in coordinating eye-contact with you as you pass by. What is included in the body movement associated with the action is
an important question, but what matters for our purposes here is that the arguments given below should work for
whichever set of movements one takes to most plausibly be identical to a particular action.
25
proceeded whether or not Martha had performed exactly this arm-movement. We will examine what
this intuition turns on presently, but first let’s get on board a few more counterfactuals.
Moving on, consider a counterfactual that concerns what happens before the greeting/arm-
movement: ‘If the neurons and nerve impulses that actually fired prior to Martha’s arm-movement
had not occurred, then Martha would not have performed the same arm-movement.’ It’s not as clear-
cut that this counterfactual is true; that depends on how fragile we are taking the event of the arm-
movement to be. We might want to say that the same arm-movement would have occurred, but in a
slightly different way, or we might want to say that it is a different arm-movement because it occurred
in a different way (or, we might think that the upshot of that misfire would be that no arm-movement
would occur). Regardless of how we answer this though, we should all agree that something is different
about the arm-movement in the closest worlds in which the neurons that proceed it do not fire exactly
as they do.
However, I maintain that the same is not true for the corresponding counterfactual concerning
the greeting: ‘If the neurons and nerve impulses that fired prior to Martha’s greeting had not occurred,
then Martha would not have performed the same greeting.’ This counterfactual is intuitively false. To
avoid backtracking counterfactuals, we hold fixed that Martha still believes that Michael is walking
towards her and that it would be polite to greet him, and that she wants to be polite. Given this, it
seems clear that the closest worlds in which Martha’s neurons do not fire exactly as they do are still
worlds in which Martha greets Michael. Further, I don’t think there’s much pressure to say that it is
in any way not the same greeting. It would be recognized by Michael in the same way (he may not
perceive a difference), and it leads to the same waving by Michael in response.
We could suggest that it is not the same token greeting, because the arm-movement would
have been slightly different, but why should we think this? At the level of interpersonal
communication, the exact same thing was communicated (e.g. “Hello”). Further, it elicited the same
response. Michael may only perceive and respond to the fact that Martha greeted him at all, and not
that she did so by moving in some particular way. So, his response will only be counterfactually
dependent upon whether Martha greets him, and it will not be sensitive to whether she greets him in
some exact way or in a slightly different way.
If my assessment of these counterfactuals is correct, then different counterfactuals hold for
the arm-movement and the greeting. These modal differences alone can be used to directly argue for
26
the non-identity of the greeting and the arm-movement.
19
If they do not share all of the same
properties (including modal properties), then they cannot be identical. However, we can also recognize
that these modal differences seem associated with intuitive causal differences between the greeting
and the arm-movement.
The greeting is caused by the belief that it would be polite to greet Michael and the desire to
be polite. This is reflected in the fact that in the counterfactuals that if these mental states differed,
then Martha may not have greeted Michael. So, Martha’s greeting depends for its occurrence on these
mental states. I take no stance here on whether these mental states are identical to or merely closely
related to certain neurological events in my brain as I have them,
20
but it seems true regardless that
Martha’s greeting does not depend on those precise neurological events (or events in my nervous
system proximately caused by those states) in the same way that it depends on the mental states.
Similarly, Michael’s wave counterfactually depends on and so seems caused by Martha’s greeting and
not her arm-movement. So, Martha’s greeting at least has effects not shared with her arm-movement,
and her arm-movement has causes not shared with her greeting.
21
Given that the greeting and arm-
movements are events, and it is not uncommon to individuate events on the basis of their
causes/effects,
22
we can again conclude that the greeting is distinct from the arm-movement.
These considerations depend on the connection between counterfactuals and causation, and
the relationship between these is controversial. It might be possible for events to counterfactually
depend on things which are not causes,
23
and it well-established that events can have causes that they
do not counterfactually depend on (at least, for their existence).
24
Nevertheless, what is crucial here is

19
In fact, there are many more apparent modal differences besides these that suggest the greeting is distinct from the
arm-movement. We can imagine worlds in which Martha greets Michael, but perhaps by using my other arm, for instance.
Or, we could imagine that Martha falls asleep and moves her arm just as she does in the actual world, although this would
not be her intentionally greeting.
20
Notice, however, that if these mental states are distinct, then we will not be able to say that they cause Martha’s arm-
movement without overdetermining it, since Martha’s arm-movement will have a sufficient physical cause from within her
body. So the arm-movement and the greeting would in fact not share any causes in this case.
21
There could be more causal differences besides these. As noted in the last footnote, the action and movement may not
share any causes if mental events are not reducible to physical events. Further, there will be many minute physical effects
of Martha’s arm-movement that it sounds odd to say are also effects of her greeting. It may be that they ultimately are,
and the counterfactuals might support this (if she hadn’t greeted Michael, the light would not have reflected off of her arm
just as it did), but there is something odd about this. For instances, it seems like there are many worlds in which Martha
performs the same greeting, but the light does not reflect just as it does in the actual world.
22
This is how events are individuated in Davidson (1969/2011), and it will be discussed more below. We could advocate
for a view on which more than one event could have the same causal profile, but this would also not help the Davidsonian
in maintaining the Identity Thesis.
23
For example, effects may depend upon some state of affair’s obtaining in nearby worlds, but we may be persuaded that
states of affairs are never themselves causes.
24
This is what is established by the literature on causal preemption cases – cases in which one event seems to clearly cause
another, but there is some other event that would have caused the effect if the actual cause had not occurred.
27
not whether causation is analyzable in terms of counterfactual dependence. Rather, we only need it to
be likely the case that when one event counterfactual depends upon another, then the latter is a cause
of the former. If we have this much, then we can say that the modal differences I pointed to are
correlated with or evidence for causal differences between the actions and the movements.
25

Now that we see this connection between these counterfactuals and causal facts, we are in a
better position to appreciate what this argument from the difference in counterfactuals turns on. What
is doing the work is an appeal to a difference in the apparent fragility of the greeting and the arm-
movement. Events can be more or less fragile depending on the kinds of changes they can survive.
My argument turns on the intuition that bodily movements are typically more fragile events than
actions (at least, than the act of greeting, in our example).
Bodily movements might be fragile enough as events that any difference in the speed of the
arm-movement would make it a different movement. This sounds reasonable, because particular
physical effects of the arm-movement (such as the reflection of light from Martha’s arm) are
counterfactually sensitive to and seem to be causally explained by the exact speed at which her arm
moved. If we think that events are individuated by their causal profiles, then the fact that the arm-
movement has effects that are sensitive to the exact speed at which it occurs indicates that the event
will be fragile enough to be essentially that speed. On the other hand, the event of Michael’s waving
does not seem sensitive to the exact speed at which Martha’s arm moves as she greets him. She could
have greeted him at a slightly slower rate, and it may have caused his response just as it does in the
actual world. This indicates that the greeting is less fragile than the arm-movement, because its effects
are not as sensitive to slight changes in its properties. In general, our actions and reactions can only
be sensitive to differences that we can perceive and possibly control, but our physiological bodily
movements may be sensitive to many more microphysical changes.
With these arguments on the table for the distinctness of Martha’s act of greeting from her
body movement as she greets, we can see how these arguments can generalize to many of our actions.
For instance, the way that I raise my arm when I vote on a measure is caused by muscle contractions
produced by nerve impulses originating in the way my neurons fire in my brain, but it’s plausible to
say that if my muscles had not contracted exactly as they had, then I still would have voted. This modal
difference suggests that the voting does not share (at least all of) the physical causes of the arm-

25
We might further maintain that these modal differences are in fact grounded in these causal differences. I take no stance
here on whether the grounding relation obtains here or on the status of the grounding relation more generally. However,
insofar as we might think that modal differences between the action and the movement must be grounded in some actual
difference between them, it is suggestive that these counterfactuals pattern with the causal profiles of these events.
28
movement. Further, many actions have effects that do not seem caused by the agent’s token body
movements as they actually occurred. For instance, a student could be deterred by my proctoring
regardless of my exact pattern of walking as I proctored. That is, I could move in a completely different
way (or remain still in the exam room), and the student may be deterred from cheating, but the student
may not be deterred if I were to not proctor.
It may be feasible to use these kinds of arguments to conclude that all actions are distinct from
body movements. It certainly seems as if certain actions such as my wiggling my toe is identical to my
body movement as I wiggle my toe (in this case, my toe-movement). Though perhaps we could find
similar counterfactual differences to suggest that they are in fact distinct. For example, it seems like I
could have moved my toe just when I did but as a reflex or spasm. Alternatively, we may focus on a
case where the action seems exactly as fragile as the movement, such as my walking on a tightrope. If
I were to move in any different way, then I would fall off of the rope and not be tightrope walking,
so how I move appears to be essential to my tightrope walking. Even here though, there are apparent
differences to be found. I may have to be careful, but there is not literally one way to move to
successfully walk across the tightrope. What’s more, I could have moved just as I did without it being
a tightrope walking (if I had been on solid ground instead of a rope). We may even want to say that
my action has the property of being daring, while there is nothing particularly daring about my moving
just as I did as I walked across. So, we can advert to other differences still between actions and body
movements when the action appears to be just as fragile as the movement.
These arguments are compelling, and it would be provocative if no actions can be identified
with body movements. However, this is simply a stronger claim than we need for our purposes here.
My aim is just to show that the means exist to defend a version of the causal theory of action even if
we must reject the thesis that all actions are identical to body movements. Establishing the claim that
all actions are distinct from body movements is not only more difficult to do, but it is not necessary
to achieve this aim. Instead, I will move forward allowing that some actions are body movements,
although what I will say of those actions that clearly are not would carry over if we thought that all
actions are distinct from body movements.
In the next section, I want to show how there is a package of views that will allow a
Davidsonian to answer the arguments from this section. However, I will contend that this package of
views is not only implausible, but unnecessary. If we consider how these actions non-identical to body
movements are related to body movements, an alternative view presents itself which fits nicely with
the causal approach.
29
How a Davidsonian Could Respond
These arguments are compelling, but there is a package of views that will allow a Davidsonian
to reject them. What will be important from our perspective is that even if we disagree with the
Davidsonian, rejecting this package of views seems unrelated to whether we should endorse a causal
view of action more generally. Once this is made clear, we will be in a good position to see how we
might reformulate a causal view even given a rejection of the Identity Thesis.
First though, we should recognize how a Davidsonian could respond to the kinds of
arguments given above, and how that response is natural given Davidson’s antecedent views. To
rehearse the arguments again, we began by noticing that actions (in our example, Martha’s greeting)
seem to have properties not shared with body movements. Second, we found intuitive differences in
counterfactuals between actions and body movements. This alone provides a modal argument for
their distinctness, but these counterfactuals were evidence of causal differences between the actions
and the body movements as well.
A Davidsonian response to these arguments begins with adopting Davidson’s preferred view
of events. Although Davidson (1969/2011) at first purposed that we individuate events entirely by
their causal profiles (so events were identical if they had the same causes and effects), he later came to
repudiate this view of event-individuation. Instead, in Davidson (1985/2011) he accepted the view
given in Quine (1960) that individuates events purely based on their spatio-temporal locations. On
this view, two events are distinct only if they differ in the space-time region in which they occur, and
everything that happens within a given region of space-time is part of the same event. This view is
extremely coarse-grained, because it does not distinguish events in any way besides their locations,
and it trades on the intuition that events are concrete particulars that may involve several aspects.
With this view of event-individuation, the view on which actions and body movements are
identical events falls out very naturally. For any action, there will be some corresponding body
movement that occurs in just the same spatio-temporal region,
26
and they will be identical events
because of this. Davidson recognized that events have different properties – the same token can admit
of different types. So he can say that one and the same event could instantiate both action and
movement properties, and one event could correctly be described as both an action and a movement.

26
Whether all actions actually are colocated with body movements is controversial, and the question of whether some
actions are located beyond the body will be taken up in chapter 3. There, I will argue that the linguistic evidence does not
compel us to locate actions beyond the body.
30
This gives a Davidsonian the resources to very quickly answer the challenges from the apparent
difference in properties and causal profiles.
It may seem as if Martha’s greeting is friendly while her arm-movement is not, but this intuition
will simply be mistaken on Davidson’s view of events. The arm-movement in fact is friendly, because
it is identical to the greeting, and the greeting is friendly. A Davidsonian could further explain away
our intuition to the contrary by suggesting that greetings in general (or ‘greeting’ as a type) are often
thought to be friendly. We may not associate the type ‘arm-movement’ with the property of being
friendly, but nevertheless arm-movements are friendly when they are friendly greetings.
Similarly, this view of events immediately suppresses any argument from a causal difference
between the action and the body movement, as long as we maintain that events are the causal relata
(the things that cause and are caused). An action and the corresponding body movement will be the
same event, and so they must have the same causes and effects. An event may instantiate the property
of being a greeting and the distinct property of being an arm-movement, but if it is a singular event,
then it will have only one set of causes and effects. We may want to say that that event causes in virtue
of having one or another property (in virtue of its being an act of greeting), but this still leaves the
events themselves as having all of the causes and effects of both movements and actions.
27

This view of events and the causal relata will be sufficient to answer the arguments from a
difference in properties and causes/effects, but a Davidsonian will need to appeal to yet more to
answer the argument from counterfactual differences. In order to answer this argument, Davidson
would seem to need to deny the truth of very intuitive counterfactuals. Crucial to my argument was,
for example, the intuition that it is true that if certain events in my body had not occurred, I would
not have performed the same body movement, but false that I would not have performed the same
action. Given the case of the greeting in particular, it just seems like the same greeting would have
occurred (holding fixed Martha’s reasons for greeting).
In order to answer this argument, there are a few options. The least revisionary option would
be to assert that the truths of the counterfactuals follow the causal facts, and we already have an
explanation for why the action and the movement have all and only the same causes/effects. We saw
in the last section that these counterfactuals seem correlated with certain causal claims and likely tied

27
It’s not clear what it would mean to say that events cause each other in virtue of the properties they instantiate. Events
on this view are concrete particulars that contain a multitude of property instances within a region of space-time. If these
are the things that cause (the events), and they cause each other, then how is it that the causation is explained by the
properties? Whether this coarse-grained view of events really sits well with understanding events as the causal relata is
explored more below.
31
to them in some way. If our intuitions about counterfactuals is explained by our causal intuitions, for
example, then our judgments about the counterfactual cases would be mistaken because our causal
intuitions are mistaken. This response does not call for accepting anything beyond some connection
between our causal and counterfactual intuitions, which seems plausible. However, it does have the
upshot that our intuitions are often entirely mistaken in these counterfactual judgements.
Alternatively, a second response would be to accept that the identity relation can be contingent
(ala Gibbard 1975) and is contingent in this case. If we accept this, then we can say that the action and
the movement are identical in the actual world, but that there are worlds in which they are not identical
(and one or the other may not occur). This is an uncommon view to hold, but it does have
proponents.
28
Still, holding onto the Identity Thesis does not require accepting contingent identity,
29

and this route goes very poorly with Davidson’s view of events. If events are individuated by their
spatio-temporal regions, then whatever happens in the same spatio-temporal region in different possible
worlds will be the same event. So, the event of Martha’s greeting could have been an event of air moving
slowly (in fact, this will be the case in many of the closest worlds, since there are many worlds in which
Martha does not move through exactly the same region that she does in the actual world).
Given this, and given the implausibility of transworld identity more generally on this view of
events, I think it is much more likely that a Davidsonian would instead offer a counterpart theory of
events.
30
On this view, events in the actual world only exist in the actual world, and events in other
possible worlds can only be more or less similar to events in the actual world. How similar events are
across worlds will depend on the description we use to pick them out. For example, there may be a
single event that is the arm-movement and the greeting, and it can be described as either an arm-
movement or as a greeting. Considering that event qua arm-movement, there will be a certain
collection of worlds with similar events that fit the description of being an arm-movement.
Alternatively, considering the event qua greeting, there will be a collection of worlds with events that
have similar greetings occurring in them. The key here is that although the arm-movement may be the
greeting, which collection of worlds is taken to involve counterparts to the event in the actual world
will depend on whether it is described as an arm-movement or as a greeting. Put another way, the
description of events is what determines which events in other worlds are similar.

28
Gibbard (op cit.) appealed to contingent identity in the case of the statue and the clay, so it would be natural to extend it
to the case of action. For a discussion of the view’s motivation and proponents, see Schwarz (2013).
29
Francescotti (2005) argues for this explicitly.
30
This would be in the spirit of the counterpart theory of objects given in Lewis (1968, 1971).
32
Using this framework, a Davidsonian has the resources to both maintain the Identity Thesis
and deliver the right intuitions regarding the counterfactuals. Martha’s arm-movement and greeting
only occur in the actual world for the counterpart theorist, but they could evaluate counterfactuals by
determining the counterpart events based on the descriptions used. For example, we said before that
it seemed true that ‘If Martha had not greeted Michael, then Michael would not have waved to Martha.’
The counterpart theorist can look to worlds where Martha passes Michael and there is not an event
that can be described as Martha’s greeting Michael, and she can see that in these worlds there also fails
to be an event appropriately described as Michael’s waving to Martha. Similarly, the counterpart
theorist can also confirm the falsity of the counterfactual: ‘If Martha had not performed that arm-
movement, then Michael would not have waved to Martha.’ In close by worlds in which there is no
event that can be described as (and so similar to) Martha’s arm-movement, it may well be that there is
an event similar to Michael’s waving (if Martha had greeted in a slightly different way in those worlds).
By accepting counterpart theory for events, a Davidsonian can accept our counterfactual judgments
without concluding from them that actions are distinct from body movements.
So, there are ways of maintaining the Identity Thesis. That being said, I do not think the bundle
of views that best supports the Identity Thesis is very plausible. There is a tradition of philosophers
extending back at least to Kripke’s Naming and Necessity that takes us to be able to specify how one and
the same thing could have existed in slightly different ways in different possible worlds. I subscribe to
this tradition, and it compels us to reject counterpart theory. Further, although Davidson was
motivated to accept his view of events, it does seem too coarse-grained to individuate events based
solely on their spatio-temporal location. We regularly talk about two things happening in the same
place. If events really are to be the causal relata, then their causal profiles would seem more relevant
to their individuation than their locations.
As an alternative, suppose we accept a view of events that exactly tracks our causal intuitions.
For example, we might accept the view on which events are sui generis entities on a par with objects
and properties and not to be reduced to what happens in space-time regions, or properties of space-
time regions,
31
or the exemplification of properties by things at times.
32
On this view, events are best
thought of as occurrences, or things that happen. They certainly will involve objects and properties,

31
See Lewis (1987) for this view of events.
32
See Kim (1976) for this view of events. We will discuss this view in more detail in chapter two, where our question will
concern how we might think of omissions as events and as fitting into the CTA. Although I find the view I endorse below
to be more plausible than the Kimian view, it is more convincing that omissions are events if they can be shown to be
events on this respectable account.
33
and they will occur in regions. However, they will not be identified with any of these items. This view
of events is more fine-grained than Davidson’s view, and it does not seem to be too fine-grained to be
plausible. On this view, actions and movements will be events. Further, actions and movements may
occur in the same place without our being compelled to identify them. Instead, we can be guided by
our intuitions concerning the differences in counterfactuals and causes between actions and
movements to conclude their distinctness as events.
This is not the place to argue for or against a particular view of events or of transworld identity.
Rather, my aim is to show that plausible views lead us to deny the Identity Thesis. More importantly
though, accepting these views need not recommend rejecting some version of the CTA. The standard
view of transworld identity and my above-sketched view of events are not so exotic as to suggest that
actions are not to be analyzed in terms of their causal history. After all, these are just views concerning
modality and event-individuation. These are not the sorts of changes that, by themselves, should lead
us to question the fundamental intuition behind a causal approach.
33
If actions are not always
movements of the body, this does not automatically put pressure on the idea that the CTA is
fundamentally misguided. There still seems to be something causally distinctive about actions, or we
may still think that reasons cause actions and that events are actions in virtue of this.
So there is cause for optimism for the proponent of the CTA. There is a package of views that
would allow the Davidsonian to maintain the Identity Thesis. However, if we do not find this package
compelling, as I do not, then having to reject the Identity Thesis will not require or even suggest
abandoning the causal approach. Instead, we should suspect that there is a way to reformulate the
CTA. Given this, I want to pursue how we might reformulate the CTA if we are persuaded by the
arguments given and taking on the more fine-grained view of events suggested above. If not all actions
are body movements, then we must find another way to pick out all and only those things that are
actions. In the next section, I will consider how actions are related to body movements when they are
not identical to them, and this will naturally suggest a new formulation for the Causal Theory of
Action.

33
Contrast this with problems for the CTA that come from accepting a different non-causal view of the nature of reasons
explanations, for instance.
34
If Not Identity, then What?
If we do not accept the package of views necessary to defend the Identity Thesis, then we are
left with the modest conclusion that some actions are not movements. This is to say nothing, however,
of how many actions are not movements (if not all of them), of what kinds of actions are not
movements, of what actions are if they are not movements, or of how these actions are related to
movements.
For all that I have said, we can still assume that all actions are events. Some philosophers have
rejected this.
34
The views they offered instead are interesting, however, none of the arguments that we
gave against Davidson’s view were of a sort to persuade us that actions are not events. So, I will
continue to assume this.
Further, although it will the topic discussed later in the dissertation, it may still seem that
Davidson’s view about the location of actions is correct: all actions are at least colocated with certain
body movements even when they are not identical to them. After all, insofar as we see actions, we
seem to see them happening just where our body is moving. For example, if I vote on a measure by
raising my arm, someone watching me vote might claim that they saw me voting just where my arm
went up. Some of the biggest proponents of other views in the philosophy of action are motivated in
part by a denial of this claim of colocation. These kinds of views deserve serious consideration,
35
but
in this chapter we will just take on board the idea that actions that cannot be body movements are
nevertheless located just where the body moves.
Beyond this, we can still appreciate how actions depend on body movements in many ways
even if they are not in all cases identical to those movements. At the very least, our actions seem to
supervene on our movements. We cannot perform many actions without moving our body in specific
ways, and there is no difference in how we perform those actions without some difference in our
movements. Further, in many cases, it may be that the success of our action depends on the exactness
of those movements. The success of a dance may depend on the pointedness of the toes, for instance.
Or whether a basketball defender can block a shot successfully without fouling depends on the precise
location of their hands.

34
Von Wright (1963), Bach (1980), and Bishop (2010) all deny that actions are events. Each of them is motivated in part
by a rejection of identifying actions with the events that those actions result in, although they give slightly different pictures
of what actions are instead and different reasons for adopting those pictures. It is worth noting that Bach and Bishop both
maintain that their views can nevertheless be reconciled with some version of the CTA.
35
Chapter three is concerned with arguments made by those that claim that actions extend out beyond the body. I deal
with those views in particular, because this presents a greater challenge for a proponent of the CTA.
35
More directly, actions seem to inherit many of their properties from the corresponding
movements. That is, actions seem to share in or have certain properties in virtue of the movement’s
having those properties.
36
The speed of the 40-yard dash depends on the speed of the athlete’s legs,
and we might think that the dash has its speed in virtue of the speed of the athlete’s legs. Similarly, the
strength of the powerlift depends on the strength of the arms and back movements. As another
example, the coyness of the wink depends on the exaggeratedness of the face and eye movements.
The cases go on and on.
37
In general, even if certain actions can be shown to be distinct from
movements of the body, they will seem to inherit many of their physical properties (how they appear,
the speed at which they occur, etc.) from those body movements.
In contrast, we do not take movements of our body to depend in the same ways upon our
actions. The speed of the leg-movements does not depend upon the speed of the run, for instance.
That being said, how we move when acting may depend on how an action needs to occur. So, how
exaggerated our face contorts as we wink may depend on how coy we want the wink to be. But the
coyness depends on the exaggeratedness of the movements, not vice versa. Further, actions don’t
seem to depend upon themselves in the same way that they depend on the body movements to which
they are closely related. The run does not inherit its speed from its speed. If we put all of these thoughts
together, then we can realize that the relation of dependence between actions and our movements is
asymmetric and irreflexive.
Once we have gone this far in characterizing the relationship between actions and movements,
we may have said enough to reformulate the CTA. Actions, on this account, will either be body
movements or else events colocated with and asymmetrically dependent upon certain body
movements caused in the right way by certain mental states. This reformulation gets around the above
arguments to the Identity Thesis, and it preserves the intuition that events are actions in virtue of their
causal history. Still, I think we can say more about the relationship between actions and body
movements (when they are not identical), and I think doing so will allow us to reformulate the CTA
even more succinctly.

36
It is not clear whether we should say that the actions share or have in common some of the exact token properties of
the movement or if we should say that the action has the property that it has in virtue of the movement’s instantiating a
property of the same type. What I say here should not turn on how we understand property dependence, although
proponents of the constitution relation to be discussed below may have reasons for preferring the former view.
37
I claimed above that some actions were distinct from body movements precisely because they did not share some of
these more fine-grained properties. However, the real difference between actions and body movements is not that they do
not share these properties (though they may not) as much that it is that actions may have some of these properties
accidentally, rather than essentially. This is how actions can depend on movements for these properties without it being
true that these same actions could not have existed without having these fine-grained properties.
36
For all that we have said, we might be skeptical that there is a relationship with these
characteristics that could obtain between the action and the movement. It may be that in the past the
case for actions as movements, combined with there being no clear alternative to the relation of
identity that might obtain between the two, explained the force of the Davidsonian view. Alternatively,
it might be that in the past those dissatisfied with this view felt compelled to reject the claim that
actions are even colocated with body movements or that they are events at all. Now, however, we can
realize is that the relation between actions and body movements might be similar to other relations
that have been much more thoroughly discussed, characterized, and legitimized in metaphysics. In
particular, it can seem that the way in which my act of greeting is related to my body movements as I
greet is very similar to the way in which the statue is related to the lump of clay that makes it up.
This relation between the statue and the clay has received a lot of attention in the past few
decades in the metaphysics of ordinary objects.
38
When examining a statue, we may think that it just
is the lump of clay that somehow makes it up (after all, we would take ourselves to just be looking at
one thing); however, there are good reasons for thinking that the statue is distinct from the lump of
clay. Many of those reasons carry over to thinking that actions are distinct from body movements (e.g.
causal and modal differences between the statue and the lump).
Once we distinguish the statue from the lump, we can see that their relationship shares many
of the features of the relation between the action and the movement. First, they are colocated, distinct
objects. Second, the statue depends on the lump of clay for many of its properties (for example, its
weight, height, solubility, color, etc.). Further, he way that it depends on the lump for these properties
is immediate as opposed to causal (in fact, we might even think that the lump and the statue
simultaneously share these properties
39
). What’s more, we can see that the statue doesn’t depend on
itself in the same way that it depends on the lump, and that the lump also does not depend on the
statue in the same way that the statue depends on it.
40, 41
So, the relation between the statue and the
lump also appears to be one of synchronic, asymmetric dependence.

38
Wiggins (1968), Lowe (1983), Thomson (1998), Baker (2000), and Fine (1982, 2003) among others have defended this
idea that distinct objects can be colocated and stand in an asymmetric relation of dependence distinct from identity.
39
See Paul (2007) for this suggestion. I mentioned above that actions and body movements may share many of their
properties as well.
40
This doesn’t rule out the possibility, however, of there being ways in which the lump does depend on the statue. Perhaps,
if it weren’t for the beauty of the statue, the lump would have been destroyed long ago, for example. Still, the dependence
between the lump and the statue is asymmetric, in that they do not depend upon each other in the same ways.
41
Beyond irreflexivity and asymmetry, there is a question as to whether or not the relation between the statue and the
lump is transitive. It may seem plausible to say that if the statue depends on the lump in these ways, and the lump depends
on some further object (a composition of atoms perhaps) in just the same ways, then the statue will seem to depend on
this further object in the same ways as well. Whether or not the relation of constitution should really be thought of as
37
In the literature on material objects, the relation that is taken to obtain between the statue and
the lump is known as ‘constitution,’ and we would say that the statue is constituted by the lump. Given
how similar this relationship is to the relation between certain actions and body movements, we might
wonder whether it is appropriate to think that the relation between them is the same relation, or whether
certain actions are constituted by body movements. It sometimes gets said in the philosophy of action
literature that action is constituted,
42
but this is rarely with the relation of constitution as discussed by
metaphysicians in mind.
43
However, there are several clear examples of metaphysicians who took their
views of constitution to be directly applicable to the case of action.
Fine (1982) gives a hylomorphic view of objects and a compositional conception of
constitution in response to similar identification concerns about ordinary objects, and he explicitly
extends his view to distinguish actions from movements. On his view, the statue has the lump as a
proper part, and the statue is a composite of the lump and the shape of the lump. Although inspired
by Aristotle, Fine’s view goes far beyond matter and form. For any object and way of predicating a
property of that object, there will be a composite object that is the object-qua-that property.
44
So, if I

transitive is controversial. For example, Baker (2000) takes constitution to not be transitive, but then Baker (2002a) reverses
her position in response to the criticism in Zimmerman (2002). However, see Wilson (2009) for a thorough and
illuminating discussion on this question of the transitivity of material constitution. We can accept the verdict there that
the different senses of constitution are transitive (once disambiguated). I omitted a discussion of whether the relation of
dependence between actions and body movements is transitive, but it does seem like it might be. If I vote on a measure
by raising, then the speed at which I vote will depend on the speed at which I raise my arm, which will depend on the
speed at which my arm rises. If constitution is transitive, then I can maintain my stance that all actions are constituted by
body movements; whereas rejecting transitivity would perhaps involve claiming that some actions are constituted by body
movements and non-basic actions are constituted by other actions.
42
For example, Bishop says, “…the [Causal Theory of Action] is the thesis that the concept of event, cause, and intention
are adequate to provide a suitably naturalistic ontological account of what constitutes intentional action…” (2010:74). However,
Bishop clearly does not have the relation of constitution as distinct from identity in mind when he says this. As another
example, Brian O’Shaughnessy concludes, “…we are forced back upon thesis 4, the doctrine that the act of raising an arm
is a complex event constituted out of two causally linked simultaneous events that were ‘made for each other.’” (1973:385).
But O’Shaughnessy is clearly here thinking of constitution in a way that is closer to composition.
43
However, there are two clear exceptions to this. Fine (1982) gives a hylomorphic view of objects in response to similar
identification concerns about ordinary objects, and he explicitly extends his view to distinguish actions from movements.
Similarly, Lynne Baker is a more general proponent of distinguishing ordinary objects from the matter that makes it up,
and in Baker (1998) she extends her view to cover actions in an attempt to solve the causal exclusion problem for non-
reductive physicalists in the philosophy of mind. Both Fine and Baker’s views will be discussed more below. Another
notable exception, and the only exception I could find of someone who is primarily a philosopher of action, is Schlosser
(2009). There, Schlosser at one point entertains the idea that actions are constituted also as a means of helping to solve
the causal exclusion problem.
44
The view thus stated is implausibly plenitudinous (in that it licenses the existence of many objects). However, see
Fairchild (forthcoming) for an argument that this and similar plenitudinous views are inconsistent and must be restricted
in some way.
38
have a shirt that is blue, then there is an object that is a shirt and an object that is a shirt-qua-blueness
(of course as well as the matter that makes up my shirt qua-shirtness).
45

Far from being just a view about objects, Fine recognizes that events can instantiate properties,
and so he concludes that there will be further qua-objects that are those events-qua-properties (101-
103). Recognizing that there are modal arguments that suggest distinguishing certain actions from
body movements, he suggests that actions are just qua-objects with the body movements as a base or
‘matter’. So, my greeting will be an arm-movement-qua-greeting.
46
Insofar as he takes pains to
recognize the different descriptions that movements can be under, in some sense this view is very
close to Davidson’s view. However, instead of claiming that one movement may be described in many
ways, Fine claims that one movement may be a proper part of many events based on the ways in
which it can be described.
Second, Baker (1998) extends the relation of constitution to the case of action. Baker (2000)
gives a detailed modal analysis of the concept of constitution that we will discuss more below. In short,
her account relies on the idea that something could exist alone, but it has properties such that if it is
in the right circumstances, there will be a distinct object colocated with it. In her earlier paper, she
suggests that the concept of constitution can be applied directly to the case of action in order to say
what the difference is between an action and a body movement (thereby answering Wittgenstein’s
question, as given in fn.1 of the Introduction).
47

I will not take a stance here on the right view of constitution more generally. There are other
views of constitution besides these, and one of them may be a better model of constitution or for
action in particular. Neither will I attempt to further defend the claim that constitution is distinct from
identity.
48
Instead, I will take it for granted that there is such a relation as material constitution, and I
take the work of Fine and Baker to strongly suggest that this relation extends to the case of action. If

45
It is worth acknowledging that this view has gained support and been further refined in the past few decades. See
Koslicki (2008), Sattig (2015), and Evnine (2016).
46
Fine would actually say that the property element of this qua-object is being intentional (102), since the arm-movement
was performed with a certain intention and can be described as intentional. This is reminiscent of how Davidson conceived
of actions as body movements that can be given an intentional description in Agency.” However, given that several distinct
intentional actions could be performed with the single body movement (e.g. raising my arm and voting on a measure), I
do not think that the property of being intentional is the one that is a part of the qua-objects of these actions.
47
Even more controversially, Baker argues that recognizing that actions are constituted rather than identical to body
movements avoids the causal exclusion problem for nonreductive physicalists in the philosophy of mind. If actions are
distinct from body movements and mental events are distinct from events in the brain, then mental events can cause
actions and brain events can cause movements, and there is no overdetermination of body movements by mental events.
There are still problems with this kind of solution, and this issue is too big to discuss here; however, see Gibbons (2006)
for a similar ‘intralevelist’ solution to the causal exclusion problem.
48
Though see Johnston (1992) and Noonan (1993) for an early back on forth on this issue.
39
the relation between actions and movements is the relation of constitution, then we can reformulate
the CTA and perhaps even shed light on the nature of action.
There is a residual question I would like to acknowledge before moving on. The question is
whether the relation between actions and movements really is the same relation as the relation between
the statue and the clay, or merely an analogous relation with similar structural features, and whether
this matters. After all, constitution is a dependency relation, and several other dependency relations
only take one kind of relata. Grounding is only between facts; causation is only between events. So
should we be worried that the constitution relation only obtains between objects?
I do not see a reason why constitution would only obtain between objects and not events, but
the more important point to bring out for our discussion is that it does not matter whether we take
the relation between actions and movements to be constitution or some analogous relation.
49
Settling
this matter is of less significance because constitution is a technical notion, a philosophic term of art
that’s meant to classify a certain kind of structural relation: relations of synchronic dependence
between colocated items that are irreflexive, asymmetric, and transitive.
What is important to see is that objects when constituting aren’t really doing anything that
events would not be doing when being in a very similar relation. That is, constitution as a relation is
not quite like the relations of grounding or causation, which, regardless of how ‘out in the world’ these
relations are, at least have a lot of explanatory power. Facts that ground explain in virtue of what
certain other facts obtain, and events that cause explain why certain events occur. Constitution,
however, does not play this kind of explanatory role. So there’s no reason to think that objects that
constitute are doing or explaining anything that events that seem to constitute are unable to
do/explain. Since events can be similarly colocated and feature this brand of asymmetric dependence,
it again seems that there should be no prohibition on thinking that events are just as able to constitute
as objects.
The more important question is whether it is helpful to say that actions are constituted. This
comes down to whether the frameworks that have been developed to think about constitution can be
appropriately applied and useful for thinking about action. In the next section, I will show that they
can.

49
Although Fine himself goes as far as saying, “It is not just that the two relationships are analogous; they are the exact
same” (op. cit.:102).
40
Why Constitution Works Especially Well for the CTA
If actions are either body movements or constituted by body movements,
50
then we have the
makings for an easy revision to the original CTA. We can reformulate it as follows:

Revised Causal Theory of Action: Something is an action if and only if it is a body movement or an
event constituted by a body movement that is caused by certain mental states in the right way.

That constitution allows for such a simple emendation of the CTA recommends it,
51
but there are a
few other advantages to using the constitution relation in this account. As we will see towards the end
of the next chapter, not only do many regular actions seem to depend on body movements in a way
that suggests that they are constituted, but omissions are arguably colocated with and constituted by
body movements as well. Legitimizing omissions as events colocated with body movements is largely
the task of the next chapter, but, if this can successfully be done, then the relation of constitution can
be used to fit omissions into the CTA.
A second advantage to the view that certain actions are constituted comes from thinking about
how certain actions are related to one another. We very often take ourselves to be able to do more
than one thing at a time, and we frequently take ourselves to be doing one thing by doing another. For
example, I may ride by bicycle by moving my legs in a certain way and balancing. Further, I may
commute to work by riding my bicycle. In cases such as these, there is a longstanding question about
how to think about the relation between these actions, the by-relation.
52

Davidson of course took each of the actions related in this way to be identical to the agent’s
body movements, although those movements could be picked out via different descriptions. However,
we could use arguments of the type given above try to show that both the bicycling and the commuting

50
Again, it may be that the above arguments can be used to show that no actions are identical to body movements. That
possibility is satisfied on the below account (assuming, in that case, that body movements would never have the right
causal history to count as actions). I remain agnostic about this possibility, because the aim is merely to find a revision of
the CTA that avoids the objections to Davidson’s CTA given the Identity Thesis.
51
It is worth acknowledging that Fine’s view would not allow for this revision to the CTA. According to Fine, it is not the
action as a qua-object that is directly caused by mental events; rather, the action’s matter, the movement, is caused by
mental events. Actions inherit those causes from their movements. If the spirit of the causal view necessitates that actions
themselves are caused by reasons, then Fine’s view will not be able to accommodate this. However, if what matters is that
reasons-explanations involve causation (but may have other elements), then he could say that something is an action if it is
an event constituted by a body movement where that movement is caused in the right way by the right mental states. As
we will see below, Baker’s view does allow actions themselves to be caused by mental states (this is crucially for her), and
I prefer Baker’s view of constitution.
52
The classic discussion of how to think about cases like this is in Anscombe’s Intention. There, she advocates for the view
that the actions referred to on either side of the by-locution are identical.
41
are distinct from my body movements. What’s more, we could go one step further to show that they
are distinct from one another as well. In the closest world in which I did not want to ride my bike to
work, I might have not ridden my bicycle, but I may still have commuted to work by train instead.
53

If the by-relation is not identity, then there is a question about what it is, and the constitution-theorist
has an easy answer. We can argue that the by-relation is constitution. More formally, we could say:

If someone F’s by Y-ing, and F-ing and Y-ing are actions, then the action that is Y-ing constitutes the action that is
F-ing.
54

Once we have argued that two actions are distinct, we need only show that they are colocated,
and that one depends for many of its properties on the other in just the same way that the statue
depends on the lump (or that certain actions depend upon body movements). I do not have the space
here to fully make the case that the by-relation is that of constitution, but it is suggestive that it is
standard in the literature to take the by-relation to be asymmetric and irreflexive. Further, it is also not
hard to claim that my commute occurs just where my cycling occurs, nor to imagine that many
properties of my commuting (such as that it is relaxing) will depend upon my action of bicycling
(which may in turn depend on the speed and exertion of my leg-movements). So, the relation of
constitution offers us not only a way of revising the CTA but a way of understanding the relation
between certain actions.
A final advantage to bring out for the view that many actions are constituted is that at least
Baker’s account of constitution fits nicely with the primary causalist intuition – that events are actions
in virtue of how they are caused. To see this, let’s consider in more detail the account of constitution
given by Lynne Baker in Persons and Bodies. The fundamental suggestion regarding constitution there is
that things that constitute do so when in certain circumstances, that it’s the object’s relation to these
circumstances that make it constitute a different kind of object. The lump may constitute the statue
when it’s in an artistic community such that it can be admired or when it’s made by an artist; a piece
of plastic may constitute a driver’s license in a community with certain standards and conventions for

53
For example, Goldman argues that actions in this relation can be shown to be different actions by appealing to their
distinct causes and effects (1971:765-767).
54
Some have argued that the by-relation does not relate actions or events at all. See Hornsby (1980:7-8), Bennett (1994),
and Schneider (2009:654-655). I will be agnostic here about whether the by-relation relates events or propositions or facts.
If the by-relation does not relate events, then the constitution theorist could endorse the following principle instead: If
someone F’s by Y-ing, and F is the act which makes it the case that she F’s, and P is the act which makes it the case that she Y’s, then P
constitutes F.
42
how to use it; a piece of wood may constitute a pawn when put on a chess board and used in a game
of chess, etc.
Put more concretely, Baker realizes that objects can have relational properties that detail their
relation to their circumstances, and one object constitutes a separate object of another kind whenever
it is in those circumstances. To capture this fact in her analysis, Baker claims that whenever an object
that has the property of being a certain kind of thing is in the right kind of circumstances, there will
be a coincident object that has the property of being a different kind of thing, though it’s possible for
the former to exist without the latter in general.
This talk of ‘circumstances’ is really plausible when we are thinking about everyday objects
and just how they are different from the material that constitutes them. The pawn depends on the
piece of wood for many of its properties (e.g. its weight, reflectiveness, etc.), but it also depends on
being in a society that plays chess and does so with objects of that size and shape. How this is relevant
to defending the CTA becomes especially clear when we consider what this talk of ‘circumstances’
means in the context of events.
No part of Baker’s account seems specific to objects – it should be able to hold of events as
well. Events have properties, may have certain properties essentially or accidentally, and can have
relational properties to other events. Given this, it should be possible for one event to constitute
another. We could say that when one event has certain essential properties and stands in certain
relations to other events, there will be a coincident event with distinct essential properties. If we think
about some examples of actions, we can see that these actions seem to conform to this account of
constitution.
55, 56
I can move my arm in just the way that I do when I greet my friend, but I can’t
perform that action of greeting my friend without being in the right circumstances of seeing my friend
and being in a society with the convention of waving as a sign of friendly greeting. Similarly, I can
move just as I do when I’m proctoring without proctoring, but those movements constitute my
proctoring when I’m in the situation of being in a room with my students as they take an exam.
Without the students or exam, or even my authority to monitor them, I can’t be proctoring. So it’s my

55
In fact, this is exactly how Bennett portrays the relationship between bits of behavior and action in The Act Itself. He
specifically speaks of the action as understood in terms of the relational properties of the behavior. This is ironic, however,
because Bennett himself discusses action in this way in part to eliminate talk of action from theories and instead only use
talk of behavior.
56
Mele (1997:3) even uses the example of thinking of actions as dollar bills in his discussion of the causal theory of action,
and this is one of the examples of paradigmatic constituted objects mentioned specifically in Baker (2002b).
43
movements bearing this relation to these favorable circumstances that makes it such that I’m
proctoring.
This is suggestive, but there’s one more important thing to notice about favorable
circumstances for events. To some extent, I’m not sure how clearly we can really define which
circumstances are ‘favorable’, as this will vary for different things. Just what favorable circumstances
are is one of the strongest sources of objections to Baker’s account.
57
Baker thinks of them as states
of affair and notes that they will often involve certain societal conventions or norms. This will seem
true for both objects and events, but there is another class of favorable circumstances that may be
essential to the constitution of events.
When we are talking about events and why events are as they are, we very often understand
their ‘circumstances’ as a matter of their causes and effects, their direct causal profile. So, when one
event constitutes another, it may be that it does so in part when certain causes (or the right causal
chains) exist to cause the constituted event to occur given the presence of the event that constitutes
it.
58
We can’t fully eliminate the need for certain states of affairs to be the case as well, but it’s good
that thinking of the causes of events gives us at least a clear necessary
59
kind of favorable circumstance.
Once we realize that circumstances are critical to constitution and an event’s causes may be
those circumstances necessary for the constitution of that event, then we can straightforwardly see the
relevance of the notion to a proponent of the CTA. According to a causal theorist, actions are
distinguished by their particular causal history; it is in virtue of the causal history of the action that it
is an action. So, if actions are not to be identified with the body movements as one acts, then it is only

57
Rea (2002) criticizes the coherence of the notion of favorable circumstances, and he attempts to give cases in which the
mere existence of favorable circumstances for an object entails the presence of that object. Wasserman (2004) tries to
point to cases where those favorable circumstances involve intrinsic properties that raise similar trouble (and challenge the
asymmetry of constitution). And Wasserman (2013) claims that the account will appear somewhat explanatorily backwards
in nature if the favorable circumstances involve elements that seem to come after the object in question (for instance, if
the statue is a statue because it’s admired, instead of admiring it because it’s a statue). Baker (2002) defends the analysis
against the charges made in Rea (2002), though I think there are things that could be said by way of defense against
Wasserman (2004) and Wasserman (2013) as well. At least, I think clear answers to these objections exist for constitution
of events, as discussed in the next footnote.
58
This is importantly different from how Baker herself understands the circumstances in which body movements
constitutes actions. For her, the body movement constitutes the action when the agent could have prevented the action
without doing anything else (1998:253). Given the importance of causal profiles to the individuation of events, I think
there is good reason to accept my construal of ‘favorable circumstances;’ however, it is worth noting how on Baker’s
construal constitution requires (and so an event may be an action in virtue of) the agent’s having free will (or at least the
ability to not perform that action).
59
That the causes of a constituted event are a necessary part of its circumstances is crucial for evading the objects to Baker’s
characterization of ‘favorable circumstances’ mentioned in the previous footnote. Because the favorable circumstances of
events must always include some cause or causal chain, we need not worry that the circumstances entail the event in
question, because causes do not entail their effects. Also, in response to Wasserman’s concerns, the causes that are part of
the circumstances of the constituted event will not be intrinsic to or after the event in question.
44
natural to say that the action is constituted by the body movement in the circumstances of having its
distinctive causal history. This even fits with how we disagreed with Davidson above, where we took
certain actions to have distinct causes to justify their being distinct events.
Consider again our examples. In the case of my greeting, we said that the conditions necessary
for it to truly be an act of greeting included seeing my friend and living in a society with the convention
of waving as a sign of friendly greeting. These conditions may well be conditions deeply important for
my arm movement’s constituting a greeting even as my arm’s moving, but there seems to be another
condition necessary to it being a greeting: my intention to greet. And again, though it may be necessary
to my proctoring that I’m actually in a classroom of my students taking an exam, it’s also necessary
that I intend to be proctoring.
For each intentional action we might perform, there is some antecedent mental states
(beliefs/desires/intentions) that cause that action; and it’s necessary to it being an action that it was
caused by these mental states and in the right way. According to the CTA, actions are things caused
in the right way by the right kinds of mental events, and what it is to be a constituted event is for that
event to have the right kind of causal history. So, far from it being a surprise that many actions are
constituted, we can rest assured that actions that are constituted by body movements rather than
identical to them fit just as nicely with the Causal Theory of Action.

45
Chapter Two: The Act of Omitting
We often omit to do certain things. I can promise to meet a friend for lunch and then omit to
show up because I decided I had something better to do. I can refrain from fixing some faulty
equipment that it’s my job to inspect and fix (say, purely as a matter of laziness). I can willfully abstain
from eating the cookie that’s right in front of me. In each of these cases and many more, it seems like
there was something that was my omitting, or refraining, or abstaining, and that I can be responsible
for it.
One thing we can say about omissions is that omissions often seem to be active in some way
or at least to be a means of manifesting our agency. Refraining and abstaining, for instance, seem to
be things that I do. If a vote on the measure comes up, and someone asks, “What did he do?” it is
acceptable to reply, “He abstained.” In this case, this reply is also appropriate to the question, “How
did he vote?” and so abstaining seems to be a way of voting, and voting is surely something that I do.
This linguistic data by itself is of course no proof that refraining and abstaining are manifestations of
my agency, but that we speak in this way at least suggests that we pretheoretically take omissions such
as these to be active. We also often talk about intentionally omitting, or performing an intentional
omission. For example, I might intentionally abstain from eating cookies.
That omissions seem to exist as manifestations of our agency presents a problem for the
proponent of a causal approach to action. It seems even less plausible to identify omissions with body
movements than it was to identify complex institution-involving actions with body movements. Even
for a modified version of the CTA, however, it’s not clear that omissions exist as things that can be
caused in the right way by certain mental states.
Prima facie, I think we do take omissions to be things that can cause/be caused. My desire to
leave my friend in a lurch can cause me to omit to meet him for lunch, and that omission can cause
him to be dismayed. My abstaining from eating a host’s cookies may cause the host to be insulted,
even though it may itself be caused by my desire to not have a second dessert. We regularly cite
omissions in causal explanations and can pick out the causes of our omissions. However, if we hold
on to the idea that events are the true causal relata, which is the standard view, then omission would
need to be events in order to fit into any formulation of the CTA. But few philosophers writing about
omissions have been willing to accept omissions as events.
60

60
For example, Bernstein (2014a) instead argues that omissions are possibilia, and Clarke (2014) argues that most omissions
are absences, which he argues are nothing at all.
46
There are strong objections to most of the positive views of omissions as events in the
literature, and there are other worries raised for omissions as events besides. There are at least three
common concerns for the idea of omissions as events. The first has to do with their metaphysical
nature: Omissions do not seem to be like regular physical events. How can omissions exist like regular
events without being somehow problematically negative? The second has to do with the location of
omissions: Are omissions located with me as I omit or are they located where I am omitting to be?
The third concern has to do with how many omissions there are: If we admit that I omitted to water
Jane’s flowers, and the Queen of England along with everyone else also did not water Jane’s flowers,
then are these all distinct omissions? And are we not committed to the existence of far too many
events of omission? A view of omissions as events must answer all of these concerns.
In response to these problems, we might argue that omissions are not events after all, but
events are not (or are not the sole) relata of the causation relation.
61
That way, omissions could cause
and be caused and so perhaps fit into a version of the CTA. Alternatively, we might just give up the
intuition that omissions can cause and instead claim that the CTA does not need to accommodate
omissions, because omissions are not actions.
62
In this chapter, I want to pursue a more controversial
but unifying view and claim that omissions are sui generis events that can be captured by our revised
CTA.
I will begin by considering the views of omissions as events that have been given. These
legitimize omissions as events by identifying omissions with some class of events that we already take
to exist unproblematically. The best of these views will turn out to be a Davidsonian approach, but
the problems for this approach familiar from the last chapter will resurface here. Unfortunately for
the Davidsonian, it is even easier to be persuaded that omissions are distinct from body movements
than other actions. Instead, I will show how the objections to omissions as events can be met head-
on by reifying them as sui generis events without identifying them with body movements.

61
Philosophers arguing against events as the causal relata often even use the case of omissions as leverage for their
positions. For example, Mellor (2004) has argued partially on the basis of omissions as causes but not events that facts are
the true things that cause; Beebee (2004) argues that absences can still be causally relevant without themselves being causes
(and so argues that non-events can be causally relevant without causing); and Schaffer (2005) argues that causation isn’t a
two-place relation between actual events in part on the basis of his view’s treatment of omissions. If any one of these
views is right, then the case will be even stronger for views of omissions that do not take them to be events. There are
independent criticisms we could offer against these alternative views of the causal relata, but the common view of the
causal relata as events is prima facie preferable insofar as it is both parsimonious and accords well with our intuitions (at
least in non-omissions cases).
62
Our omitting to perform some action can seem to be more a matter of not doing something than of performing some
alternative, negative action. So perhaps it should not fall to a proponent of the CTA to have to account for omissions.
Omissions may manifest our agency in some way, but if they are not actions, then they need not be events caused in the
right way by certain mental states.
47
First, by seeing what omissions would look like on a fine-grained view of events, we can
understand what their being ‘negative’ would amount to and why we should not be concerned. Second,
the view of omissions as events given will answer the question of their location. The more difficult
challenge is in explaining our conflicting intuitions about their location, but I will argue that these
intuitions can be explained by appealing to the act/result distinction in the philosophy of action.
Finally, I will address the question of the number of omissions. There is a difference between
omissions as manifestations of our agency and mere instances of not doing things, and appealing to
this difference will help to sooth the worry that if omissions are events then we will be responsible for
not doing far too many things.
Once these concerns have been laid aside, I will consider how omissions should be integrated
into the CTA. I will reflect on whether it is appropriate to say that omissions too are constituted by
body movements, and I will argue that it is. Making this final connection will allow us to provide a
complete story of how omissions fit into a revised version of the Causal Theory of Action.

Could Omissions Be Identical to Events We Already Take to Exist?
The philosophers who think omissions are events typically proceed by arguing that omissions
are identical to events that we already take to exist. This is a good strategy insofar as it reduces
omissions to normal, physical events, and it follows the more generally accepted metaphysical strategy
of reducing items in a problematic category to items in an unproblematic category. However, this is a
difficult way to proceed in this case given how varied the situations are in which we omit. For example,
in order to claim that omissions are still their own type of action, we might say that omissions are
simply identical to the actions we perform when we are actively restraining ourselves or putting some
kind of effort into omitting. A view such as this is given by Brand (1971), where omissions are
identified with the actions we perform in order to omit. Brand gives the example: “I can refrain from
raising my hand by putting it in my pocket, by sitting on it, or by keeping it at my side” (49-50). This
clearly won’t do, however, because we frequently take ourselves to be omitting (and to be responsible
for those omissions) even when we do not perform certain actions as a means of omitting (Clarke
(2014:14) makes this point). Instead of meeting my friend for lunch, I may go to the movies, but it
needn’t be that my going to the movies is the means by which I omit to meet my friend. Given that we
want a view on which we can be responsible for the many omissions that are not of this kind, we need
to look for a different view.
48
Another view we might adopt is that our omissions are identical to the actions we perform as
we are omitting, regardless of whether those actions are in the pursuit of (or as a means of) omitting.
Rather than thinking of omission as negative actions, we might instead think that we have omitted
when our actions can be given a negative description. Vermazen (1985) and Varzi (2007) both
advocate for this view. We might, for example, alternatively describe a certain action as ‘sitting still’ or
as ‘not moving.’ The idea, then, is that there’s nothing extra and spooky about my not moving; rather,
the act called ‘not moving’ is just another way to describe my action of ‘sitting still.’ In the case of my
omission to meet my friend, we may think that I perform one action we call ‘going to the movies’ that
can simply also be negatively described as ‘not meeting my friend.’ This view gets us the positive result
that our omissions will no longer seem to be paradoxical, since we are no longer saying that our not
acting is a kind of action. Instead, our omissions are just regular actions that admit of a negative
description.
This view is more intuitive than the last, and it is also a natural extension of Davidson’s view
of actions.
63
For Davidson, as we have seen, our body movements can be described in a myriad of
ways that name all of the intentional actions we can perform, but these actions are all identical to the
body movements. Omissions, then, are also identical to our body movements; we omit when our body
movements admit of descriptions of what we are not doing at that time. As such, omissions will fit
into Davidson’s version of the CTA.
Unfortunately, this view runs into a number of problems.
64
First, it can be unclear on this view
which body movement our omissions are identical to. Even on Davidson’s view, I can perform two
distinct actions simultaneously. I can simultaneously wink and wiggle my toe, for instance. For
Davidson, the winking will be identical to the movement of my face or eyelid as I wink, and the
wiggling of the toe will be identical to the toe-movement. But if I am omitting while I am doing more
than one thing, then it will not be clear on Davidson’s view which action my omission is identical to.
While omitting to meet my friend for lunch, I can listen to a podcast while I exercise. Listening
to a podcast won’t be identical to exercising on Davidson’s view, since they will involve different parts

63
This is why Vermazen proposed it, and Davidson himself accepted it as a nice extension (in Davidson 1985, later in the
same volume). For it to be a true extension, however, we need to say that the omission is identical to the body movement of
the agent as she omits, rather than the action. This will typically amount to the same thing, as any action the agent performs
as she omits will also be identical to the body movement, but speaking in terms of body movements admits the possibility
of the agent’s omitting even when there is no other intentional act description of her body movement (she could omit
while sleeping, for instance). From here on, I will take the view as identifying omissions with body movements.
64
Clarke (2014:24-27) briefly raises many of these problems, although similar appeals to, for instance, differences in modal
properties or causes and effects are common when giving arguments that use Leibniz Law to show that two items are not
identical.
49
of my body, so to which action will omitting be identical? Davidson will have to say that my omission
isn’t identical to any of the particular actions I am performing, but it is identical to how my whole
body is moving as I omit (the conjunction of all of the movements my body makes as I am omitting).
So, it’s how my body is moving that can be described as not meeting my friend for lunch. However,
this sounds like an implausible way to describe that specific cluster of actions – that big conjunctive
movement as ‘my omitting to meet my friend for lunch.’ This is especially true when we recognize
that I could have performed virtually any other cluster of body movements on the same occasion and
seemingly still performed the very same omission of not meeting my friend for lunch.
This last point turns on the fact that there are apparent modal differences between our
omissions and our actions or how our bodies are moving as we omit. Just as we saw in the last chapter,
there are a host of arguments appealing to modal, causal, and property differences that strongly suggest
that the items in question (in this case, omissions) are not identical to body movements. As they are
familiar, I will rehearse them in this context quickly. What will be important to notice is first that these
arguments seem even more compelling in the case of omissions and second that again what matters
for our discussion is not that Davidson’s view is wrong. Instead, what is crucial is that there is a
different view of omissions as events available for those compelled by these arguments.
In nearby worlds in which we are not performing the same movement, it seems like we could
still be committing the same omission. As another example, it may be true in the actual world that
when I omitted to do my homework, I moved in certain ways as I do when going to the movies.
However, I could have performed the very same omission and been moving instead as I do when I
go to get a haircut or go to the gym or take a nap.
65
While in the last chapter it may have felt question-
begging to assert that I could have made the exact same token greeting made with the other arm, it is
even more plausible here to assert that my omitting to do my homework really is the same regardless
of what I do instead. What matters to its being that instance of omitting to do my homework is that I
did not do that homework on that occasion. It sounds odd to say that crucial to its being that omission
is exactly how I happened to move as I omitted.
In addition to modal differences, our omissions also appear to have different causes and
effects from the movements to which they are supposed to be identical. In the case of my going to
the movies when I am supposed to be meeting my friend, I may have omitted to meet my friend for

65
I take it to be possible for our movements to modally outrun our omissions as well. Brand gives the interesting case of
refraining from raising his hand even in the case in which someone else lifts it up instead (50). So this is a case in which
the movement is the same as if one were not omitting, but one is still omitting.

50
the reason that I did not want to see my friend, while I may have gone to the movies separately for
the reason that it looked like it was going to be a good movie. It would be wrong to say that the reason
I decided to omit caused me to go to the movies, since my desire to avoid my friend is unrelated to
my desire to see the movie; and it would be wrong to say that my reason for going to the movies
caused me to omit to meet my friend, since my desire to see the movie may not have even entered my
mind until after I had decided to omit to meet my friend. Notice that this difference in causes does
not at all rely on any considerations about the metaphysical status of mental versus physical events,
which was an issue that came up in last chapter. Instead, these differences are purely a matter of acting
for different considerations.
Concerning effects, my omission may cause my friend to become upset at me, taking it as an
expression of my lack of care for the friendship, while my going to the movies may cause me to reflect
on the themes of the movie. Again, it seems wrong to say that my omission caused me to reflect on
the themes of the movie, and it seems wrong to say that my going to the movie caused my friend to
become angry (perhaps he didn’t even know what I did instead; it might not even matter to him).
66

Finally, we might think our omissions can come apart from what we do while omitting in
terms of the properties they instantiate. Though my omission to meet my friend for lunch may cost
me her friendship, it cost me no money. So my omission has the property of being free or of saving
me the cost of lunch. Going to the movies, however, may cost nine dollars (and so have the property
of costing nine dollars). Even if I spend the very same amount I would have spent at lunch on the
movies, it seems wrong to say that it cost me nine dollars to omit. My omission saved me money that
I instead spent at the movies, and I didn’t need to go to the movies in order to omit.
These arguments strongly suggest that omissions are not identical to body movements, and
some have been motivated to reject this Davidsonian view because of them. Unsurprisingly though,
there are ways for a Davidsonian to respond to these objections, and they will be just the same
maneuvers that were recognized in the last chapter. By appealing to Davidson’s coarse-grained view
of events and maintaining a counterpart theory of events, a Davidsonian can claim that omissions and
movements share their properties, causes/effects, and modal profiles after all.
The point here is not that a Davidsonian view cannot be maintained, but that it need not be.
Extending the Identity Thesis to cover omissions requires accepting a number of controversial
metaphysical views in order to answer arguments that are quite compelling. It may be that those

66
It may be that my friend learns that I was at a movie and this causes him to become angrier, but his anger here seems
compounded.
51
controversial metaphysical views are antecedently attractive. If not, however, an alternative is available
to a proponent of the CTA that acknowledges the apparent differences between omissions and
movements. I will outline such an alternative in the next section before moving on to the other
challenges that face a view of omissions as events.

Omissions as Sui Generis on a Fine-Grained View of Events
It is reasonable to expect there to be a view of omissions as events even if the Davidsonian
view is incorrect. The problems that we noticed for the Davidsonian view did not involve anything
unique or mysterious about omissions; rather, they were problems that we identified in the last chapter
with his view of the metaphysics of action as a whole. And so just as we were able to find a view of
actions as events even when not identified with bodily movements, we should expect to be able to
give a view of omissions as sui generis events as well. In other words, that the Davidsonian view
cannot convincingly be extended to omissions is a problem for the Davidsonian view, not for the
existence of omissions.
In order to see how it could be appropriate to think of omissions as events, I suggest that we
consider what omissions would be on a particular fine-grained and reductive view of events. The view
endorsed in the first chapter took events to be individuated by their causal profiles and not be
reducible to any other ontological kind. While I think omissions can be events on this plausible view,
it will not be as clear why some people think omissions are problematically negative or where they are
located.
Instead, I want to consider how omissions can be events on a popular reductive view of events,
as presented in in Kim (1976). There, Kim gives a view of events on which events are exemplifications
of properties by individuals at times. Put another way, events are ordered triples of individuals,
properties, and times. So, for instance, if a stoplight turns green at noon, then there is an event that is
the stoplight’s exemplifying the property of ‘greenness’ at noon, and the event just is the exemplifying
of that property at that time by the stoplight. This view is far more fine-grained than Davidson’s, but
it’s not necessarily too fine-grained to be plausible. There may be questions concerning whether every
52
instance of a property is an event, or how we know which predicates pick out distinct properties,
67
but
as a theory of event-individuation it comports much more closely with our intuitions.
Before moving to omissions, we should pause to recognize that this view of events seems just
as able to capture the relation of constitution between events as the view of events given in the last
chapter. After all, we ultimately want to not only legitimize omissions as events, but to show how they
fit into the CTA. So, although Kim himself did not demonstrate how constitution would work on this
account, we can imagine how it could go. Given the event of my moving my arm and the event of my
greeting, we can say that there is a property of being an arm-movement that is exemplified by me at a
particular time t, and there is a property of being a greeting that is exemplified by me at the same time
t. Since the arm-movement and the greeting are both exemplified by me at the same time, the
movement’s constituting the greeting will amount to the properties of ‘being a movement’ and ‘being
a greeting’ being appropriately related. That is, constitution between events will be a matter of the
relation between property exemplifications. Since Kim argued that events on his view could
themselves have properties, and presumably these properties can be dependent upon the properties
of other events, there seems to be no reason why constitution cannot obtain on this Kimian view of
events.
Now we turn to the task of seeing how the account might apply to omissions. Take the case
of my omitting to pick up my friend from the airport yesterday. This is an event of which I am the
subject; I exemplify the property of omitting to pick up my friend from the airport;
68
and this occurred
on the particular day that is now yesterday. On this account of events, omissions can be events if they
are understood in terms of subjects’ exemplifying certain properties. Just as an agent may exemplify
the property of raising her arm, the agent may exemplify the property of omitting to raise her arm if
she decides not to raise it. Prima facie, it will be no more difficult to have omissions as events than
more run-of-the-mill actions.
We can also immediately again see how taking on a Kimian view of events avoids the Leibniz’s
Law arguments against the Davidsonian view. Since, on this view of events, omissions are not identical

67
For example, Bennett (1988) argues that Kim’s view was much too fine-grained because he took every predicate to pick
out a distinct property. Instead, Bennett recommended that we individuate exemplifications of properties (and so events)
by their causal profiles. I agree with this approach. Bennett was also thinking of property instances as tropes, whereas Kim
was thinking of properties as universals to be exemplified. I don’t think anything that I will say here will require taking a
stand one way or another on the issue of which view we should take.
68
Exactly what this property amounts to will be taken up below.
53
to movements of the body,
69
we should not be surprised to find that they have different properties,
causes and effects, and modal profiles. And this does not make omissions into very mysterious things
as events. Like body movements or other actions, they will involve the exemplification of properties
by agents at times. As we will discuss below, it may still be that omissions on this view heavily depend
on what the agent is doing or how the agent is moving as she omits. So, this view may ultimately be
quite similar to Davidson’s view. It again merely involves accepting a fine-grained enough view of
events to answer the objections to the Davidsonian view.
One problem with this approach is that the Kimian view of events is controversial in a way
that that might be thought to spell trouble for omissions. It is often argued that the Kimian view is
too permissive in what it counts as events. For Kim, every property that is exemplified at a given time
by a subject picks out a distinct event, but this can lead to accepting far more events than we might
have thought plausible. For example, we might say that I wave to you or that I wave to you
energetically or that I wave to you enthusiastically. Regardless of how I describe my wave, it seems
like a single event, but Kim will count as at least three events because of these three distinct
descriptions that pick out three distinct properties. As a different kind of example, we might say that
right now I exemplify the property of being identical to myself, or of being one person, but it would
be very odd to say that there is an event of my being identical to myself or of being a single person
that is occurring right now. So, Kim allows for a staggering plentitude of events. This is troublesome
if we want to show that omissions are events, because now we might think that one of the ways that
Kim’s view is too permissive is that it reifies omissions.
Kim’s view of events may be too permissive, but it does not allow for omissions simply
because it’s so fine-grained. We saw that a more coarse-grained view like Davidson’s also makes space
for omissions, so omissions are not the kinds of things to be ruled out or in depending on how finely
we individuate events. We should no more expect our account of events to rule out omissions than
we should expect it to rule out energetic waving. This issue is only relevant to whether it is appropriate
to think of omissions as sui generis events or as identical to body movements, and the arguments
appealing to Leibniz’s Law strongly suggest that our view of events should be at least as fine-grained
as allowing that omissions are distinct from body movements.

69
Well, there may still be some omissions that are identical to movements of the body. For instance, my omitting to move
at a moment may be identical to my act of sitting still then. Whether these are taken to be one or two events will depend
entirely upon how fine-grained our view of events is. If we are so fine-grained that any difference in predicates picks out
a different event, then no omission will be identical to any body movement; however, this is more fine-grained than we
are likely to accept or need. This issue is discussed more below.
54
There are plausible restrictions to Kim’s account that we might make without being driven to
a view as coarse-grained as Davidson’s, but our task here is to legitimize the idea that omissions are
best understood as events, not to deliver the correct account of events. As we will see below, we might
think there is a problem with the number of omissions if we claim that they are events, but this is
separate from this question about the permissiveness of Kim’s view. So, let’s again take for granted
the Kimian view moving forward, and we can instead recognize that the better challenge to omissions
is more a matter of what omissions on this view of events would involve.

Omissions and Negative Properties
The view of omissions as fine-grained events avoids the objections to the Davidsonian view,
and it makes omissions less metaphysically mysterious by telling us exactly what they are (they are
exemplifications of properties by subjects at times). That being said, those against thinking that
omissions are events may not think that it is mysterious how omissions could be events. Instead, they
may object to the kinds of properties that omissions would require. We might think that if omissions
were events, then they would have to involve the exemplification of certain negative properties, and
we might have reasons for thinking that negative properties do not exist.
70

If I omit to meet you for lunch, then the omission will seem to be a matter of my exemplifying
the property of not-meeting. Or, if I refrain from eating a cookie, then that omission will be a matter
of my exemplifying the property of not-eating. However, while we would accept positive properties
such as being the eating of a cookie, it may sound like there is no property that is a not-eating, for
instance. To be clear, there’s two questions here: First, must omissions involve the exemplification of
negative properties? And, second, can there be negative properties? It would be too much to try to
answer the latter question here, as this would involve engaging with the larger debate about the nature
of properties.
71
Instead, I would like to put pressure on the idea that omissions must involve the
exemplification of negative properties.

70
I will just discuss here the possibility that the properties instantiated are themselves negative, but there are other views
one might have about how negativity can be fit into the situation. For example, negative properties might be a matter of
unproblematic regular properties that are related to the object by ‘non-instantiation.’ See Clarke (2014:40-44) for an
overview of and challenges to the different options as they have been presented in the literature.
71
I will say, however, that the burden is on those who reject negative properties. Nearly everyone accepts some negativity
at least insofar as we accept that there are negative descriptions, so there must be something about properties such that
they cannot be negative. Even if we grant that there are no negative fundamental properties, we might think that non-
fundamental properties can be negative as long as they are grounded in positive properties. For example, Zangwill (2011)
argues that negative properties exist derivatively. And if omissions did involve the exemplification of negative properties,
55
If I refrain from eating a cookie right in front of me, then we may describe this omission as
an instance of ‘not eating,’ but why should we think that the properties of ‘refraining from eating’ or
of ‘being a refraining’ must be negative? It may be thought that what is different in this case is that it
is essential to a refraining that the agent not do something; however, it is also essential to positive
actions that the agent not be doing certain other things. For instance, it’s essential to voting yea on a
proposition that the agent not abstain. So it’s not clear to me how recognizing that omissions merely
can always be described negatively forces us to conclude that they ever involve the instantiation of
negative properties.
The opponent may push harder here, saying that while it is true that when an agent votes for
some measure she does not abstain, we do not think that voting involves the exemplification of the
property of ‘being a not-abstaining.’ On the other hand, continues the opponent, abstaining seems to
involve the exemplification of ‘being a not-voting.’ But does it? I certainly never claimed that ‘being a
not-voting’ or ‘being a not-meeting’ were properties that must be exemplified for omissions to obtain.
Instead, I claim that while it is true that an agent that abstains does not vote, abstaining does not
involve the exemplification of the property of not-voting. Abstaining just involves the exemplification
of the property of being an abstaining, just as voting involves the exemplification of the property of
being a voting. Now, there is a further question about what these properties are and how they relate
to how I am moving as I am exemplifying them, but we have not yet said anything to demonstrate
why abstaining would be negative while voting would not.
Aside from these points, more support comes from recognizing a distinction between
omissions and instances of mere inaction or non-doings/failings. It is much harder to support the
claim that my not eating a cookie or my failing to eat a cookie do not involve the exemplification of
negative properties (of being a not-eating, in this case), but my refraining or abstaining from eating
the cookie is not merely my not eating the cookie. Consider the difference between refraining on the
one hand and merely failing to do something on the other. It’s true to say that George Washington
did not come to the meeting last Friday, and it’s even true that Washington failed to come to the
meeting (even though we wouldn’t blame Washington or consider it a failing of him). But Washington
certainly did not omit to come or refrain from coming to the meeting. The dead can’t refrain, and in
general we think one must be able to do what one omits or refrains from doing in order to be

these properties would not be fundamental properties. Their existence would be dependent upon positive properties such
as the agent’s moving in such-and-such a way as she omits. Of course, this will not satisfy everyone (such as a proponent
of a sparse view of properties that takes there to only be fundamental properties), but I do not have the space to consider
every problematic view.
56
omitting.
72
This does not hold of failing or of merely not doing something, and this suggests both that
there is an important difference between omissions and mere inaction and that omissions are more
closely tied to actions. It may be unclear exactly in what the difference consists, but if omissions are
distinct from instances of mere inaction, then space opens up to say that omissions do not necessarily
involve negative properties and so could be events even if instances of inaction did involve negative
properties and could not be events.
As a final reason for why we should not think that omissions involve anything that is
worryingly negative, we should remember again just how close my alternative account is to the
Davidsonian view. My view involves accepting a more fine-grained view of events that distinguishes
most of our omissions from our body movements, but it does not involve the injection of anything
negative. If my view involves negative properties, then they should already be present and problematic
for the Davidsonian view. However, no one has claimed that the Davidsonian view requires accepting
negative properties, even though on this view for any given omission there will be one event that
instantiates both the property of being a body movement and the property of being an omission.
So, my fine-grained view of omissions answers the problems given for the Davidsonian
account in a way that does not involve the further stipulation of anything problematically negative.
Now that we can see this, we can move on to discuss the other problems that have faced accounts of
omissions as events more generally; namely, puzzles concerning the location and the number of
omissions.

The Location of Omissions
Even if we have what seems to be a decent view of omissions as events, there is at least two
puzzles about where they should be located and why. It’s not that omissions do not seem to be the
kinds of things that could be located. In fact, on both the Davidsonian view and the alternative fine-
grained view, omissions have locations. On the Davidsonian view, my omission is located just where
my body is moving as I omit, because my omission just is my body movement. Matters may seem
slightly more complicated on my alternative view, but it delivers the same result. On the alternative

72
This is part of what’s so appealing about the account given in Bernstein (2014a), which suggests that omissions are
possibilia – a matter of what the agent could have done but did not actually do. Her account nicely explains why there
seems to be an ability-constraint on omissions (that you must have been able to do the thing you are omitting to do),
although it is unclear how you could have such an account and hold on to the intuition that omissions cause without
accepting a very outré account of causation like Schaffer’s contrastive causation.
57
view, my omission is a matter of my exemplifying the property of omitting at a time. Given the
assumption that events are located where the substance that exemplifies them is located, then
omissions are again located with the agents that exemplify them.
One puzzle concerns whether our omissions are located at a part of our body or our whole
body. To borrow an example from Clarke, if I am standing in front of a painting and I refrain from
reaching out and touching it with my arm at my side, then is my omission by my side with my arm or
is it at my whole body? On my view, this will come down to whether we take the substance that
exemplifies the property of omitting to be me or my arm. And although I may refrain with my arm, it
seems that I am the one that refrains, not my arm. So, on my view, omissions will be located with the
whole body of the omitting agent, not merely with the part that would be active if the agent were not
omitting.
This may seem like an odd result. After all, it is hard to imagine that my omission to touch the
painting is partially located with my right foot. Nevertheless, there are two things to consider to help
us accept this implication. First, while my omission is exactly located with my whole body, it is not
that part of the omission is exactly located with any particular part of my body. Second, this difficulty
is no more a problem for omissions than it is for typical actions. Suppose I do reach out and touch
the painting. Then we will again be left with the question: Is my touching the painting located with
my arm or my whole body? And, if the latter, does that mean that my touching the painting is partially
located with my right foot? It may be that I touch the painting with my arm, but it would be odd to
entirely locate my touching the painting with where my arm touches the painting, given that this is an
action I intentionally perform. I am willing to claim that omissions are exactly located with the omitting
agent’s whole body, but even if I am wrong, the fact that this is a problem faced for both omissions
and actions shows that this location puzzle alone is insufficient to rule out omissions as events. There
is a second location puzzle, however, that we might think is more worrisome.
The puzzle is to explain why our omissions are located with us given that we often have strong
intuitions that our omissions can be located far from us. Prima facie, it is not clear whether we should
think our omissions are always located where we are or whether our omissions are at the place we are
omitting to be. For example, I can refrain from meeting my friend for lunch and instead go to the
beach, but then it is not clear whether I omit at the beach to be at lunch or whether my omission is at
the restaurant, causing my friend’s disappointment, though I am far away. Wherever we say my
omission is, we are forced to say something problematic. If I can actively decide to omit and then omit
from where I am, then how I can control my omission, or how can it depend in any way on what I
58
am doing far from where I am omitting? However, if I can omit just where I am as I omit, then it’s
not clear how those omissions can be immediately causing something far from where they are located.
So, we are faced with a dilemma.
Given my metaphysical view of omissions, I am committed to grabbing the horn of the
dilemma that claims that omissions are always located with the agent omitting. The challenge, then, is
to say how our omissions can be causally relevant to events that are far from where they occur and
why we are drawn to thinking that they can occur far from us. To answer this, I think we must first
recognize that an omission by an agent to be at a certain place results in an absence of the agent from
that place. It’s familiar territory in the philosophy of action to distinguish between an action and the
result of that action – for example, we often distinguish the act of raising one’s arm from the result of
that arm’s rising. McCann (1974) spells out the distinction between actions and results, where results
of action are understood to be those entities (for him, events) necessary for the act to have occurred,
though not sufficient. If we take omissions to be events and a subspecies of our actions, then we
should expect our omissions to have results. Further, I think we should expect the act of omission to
result in an absence.
73

So, to go back to our example, my omission stays with me at the beach, while the state of
affairs of my absence from lunch is located where I am not.
74
The absence is entailed by my omission,
since it’s a necessary condition on my so omitting that I am absent from the lunch.
75
As another
example of this, if John omits to water Mary’s flowers, then John omits wherever he is, though there
is an absence of water in the soil of the plant. Since absences such as these are not identical to our
omissions or any of our actions, they may be located far from us (or not located at all) without
generating a location problem.
This answer requires the use of an uncontroversial distinction in the philosophy of action, but
it also requires the more controversial assumption that absences exist along with omissions. Absences
and omissions have many of the same problems, so perhaps if we are satisfied that omissions could

73
The relationship between absences and omissions is clearly a close one, but how closely they are related is harder to say,
and I will leave this issue largely untouched. Whereas I have argued that omissions are events, I will, following Thomson
(2003), take absences to be states of affairs, where states of affairs cannot cause without being connected to some event
that causes. Thomson also suggests that omissions are states of affairs as well, but I disagree.
74
I am certainly absent from the lunch, but it is less clear that my absence is itself located at the lunch. I do not wish to
take a stance on whether states of affairs are located and where, but what is crucial here is simply that if we can distinguish
the absence from the omission, then we can see that my absence needn’t be with me at the beach.
75
If the absence is taken to be located at the lunch, then this view has the non-Humean upshot that items in two different
spatiotemporal regions are logically connected. This is an odd consequence, and I am not sure how to answer someone
who is convinced both that the state of affairs of the absence is located at the lunch and that Hume’s prohibition of
necessary connection between distinct existences is correct.
59
exist as events, it is not such a leap to claim that absences may exist as well. Regardless of whether we
think it takes more to show that absences would exist along with omissions or whether the fact that
we would have to include absences into our ontology is further evidence that omissions do not exist,
it is at least interesting to see how absences can be used to help solve this problem for the location of
omission.

The Number of Omissions
It is often claimed that if we allowed omissions, there would be entirely too many of them.
Beebee (2004), McGrath (2005), Dowe (2010), Sartorio (2010), and Bernstein (2014b) each discuss
this problem and different possible answers to it. Suppose I refrain from eating the cookie on the
platter, and suppose this is a matter of my exemplifying the property of not eating the cookie on this
occasion. If that’s all it takes to omit in this case (or for there to be an event that is an omission), then
the class of people who will count as omitting to eat the cookies on that platter will include everybody
who does not eat one of those cookies, and that will include almost all humans. But it does not seem
as if almost all humans are omitting to eat the cookies that just happen to be in front of me. The
problem is even worse when we consider a morally important omission. Suppose I have omitted to as
save a drowning child right in front of me. I am clearly responsible for this omission. In this case, we
don’t want to say that the Queen of England also omitted to save the drowning child, and we would
not want to say this even if we could excuse her omission.
This may seem to be a problem especially for someone who accepts a Kimian view of events.
After all, if we are serious about accepting Kim’s view of events, then all of these not-doings will be
distinct events. Nevertheless, this problem of too many omissions is taken to be a problem for all of
the views of omissions as events. We want to say that I omitted to save the child in front of me and
Queen Elizabeth did not refrain, but none of the views of omissions as events are thought to have the
resources for making this distinction.
I think we can at least give some reasons why not all of these not-doings are omissions, even if
we did have to accept that these are all events. What we are concerned with here are not simply those
instances where agents do not do certain things; rather, we are concerned with those times when it
seems like the agent’s not doing something is a manifestation of her agency in just the same sort of way
that doing something would be. That is, we are concerned with omissions that seem to be actions in
some sense, or at least events for which the agent seems to be responsible (and we do typically use the
60
term ‘omission’ in this sense).
76
To see how this helps with the problem, let’s think about normal
actions.
All actions may be related or identical to certain body movements, but we do not think that
all body movements are related or identical to some action. That is, only some body movements seem
to be related to manifestations of our agency, and we usually don’t have any trouble figuring out which
ones. The proponent of the CTA distinguishes actions from mere movements in terms of having the
right causal history, and we will see how a similar maneuver can help us distinguish omissions from
mere non-doings.
If we accept that there are properties being exemplified when agents are not doing things and
we accept a view of events as property exemplifications, then it might be that there are many events
of different agents not doing certain things. Still, these events may not be omissions in the sense we
are concerned with. We are concerned with the sense in which I can omit to eat a cookie right in front
of me while not omitting to eat a cookie across the world, although I do not eat either cookie.
77

We saw above how omissions come apart from other non-doings. Instances of inaction and
failings are ubiquitous. Even the dead can fail. Refrainings, abstainings, and omissions, however, do
not occur nearly as frequently. If there is a distinction to be drawn between mere movements and
actions, then I think we should be just as willing to draw a distinction between instances of mere
inaction and omissions. Just as in the case of regular intentional actions, a proponent of the CTA will
have a tidy explanation for which events of non-doings are omissions – they are those with the right
causal history.
This distinction not only shows us why there are not too many omissions, but it also helps us
make the right claims of responsibility. Going back to the example of the drowning child, I omitted
to save the child and I will be responsible for this omission, whereas the Queen of England did not

76
I leave it open here whether all omissions are actions of a kind. Even if we thought that intentional omissions were
actions, we may think that unintentional omissions are not (or need not be) actions. The standard view in action theory is
that unintentional actions are intentional under some description, but we can imagine cases where an agent unintentionally
omits to do something without intentionally acting at the time of that omission. Suppose I omit to meet my friend lunch
because I was accidentally asleep. We often take ourselves to be responsible for unintentional omissions of this kind, and
they do seem to often be manifestations of our agency insofar as they express what we care about, are focused on, or our
attitudes. My omission may express a lack of regard for my friend, or at least a lack of care for trying to stay awake. We
take ourselves to be responsible for omissions like this because of what they express about us, not because they are actions.
What matters for our purposes here is just that there will still be a difference between unintentionally omitting to do
something where that omission expresses our lack of concern for that activity and merely not doing the activity.
77
Clarke (2014:29-33) claims that we can advert to some measure of context and what we would have been able to do
when considering how we apply the term ‘omission,’ and I could use much the same story for determining when omissions
actually occur. In fact, that one must be in the right context in order to omit fits very well with the suggestion given below
about how omissions are related to body movements.
61
omit and is not responsible. It is not that the Queen did omit from saving the child but she can be
excused for so omitting. It’s inappropriate to attribute any sort of responsibility in this case to her,
because the drowning was in no way a reflection of her will or a manifestation of her agency.
Distinguishing omissions from mere inaction allows us to admit that the Queen did not save the
drowning child without claiming that she omitted to or refrained from saving the child, and this is in
accordance with our intuitions.

Omissions and Constitution: The CTA and Other Advantages
I have argued that we can show how omissions fit into a fine-grained account of events, even
if they cannot be identified with how our body is moving as we omit. Still, we might wonder: If a
proponent of a causal approach to actions is to take on this view of omissions, will she have to revise
the formulation of the Causal Theory of Action in order to accommodate omissions? In the first
chapter, I gave the following formulation:

Revised Causal Theory of Action: Something is an action if and only if it is a body movement or
an event constituted by a body movement that is caused by certain mental states in the right way.

There are a few different ways that we might try to revise this account again to incorporate omissions
into the CTA; however, I do not think a further revision is necessary. If some
78
omissions are not
identical to body movements as I argued, then I claim that these omissions will be events constituted by
body movements. If so, and if intentional omissions are a species of actions more generally, then they
will already be captured by the above account. Alternatively, if omissions are not a species of action,
then we need not augment the CTA, since after all the CTA is a theory of action.
79

I follow Davidson (1963:fn.2) in maintaining that intentional omissions are actions. Regardless
of whether or not we ultimately take intentional omissions to be actions, however, the more
contentious idea to discuss here is whether or not it takes the relation of constitution too far to think
that omissions are constituted. Should we really think that the relation between my omission and my

78
I say ‘some’ omissions rather than ‘all’ omissions because whether we take all omissions to be distinct from body
movements will depend on exactly how fine-grained a view of events we take on, as discussed in fn. 69.
79
If, however, what the causal theorist cared about is not action per se, but the intentional expression of agency more
generally, then CTA could be re-written to capture both actions and omissions.
62
movements is the same relation as that between the statue and the lump? I made the case in the last
chapter for thinking that certain actions – especially those in the context of social institutions – have
a lot in common with how statues are related to the lumps that make them. Further, there are different
conceptions of constitution that have been taken to be extendable to the case of actions. We might
think, however, that claiming that omissions are constituted as well is asking too much of the relation
of constitution.
I disagree. What is important is not that constitution is a relation out in the world that brings
actions and omissions instead existence. Instead, what is important is that the actions and omissions
not identical to body movements stand in a relation to movements that has all of the same features as
the structural relation between the statue and the lump. It’s merely the relation with these features that
we call constitution. That is, omissions and certain body movements stand in a relation of colocation
and asymmetric dependence, where omissions only exist in the right kind of context.
It is worth acknowledging that this view of omissions as events that may be constituted by
body movements can accrue other advantages that we saw to the constitution relation. For example,
in the last chapter I argued that constitution could be used to explain the relation between certain
actions as well as between those actions and body movements. So, when we say that, for example, I
greet my neighbor by raising my arm, we can say that there are two actions that I perform: greeting
my neighbor and raising my arm, where the latter constitutes the former. If it is appropriate to think
that omissions are constituted, then we are able to distinguish different cases concerning how these
omissions are related to our other actions. For example, we can imagine one case in which I decide to
omit to meet you for lunch and where I separately decide to go to the movies during the time that I
was supposed to meet you. Alternatively, we can imagine the case such that although it is with good
reason that I must omit to meet you for lunch, the guilt of my omission drives me to go to the movies
in order to stop myself from going to meet you.
How are we to distinguish these cases? On the constitution view, we can recognize that in the
former case, my action of going to the movies and my omission to meet you are both intentional
expressions of my agency with distinct causes and effects. Both will be distinctly constituted by my
movements. In the latter case, however, my omission is importantly related to my going to the movies;
I omit by going to the movies. In this case, we can say that while my omission and my going to the
movies are both constituted by my movements, my omission is also constituted by my action. This
framework allows us to consider other cases in which the constitution may go in the other direction.
For instance, perhaps I intentionally offend you by omitting to greet you. The constitution theorist
63
can say in this case that my action is constituted by my intentional omission which is itself constituted
by my movements as I omit.

Omissions and Constitution: More Discussion
This view that omissions are events constituted by our body movements has been given before
in the literature,
80
and it has received ample criticism,
81
but I don’t think it has been given an adequate
defense. Hopefully, however, the foregoing gestures towards many ways in which it can be defended.
Constitution is a relation that requires colocation, and I have claimed that omissions are always at least
colocated with body movements. Constitution also requires that the constituting entity be in certain
favorable circumstances to be constituted. We can again advert to the framework of the CTA to claim
that omissions are constituted by body movements when they are preceded by the right causal history.
For example, my movements as I go to the movies may constitute my omission to meet my friend for
lunch when I promised to meet him and my omission is caused by my desire to leave my friend in a
lurch.
In order to defend this idea more thoroughly, I want to explicitly address two objections to
the view that omissions are constituted by movements of the body given by Randolph Clarke.
Answering these objections will help us to recognize that omissions depend on body movements in
the right way to think that they are constituted by them. In Omissions: Agency, Metaphysics, and
Responsibility, Clarke gives the most sustained criticism of the constitution view. Several of his
objections (e.g. concerning the location of omissions) have already been answered above, but two
larger objections call us to consider whether omissions are the same kind of thing as body movements
and whether omissions depend on body movements to the right degree.
Clarke suggests that omissions are not the same kind of thing as movements in the way that
statues are the same kind of thing as lumps of clay (9). In the case of the statue and the clay, they both
seem to be physical objects, but, though the movements of our bodies may be physical events, it’s not

80
According to Clarke (2011, 2014), this view has been taken up in Fischer and Ravizza (1998). However, though Fischer
and Ravizza do say that omissions are constituted by our body movements, they do not defend this as a substantive
metaphysical view.
81
Clarke offers extensive criticisms of this view that I cannot hope to entirely address here. I have anticipated some of the
objections he has given in my discussion of omissions already, but below I will explicitly discuss how to answer two
challenges he gives.
64
clear that omissions are physical events.
82
This may be a problem if we have a conception of
constitution on which things can only constitute other things of the same kind.
83
And, in fact, in
Baker’s account of constitution, one of the necessary conditions for x’s constituting y is “If y is
immaterial, then x is also immaterial” (2000:43).
84
For Baker, this condition is in place specifically to
avoid thinking that physical objects can constitute immaterial ones. So, if we thought that omissions
were actually immaterial, then they could not be constituted by body movements, at least on this view
of constitution.
This may be, but again this is just one of several views of constitution. Even considering this
account, why should we think that omissions are immaterial? They may appear to lack some sense of
the physicality had by our actual movements (perhaps in that how we perceive them is more
complicated), but we may still think that they are in some sense of the physical world, or else that they
are derivatively physical.
85
As we have seen, we speak as if omissions can have physical effects, and we
only expect physical entities to have physical effects. For example, we may say things such as that my
omitting to fix the brakes on a car caused the brakes to fail. Omissions may be spooky in some sense,
but it’s not clear to me that we need to admit that they are as spooky as immaterial entities.
There is, however, a more formidable problem that Clarke raises against the idea that our body
movements constitute omissions concerning the extent to which omissions really depend on body
movements. Constituted objects like statues inherit properties from and be sensitive to changes in the
properties of the lumps that constitute them. For instance, the statue inherits its mass from the mass
of the lump, and changes to the mass of the lump are reflected in changes in mass of the statue. We

82
This isn’t quite how Clarke puts this worry. He says, “Anything wholly constituted by a material object is, it seems, an
object, and anything fully constituted by a neural event is an event. But consider a case in which, when omitting to raise
one’s arm, one is moving it in some other way…If the omission were wholly constituted by this bodily movement, then,
it seems, the omission would itself be a movement. But that seems wrong…” (9). This way of putting the worry, however,
seems far less problematic than it might be, and perhaps question-begging. Why should omissions have to be movements
just because they are constituted by movements? If omissions must be movements in the sense that they must involve
ways of moving, then omissions are movements on my view – they inherit the property of being a movement from the
body movement that constitutes them. If this is not enough, however, and actions cannot be movements because they are
distinct from body movements, then this begs the question against my view, since my view requires that actions and
movements be distinct events.
83
This restriction will not be found on all views of constitution. On Fine’s view, for instance, I see no reason to think that
an omission could be a movement-qua-refraining regardless of whether the omission itself is immaterial.
84
If something can asymmetrically depend upon something else in just the way that the statue depends upon the lump,
then I am not sure it should matter whether or not it is immaterial. In fact, we might think that level of dependence itself
requires both items to be material or immaterial.
85
Similarly, defenders of non-reductive physicalism in the philosophy of mind maintain that mental properties are still in
some sense physical (or material) properties, despite their being distinct from properties of the brain. Of course, a Dualist
may more naturally claim that mental properties are immaterial, but she will have a much harder time of explaining the
apparent dependence upon processes in the brain.
65
were compelled by Leibniz’s Law arguments to say that omissions are typically not identical to your
body movements as you omit. Omissions though are so different from body movements that we might
think that they do not have enough in common with or inherit enough properties from body
movements to count as constituted by them.
We saw from the modal argument that we might perform one and the same omission by doing
entirely different body movements. So my omission does not seem sensitive at all to how I am moving
my body as I omit. That is, if I am omitting to save the drowning child and instead eating ice cream,
it seems to be exactly the same omission whether I eat my ice cream quickly or slowly or loudly or etc.
There are any number of ways that I might be doing the same movements, though my omission will
not seem to be any different whatsoever. This is a problem, because we expect constituted entities to
inherit certain (or at least some) properties from that which constitutes them, or to have properties
that counterfactually depend on the differences in properties had by that which constitutes them. If
we can’t show how omissions depend on our movements for certain properties (and are sensitive to
changes in those properties), then it is not clear how omissions depend on our movements at all.
Towards answering this problem, I think we need to first note that there are at least two clear
senses in which omissions do depend on our body movements. First, as I argued above, omissions
are colocated with our body movements. So omissions inherit their location properties from body
movements. Second, whether we omit to s depends on whether we move our body in one of the ways
that would constitute s-ing. That is, I am omitting to s only if my body is specifically not moving
certain ways, and my body’s moving any one of those ways would make it such that I’m not omitting
to s. For example, my omitting to save the drowning child depends on whether my body jumps into
the water (without someone forcing my body to move this way).
Also, we should notice that at least some of our omissions really do depend on how we are
moving. For example, if I omit to help set up the dinner table, then whether my omission is
inconsiderate or acceptable may depend on exactly what I am doing instead. If I am playing video
games, then my omission will be inconsiderate, whereas if I am hard at work on the finances of a
charitable organization, then my omission may seem acceptable (unless I routinely use this as an
excuse). So, our omissions can depend on what we are doing for some or many of their properties.
It may still be objected, however, that inheriting some properties is not enough. One might
point out that the statue, after all, seems to inherit all of the physical properties of the lump of clay.
The statue weighs the same as the clay, is as opaque as the clay, smells the same, etc. What can we say
then about our omissions, which don’t seem to inherit all of the properties of our movements? I think
66
we can answer this question in two ways: the first is to claim that it’s not actually true that constituted
things always inherit all of the properties of that which constitutes them, even in the statue case. The
second way to answer this question is to say something towards what’s special about omissions such
that we shouldn’t expect them to inherit many properties while still being constituted.
The statue clearly does not inherit all of the properties of the clay, since we recognize that the
lump of clay may have certain historical properties (e.g. where it’s from, how long it’s existed, etc.)
that are not shared by the statue. It’s also true that many of the physical properties had by the clay
essentially will not be essential to the statue. For example, the clay may be disposed to sinking in water
essentially (as a matter of the kind of clay that it is), whereas the statue may only be so disposed in that
it’s accidentally constituted by this kind of clay. Also, if we were to deny the transitivity of parthood
(though it would be very controversial to do so), we think that the molecules of clay are surely parts
of the clay, though it sounds odd to say that the molecules of the clay are parts of the statue.
Even if statues did inherit all of the physical properties of clay, perhaps it’s alright that
omissions are insensitive to certain kinds of changes in what constitutes them. Clarke mentions that
we can move slower or more quickly without the omission thereby changing and occurring slower or
more quickly.
86
But I don’t think this shows that omissions are totally insensitive to changes in body
movements. Instead, I think it shows that in general omissions are only be sensitive to physical events
in certain respects. Omissions do not normally depend on our body movements for how quickly they
occur, because omissions simply are not the kinds of things that occur quickly or slowly. Omissions
in general are the kinds of things that are only physically sensitive to whether or not the body is moving
in such a way as to be omitting.
That being said, there may actually be some instances where a more specific kind of omission
can be sensitive to the speed with which our bodies move. For example, perhaps I can quickly refrain
from talking to someone at a party by quickly turning and walking away. Or perhaps I can slowly
refrain from eating a cookie if I reach for it and then slowly move my hand back as I battle
temptation.
87
However, notice that these refrainings are slow or fast depending on my body
movements specifically because in these cases I was directly refraining, as opposed to merely doing
other things while omitting. That is, these are cases discussed by Brand and Vermazen where I omitted

86
Thomson makes this point as well and uses it to suggest that omissions are states of affairs (83).
87
However, notice that these refrainings are slow or fast depending on my body movements specifically because in these
cases I was directly refraining, as opposed to merely doing other things while omitting. That is, these are cases discussed
by Brand and Vermazen where I omitted by doing something else instead (where what I did instead was my means of
omitting).
67
by doing something else instead (where what I did instead was my means of omitting). So, it may be that
there are omissions of this kind that actually are sensitive to the physical features of our body
movements after all, though it should not be surprising or worthy of objection if most of our
omissions are not. As we saw above, these are omissions not merely constituted by our body
movements along with whatever actions we perform instead; these omissions are constituted by the
actions we perform which are themselves constituted by movements. So these omissions will depend
for their properties not only on the movements but on the actions that directly constitute them.
Our omissions are clearly closely related to our body movements. Seeing how closely related
they are was crucial to avoiding some of the biggest challenges for thinking of omissions as events.
Omissions are not fundamental existences; they depend on our body movements for their existence,
some of their properties, and even their location. Recognizing this helped us to see where omissions
are and why they do not involve any kind of problematic negative existence. Once we have
characterized exactly how omissions are related to body movements, however, we can step back and
recognize that it is appropriate to say that those omissions not identical to body movements are
constituted by them.

Conclusion
If we accept the causal theory of action, then it can be troublesome what we should say about
omissions. If omissions are to be a species of actions (or at least things that can be done intentionally),
then they must be things that are caused in some way by our mental states. However, it will be hard
to say how they can be caused if they are not events. We could deny that omissions and refrainings
are manifestations of our agency, or we could claim that they are not events but have a different view
of the causal relata such that they could still be caused.
There are also several recent alternatives that have been given by those who are determined
to defend some version of the CTA but do not take most
88
omissions to be events. Clarke (2014: ch.3)
gives a particularly interesting solution to this problem by claiming that what our mental states cause
is something inconsistent with (and so to entail) what we omit, rather than claiming that our omissions
exist as events to be caused themselves. Alternatively, Shepherd (2014) argues that although most

88
I say ‘most’ here because for both Clarke and Shepherd some omissions will be identical to what the agent is doing as
she omits, as in the case of ‘not moving’ and ‘standing still.’
68
omissions are not events to be caused, our intentions to omit to A do cause a relevant disposition to
not A. Both Clarke and Shepherd’s response attempt to capture intentional omissions without reifying
omissions themselves, but rather by finding something else to be caused in the right way by the
intention to omit.
If one is convinced that most omissions cannot be events, then it is open to one to pursue
one of the above-mentioned options. I do not have the space here to argue against these particular
views except to say that they necessarily involve somewhat complex maneuvers to obtain the same
intuitive result. If omissions can be shown to be events, however, then the answer is alluringly simple:
omissions, like other actions, are events that are caused in the right way by the right kinds of mental
states. That is, omissions may be distinct in that they involve an agent’s not doing a certain thing, but
they are no different from other actions insofar as they satisfy the same causal criterion. Further, the
CTA can be easily reformulated to include omissions, since omissions also will be events constituted
by body movements.
Of course, a lot has to go right in order for this picture to work. We have to be convinced that
omissions are events. It may also be that there are particular challenges to fitting omissions into the
causal theory of action. For example, Sartorio (2009) claims that there is a causal exclusion problem
unique to omissions on the causal theory. Nevertheless, by showing how we can accept omissions on
a reasonably fine-grained view of events, and by showing where we can find them, I hope to have
gone part of the way towards fitting omissions into a causal account of action.

69
Chapter Three: The Vague Time of a Killing
In the first two chapters, we saw two different challenges to Davidson’s Identity Thesis. What
these two challenges had in common was that they involved instances of agency that seemed very
different in kind from body movements. My body will surely move as I greet someone or intentionally
omit to pay my taxes, but my doing these things does not seem identical to those body movements.
What I suggested was that the arguments behind these intuitions could persuade us to deny the
Identity Thesis without pushing us to deny some version of the Causal Theory of Action. Actions and
omissions could still exist as events closely related to our body movements, and they could be actions
in virtue of having the right kind of causal history. What’s more, if it is appropriate to think that actions
and omissions are constituted by those body movements, then we can successfully revise the CTA.
Constitution, however, requires colocation. If actions are to be constituted by body
movements when they are not identical to them, then they must at least occur at the same time and
place. However, a number of philosophers have objected to the Identity Thesis exactly because they
deny that actions are colocated with body movements. The case can be made that many (and perhaps
all) actions do not occur when and where our body moves. Even if we let go of the claim that many
actions are constituted by body movements, we might still worry that these challenges pose a threat
to any successful reformulation of the CTA.
Some, namely Volitionists, have argued that many or all actions are located within the body.
On these views, actions may be under the skin or in the mind (as mental acts). In either case, they will
precede our physical body movements rather than being colocated with them. There are many
different proponents of this broad idea, and there are many different versions of Volitionism.
However, I will put the Volitionist to one side in this chapter. Recognizing and answering the
arguments of Volitionists is essential for maintaining the stronger claim that some actions are
constituted by body movements. But my aim in this dissertation is not to defend that claim. Rather, I
am only concerned with showing how a proponent of the CTA has the resources for reformulating it
given objections to the Identity Thesis. Although the CTA would need to be reformulated in some
way if objections of Volitionists could not be answered, how it would be reformulated is fairly
straightforward. For whichever events internal to the body that a particular Volitionist takes to fully
comprise the set of her actions, we could say that an event is an action when it is among those internal
70
events caused in the right way by the right mental states. So, the views that suggest that actions occur
before the body moves do not pose a special challenge for the Causal Theory of Action.
89

Alternatively, however, several philosophers have suggested that a slew of actions are not
colocated with body movements because they occur beyond and after our body moves, or they include
elements beyond our body. It may be plausible that my action of moving my finger occurs just when
and where my finger moves. If I thereby shoot someone in a duel, however, then it might seem that
my shooting them occurs (at least in part) where they are shot, which is distant from my finger.
Similarly, if I make a three-point shot in basketball, then it seems that my shot must be located (at
least in part) when and where the ball went into the basket.
As we will see, cases such as these are ubiquitous in our active lives, and they should be
especially worrying for the proponent of the CTA. Although they do not directly suggest that it is
inappropriate to understand actions in terms of having the right causal history, it will be extremely
difficult to give an account of how our actions include such causally distant events. We will be unable
to give an analysis of action that tells us which events, when caused in the right way, are actions. We
will also have an even more challenging time of specifying just what ‘the right way’ is that actions are
caused.
In what follows, I will take on this challenge and argue that these actions can actually be located
when and where the body moves after all. Let’s start, however, by considering more deeply a particular
case of apparently distant action that has been widely discussed in the literature. After seeing the
specific problem that it creates, I will step back to show how this problem is a problem for everyone,
how it arises for many of our actions, and how we should go about solving it. Consider the following
case, often referenced in the literature on the metaphysics of action:

A shoots B at a certain time and from a certain place (say, t
0
and p
0
). B dies of the wound, but not for
several days, say at t
2
, and after being moved to a hospital in a different city, say at p
2
.
90

89
This is not to say, however, that all Volitionists have endorsed the causal theory. Ginet (1990), for example, argues that
all basic actions are mental events, and these mental events occur contemporaneously with and are explained by an
intention to perform that action.
90
One often encounters in the literature the example of Booth’s shooting Lincoln, where Lincoln died several days later
from the gunshot wound. However, an even more startling recent case is that of Hinckley’s assassination attempt of
President Raegan. During that attempt in 1981, he shot and wounded James Brady, who continued living but eventually
died of the gunshot wound a full thirty-three years later in 2014. The death was ruled a homicide, although this ruling was
controversial at the time. I use ‘A’ and ‘B’ here to avoid trivializing these cases.
71
In this case, it sounds appropriate to say that A killed B, but the question is: When and where
did A kill B? It may seem that A has done his work towards killing B by pulling the trigger, and so A’s
action of killing B may seem completed at t
0
and p
0
; however, since the death occurs at t
2
, and t
0

precedes t
2
, this entails that the killing occurred before the death. And, crucially, it sounds wrong to
say that A killed B before B died. The problem of locating the time and place of killing is so pernicious
and oft discussed that it is often referred to as the problem of the time of a killing.
91
We can first see
how it raises an immediate problem for Davidson and my view of action, but it is not hard to see how
this case raises a similar problem for nearly any theory of the location of actions, and there are many
such cases.
This case on its own provides an objection to the Identity Thesis. If the killing does not occur
before the death, and the death occurs after the body movement, then the killing cannot be identical
to the body movement. It also is an objection to the revised view that I advanced in chapter one, that
actions are body movements or constituted by body movements. Even if we suggest that ‘the killing
of B by A’ refers to an action of A that is constituted by A’s finger-movement, we will still be forced
to say that the killing occurs long before the death, and this sounds false. This case is also an objection
to anyone (such as the many of the Volitionists) that claims that actions must be entirely located within
the body or somehow prior to its movement. For a Volitionist such as Jennifer Hornsby, for example,
the finger-movement, the trigger-pulling, the shooting, and the killing will all be located with some act
of trying by the agent that precedes even the finger-movement. If we find it compelling that you
cannot be killed before you have died, then this intuition will tell just as much against Hornsby’s view
and others that locate all actions internally.
However, the time of a killing is not only a problem for views that locate actions with or within
the body movements. Some problem involving killing can be generated for nearly any view of action
that we might offer. Given what I have said of the case so far, it may seem that A’s killing of B cannot
be located where A pulls the trigger, and it must be located at least in part later, when and where B
dies. But matters are not quite this simple. If we say that A kills B after A pulls the trigger and
when/where B dies, and we suppose further that A dies at p
1
and t
1
, before B dies, then we seem
forced to say that A kills B after A dies. That is, we seem forced to say that A is dead but still acting,
92

and this sounds inappropriate as well. Thus, we are left with a dilemma. Either we hold that A kills B
when A pulls the trigger, or we hold that A kills B at least in part when B dies. If we hold the former,

91
For a classic discussion of this question of the time of a killing, see Thomson (1971).
92
Hence why this problem is sometimes called ‘the problem of the acting dead.’
72
then A kills B before B dies; but if we hold the latter, then A kills B after A’s already dead. Neither
option seems palatable, but the killing has to occur somewhere. Pointing out this dilemma doesn’t
help us resolve in it Davidson’s or my favor, but it shows that we are left with a hard choice, and any
answer will involve biting a bullet and explaining away some intuition about what it seems
inappropriate to say.
Not only are we presented with a problem regardless of what view of the location of action
we take, but this kind of problem case can be found far beyond the much-discussed verb of ‘killing.’
It is again easy to see that less violent cases of this structure are ubiquitous. If I write and mail you an
inflammatory letter on Monday that you receive, open, read, and are offended by on Thursday, then
it seems as if I have offended you on Thursday. I may have written the letter in order to offend you,
and I may consider offending you to be an action that I have intentionally performed; nevertheless, it
would sound inappropriate to say that I offended you before you were offended. It would also sound
inappropriate to say that I offend you after I’m already dead (if we add to the story that I died on
Tuesday). And countless action verbs can be shown to lead to just the same problem. If I bake the
bread, is my action seems only complete when the bread is baked (even if I am taking a nap), or must
we say that I baked the bread when I put it into the oven, before it is baked? If I print the paper, have
I completed the action of printing when I press print, or does the paper need to be printed? The
examples go on.
Each of these verbs provides the same kind of dilemma for a view of the location of action.
To give a solution to this general problem, I will focus on killing and pursue the answers to two simple
questions: What is the definition of ‘kill’? And what event does the definition of ‘kill’ indicate is the
killing? Although there has been some discussion and disagreement about the correct definition of
‘kill,’ much less has gone into arguing that the correct definition indicates that a particular event is the
killing. After surveying the attempts that have been made to be guided to the right metaphysics by the
semantics of ‘kill,’ I will argue that the semantics of ‘kill’ leave it vague which event is the referent of
the description of ‘killing,’ and this vagueness explains different intuitions we have in different
contexts concerning which event is the killing. If it is vague when A kills B, then it is inappropriate to
assert that it determinately occurs at any one time or interval. I will contrast an appeal to linguistic
vagueness with an appeal to ambiguity to showcase the advantages of accepting vagueness, and I will
show how using vagueness in this way is amenable to the views of action of all parties. Finally, I will
conclude by considering how the revised view of the CTA that has already been given provides the
framework for incorporating these seemingly distant actions.
73
The Definition of ‘To Kill’
To know when A kills B, or when the event of A’s killing B occurs, we first need a definition
for ‘to kill.’ However, even giving the right definition here has proven difficult in the literature. A
naïve first attempt would be to define ‘to kill’ simply as ‘to cause to die.’ However, we can very quickly
see problems with this definition. Fodor (1970) argues that ‘kill’ could not be derived from ‘cause to
die,’ because he thinks there is no non-ad hoc transformations that map phrases onto words.
Alternatively, Morreall (1976) argued directly for the nonsynonymy of ‘kill’ and ‘cause to die,’ because
killing is an action, while causing is not. He further adduced linguistic evidence to show that we treat
these terms differently because of this. I agree with this line of reasoning, since this naïve definition
doesn’t make any mention of an agent that is doing anything. Davidson himself noticed this problem
with the naïve definition, saying,
…although we say the agent caused the death of the victim, that is, that he killed him, this is
an elliptical way of saying that some act of the agent – something he did, such as put poison
in the grapefruit – caused the death of the victim. (1971/2001:49)
To put the agent back into the picture, we can amend the definition thusly: ‘to kill’ means ‘to
perform some action that causes a death.’ This definition seems much better, since it can replace ‘to
kill’ in sentences without losing the sense of activity involved, or the fact that an action has been done.
However, it still seems as if there may be cases in which the agent performs an action that causes a
death without thereby killing anyone. For instance, suppose Ian writes Patrick a sonnet, and the sonnet
unexpectedly drives Patrick into a rage during which he picks up a gun and shoots Jimmy. In this case,
Ian has performed some action (writing/delivering a sonnet to Patrick) that ultimately causes a death
(the death of Jimmy). Of course, Ian doesn’t intend by his actions to cause Jimmy’s death, but intent
is not required for merely killing someone (as opposed to murdering them). And of course Ian’s
actions do not immediately cause Jimmy’s death, but proximate causation surely is not necessary for
killing someone either.
What this case rests on is the transitivity of causation. Ian’s sonnet writing causes Patrick to
fly into a rage which causes Jimmy to be shot to death. Now, given this, we could just claim that the
relation of causation is not transitive (or at least that it’s not always true that if A causes B and B causes
C, then A causes C). In fact, cases like the above have separately been given in the literature of
74
causation to argue against the transitivity of causation.
93
However, the problems for this definition of
‘to kill’ go beyond issues with the transitivity of causation.
Suppose that Ian sees Patrick pick up the gun in a rage and Ian grabs Patrick’s arm. Ian tries
to pull Patrick’s arm so the gun will not point at Jimmy, but Patrick is stronger and still shoots at
Jimmy, resulting in Jimmy’s death. Further, suppose that in this case, if Ian had not tried to stop
Patrick, Patrick’s shot ultimately would not have killed Jimmy (since he would have only hit Jimmy’s
leg). In this augmented case, it’s true that if Ian had not pulled at Patrick’s arm, then Jimmy would not
have died. It’s also true that Ian himself did not shoot Jimmy, and we would not say that Ian killed
Jimmy, but on some accounts of causation (such as counterfactual accounts), we would say that Ian
did something that was a cause of the death of Jimmy. After all, Jimmy would not have died if Ian had
not pulled on Patrick’s arm. What this case shows is that the definition of ‘to kill’ wrongly classifies
certain actions as killings because they in some way causally contribute to a death, even if they are not
the most important factor.
I think there is a way of re-writing the definition of ‘to kill’ that still uses the concept of
causation, but all of these causal challenges may push us to look for a definition that avoids using the
concept of causation altogether. In fact, a fairly comprehensive definition for ‘to kill’ has been given
by Wierzbicka (1975) that does not mention causation at all. When considering a case in which Peter
killed a cat, Wierzbicka offers the following explication:
At one time Peter and the cat were in some one place; Peter was doing something and
something was becoming in contact with the cat’s body because of that; the cat was dying
because of that; after that, the cat was dead because of that. (498-9)
Pols (2013:729) gives a definition for ‘to kill’ along the same lines as Wierzbicka, but with a few small
qualifications: he thinks we should remove the requirement of something in contact with the victim’s
body (since the cat could be killed via oxygen deprivation), and that we need not mention these events
occurring at ‘one time’ or ‘one place.’ These definitions involved here seem somewhat more
complicated, but I think someone who accepted the general idea behind them might advocate for the
definition of ‘to kill’ to simply be ‘to perform some action after which there is a death because of that
action.’ Put even more simply, we might define ‘to kill’ as ‘to perform some action that explains a
death.’

93
See Hall (2000) for a discussion of whether and under what conception causation is transitive.
75
Since Wierzbicka and Pols define verbs like ‘kill’ in terms of explanation more generally –
using ‘because’ instead of ‘cause’ – their view avoids commitment to any picture of what’s going on
causally in these cases.
94
Instead, they can rest on the intuition that killing involves how an agent’s
actions explain something’s dying. The intuitiveness is a virtue of the view, and it could be nice to
avoid the headache that comes with the difficult causal cases. Unfortunately, though, I think this is a
headache we shouldn’t try to avoid.
We may be talking about the single case of defining ‘to kill,’ but Wierzbicka and Pols purpose
we explicate a whole class of verbs in this way. The verbs in question, however, are causal verbs. Given
this, it would be better to give definitions for these words in terms of causation. It is also mysterious
how the actions of these agents could explain these deaths if not by causing them. The explanation in
question is clearly causal explanation, so we should expect that there is some correct definition that
uses ‘cause’ instead of ‘because.’ Speaking in terms of explanation more generally simply avoids this
necessary issue.
To get the right definition for ‘to kill,’ I think we need to incorporate the notion of a primary
cause. We saw that simply having an action that is a cause of a death is insufficient to guarantee that a
killing has occurred. The action has to somehow be a significant or important cause of the death for
us to think that the acting agent has killed the victim, and this notion of a primary cause is often used
to get at that idea of significance. For example, if I throw a brick that breaks a window, then my brick-
throwing is the primary cause of the window-breaking, even if the window-breaking may technically
have other causes (such as someone’s hitting the window the day before or the wind’s moving the
brick to a slightly different angle). Of course, just what it takes for an event to be the primary cause
of some effect (as opposed to a mere background condition) is difficult to say. It is also difficult to
determine whether this is a metaphysically real distinction or one that is merely sensitive to the context
in which we are speaking. Nevertheless, supposing that there is something to the notion of a primary
cause gives us the tools necessary to appropriately define many of these causal verbs.
Given this, I think we should understand ‘to kill’ as meaning ‘to perform some action that is
the primary cause of a death.’ In the case I gave, Ian’s writing/delivering a sonnet to Patrick would
not count as the primary cause of the death of Jimmy, and neither would Ian’s pulling at Patrick’s arm.
In both cases, Ian’s actions may count as causes of the death of Jimmy, but they are not the primary
or motivating force that brings about the effect. This shows why it’s fitting to use the concept of a

94
However, this is not part of the reason that Wierzbicka gives for using ‘because’ instead of ‘to cause.’ She claims using
‘because’ allows us to see the underlying structure better (498).
76
primary cause for this definition, because we take killing to involve actively bringing about the death
of someone, where that action is the most important factor that leads to the death.
This definition appears to get around the problems for the other definitions that have been
given, but there is a last worry that I want to consider. White (1980) recognized the distinction between
killing and fatally wounding, and we might worry that my definition is more fitting of ‘to fatally wound’
rather than ‘to kill.’ After all, every instance of actively fatally wounding someone will involve
performing an action that is a cause of a death. Nevertheless, I think our definition gives us the tools
to distinguish ‘killing’ from ‘fatally wounding.’
First, I hesitate to say that fatally wounding involves performing an action that is the primary
cause of a death. The action is surely the primary cause of the patient’s being wounded, and events
involving the patient’s wounds are surely the primary cause of the patient’s death.
95
Even if we thought
that causation was transitive though, it is not clear to me that we can assume that being a primary
cause is transitive. So it may never be true that fatally wounding involves performing an action that is
the primary cause of death.
Second, and more importantly however, not every instance of an action’s causing a death is a
fatal wounding. For example, if A strangles B to death, A kills B, but A does not fatally wound B.
Given these considerations, a better definition for ‘to fatally wound’ might be ‘to perform an action
that is the primary cause of the patient’s being in a physically damaged state that will cause the death
of the patient.’ This definition shows how fatally wounding is distinct from killing, but it still has the
same definitional schema – involving an action that is the primary cause of some result.
96

If we accept my definition for ‘to kill’ (as well as ‘to fatally wound’), then we can take a step
back and recognize how many such action verbs can be defined as an instantiation of the same general
schema. For instance, if I insulted you via mail, then I have performed some action (writing and
sending you a letter with insulting content) that is the primary cause of your being insulted. It’s

95
Even this may not be right though. Perhaps I encounter a fatally wounded animal and kill it so as to save it from the
misery it would have in the time between then and when it dies from the wound. In this case, what I do is the primary
cause of the death of the animal, not the state that it is in from the wound. However, although it does not die from the
wound, it still sounds appropriate to say that it was fatally wounded when I encountered it. The solution to this appears
to be that the mark of being fatally wounded is that the wound inexorably will cause the death (unless something else
intervenes to cause the death first).
96
White uses the apparent difference between fatally wounding and killing to suggest that fatally wounding is much more
plausibly located with the shooting than is the killing, and that the killing does not occur until the death. However, although
the result of the fatally wounding (the patient’s being wounded) will occur before the death, we have exactly the same
puzzle about whether the fatally wounding is located when the shot is fired or when the patient is struck. The solution
that I offer below for killing will extend to fatally wounding, and I will explain the sense in which it is true that B was not
killed until after the death. Thanks to an anonymous referee for prompting me to consider how we should think about the
difference between killing and fatally wounding.
77
important to note here though that the proposal is not to understand all action verbs in this way. Basic
actions, or those actions that can be performed directly without doing anything else first,
97
may be
picked out by verbs that do not fit this schema. For instance, ‘to imagine’ is not ‘to perform an action
that is the primary cause of one’s imaging something.’ If imagining a particular thing is an event, then
it will have a cause, but my act of imagining a horse need not involve my doing anything that causes
me to picture a horse. So, this kind of definition will not work for all action verbs, but it does seem to
work for many action verbs, including ‘kill.’
In contrast, I think it is appropriate to say that all causal verbs can be defined using a more
general version of this schema.
98
In fact, this simply falls out of how we understand causal verbs. As
a category, causal verbs are typically grouped in the following way: if F is a causal verb, then ‘A F’s
B’ can be paraphrased as ‘A causes B to be F’d.’ For example, ‘Johnny prints the paper’ can be
rewritten as ‘Johnny causes the paper to be printed.’ Given this, assuming that events are the true
causal relata, and internalizing my suggestion that what matters is being the primary cause, then we
arrive at the following: where C and D are objects, ‘C F’s D’ can be paraphrased as ‘There is some
event with C as a subject that is the primary cause of the event of D’s being F’d.’ Once we have this
schema, we need only recognize that actions are events of which the agent is a subject
99
to apply it to
those causal verbs that pick out actions: Where F is a causal verb referring to an action of agent A,
the definition of ‘A F’s B’ is ‘an action of A’s is the primary cause of the event of B’s being F’d.’
From this, we can substitute and derive our definition of ‘to kill’ as ‘to perform some action that is
the primary cause of a death.’

Given the Definition, To What Event Does the Word Refer?
Suppose we like our definition. Can we get from a definition to an answer of when a particular
killing occurs? Everyone writing on this problem seems to think so. There is real trouble though with

97
The best way to characterize basic actions, or whether there is a meaningful distinction between basic and non-basic
actions, is controversial in the philosophy of action. The point here is just that I do not expect all action verbs to be defined
in this way, and there will likely be principled reasons why certain action verbs will not be defined in this way.
98
Given this, it should not be surprising that not all action verbs can be given this kind of definition, because not all action
verbs are causal verbs. See Thomson (1987) for a thorough characterization of causal verbs and discussion of which action
verbs will and will not be causal verbs (and why).
99
This is not to say that every event of which the agent is a subject is an action. There is an important distinction between
things that the agent does (perhaps she sleepwalks) versus actions that the agent performs.
78
saying exactly which event is the killing. The trouble comes from the fact that our definition (and most
of the ones given above) mentions two events. There is the event of the action that the agent performs,
and then there is the event of the death of the victim that is primarily caused or explained by that
action. Of course, very few are tempted to say that the event of the agent’s killing the victim is identical
to this second event
100, 101
– the death of victim – but these two events suggest two views that have
dominated the literature. First, the killing may be thought to be identical with the event that the agent
performs. Alternatively, we may think that the killing is identical to the complex event composed of
both the action of the agent and the death of the victim (also including any causal intermediaries from
the action to the death).
Concerning these views, I think we must ask two questions: First, can either of these views be
accompanied with some story that actually avoids having to accept counterintuitive sentences as true?
And second, can any good reason be given for thinking that the event referred to in ‘the killing of B
by A’ is one rather than the other? We saw in the introduction how the time of a killing raises prima
facie problems regardless of your view of when the killing occurs, but I will show that even the most
refined answers to this do not fully avoid the problem. Further, none of the views that have been
offered can satisfyingly answer the second question. However, careful attention to this question of
what event is picked out by the definition can guide us to an answer to the time of a killing that can
avoid all of the counterintuitive results.
Davidson is known for the first view, of identifying the killing with the action of the agent that
causes the death (for him, the agent’s body movement).
102
Davidson’s reasons for accepting this view,
however, have more to do with his overall picture of action and of how we use descriptions to pick
out actions than with the definition of these action verbs. So no attempt is made, at least in his earlier
work, to claim that the definition of ‘to kill’ in any way suggests that the event referred to is the action
that causes the death. Further, as we saw in the introduction, this view has the original problem of
entailing that someone can be killed before they have died. In fact, this will be the typical situation,
given that the actions of agents that cause death typically precede the death. The strategy that Davidson

100
For example, Price (1982) argues that the killing is not only colocated with but identical to the death of the victim.
101
Interestingly, however, our intuitions might change depending on the action verb under discussion. It sounds unintuitive
to say that A kills B entirely when B dies (and not even partially before), but it sounds a lot more intuitive to say that A
insults B entirely when B reads the letter from A and is insulted.
102
In later work, however, Davidson actually rejects this view and instead seems to suggest that ‘killing’ is ambiguous in a
way to be discussed below (1985/2001:300). It is hard to know when a killing occurs absent any specification. However,
if we specify that we want to know when the action of killing occurred, then he again identifies it with the action performed
by the agent that causes the death of the victim.
79
took given this view was to simply bite the bullet and say that yes, it is possible to kill a man before he
dies (at least, as long as he dies). Davidson admits that it sounds paradoxical to have killings before
deaths, but he says several things towards helping us cope with this:
First, we should observe that we may easily know that an event is a [shooting] without knowing
that it is a killing…
…as it becomes more certain that a death will result from an action, we feel less paradox in
saying, ‘You have killed him.’… (1969/2001:177)
Unsurprisingly, few have been placated by this response. It’s not clear at all that knowing with
certainty about the impending death of B eases our sense of paradox about saying that B has been
killed before B died. So, someone who advocates for this Davidsonian view better have more to say
about this intuition.
103
Towards this end, Jonathan Bennett offers a metaphysical view that purports
to vindicate the intuition that killing cannot precede dying. When considering the case of killing by
shooting, he claims that the killing is the shooting, but the shooting does not become a killing until the
victim has died. That is, the event of the shooting gains a property once the victim dies, and it is in
virtue of that property that it is a killing. He says,
On this theory, an action acquires a new characteristic long after it has been completed. I
distinguish (a) an action’s immediate characteristics, which it has at the time when it occurs,
from (b) it’s delayed characteristics, which is acquires at some later time.
…to call an action a ‘killing of B’ is to say, in part, that it causes B’s dying; this is a relational
property of it, which it may acquire long after the action has been performed and in that sense
after it has ceased to exist. (1973:317)
We very often talk of objects which can gain these delayed characteristics even after they no
longer exist.
104
For instance, my grandmother would become a great grandmother if I were to have a
child, even though she no longer exists. Or, as the classic example goes, students may come to admire
Socrates long after he died, and so we might say that Socrates has these characteristics of having

103
For example, Lombard (1978) argues for the Davidsonian view and attempts to directly address the unintuitiveness of
claiming that killings can occur before deaths.
104
These are typically called mere ‘Cambridge changes.’ Bennett does not refer to them in this way, although I think this
is what he has in mind. For an alternative view that explicitly uses Cambridge changes, see Ruben (1999). Ruben claims
that actions like killings are the actions they are in virtue of these properties, but that these actions are in fact located where
the Cambridge change occurs. So the killing is located with the death, not the shooting. However, while it is true that my
mother becomes a grandmother when I have child, what’s important to stress is that my mother still exists when she exists,
not when I have a child. Similarly, although an action becomes a killing only when someone dies, that action (the killing)
still occurs when it occurred (which, for Bennett, will be before the death). See Lombard (2003) for other significant
criticisms of Ruben’s view, including pressure on the idea that a Cambridge change in an individual entails the existence
of an event with that individual as the subject.
80
admirers. Bennett recognizes this and extends this thought naturally to events, so that events can be
seen as gaining features after they no longer exist as well. For instance, the event of Socrates’s defense
can gain the property of being admired by readers of The Apology, or the event of the birth of [omitted
for anonymity] can gain the property of being the birth of my mother once I have been born. If events
can have delayed properties in this way, and actions are events, then it seems that actions can have
delayed properties as well. Thus, according to Bennett, the killing may be identical to the shooting,
but the description isn’t fitting until after the death, because it is only in relation to the death that the
killing is the shooting.
This is an intriguing idea, and one that we will take up towards the end of this chapter. Still,
this view fails to explain why ‘the killing of B by A’ would refer to the action that A performs (rather
than the whole sequence that includes the action and the death). It also has a number of other
problems. While there will be no killing unless the death occurs, and we may not know that the
shooting is the primary cause of a death until the death has occurred, we may not be able to infer that
the shooting does not become a killing until the death occurs.
105
If Bennett cannot say this, then his
view loses the apparent intuitive advantage it has over Davidson’s view. That is, it may true after all
on his view that ‘A kills B before B dies.’ Even if we granted that his view could count this particular
sentence as false, however, he may still have to counterintuitively count closely related sentences as
true. We will be able to truly say on Bennett’s view, looking back after the death, ‘A killed B before B
died,’ and it sounds just as inappropriate to say this. For Bennett, it is not yet true that A has killed B
at the time that A shoots B, because B must die before the killing can be identified with the shooting.
Once B is dead though, if the killing really is colocated with the shooting, then it will now, contra our
intuitions, be appropriate to say that A killed B days before B died.
106

So, although Bennett offers a metaphysical picture for how the predicate ‘is a killing’ could
pick out the event of shooting, he doesn’t show how the definition of ‘to kill’ compels us to accept
that ‘the killing’ refers to the agent’s action. What’s more, his view is insufficient to avoid and cannot
explain the counterintuitiveness of saying that killing can occur before death.
To try to have our cake and eat it too, let’s consider a final view that pushes the Davidsonian
line. To answer the seeming inappropriateness of saying that A killed B before B died, Wierzbicka and
Pols both suggest that instead of wondering which of the two times of those specific events is the time

105
Lombard (1989) argues that several views on the time of a killing including Bennett’s are guilty of fallaciously inferring
from the fact that the killing will not occur unless the death occurs to the fact that the killing will not occur until the death
occurs. He claims that this mistake trades on a subtle ambiguity in the meaning of ‘unless.’
106
Mossel (2001:262) presents this objection.
81
of the killing, we should recognize that there is a broader timeframe during which both of those events
occur and in which both events can be thought of as occurring simultaneously. For example, if A
shoots B on Saturday, and B dies from the gunshot wound on Sunday, then what we should say is that
A killed B during the weekend. Since the death is required for it to be a killing, A can’t kill B before B
dies. This will be false, just as intuition suggests. But it also won’t be that A is killing B even when A
is no longer doing anything. Instead, we can only speak of the killing as occurring at a particular time
that is wide enough to include both events.
Pols takes his view of the meaning of ‘to kill’ to allow for the defense of the Davidsonian view,
according to which the act of killing will be identical to the shooting and the moving of the finger;
however, it can be hard to see how this is the case, because it is hard to see on this view how the killing
and the shooting can be colocated. The shooting happens at a particular time, but the killing is said to
occur within some stretch of time that only partially includes the shooting. It may be that killing need
not be an event that occurs solely at the time that the person killed dies, but it is hard to see why the
killing would not be a separate event from the shooting, since it seems that the killing is only partly
and not entirely colocated with the shooting. Indeed, the killing seems to stretch beyond the shooting
to include both the shooting and the death. In response to this very concern, Pols says the following:
The important point here, as already noted by Anscombe (1957), is that one action is (or can
be described as) another only in certain circumstances, though it does not include those
circumstances…Wierzbicka has spelled out the circumstances in which we can describe an
action as ‘to kill,’ and it turns out that the description ‘killing’ can only be applied to the action
if the time frame (and place) referred to when using the description…contains both the action
(the shooting) and its consequence (the death). (730-731)
So, the view is that the instance of killing can occur exactly when the shooting occurs, although
the killing can only be referred to as a killing when someone is using a timeframe that includes both
the shooting and the death. Despite what Pols says here, his view of the semantics of ‘kill’ can still
sound like it goes much more naturally with what we recognized as the other easy possibility for the
referent of ‘the killing’: the complex event that includes both the shooting and the death. So, just as
we seem to get a satisfactory answer to the question “When did A kill B?” we lose our grip on which
event is picked out by the word ‘kill.’ Worse still, this problem of the time of a killing can again be re-
written to arise for this view. Although it gets the right result that we cannot felicitously say that ‘A
killed B before B died,’ on this view it will still be true to say that ‘the event that is the killing occurred
before the event that is the death.’ This is not quite as unintuitive as saying that A killed B before B
82
died, but it may only fair better because it uses the jargon of events. If we are not comfortable saying
that the event of the killing occurred before the event of the death, then this will encourage us to look
elsewhere for an answer to the problem.
Turning away from this Davidsonian answer, we might embrace the view that the killing is the
complex event which includes both the action done by the agent and the death that that action causes.
However, we can again quickly see that the best versions of this view will both not be able to get all
of our intuitions about sentences right, nor will they be able to adequately argue that ‘the killing’ picks
out the whole causal sequence from action to death.
Thomson (1971) offers the classic view according to which killing is an action that extends
from the movements of the agent to the death of the victim. This is an interesting view that succeeds
in making it impossible for someone to be killed before they have died, but it also fails in both of the
mentioned respects. First, Thomson is motivated to accept her view via a rejection of Davidson’s
conservative view that all actions are mere body movements. While this may be a good reason to push
for a different view of action, this does not explain why killing itself would be identical to the complex
event that begins with the body movements of the agent and ends with the death of the victim. Second,
this view has the problem of the acting dead mentioned above: this view may count agents as still
acting although they themselves have perished. For instance, if A dies after shooting B but before B
dies, then on Thomson’s view A will nevertheless still be acting.
As one final view to consider, we can examine the view given in Pietroski (1998). Pietroski
argues that the killing is identical to the complex event that begins with the agent’s action and
eventuates in the death of the victim. Since he does not claim that the complex event in question is an
action, he can avoid the problem of the acting dead. Someone may kill after they have died, but they
are not technically acting after they have died, so we do not have to worry about zombies. This very
feature of his view is problematic, because now we have act descriptions like ‘the killing’ which do not
refer to actions. Killing on his view only picks out the complex event that begins with the action; it
specifically does not refer to the act itself. Perhaps Pietroski could say that killing is something that
the agent does, but it will be an unwelcome result if ‘the killing’ and many other act descriptions do
not refer to actions.
107
This would be a consequence we would nevertheless have to accept if we had

107
Pietroski is a Volitionist, however, and this may afford him a way around this problem. Volitionists claim that all actions
are located within the body (or mind), and Pietroski argues that all actions are identical to instances of trying (which are
within the body). Given this, he may accept that strictly speaking it’s true that all we can ‘do’ is try, and so technically killing
is not an action that we can perform. I take this to go against our normal understanding of these action verbs as referring
to actions.
83
a compelling reason to thinking that ‘the killing’ had to refer to the complex event, rather than the
action that is the first part of that complex event, but again we are not given a reason stemming from
the semantics of ‘kill.’ We want to know what it is that the nominal ‘the killing’ derived from the verb
‘to kill’ refers to.

Vagueness
All of the views discussed, and nearly every view given on this topic, takes the stance that ‘the
killing of B by A’ refers to either the action that A performs that is the primary cause of B’s death or
the complex event that includes both A’s action and B’s death. However, if we just consider the
definition that we have for ‘to kill,’ we can easily see that this definition suggests neither of those
views. Instead, I want to purpose that it is vague which event is the event of A’s killing B. That is, it is
indeterminate to which event ‘the killing of B by A’ refers.
The definition that we found, again, is that ‘to kill’ means ‘to perform an action that is the
primary cause of a death.’ The definition mentions the event that is the action, the event that is the
death, and the causal process from the former to the latter (which can itself be considered a complex
event). Further, it is clear that sentences that use the verb ‘kill’ pick out events: ‘A kills B’ refers to the
event that is the killing of B by A. Similarly, ‘A killed B’ suggests that there is an event that is the killing
of B by A, and that that event occurred at some time in the past of the speaker. We can also infer
from this that these sentences mean to refer to or pick out a single event (whether or not that event is
complex). So, we have a term that refers to a single event, but the definition mentions three distinct
events.
Looking at the definition, it is plain that it gives no indication as to which of those three events
is the singular event of killing. It would be one thing if ‘to kill’ simply meant ‘to cause a death.’ Then
we would say that the use of ‘kill’ in a sentence refers to that action that causes the death. But, with
our definition, no easy answer presents itself. We cannot just christen one of the events as the referent
simply because it goes with our preferred theory of action. Instead, we have to recognize that when
the definition does not determine which of several events is the event referred to in sentences with
‘kill,’ then it is semantically indeterminate which event is the referent. That is, we must accept
indeterminacy of reference, and it is linguistically vague which event is the killing.
I do not make this suggestion lightly given how many philosophers have such a strong aversion
to vagueness. However, I feel that a number of factors should strongly compel us to accept it here.
84
First, as we have just seen, the best definition
108
does not determine which of several events is the
killing, and the killing can only be one event. Given this, what else could it be but indeterminate which
event is the killing? To accept this indeterminacy is all that it is to accept this kind of vagueness.
Second, it is important to emphasize that we are only talking about accepting linguistic
vagueness. To say that some terms are vague is not to say that clouds have metaphysically fuzzy
boundaries or that it’s vague when a fetus becomes a person.
109
While relatively few are willing to
accept ontological vagueness, many philosophers are willing to accept some vagueness when it comes
to language or our concepts, and this vagueness is often cashed out in terms of indeterminacy of
reference.
110
Sometimes, the line goes, our language is just too imprecise to refer to a particular among
several alternatives even when it can only be referring to a singular thing. I am suggesting that ‘kill’
and other action verbs in English are vague in this restricted sense.
Third, accepting the vagueness view helps to solve the problem of the time of a killing, and it
is a solution for all parties. On this view, the answer to the question of when the killing of B by A
occurred is that it is vague. This can sound non-committal or to be merely avoiding the problem. To
be frank, it sounds cheap. In fact, though, it is indeterminate which event is picked out as the killing
of B by A, and the different events it could be occur at different times. How this view helps, however,
really comes out when we consider our uneasiness with the different sentences involving killing.
We have throughout recognized that it is at least prima facie unintuitive to say that ‘A killed B
before B died.’ Now we have an explanation for this. If it is indeterminate whether our use of ‘the
killing’ refers to an event that concludes before the death and a complex event that includes the death,
then it cannot be definitely true that the killing concludes before the death. So it is infelicitous to use
sentences that imply that the killing definitely does not include the death. By the same token, it will be
infelicitous to utter sentences that imply that the killing definitely does include the death. We are not
committed to the idea that the killing is an action that may extend beyond the agent and continue past
her death, and so it will never be definitely true that ‘A killed B after A died.’ Thus, we have the
resources to avoid the problem of the acting dead.

108
Not just my preferred definition, but in fact several of the definitions mentioned do not clearly pick out a single event
as the event that is the killing.
109
You may think that accepting any vagueness, even linguistic, commits you to some form of ontological vagueness.
Insofar as we and the propositions we express are items in the natural world, I am inclined to agree with this. Still, it is a
very different thing to accept that some referent is underdetermined than it is to accept that certain molecules
indeterminately bear the parthood relation to certain macroscopic objects.
110
There is a way for my claims to technically be compatible even with those who think that there is no indeterminacy in
the world or language, such as epistemicists. An epistemicist can deny that it is indeterminate which event is the killing,
and instead claim that it is determinate but unknowable.
85
As a final linguistic advantage of this view, we can explain the sense in which it sounds right
to say that ‘B has not been killed until B dies.’
111
Although ‘the killing of B by A’ indeterministically
refers to certain events that are completed before the death, it is not until the death that all of the
events to which this indeterministically refers have been completed. So, it will not be definitely true
that the killing has occurred in the past until the death of B.
However, we might worry that this raises a different problem for the vagueness view. This
view does not say that ‘the killing’ determinately refers to the act the agent performs, and it also does
not say that the complex event is an action (as Thomson’s view did). So, since the use of ‘the killing’
indeterministically refers to at least one event that is not an action (say, the complex event), then we
risk falling back into the problem we raised for Pietroski. Pietroski only avoided the problem of the
acting dead by claiming that killing was not an action after all, and we rejected it for this reason.
However, according to the vagueness view, we cannot say that killing is definitely is an action. It is
indeterministic whether the use of ‘the killing’ refers to an action or to a complex event, and that’s all
we can say about it. Given that it can seem so clear to us that killing is an action, the worry is that
indeterminately referring to an action is not enough.
Though this may seem problematic, this is a far cry from saying that killing is not an action.
We are only saying that when ‘the killing’ is used in sentences, it is indeterminate whether it is referring
to an action or a complex event partially composed of an action. Given that, it is hard to assess how
much of a worse position this puts us in intuitively. That the use of ‘the killing’ is vague in this way
does not show that we cannot kill intentionally, nor does it show that killing is not something we can
try to do. So it does not seem to run afoul of the intuitions connected with action.
What’s more, the vagueness view seems to be one that could be adopted regardless of one’s
view of the metaphysics of action. For instance, Davidson should be able to adopt it. It may be true
that in a sense insofar as Davidson’s view is that the shooting is the killing, this won’t quite be right.
However, he could instead say that when ‘the shooting’ indeterminately refers to an action and when
‘the killing’ indeterminately refers to an action, then ‘the shooting’ is ‘the killing’ in that they
indeterminately refer to the same action. This isn’t so bad though, because there is still only one event
that is taken to be the singular action performed, and this view does not require saying that there are
actions beyond movements of the body. After all, it is not as if the claim is that killing is a vague event

111
So I think we can capture the advantage that Bennett’s view is supposed to have.
86
while moving is not; it’s that the definition of ‘to kill’ is such that ‘the killing’ could be referring to
several distinct events, one of which will be the movement.

An Unambiguous Solution
As a final note helpful in clarifying my position, it is crucial to recognize that my proposal is
not that ‘to kill’ is ambiguous.
112
I am claiming that there is one definition for ‘to kill,’ which is ‘to
perform an action that is the primary cause of a death.’ It’s just that this single definition does not
determine amongst several events mentioned in the definition which of those events is the event of
killing. Contrast this with a view on which we take ‘to kill’ to have several distinct definitions. This is
a view on which the term ‘the killing’ could ambiguously refer. If we had this view, then we would
have an explanation for the dilemma of intuitions with which we began. It seems both inappropriate
for killings to be able to occur before deaths and for killings to be able to occur (at least in part) after
the death of the acting agent, and the proponent of the ambiguous view can say that one definition
for ‘to kill’ killings cannot occur before the death of the person killed, and on the other definition of
‘to kill’ killings cannot be done by those that have already died.
As an example of this ambiguity view, Weintraub (2003) suggests that a conflation of two
possible meanings of ‘kill’ leads our intuitions in this case astray. She says,
Davidson (1969:177) slides between viewing the killing as ‘an event that caused a death’ and
‘the causing of a death.’ But the difference is crucial.
On the first reading, the location is clearly at [t 0]. And there is nothing paradoxical about the
temporal gap between the killing – thus construed – and the death: causes can be distant…
On the second reading, the question is about the temporal location of the causal interaction
between the two events, and thus interpreted, the first suggested reply…is inappropriate. Since
both cause and effect are involved in the interaction, they must both overlap with it.
…Because we conflate the two readings, neither response seems apt. (179)
This response is suggesting a way of biting the bullet in the case at hand. Killing can occur before
death; it just seems unintuitive to say so because we conflate another possible definition of the word
‘killing’ that does require the killing to be colocated with the death. This is an interesting answer to
the problem, and it seems to have several of the virtues of the vagueness view. However, I think there
are decisive criticisms of it that are not shared by the vagueness view.

112
See Dunbar (2001) for a clear discussion of the distinction between vagueness and ambiguity.
87
This ambiguity view does raise an interesting issue about the relationship between being a cause
of some effect and being the causing of that effect — both in terms of what and where these items are
metaphysically and how verbs work with referring to them.
113
Nevertheless, we should not
underestimate how big of a mistake Weintraub is accusing us of making in our everyday language. The
cases that lead to this question of timing are ubiquitous in English. If Weintraub is right, then in all of
these cases regular speakers only hear these sentences as incorrect because they are regularly conflating
the event that causes some effect with the causing of that effect. If this view is correct, then ‘offending’
must always be pulled apart into ‘an action that causes offense’ and ‘the causing of offense,’ and
‘baking’ must be separated into ‘an action that causes something to be baked’ and ‘the causing of
something being baked.’ This may just be how English works, but it seems to be a rather big
assumption to say that speakers are so regularly making this kind of conflation.
114
On the other hand,
it is a much smaller fault, I submit, to blame regular speakers of being vague in their use of language.
Not only does the ambiguity view attribute a mistake to speakers, it attributes a mistake that
speakers regularly and easily avoid making. Consider: we do not think that it is false to say ‘The bank
is open for business’ because riverbanks are not businesses that can be open. Instead, when we hear
that sentence, we immediately take the speaker to be referring to a particular financial institution. In
general, when we confront ambiguity in language, we look to the context and straight-away charitably
disambiguate to get a true reading. So, a sure sign that ‘to kill’ is not ambiguous is our failure to
disambiguate in different contexts to avoid an intuition of falsehood. By contrast, that we get the
intuition that it is false to say ‘A killed B before B died’ is explained (as opposed to explained away)
by the fact that the sentence implies that the killing is complete before the death, but one of the events
indeterminately referred to includes the death.
Ultimately, we are left with what appears to be the typical exasperated answer of a non-
philosopher, “Oh, well it’s just vague!” Apart from debates concerning the nature and prevalence of
vagueness in the world, philosophers tend to avoid appealing to vagueness, because it can seem
unhelpful or to be simply giving up on answering a hard problem. In this case, however, an appeal to
vagueness is neither of these things. Far from giving up, we see that linguistic vagueness is
straightforwardly the right thing to appeal to when the semantics of a term do not determine to which

113
Weintraub herself thinks that although we are conflating the action that causes the death with the causing of the death,
there in fact is no event that is the causing of the death. I take no stance here on the metaphysical status or ontological
category of causings; however, I do not think the causing of a death is as easy to locate as Weintraub maintains.
114
Pols gives this worry for Weintraub’s solution. He says, “…it would mean assuming that our intuitions are structurally,
as opposed to incidentally misguided” (728).
88
event it refers. If the reference is not determined between several possibilities, then we have
indeterminacy of reference.

The Time of a Killing and Constitution
If the time of a killing problem is solved by appreciating that it is vague when ‘the killing of B
by A’ occurs, then the problem cannot be used to argue against the Davidsonian view that all actions
are body movements. It also cannot be used to argue against my revised view that actions are either
identical to or constituted by body movements. Still, it would be nice to see how these seemingly
distant actions are to be accommodated on the CTA. Although the Davidsonian has a simple story
for how these actions fit into Davidson’s original version of the CTA, I want to suggest here that my
revised account’s appeal to constitution works especially well for incorporating these actions.
A Davidsonian could appeal to my semantics for these verbs and claim that actions such as
killing and insulting are identical to certain movements of the body after all. My semantics explains
why we have intuitions against saying that ‘A killed B before B died,’ but how it nevertheless might be
true that the action that is the killing just is the action of A’s finger-movement as he pulls the trigger.
Given this, if the Identity Thesis were only facing pressure from the problem of the time of a killing,
then we could use my semantics to maintain that thesis. However, since we have adduced good reasons
for rejecting the Identity Thesis in the first two chapters, there is a question about how we should
think about acts such as killing on my revised version of the CTA, or whether the CTA will require
further revision.
Again, according to my revised CTA, something is an action if and only if it is an event either
identical to or constituted by a body movement caused in the right way by certain mental states. I
believe this account can capture those actions such as killing that are seemingly distal from our body
movements. For these actions, we should say that they are constituted by body movements rather
than identical to them, and what is crucial to their being constituted is the fact that the body
movements have the right effects.
We saw in the first chapter that one compelling account of constitution
115
prominently
featured the idea that one thing constitutes another when it is in the right circumstances. More formally,
this was understood to be a matter of (for instance) the lump’s having certain relational properties to

115
This is the account of constitution given in Baker (2000).
89
its surroundings (perhaps the artworld or the artist that made it) that make it constitute a statue in
those surroundings. We also saw how this idea fit naturally with and could be fleshed out in terms of
a causal theory of action. We can say that a body movement constitutes some action when it is in the
right circumstances of having the right causal history. For example, my finger-movement constitutes
my trigger-pulling given both the existence of the trigger and my intention to pull the trigger that
causes me to do it.
Given this, we can see a way of accounting for apparently distal actions like killings: a body
movement (say, the finger-movement) constitutes the intentional killing given the circumstances of
the intention to kill and the death of the patient. It is not an intentional killing without either of these
components, but this is not to say that the killing includes or extends until the death. Rather, it is just a
necessary feature of the killing that my body movement is the primary cause of the death.
116
In this
way, this view has a lot in common with Bennett’s. We can be happy to recognize, as Bennett did, that
the body movement has this relational property of causing the death of the patient. However, whereas
Bennett would say that the movement became a killing (and did not become a killing until the death),
we can say that the body movement constitutes a killing in virtue of causing a death. Or, put another
way, the event that is the action of A’s killing B is constituted by A’s finger-movement in virtue of the
fact that it causes the death of B. So, more generally, all actions will be body movements or constituted
by body movements, and some of those actions will be constituted in virtue of the body movements
having certain effects.
It might be argued that my view involves some kind of problematic backwards causation or
determination. On my view, the victim dies, and as a result the shooting constitutes a killing in the
past. I agree that this can be made to sound problematic, but I don’t think it involves anything
metaphysically unacceptable. The death does not cause the killing, even on my picture, so there is no
backwards causation. Further, the constitution relation is not a generative or determination relation;
the statue doesn’t come out of the lump. Constitution is merely a structural relation of synchronic
dependence. So the future event of the death does not cause extra causing, or generating, or
determining to be going on in the past. It’s true that on my view the killing is constituted by the

116
Well, the killing will be constituted given the intention to kill and something’s causing the death that constitutes it. On
the constitution framework, we can say that the finger-movement constitutes the trigger-pulling, which constitutes the
shooting, which constitutes the killing. What is important and clear is that the effect of the death is among the
circumstances necessary for constituting the act of killing, but what is less clear is which action has the death as an effect.
It is natural here to say that it is an effect of the finger-movement, but it may be more appropriate to say that it is an effect
of the shooting. I leave this issue unresolved here.
90
shooting in virtue of the shooting’s causing the death, but this doesn’t require saying that the death
made the killing exist in the past in any sense.

Conclusion
The view that actions are colocated with our body movements is very appealing, since what is
beyond our body movements seems to be in an importance sense beyond our control (or more
literally: beyond our reach). Davidson’s view of action is one view that does justice to this intuition,
and I have sketched one other such view, but this intuition comes under fire when we confront the
problem of the time of a killing. Although this problem is thorny and provides a problem for
everybody, I think there is a way to answer it and save this intuition about the location of action. I
have argued that the means necessary for this can be used regardless of one’s theory of action, so it is
available to both Davidson and myself. However, I further suggested that it is reasonable to say that
actions such as killing are constituted by body movements in the context of those movements having
certain effects.

91
Concluding Remarks
It is plausible to think that all actions (or at least all intentional actions) are done for and
explained by reasons. Causal theories of action offer analyses of the concept of action that attempt to
capture this plausible idea by construing this explanation causally. Events are actions, then, in virtue
of being caused in the right way by the right mental events. This is compelling, but it is not enough to
give us necessary and sufficient conditions for when events will be actions.
117
We need a picture of
which events, when caused in the right way by the right mental events, are actions.
As we saw, Donald Davidson gave us a picture of which events are actions: actions will be
identical to certain movements of the body (those caused in the right way by the right mental events).
While this view was initially plausible, we saw that there are a series of objections that can be raised to
it. I argued that Davidson’s view was incorrect, but these objections could suitably be answered in
ways that did not threaten (and in fact suggested) a revised version of the Causal Theory of Action. I
motivated and gave such a revised version, and I showcased some of its advantages. In these
concluding remarks, I want to say in more detail what I take myself to have shown in this dissertation.
Further, I want to consider how what I have shown should influence our thinking about the Causal
Theory of Action more broadly.
Although Davidson’s Identity Thesis is parsimonious and motivated, in chapter one I offered
several arguments (from modal, causal, and property differences) to conclude that many regular
actions cannot be identified with body movements. There is a package of views involving a coarse-
grained theory of events and counterpart theory that a Davidsonian could use to defend the Identity
Thesis, but this is not a plausible package of views. More importantly though, I suggested that we did
not need the Identity Thesis to maintain a causal theory. Instead, I suggested that if actions are not
identical to body movements, they nevertheless could be constituted by them, just as the statue is
constituted by the lump of clay. We saw how the relation between actions and movements seems to
share many of the features of the constitution relation, and that Kit Fine and Lynne Baker even
explicitly took their views of constitution to extend to action. My contribution was to recognize how
this relation allows for an easy revision of the CTA. On this view, an event is an action iff it is a body
movement or constituted by a body movement and caused in the right way by the right mental events.

117
It may be possible to spell out ‘the right way’ that actions are caused by mental events that is sufficient to pick out all
and only actions. I leave this possibility to the side, however, first because it involves giving a full solution to the problem
of deviant causal chains, and, second, because the bigger question we were engaged with concerned how actions relate to
other events such as body movements.
92
Moreover, that items are constituted in the right circumstances is suggestive of the causalist intuition
that events are actions in the right causal circumstances.
Davidson’s view can be extended to include omissions – they will be identical to however the
agent is moving as she omits. However, the Identity Thesis is perhaps least plausible in this context.
Further, it is unclear what intentional omissions are such that they can be captured by some version
of the CTA. Given this, chapter two was largely dedicated to the project of legitimizing omissions as
events. I showed how omissions could be captured on Kim’s fine-grained view of events, how the
act/result distinction could be used to locate omissions with the agent, and how we will not count as
responsible for omitting to do too many things. Once these issues were resolved, we were in a position
to see that omissions are events able to be caused in the right way by the right mental states. I argued
that it is appropriate to think that omissions are typically constituted by body movements as well, and
so they will fit into my revised CTA.
A final problem for Davidson’s Identity Thesis as well as the CTA concerned actions such as
killing or insulting. These seem to involve some event that is distant from the agent. These cases are
ubiquitous, and holding onto a view that actions are at least colocated with movements of the body
seems to lead us to say unintuitive things, such as that ‘A killed B before B died.’ To answer this, I
sought the correct definition of ‘to kill’ and other causal verbs, and I argued that the definition did not
determine amongst several options which event was the referent of expressions such as ‘A’s killing of
B.’ This indeterminacy of reference explained the unintuitiveness of locating the killing solely with the
agent, solely with the patient, or with some event that includes both. However, although it is
indeterminate to which event ‘the killing’ refers, it could still be that the action of the agent was
definitely located somewhere. I suggested that this solution could be used by proponents of different
views about the location of action, including Davidson and myself. I concluded by showing how these
apparently distant actions could be captured on my revised CTA.
By using various metaphysical and linguistic resources, I showed how we can be led to a view
of actions and omissions that avoids the problems of Davidson’s Identity Thesis. Recognizing that
actions (and omissions) may be constituted by body movements allowed us to give a revised version
of the CTA, but it does more than this. It gives us a more fine-grained picture of action amenable for
understanding how actions can be related to each other as well as body movements, and it also gives
a cohesive picture of both actions and omissions as intentional expressions of agency.
It may be that my particular revision of the CTA itself needs revision. After all, I did not argue
at length for the constitution relation itself or against direct objections to the claim that some actions
93
are constituted. However, what we have seen in the dissertation may change our expectations about
whether such a revision is likely to be found. At the start, the fear was that rejecting Davidson’s
Identity Thesis would quickly lead us to reject the CTA itself. That fear can hopefully now be seen to
be unfounded. We saw in the first chapter that rejecting the Identity Thesis and finding an alternative
turned not on any deep truth about the nature of purposive action; rather, it was a matter of event-
individuation and synchronic relations. Similarly, in the third chapter, what was needed was adducing
the right linguistic tools, not suggesting a non-causal mode of the explanation of action.
There are certain legitimate concerns that a proponent of the CTA must face. She must argue
that reasons can explain actions by causing them, that a non-deviant causal explanation can be found,
and that a casual explanation is a better expression of reasons-explanation than possible non-causal
explanations. If we accept actions as events and take events as the causal relata, however, then it should
not be surprising that there will be some way or other to reformulate the CTA. The opponent of the
CTA too will need to say which events explained in the right (non-causal) way are actions, after all. Of
course, I think all actions will be body movements or constituted by body movements. Even if I am
mistaken though, I hope to have gone some way towards refuting the idea that certain actions by their
nature or location are not up for being causally explained by reasons.

94
References

Introduction
Aguilar, J. & Buckareff, A. (2010) “The CTA: Origins and Issues” in Causing Human Action. Aguilar,
J. & Buckareff, A. (Ed.). MIT Press. pg. 1-26.
Anscombe, E. (1957) Intention. 2
nd
Edition. Cornell University Press.
Bishop, J. (1989) Natural Agency: An Essay on the Causal Theory of Action. Cambridge University Press.
Davidson, D. (1963/2001) “Actions, Reasons, and Causes” in Essays on Actions and Events. Oxford
University Press pg. 3-20.
Davidson, D. (1971/2001) “Agency” in Essays on Actions and Events. Oxford University Press pg. 43-
62.
Davidson, D. (1973/2001) “Freedom to Act” in Essays on Actions and Events. Oxford University Press
pg. 63-82.
Dretske, F. (1988) Explaining Behavior: Reasons in a World of Causes. MIT Press.
Frankfurt, H. (1978) “The Problem of Action,” American Philosophical Quarterly, 15:2, pg. 157-162.
Ginet, C. (1990) On Action. Cambridge University Press.
Hampshire, S. (1959) Thought and Action. Chatto and Windus.
Kenny, A. (1963) Action, Emotion, and Will. Routledge & Kegan Paul Limited.
Melden, A.I. (1961) Free Action. Routledge & Kegan Paul Limited.
O’Brien, L. (2012) “Deviance and Causalism,” Pacific Philosophical Quarterly, 93:2, pg. 175-196.
Schueler, G.F. (2003) Reasons and Purposes: Human Rationality and the Teoleological Explanation of Action.
Oxford University Press.
Sehon, S. (1998) “Deviant Causal Chains and the Irreducibility of Teleological Explanation,” Pacific
Philosophical Quarterly, 78, pg. 195-213.
Sehon, S. (2005) Teleological Realism: Mind, Agency, and Explanation. Cambridge, MA: MIT Press.
Shepherd, J. (2014) “The Contours of Control,” Philosophical Studies, 170:3, pg. 395-411.
Wilson, G. (1989) The Intentionality of Human Agency. Stanford, CA: Stanford University Press.
Wittgenstein, L. (1953) Philosophical Investigations.

Chapter One
Anscombe, G.E.M. (1957) Intention. Basil Blackwell.
Bach, K. (1980) “Actions Are Not Events,” Mind, 89:353, pg. 114-120.
Baker, L. (1997) “Constitution is not Identity,” The Journal of Philosophy, 94:12, pg. 599-621.
Baker, L. (1998) “What We Do: A Nonreductive Approach to Human Action” in Human Action,
Deliberation, and Causation, Eds. Branson, J. & Cuypers, S. E. Philosophical Studies Series, 77.
Kluwer Academic Publishers. pg. 249-270.
Baker, L. (2000) Persons and Bodies. Cambridge: Cambridge University Press.
Baker, L. (2002a) “Replies,” Philosophy and Phenomenological Research, 64:3, pg. 623-635.
Baker, L. (2002b) “On Making Things Up: Constitution and Its Critics,” Philosophical Topics, 30:1, pg.
31-51.
Bennett, J. (1994) “The ‘Namely’ Analysis of the ‘By’-Locution,” Linguistics and Philosophy, 17, pg. 29-
51.
Bennett, J. (1995) The Act Itself. Oxford University Press.
Bishop, J. (1989) Natural Agency. Cambridge University Press.
95
Bishop, J. (2010) “Skepticism about Natural Agency and the Causal Theory of Action” in Causing
Human Actions: New Perspectives on the Causal Theory of Action, Eds. Aguilar, J. and Buckareff, A.
MIT Press. pg. 69-83.
Brower, J. (2011) “Aristotelian Endurantism: A New Solution to the Problem of Temporary
Intrinsics,” Mind, 119: 883-905.
Davidson, D. (1963) “Actions, Reasons, and Causes,” The Journal of Philosophy, 60:23 pg. 685-700.
Davidson, D. (1971) “Agency” in Essays on Actions and Events. Oxford University Press pg. 43-62.
Davidson, D. (1969/2011) “The Individuation of Events” in Essays on Actions and Events. Oxford
University Press pg. 163-180.
Davidson, D. (1985/2011) “Reply to Quine on Events” in Essays on Actions and Events. Oxford
University Press, pg. 305-312.
Evnine, S. (2016) Making Objects and Events: A Hylomorphic Theory of Artificacts, Actions, and Organisms.
Oxford University Press.
Fairchild, M. (forthcoming) “A Paradox of Matter and Form,” Thought.
Fine, K. (1982) “Acts, Events, and Things” in Sprache und Ontologie. Vienna: Hoelder-Pichler-Tempsky,
pg. 97-105.
Fine, K. (2003) “The Non-Identity of a Thing and its Matter,” Mind, 112, pg. 195-234.
Francescotti, R. (2005) “Constitution and the Necessity of Identity,” Logique & Analyse, 48:192, pg.
311-321.
Gibbard, A. (1975) “Contingent Identity,” Journal of Philosophical Logic, 4, pg. 187-221.
Gibbons, J. (2006) “Mental Causation without Downward Causation,” Philosophical Review, 115:1, pg.
79-103.
Goldman, A. (1971) “The Individuation of Action,” The Journal of Philosophy, 68:21, pg. 761-774.
Horsby, J. (1980) Actions. London: Routledge & Kegan Paul.
Johnston, M. (1992) “Constitution is Not Identity,” Mind, 101:401, pg. 89-105.
Kim, J. (1976) “Events as Property Exemplifications” in Action Theory, Eds. Brand, M. and Walton, D.
Dordrecht, Holland: Reidel, pg. 159-177.
Koslicki, K. (2008) The Structure of Objects. Oxford University Press.
Kripke, S. (1980) Naming and Necessity. Harvard University Press.
Lewis, D. (1968) “Counterpart Theory and Quantified Modal Logic,” The Journal of Philosophy, 65, pg.
113-26.
Lewis, D. (1971) “Counterparts of Persons and Their Bodies,” The Journal of Philosophy, 68, pg. 203-11.
Lewis, D. (1987) “Events” in Philosophical Papers, Vol. II. Oxford University Press.
Lowe, E. J. (1983) “Instantiation, Identity, and Constitution,” Philosophical Studies, 44, pg. 45-59.
Mele, A. (1997) “Introduction” in The Philosophy of Action. Ed. Mele, A. Oxford University Press.
pg.1-26.
Noonan, H. W. (1993) “Constitution is Identity,” Mind, 102:405, pg. 133-146.
O’Shaughnessy, B. (1973) “Trying (as the Mental Pineal Gland’),” The Journal of Philosophy, 70:13, pg.
365-386.
Paul, L. A. (2007) “Constitutive Overdetermination” In Topics in Contemporary Philosophy, Volume IV:
Causation and Explanation, pg. 265-290.
Pereboom, D. (2011) Consciousness and the Prospects of Physicalism. Oxford University Press.
Pickel, B. (2010) “There is No ‘Is’ of Constitution,” Philosophical Studies, 147, pg. 193-211.
Quine, W.V.O. (1960) Word and Object. Cambridge, Mass.: MIT Press.
Sattig, T. (2015) The Double Lives of Objects: An Essay in the Metaphysics of the Ordinary World. Oxford
University Press.
96
Schlosser, M. (2009) “Non-Reductive Physicalism, Mental Causation, and the Nature of Actions” in
Reductionism: Between the Mind and the Brain, Eds. Hieke, A. and Leitgeb, H., Transaction Books:
Rutgers University, pg. 73-90.
Schneider, B. (2009) “‘By’ A Refutation of the Anscombe Thesis,” Linguistics and Philosophy, 31, pg.
649-669.
Schwarz, W. (2013) “Contingent Identity,” Philosophy Compass, 8:5, pg. 486-495.
Thomasson, A. (1998) “A Nonreductive Solution to Mental Causation,” Philosophical Studies, 89,
pg.181-195.
Wasserman (2004) “The Constitution Question,” Noûs, 38, pg. 693-710.
Wasserman (2013) “Material Constitution” in The Stanford Encyclopedia of Philosophy, Ed. Zalta, E., URL
= http://plato.stanford.edu/entries/material-constitution/.
Wiggins, D. (1968) “On Being in the Same Place at the Same Time,” Philosophical Review, 77:1, pg. 90-
95.
Wiggins, D. (2004) Sameness and Substance Renewed. Cambridge, UK: Cambridge University Press.
Wilson, R. (2009) “The Transitivity of Material Constitution,” Noûs, 43:2, pg. 363-377.
Wittgenstein, L. (1958) Philosophical Investigations. New York: The Macmillan Company.
Von Wright, G. (1963) Norm and Action: A Logical Inquiry. Humanities Press.
Zimmerman, D. (2002) “Persons and Bodies: Constitution Without Mereology?” Philosophy and
Phenomenological Research, 64:3, pg.599-606.

Chapter Two
Anscombe, G.E.M. (1957) Intention. Oxford, UK: Basil Blackwell.
Bach, K. (1980) “Actions are Not Events,” Mind, 89:353, pg. 114-120.
Baker, L. R. (2000) Persons and Bodies. Cambridge, UK: Cambridge University Press.
Beebee, H. (2004) “Causing and Nothingness” in N. Hall, J. Collins, and L.A. Paul (Eds.), Causation
and Counterfactuals. Cambridge, MA: MIT Press. pg. 291-308.
Bennett, J. (1988) Events and Their Names. Indianapolis, IN: Hackett Publishing Company.
Bernstein, S. (2014a) “Omissions as Possibilities,” Philosophical Studies, 167:1, pg. 1-23.
Bernstein, S. (2014b) “Two Problems for Proportionality about Omission,” Dialectica, 68:3, pg. 429-
441.
Bishop, J. (2010) “Skepticism about Natural Agency and the Causal Theory of Action” in J. Aguilar
and A. Buckareff (Eds.), Causing Human Actions: New Perspectives on the Causal Theory of Action.
Cambridge, MA: MIT Press. pg. 69-83.
Brand, M. (1971) “The Language of Not Doing,” American Philosophical Quarterly, 8:1, pg. 45-53.
Clarke, R. (2011) “Omissions, Responsibility, and Symmetry,” Philosophy and Phenomenological Research,
82:3, pg. 594-624.
Clarke, R. (2014) Omissions: Agency, Metaphysics, and Responsibility. New York, NY: Oxford University
Press.
Davidson, D. (1963) “Actions, Reasons, and Causes,” The Journal of Philosophy, 60:23, pg. 685-700.
Davidson, D. (1971/2011) “Agency” in Essays on Actions and Events, 2
nd
edition. Oxford: Oxford
University Press. pg. 43-61.
Davidson, D. (1985) “Replies to Essays I-IX” in B. Vermazen and M. Hintikka (Eds.), Essays on
Davidson: Actions and Events. New York, NY: Oxford University Press. pg. 195-230.
Davidson, D. (1985/2011) “Reply to Quine on Events” in Essays on Actions and Events, 2
nd
edition.
Oxford: Oxford University Press. pg. 305-312.
Dowe, P. (2010) “Proportionality and Omissions,” Analysis, 70:3, pg. 446-451.
97
Fischer, J. M. & Ravizza, M. (1998) Responsibility and Control. Cambridge, UK: Cambridge University
Press.
Kim, J. (1976) Events as Property Exemplifications” in M. Brand and D. Walton (Eds.), Action Theory.
Dordrecht, Holland: Reidel. pg. 159-177.
Lewis, D. (2004) “Void and Object” in N. Hall, J. Collins, and L.A. Paul (Eds.), Causation and
Counterfactuals. Cambridge, MA: MIT Press. pg. 277-290.
McCann, H. (1974) “Volition and Basic Action,” The Philosophical Review, 83:4, pg. 451-473.
McGrath, S. (2005) “Causation by Omission: A Dilemma,” Philosophical Studies, 123:1/2, pg. 125-148.
Mellor, D.H. (2004) “For Facts as Causes and Effects” in N. Hall, J. Collins, and L.A. Paul (Eds.),
Causation and Counterfactuals. Cambridge, MA: MIT Press. pg. 309-323.
Quine, W.V.O. (1960) Word and Object. Cambridge, MA: MIT Press.
Sartorio, C. (2009) “Omissions and Causalism,” Noûs, 43:3, pg. 513-530.
Sartorio, C. (2010) “The Prince of Wales Problem for the Counterfactual Theory of Causation” in A.
Hazlett (Ed.), New Waves in Metaphysics. New York, NY: Palgrave Macmillan. pg. 250-276.
Schaffer, J. (2005) “Contrastive Causation,” The Philosophical Review, 114:3, pg. 327-358.
Shepherd, J. (2014) “Causalism and Intentional Omission,” American Philosophical Quarterly, 51, pg. 15-
26.
Thomson, J.J.T. (2003) “Causation: Omissions,” Philosophy and Phenomenological Research, 66:1, pg. 81-
103.
Varzi, A. (2007) “Omissions and Causal Explanations” in F. Castellani and J. Quitterer (Eds.), Agency
and Causation in the Human Sciences. Paderborn, Germany: Mentis Verlag. pg. 155-167.
Vermazen, B. (1985) “Negative Acts” in B. Vermazen and M. Hintikka (Eds.), Essays on Davidson:
Actions and Events. New York, NY: Oxford University Press. pg. 93-104.
von Wright, G. H. (1963) Norm and Action: A Logical Inquiry. London, UK: The Humanities Press.
Zangwill, N. (2011) “Negative Properties,” Noûs, 45:3, pg. 528–556.

Chapter Three
Anscombe, G.E.M. (1957) Intention. Basil Blackwell.
Baker, L. (2000) Persons and Bodies. Cambridge: Cambridge University Press.
Bennett, J. (1973) “Shooting, Killing and Dying,” Canadian Journal of Philosophy, 2:3, pg.315-323.
Davidson, D. (1969/2001) “The Individuation of Events” in Essays on Actions and Events, 2
nd
Edition.
Oxford University Press, pg. 163-180.
Davidson, D. (1971/2001) “Agency” in Essays on Actions and Events, 2
nd
Edition. Oxford University
Press, pg. 43-62.
Davidson, D. (1985/2001) “Adverbs of Action” in Essays on Actions and Events, 2
nd
Edition. Oxford
University Press, pg. 293-304.
Dunbar, G. (2001) “Towards a Cognitive Analysis of Polysemy, Ambiguity, and Vagueness,” Cognitive
Linguistics, 12:1, pg.1-14.
Fodor, J. (1970) “Three Reasons for Not Deriving ‘Kill’ From ‘Cause to Die,’” Linguistic Inquiry, 1:4,
pg. 429-438.
Hall, N. (2000) “Causation and the Price of Transitivity,” Journal of Philosophy, 97, pg. 198-222.
Lombard, L. (1978) “Actions, Results, and the Time of a Killing,” Philosophia: Philosophical Quarterly of
Israel, 8:2-3, pg. 341-354.
Lombard, L. (1989) “‘Unless’, ‘Until’, and the Time of a Killing,” Pacific Philosophical Quarterly, 70:2, pg.
135-154.
98
Lombard, L. (2003) “The Cambridge Solution to the Time of a Killing,” Philosophia: Philosophical
Quarterly of Israel, 31:1-2, pg. 93-106.
Morreall, J. (1976) “The Nonsynonymy of Kill and Cause to Die,” Linguistic Inquiry, 7:3, pg. 516-518.
Mossel, B. (2001) “The Individuation of Actions,” Australasian Journal of Philosophy, 79:2, pg. 258-278.
Pietroski, P. (1998) “Actions, Adjuncts, and Agency,” Mind, 107:425, pg. 73-110.
Pols, A. J. K. (2013) “Choosing Your Poison and the Time of a Killing,” Philosophical Studies, 165, pg.
719-733.
Price, M. (1982) “Causal Verbs and the Individuation of Actions,” Southern Journal of Philosophy, 20:3,
pg. 367-374.
Ruben, D. (1999) “Act Individuation: The Cambridge Theory,” Analysis, 59:4, pg. 276-283.
Thomson, J. J. T. (1971) “The Time of a Killing,” The Journal of Philosophy, 68:5, pg. 115-132.
Thomson, J. J. (1987) “Verbs of Action,” Synthese, 72:1, pg. 103-122.
Weintraub, R. (2003) “The Time of a Killing,” Analysis, 63:3, pg. 178-182.
Wierzbicka, A. (1975) “Why ‘Kill’ Does Not Mean ‘Cause to Die’: The Semantics of Action
Sentences,” Foundations of Language, 13:4, pg. 491-528.
White, A. (1980) “Shooting, Killing and Fatally Wounding,” Proceedings of the Aristotelian Society, 80, pg.
1-15.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Unlimited ontology and its limits

PDF

What it is to be located

PDF

The boundaries of the intentional

PDF

Language and the structure of Berkeley's world

PDF

The minimal approval view of attributional-responsibility

PDF

Feeling good and living well: on the nature of pleasure and its role in well-being

PDF

The metaphysics of social groups: structure, location, and change

PDF

A perceptual model of evaluative knowledge

PDF

Comparative iIlusions at the syntax-semantics interface

PDF

Responsibility and rights: a search for a principled distinction between criminal law and tort law

PDF

The morphosyntax of states: deriving aspect and event roles from argument structure

PDF

Positivist realism

PDF

Cognitive boundaries for rational coherence requirements

PDF

Rationality and the primacy of the occurrent

PDF

Aggregating complaints

PDF

Representation, truth, and the metaphysics of propositions

PDF

Contract, from promise to commodity

PDF

Minimal sensory modalities and spatial perception

PDF

Defending Rawls's Law of Peoples against cosmopolitan critiques

PDF

Negotiating pluralism and tribalism in liberal democratic societies

Asset Metadata

Creator Silver, Kenneth (author)

Core Title The constitution of action

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Philosophy

Publication Date 08/23/2017

Defense Date 08/23/2017

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Donald Davidson,metaphysics,OAI-PMH Harvest,omissions,philosophy of action,time of a killing

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Hawthorne, John (committee chair), Uzquiano, Gabriel (committee chair), Artstein, Ron (committee member), Vihvelin, Kadri (committee member), Watson, Gary (committee member)

Creator Email kennethmsilver@gmail.com,kmsilver@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-423675

Unique identifier UC11265462

Identifier etd-SilverKenn-5689.pdf (filename),usctheses-c40-423675 (legacy record id)

Legacy Identifier etd-SilverKenn-5689.pdf

Dmrecord 423675

Document Type Dissertation

Rights Silver, Kenneth

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA