University of Southern California Dissertations and Theses

The minimal approval view of attributional-responsibility

(USC Thesis Other)

The minimal approval view of attributional-responsibility

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content -1-

THE MINIMAL APPROVAL VIEW OF
ATTRIBUTIONAL-RESPONSIBILITY

by

August G. Gorman

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA

In partial fulfillment of the requirements of the degree of
DOCTOR OF PHILOSOPHY
(PHILOSOPHY)

December 2018

Copyright 2018 August G. Gorman
-2-

to Sam, for believing in my dreams, for believing in us, and for believ-
ing that those two things were compatible.
-3-
Acknowledgments

I do not come from a family of academics. I don’t think my mother un-
derstood why, after breaking my leg on the free trip to Aruba she’d won
and had taken me with her on after my second year of grad school, I was
content to have her wheel me out to read Susan Wolf’s Freedom Within
Reason on the beach. Nor do I think she understood my excitement when I
read a passage in that book in which Wolf briefly floats the idea that de-
sires cannot speak for us as agents because “the prospect of satisfying
such desires may not be preferable to the prospect of eliminating these de-
sires in other ways.”
1
(I thought I had discovered right then and there the
answer to the question of attributable agency. I still think so, but it turns
out that fleshing out the account was a bit more complicated, and occu-
pied me plenty for the next three years.) My parents have a running joke
about which of the two of them I got my “smarts” from, the joke being
that my interest in abstract conceptual matters seems to come completely
out of left field. I don’t know the degree to which they realize just how
much of my ability to carry out a project like this was shaped by them:
how I learned to have a critical eye and a dogged persistence from my
dad, how I learned to seek out only what actually interests me and to
forge a path that is unequivocally my own from my mom. For that, and
for their unwavering emotional and financial support over the course of
this journey, I am truly grateful.
I didn’t have the slightest idea what philosophy was when I arrived at
college at the University of Mary Washington, but for whatever reason
had a hunch that I might like it. I owe a tremendous debt to the professor
who taught my first philosophy class, Nina Mikhalevsky, who made it
seem like the world was endlessly full of puzzles, the pursuit of any of
which might be worth devoting a lifetime to solving. My grad school aspi-
rations were cemented by Sam Emswiler, who broke her personal ban on
ever recommending that a student go on to grad school in philosophy to
suggest that despite the odds, in my case I absolutely had to. I was given
the autonomy to pursue a fairly idiosyncratic research project in thinking

1
Wolf (1990): 31.
-4-
about how first-personal accounts of the lived experience of Tourette syn-
drome line up with Harry Frankfurt’s discussion of agential alienation
under the guise of a senior independent study with my ever-supportive
undergrad advisor, Jason Matzke. For that I am especially grateful, as
much of that research informs my work even today.
I wouldn’t trade my experiences at the University of Mary Washington
for the world, but there was a learning curve between undergrad and grad
school that at many times felt insurmountably steep. Having come from a
relatively unknown public school in Fredericksburg, Virginia, the first
hurdle was merely to actually get in. I count myself incredibly lucky to
have come across an application online for the summer seminar Bob
Pasnau runs for students from relatively unknown colleges with philoso-
phy grad school aspirations. Through working with Bob, who I must
thank for seeing early signs of potential in my work, I honed a writing
sample and gained a letter of recommendation that I am sure I have, in
great part, to thank for my acceptances.
After visiting six other schools, I was sold on USC after a thirty minute
Skype call with Mark Schroeder. From working with me on issues of ac-
cessibility my first year, to coordinating my job market letters with a very
quick turnaround time, I am thankful for Mark’s continued support and
commitment to making sure no USC grad student ever slips through the
cracks. Thanks are due to all of the professors I worked with at USC, but
especially to Kadri Vihvelin, Elyn Saks, and Janet Levin, who each influ-
enced the early stages of development of this dissertation in different
ways. Kadri Vihvelin wrote a comment on a draft of mine that read
“weakness of will is unfreedom?!!!”Answering that worry grew into
Chapter 3 of this dissertation. Working with Elyn Saks, whose impressive
body of work builds off of her own experiences with mental health, in-
spired me to have the courage to explicitly write about my own disorder,
misophonia, in this dissertation, and I think my discussion is greatly en-
riched for it. Janet Levin served as a member of my qualifying committee
and gave very helpful comments on early drafts of several chapters.
Of course my biggest debt of gratitude at USC is to my committee.
Thanks to John Monterosso for serving as my external committee member,
and for encouraging me to clarify the methodological underpinnings of
-5-
my project. Thanks to Steve Finlay for playing the role of my greatest critic
and for giving me incredibly detailed comments on several chapters.
Knowing that this dissertation would be read by at least one person who
is deeply suspicious of the very idea of a Deep Self view has been instru-
mental in broadening the audience I aim to speak to in my work. Thanks
are due to Dave Shoemaker, first, for writing Responsibility from the Mar-
gins, a book I read closely and excitedly in a reading group my third year
that showed me what an exemplar of the kind of work I wanted to do
might look like. Dave’s work continually shows a commitment, first and
foremost, to paying careful attention to what our responsibility practices
imply for how we treat the real and imperfect people in our lives, one
which I have tried to emulate in my own work. Thanks also for going out
of your way to support my work and serve as a member of my committee
even amidst your research and teaching obligations at your own universi-
ty.
Thanks to Gary Watson for several semesters of independent studies
and countless long meetings, helping me both to chart the lay of the land
in the free will and moral responsibility literature and to find my own
voice within it. Thank you for helping me learn how to emulate you in
prioritizing wisdom over cleverness, and for being a sounding board for
the earliest incarnations of the ideas this dissertation builds upon.
I owe a larger debt of gratitude to my advisor, Jon Quong, than I can
adequately put into words. On the advice of other graduate students I ap-
proached Jon at the end of my third year with what I knew to be an out-
landish request. I asked him to be my advisor despite the fact that I:
1. hadn’t so much as introduced myself to him in the three years I’d
been at USC,
2. didn’t work in his area, and
3. was moving, and so would have to work entirely remotely via
Skype.
For whatever reason, he agreed instantly, and I could not be more thank-
ful that he did. Over the years he has proven himself to be nothing short
of the Platonic ideal of an advisor: critical, kind, thorough, encouraging,
and trusting. He has read and given me extremely helpful feedback on
more drafts of this material than I can count. I am not sure whether our
-6-
working styles just happen to line up perfectly or if he has an uncanny
ability to intuit exactly what would best suit each of his advisees, but
through a combination of faith in my ability to set my own goals and
deadlines and encouragement that my projects were viable even in the
face of criticism, he helped me find my confidence at a time when it was
sorely needed. This dissertation simply would not have come into exist-
ence without his guidance and encouragement.
I am thankful also for the community of grad students at USC. In par-
ticular, I don’t know where I would be without heart-to-heart conversa-
tions over the years with Aness Webster, Kenneth Silver, and Ara Astou-
rian or without having Rima Basu, Renee Jorgensen Bolinger, and Maegan
Fairchild in the cohorts above me to give me advice. To Caleb Perl, Abe-
lard Podgorski, Nick Laskowski, Alex Dietz, Joe Horton, Nathan Robert
Howard, Jonathan Charles Wright, and anyone else whose money I took
at our weekly poker game, thank you for fostering a sense of camaraderie
and for being a part of my favorite thing in LA. I am especially grateful for
Tanya Kostochka, for making sure I didn’t stay holed up in my apartment
doing work 24-7, for going through the whole grad school experience as
thick as thieves with me from day one.
Outside of USC, I am truly grateful for the incredibly kind philoso-
phers in the agency and responsibility subfield who instantly welcomed
me into their ranks and made me feel like I had something important to
contribute to the field. Without their encouragement I doubt I would have
had the nerve to take on a project this ambitious. Special thanks to Mi-
chael McKenna, Tamler Sommers, Phil Swenson, Matt Talbert, Justin
Coates, Neal Tognazzini, Gunnar Björnson, Ben Mitchell-Yellin, and Ben
Matheson for all the conversations at conferences, often over beers, that in
one way or another informed my perspective on this material.
As for my chosen family from the East Coast, I never could have done
this without your support. Thanks to Eugene Chislenko for bringing me in
to guest lecture on this material at Temple, and for helping fund my last
summer of writing by letting me be your research assistant. Thanks to
Geena Cain for being the least excited person about me going off to grad
school in LA, but supporting me every step of the way nonetheless.
Thanks to Eben Hilpert for helping me learn how to take small breaks and
-7-
relax. And thanks to Elsie Bartlett for helping me get through those first
couple years and for making me a home-cooked dinner every single night
of my most difficult semester.
Kari David Collins is the person I narrate my life to in real-time, and so
over the past few years has heard more than any layperson would ever
reasonably want to know about the day-to-day grind of writing a disserta-
tion on attributional-responsibility. I have so much to thank you for, but
especially for always being my listening ear and for every single trophy
emoji that kept me going. (And for playing the Galinda to my Elphaba on
head-clearing sing-alongs to the Wicked cast album on our cross-country
road-trips.)
Finally, thank you to my partner, Sam Cirulis. It’s been quite a journey,
to say the least. Thank you for trusting that we could make a long-distance
relationship work despite meeting me right before I flew off to LA for a
PhD program. Thank you for flying out to see me even that one time it
meant getting stranded in Dallas due to Spirit Airlines, thank you for
bringing joy into my life during the toughest times along this journey, and
thank you for helping me land my postdoc by pretending to be every pos-
sible archetype of job interviewer while I was practicing. There have been
many days when you single-handedly kept me going, and I hope you
know that I intend to spend the rest of our lives together returning the fa-
vor.

-A.G.G.

-8-
Table of Contents
Chapter 1: A Common Necessary Condition for Attributional-Responsibility ................ 10
1. Introduction ......................................................................................................................... 10
2. Actual Sequence Compatibilism ....................................................................................... 11
2.1 Dialectical Motivation for Actual Sequence Compatibilism ................................... 11
2.2 Support for Actual Sequence Compatibilism ........................................................... 17
3. Deep Self Theory ................................................................................................................. 21
3.1 The “Deep” in “Deep Self” .......................................................................................... 25
4. Attributional-Responsibility .............................................................................................. 28
5. B-Tradition versus H-Tradition ........................................................................................ 31
6. Internality’s Role in Deep Self Views ............................................................................... 33
7. Against Skepticism about Internality .............................................................................. 37
8. Internality is Approval ....................................................................................................... 41
9. Approval as a Common Necessary Condition for Attributional-
Responsibility .......................................................................................................................... 42
9.1 Approving is Necessary on the Endorsing View ..................................................... 42
9.2 Approving is Necessary on the Valuing View ......................................................... 43
9.3 Approving is Necessary on the Planning View ........................................................ 46
9.4 Approving is Necessary on the Caring View ........................................................... 48
Chapter 2: A Sufficient Condition for Attributional-Responsibility ................................... 52
1. Introduction ......................................................................................................................... 52
2. Unnecessary Components of Deep Self Views ............................................................... 53
2.1 Valuing is Not Necessary for Attributional-Responsibility .................................... 54
2.2 Neither Planning nor Caring is Necessary for Attributional-
Responsibility ...................................................................................................................... 58
3. From Type-Disjunctive to Token-Disjunctive Views .................................................... 61
4. Partial Identification ........................................................................................................... 64
4.1 Partial Identification and Desire Individuation ....................................................... 70
5. Hypothetical Versus Explicit Endorsement .................................................................... 71
6. Approving as Endorsement with a Further Aim than Elimination ............................ 75
7. Minimal Approval and the Mechanism of Action ......................................................... 82
8. Advantages of the View ..................................................................................................... 86
8.1 Solving Frankfurt’s Infinite Hierarchy Problems ..................................................... 86
8.2 Capacity Without Process ............................................................................................ 89
8.3 Criterion Operates Independently from the Type of Mental State that
Causes Action ...................................................................................................................... 91

-9-
Chapter 3: How Should Deep Self Theorists Account for Weakness of Will? .................. 97
1. Introduction ......................................................................................................................... 97
2. The Weakness of Will Problem ......................................................................................... 98
3. Current Solutions and Their Problems .......................................................................... 101
3.1 Mismatched Accounts ................................................................................................ 102
3.2. Skepticism About Weakness of Will ....................................................................... 104
3.3. Attributability for Failure to Exercise a Deep Self Capacity ................................ 106
4. Adopting a Mosaic Conception of the Deep Self ......................................................... 115
4.1. The Minimal Approval View as a Mosaic Deep Self View that Can Solve
the Weakness of Will Problem ........................................................................................ 117
4.2 Can Other Deep Self Views Adopt a Mosaic Conception to Solve the
Weakness of Will Problem? ............................................................................................. 118
5. Responses to Fairness Objections ................................................................................... 124
Chapter 4: The Finely Individuated Trait View of Blame’s Content ................................ 127
1. Introduction ....................................................................................................................... 127
2. From Attributability to Blameworthiness ..................................................................... 128
3. Robust Traits and Implicit Judgments: Between Scylla and Charybdis ................... 130
4. The Ledger View ............................................................................................................... 133
5. The Finely Individuated Trait View ............................................................................... 134
5.1 A Paradigm Shift for Thinking About Aretaic Traits ............................................ 134
5. The Time-Slice Property Objection ................................................................................. 140
6. The Minority Report objection: Pre-Blame .................................................................... 143
6.2 Victim Relation and the Standing to Blame ............................................................ 145
6.3 Stubborn Incompatibilist Intuitions ......................................................................... 147
6.4 Denying the Existence of Resultant Moral Luck .................................................... 148
Chapter 5: The Case of Forgetting .......................................................................................... 151
1. Introduction ....................................................................................................................... 151
2. The Minimal Approval View and The Case of Forgetting ......................................... 152
3. The Tracing Strategy ........................................................................................................ 157
4. An Argument Against the Minimal Approval View as the Correct View of
Attributability? ...................................................................................................................... 160
4.1 Sher’s Agent-Constituting Mechanism View .......................................................... 161
4.2 Smith’s Rational Relations View ............................................................................... 162
5. Accountability Without Attributability ......................................................................... 167
Bibliography .............................................................................................................................. 176
-10-
Chapter 1: A Common Necessary Condition for Attributional-
Responsibility

1. Introduction
Imagine that your friend Corey has promised to come over and support
you after the death of your pet. You sit waiting for him to appear at the
time he said he’d be there, but time continues to tick by and he does not
show up. As it becomes clear that he’s not going to show up, you might
blame Corey. In response to the perceived slight, you might start to feel
angry with him, come to judge him as being insensitive, or even begin to
question your friendship. But now, suppose you find out that Corey suf-
fers from OCD and his absence is explained by the fact that while he des-
perately wanted to come support you, he felt the need to instead act on a
compulsive desire to stay at home repeatedly performing rituals, turning
the lights in each room of his house on and off several times.
It seems that you have reason to suspend your reactions to your friend
in light of finding out this information. You shouldn’t blame Corey be-
cause it seems he wasn’t, after all, being insensitive about your situation,
since his absence was caused by his compulsive behavior. But why do
conditions like Corey’s OCD exempt agents from moral responsibility in
these sorts of situations?
Philosophers often ask questions like this in the context of a skeptical
worry about how it could be the case that anyone is ever responsible for
what they do at all. One commonplace thought is that when agents’ be-
haviors are due to compulsion their “brains make them do it.” But, in a
literal sense, it might seem that our brains make all of us do everything
that we do.
Given a commitment to the veridicality of this important intuitive dis-
tinction between ordinary responsible action and compulsive behaviors,
there are two central kinds of approaches to these worries in the moral re-
-11-
sponsibility literature. One approach is to hold that what we really mean
is that agents like Corey don’t have sufficient control over their compul-
sive actions; being morally responsible is largely a matter of control. An-
other approach is that what we really mean is that when agents like Corey
act compulsively they don’t act in accordance with what they really want
to be doing; moral responsibility is a matter of agents having a certain pro-
file of mental states that contribute to their actions.
In this dissertation I advance an account of moral responsibility that
aims to solve several of the biggest problems facing current accounts that
take this latter approach, which are often known as Deep Self accounts. In
the first part of this chapter I explain and motivate several key aspects of
Deep Self accounts. While the view I advance diverges in crucial ways
from traditional Deep Self views, it nevertheless shares many of their
broad theoretical commitments. My discussion here serves both to shed
light on the general theoretical orientation that underpins my project and
to highlight the specific contribution that my view makes to the dialectic.
Each traditional Deep Self view relies on identifying a type of mental
state that is invariably internal. In the second part of this chapter I argue
that internality is best understood on each view in terms of the agent ap-
proving of being motivated in the way that she is to some degree. This
helps locate a commonality among Deep Self views: they all seem to hold
that approving of one’s action is a necessary condition for attributional-
responsibility. In Chapter 2, I build on this foundation to develop a new
view of attributional-responsibility.
2. Actual Sequence Compatibilism
2.1 Dialectical Motivation for Actual Sequence Compatibilism
How can we justify our system of moral responsibility practices given a
scientific picture of our world in which it seems likely that our choices are
explicable by means of causal chains that stretch back to events that hap-
pened before we were even born? Against a backdrop of scientific under-
-12-
standing according to which we recognize that our brains make us do eve-
rything that we do, and that these events are reducible to neurochemical
reactions, how can we preserve the distinctions we want to make between
agents who are blameworthy for their actions and those who are not due
to conditions like Tourette syndrome or OCD?
One response to these initial concerns is to claim that, in fact, we can-
not justify our practices and intuitive distinctions. The most popular ar-
gument for this conclusion runs as follows:

1. It’s only appropriate to blame agents who act freely.
2. Acting freely requires the presence of alternative possibilities for ac-
tion.
3. Having alternative possibilities involves having metaphysically ro-
bust options available at the moment of choice.
4. Our scientific picture of the world is correct, and it crowds out the
space for metaphysically robust options to ever exist at the moment of
choice.
2

5.Therefore, it is never appropriate to blame anyone.
3

2
Another route to Moral Responsibility Skepticism is to take no stand on whether or not
we live in a deterministic world, and instead show that the kind of free will that could
ground moral responsibility is incompatible with both determinism and indeterminism.
See, for example, Pereboom (2005).
3
While this argument is very popular, it is possible to be an Incompatibilist about moral
responsibility and determinism without thinking that alternative possibilities are re-
quired for free will. For example, some Incompatibilists think that the kind of free will
required for moral responsibility requires that an agent herself be the ultimate source of
the chain of events leading to her actions in a way that is not possible given the truth of
determinism. For an overview of several different views in this family, see Tognazzini
(2011).
-13-
While Moral Responsibility Skeptics are happy to accept this conclu-
sion, each of the premises of this argument has been disputed somewhere
in the literature.
Libertarians generally dispute Premise 4. They believe that for an
agent to have the requisite kind of alternative possibilities when she acts
her action must be non-deterministically caused such that at the moment
of action there is another action she could have undertaken (and that af-
terwards it will be true to say of her that she “could have done other-
wise”). While this sounds like a mysterious sort of power to ascribe to
agents, Libertarians have advanced several different pictures of the pro-
duction of action that attempt to mitigate these worries.
4
Libertarian views
are also subject to luck objections.
5
If an agent’s prior plans, values, com-
mitments, etc. cannot fully determine her course of action since, according
to the Libertarian, the choice must remain open at the very moment of ac-
tion, there can be nothing about the agent’s mental states that determines
whether she will undertake one action or the other. This agent actually
seems to lack an important form of control over what she does, and so, it
is often argued, this cannot be the sense of alternative possibilities re-
quired for the kind of free will required for moral responsibility.
For a long time the most popular strategy for those who wanted to
preserve both the scientific picture of action production and justification
of our practices of responsibility was to target Premise 3. Classical Com-
patibilists accepted that moral responsibility requires free will and that
free will requires the ability to do otherwise, but rejected that this is in-
compatible with determinism. Instead, they attempted to posit less meta-
physically robust senses of the ability to do otherwise that were both
compatible with causal determinism and convincingly able to ground as-
criptions of free will and moral responsibility. Initially, these compatibil-

4
See Clarke (2003) for an overview.
5
See Mele (1999, 2006).
-14-
ists put forth conditional analyses of the ability to do otherwise that took
the following form: If the agent had chosen/wanted to/decided to do oth-
erwise, she would have done otherwise. In this way, they hoped to secure
the conditions for free will while retaining the understanding that our ac-
tions are controlled by fixed chains of mental events.
These analyses are thought to have definitively failed, however, since
they appear to predict that agents have the ability to do otherwise in cases
in which we think they clearly don’t. For example, imagine a girl who has
such a psychological aversion to picking up blonde puppies that she could
never become such that she would pick up a blonde puppy. Now suppose
that when offered a choice of a blonde or black puppy, she picks up the
black puppy. The conditional analysis says of this case that if she had
wanted to pick up the blonde puppy she would have picked up the
blonde puppy and so it returns the verdict that she does have alternative
possibilities in the Classical Compatibilist sense. But since she is unable to
become psychologically such that she would ever pick up the blonde
puppy, this seems clearly incorrect. In what sense does she have the abil-
ity to pick up the blonde puppy if she could never be or become such that
she would actually do so?
6

More recently, some Compatibilists, sometimes called New Disposi-
tionalists, have tried to revive the spirit of this project by positing that the
ability to do otherwise is grounded in more complex agential dispositions,
or bundles of dispositions.
7
For example, following David Lewis, Kadri
Vihvelin suggests that rather than simple conditionals, the dispositions at
issue should be taken to involve intrinsic properties that are the causal ba-

6
For more on this example, see McKenna and Coates (2015). This problem is, arguably,
just an application of a more general problem for analyzing dispositions in terms of sim-
ple conditionals, known as the problem of finked dispositions. See Vihvelin (2004) for
discussion.
7
This label first appears in Clarke (2009). Prominent defenses include Smith (2003); Fara
(2008); and Vihvelin (2004, 2013).
-15-
ses of those dispositions. She thinks that in order to assess claims about
whether an agent has alternative possibilities, we must consider various
counterfactual scenarios in which the causal base of the pertinent disposi-
tion operates unimpaired. In the spirit of Classical Compatibilism, Vihve-
lin argues that the relevant sorts of abilities are dispositions to make
choices on the basis of reasons.
8
While these sorts of accounts solve some
of the problems of the more simplistic conditional analyses, they continue
to face a less resolvable challenge from defenders of Premise 3.
Defenders of Premise 3 argue that the New Dispositionalists’ accounts
do not get us to the kind of alternative possibilities required for free will,
since what’s valuable about us having alternative possibilities at the mo-
ment of choice is that they make it so that our choices are “up to us.” And,
as they argue, it’s just not clear that the New Dispositionalists’ notion of
ability to do otherwise captures this sense. Randolph Clarke puts the point
as follows:

An agent with an interesting bundle of dispositions and in friendly
surroundings might have a rich array of narrow and wide abilities to
do things that she doesn’t in fact do. That an agent might have such
abilities even if determinism is true is an important fact. But it will take
further argument to show that having such dispositions and being in
such surroundings suffices for its being up to you, on some occasions,
whether one or another of these dispositions is manifested, and hence
whether you do this or that.
9

8
Vihvelin (2004) suggests that the collection of dispositions might include, among others,
dispositions to form and revise beliefs in response to evidence and argument, to form in-
tentions in response to desires and form beliefs about how to achieve those desires, to
engage in practical reasoning in response to one’s intention to make a rational decision
about what to do, and to believe that by engaging in practical reasoning one will succeed
in making such a decision.
9
Clarke (Forthcoming): 26-27.
-16-

Here, New Dispositionalists find themselves in a battle of intuitions; they
allege that having the sorts of dispositions they put forth as an analysis of
the ability to do otherwise just is what it means for our actions to be “up to
us.”
10
It is unclear how the clash of this particular set of intuitions could be
resolved, and this leaves the debate at a standstill. Some have even taken
this clash of intuitions as reason to conclude that our notions of free will
and moral responsibility are confused concepts we should try to do with-
out in order to make progress.
11

I adopt what I take to be a more promising strategy for those who
want to preserve both our moral responsibility practices as we know them
and the scientific picture of action: Actual Sequence Compatibilism. Actu-
al Sequence Compatibilism gives up on the Classical Compatibilist at-
tempt to redefine the sense of alternative possibilities that is required for
freedom and responsibility and instead resists the notion that the condi-
tions for moral responsibility have anything to do with control or alterna-
tive possibilities at all, thus targeting Premise 2 (and sometimes also
Premise 1 in the process).
12
The idea is that the criterion for an agent to

10
It is sometimes alleged that the incompatibilist intuition about the relevance of alterna-
tive possibilities is a common sense intuition, while others argue that compatibilist no-
tions of alternative possibilities are the more predominant common sense intuition. There
is an extensive but inconclusive experimental philosophy literature on this topic. See
Björnsson and Pereboom (Forthcoming) for an overview.
11
See Vargas (2011) for an overview of Revisionist approaches to free will and moral re-
sponsibility.
12
In this dissertation I reject premise 2 explicitly and remain agnostic about premise 1. I
propose conditions for one form of moral responsibility, and say relatively little about
free will. I leave the question of whether or not my account is best thought of as a non-
freedom-requiring account of responsibility (“Semicompatibilism”), or as an account of
the kind of freedom required for responsibility to the reader (as the answer seems to me
to be determined in large part by what one wants their conception of “free will” to do),
although I suspect the former may be the better way of understanding the project.
-17-
count as having acted (freely and) responsibly can be found within the ac-
tual sequence that leads to her action. In arguing that exemptions from
moral responsibility can be explained without reference to alternative pos-
sibilities, the Actual Sequence Compatibilist boldly forges past the stale-
mate and frees up the opportunity to propose quite different kinds of cri-
teria for responsibility. This gives the Actual Sequence Compatibilist a
significant dialectical advantage in the debate.
2.2 Support for Actual Sequence Compatibilism
Dialectical advantages aside though, are there any reasons to believe in
the truth of Actual Sequence Compatibilism? There is, admittedly, at least
something initially strange about the thought that whether or not an agent
could have done otherwise has no bearing on whether or not she acts
freely and responsibly.
This strangeness, though, is mitigated by the fact that there is some-
thing equally strange about views that do not focus on the actual causal
sequence that leads to the agent’s action. As Carolina Sartorio emphasizes,
it would seem quite inappropriate for an agent to attempt to absolve her-
self of responsibility for some action by pointing to factors that were not in
any way explanatory of why she acted in such a way. Sartorio puts this
point succinctly: “if a factor is completely irrelevant to why you acted, it
seems that it cannot be used to excuse your behavior.”
13
Since there are in-
tuitions that tell against the relevance of alternative possibilities and the
irrelevance of alternative possibilities, it’s clear the debate cannot be set-
tled by appeal to initial intuition alone.
In various places throughout the body of his work, Harry Frankfurt
provides support for the view that alternative possibilities are irrelevant to
moral responsibility. Most famously, Frankfurt offers the case of Jones and
Black.
14
He has us imagine that Jones is an agent who is going to perform

13
Sartorio (2016): 2. See also Mele (2006).
14
Frankfurt (1969).
-18-
an action, ϕ-ing, for which we would all agree he would intuitively be
clearly morally responsible, were he to go through with it. Unbeknownst
to him, however, Black is waiting in the wings and has placed a chip in
Jones’ brain that will not activate unless Jones is not independently going
to go through with ϕ-ing, in which case it will cause him to ϕ anyway. As
it happens, Jones decides to go through with ϕ-ing on the basis of his own
deliberation and Black never needs to, and so does not, interfere. Jones
lacks alternative possibilities in any reasonable sense; no matter what, he
was going to ϕ at t. And yet he still appears to be just as morally responsi-
ble for his action as he would have been if Black had never been waiting in
the wings at all.
Cases that follow this general format have been termed “Frankfurt cas-
es,” and a very large literature has emerged over the years that questions
the adequacy of the thought experiment for proving that the presence of
alternative possibilities is not required for moral responsibility.
15
For ex-
ample, some question whether or not it is methodologically appropriate to
make the assumption that Black knows what Jones will do before he does
it, and others question whether or not the action Black would cause would
be identical to Jones’ actual act.
Whether or not any particular Frankfurt case is successful at decisively
establishing a foolproof example of an agent who has no alternative pos-
sibilities but is nevertheless responsible for her action, there is a broader
lesson that these cases do help to illustrate: alternative possibilities appear
to play no explanatory role in action.
16
In Frankfurt’s words:

The fact that a person could not have avoided doing something is a
sufficient condition of his having done it. But…this fact may play no

15
See Fischer (2010) for an overview in which he compares the state of the literature here
to the state of the literature surrounding Gettier cases.
16
See also McKenna (2008) for a similar take on the relevance of Frankfurt-cases.
-19-
role whatsoever in the explanation of why he did it. It may not figure
at all among the circumstances that actually brought it about that he
did what he did, so that his action is to be accounted for on another ba-
sis entirely. Even though the person was unable to do otherwise, that is
to say, it may not be the case that he acted as he did because he could
not have done otherwise.
17

Frankfurt cases remind us that it seems much less important to figure out
whether or not there are possible worlds in which Jones does not ϕ than it
does to figure out why he actually ϕs.
This lesson can also be drawn from Frankfurt’s cases of volitional ne-
cessity. Frankfurt draws attention to cases in which people feel that their
actions are necessitated by constraints on their wills in positive ways. For
example, agents committing acts of extreme love or loyalty may have this
feature, like a woman jumping in front of a bullet to save a friend or a man
running back into a house on fire to save his child. Doing otherwise in
these cases appears to be simply unthinkable for such agents, yet they cer-
tainly seem to be morally responsible. As Frankfurt remarks, when Martin
Luther made his famous declaration “Here I stand; I can do no other,” we
do not usually take Luther’s seeming lack of control over his course of ac-
tion to undercut his claim over his action, but rather, if anything, to inten-
sify it.
18
Whether or not Luther’s claim is taken literally, the rhetorical
force comes from the natural thought that feeling yourself to have no oth-
er option often serves to intensify your sense of ownership of your action,
not to undercut it. As with Frankfurt cases, the larger point is not that the
exact sense of alternative possibilities targeted by Classical Compatibilists,
Libertarians, or Skeptics is the exact same as what agents lack in volitional
necessity cases, but rather that the degree to which someone is able to do

17
Frankfurt (1988): 8.
18
Frankfurt (1988): 87.
-20-
otherwise has no simple correlation to how free they are to act in any
sense that we seem to care about when we make judgments of moral re-
sponsibility.
A final set of cases offered by Frankfurt helps to get to the heart of the
debate: his willing and unwilling addict cases. The unwilling addict des-
perately wishes not to be compelled to take drugs but continues taking
them, despite himself, due to his addiction. The willing addict, on the oth-
er hand, absolutely loves taking drugs, and though he would not be able
to resist if he ever tried to, he would never want to resist in the first place
since he fully endorses what he does. Those who are focused on alterna-
tive possibilities and control will not differentiate between the responsibil-
ity of willing and unwilling addicts; so long as an addict cannot sufficient-
ly control her action, she is not responsible, they’ll allege. But Frankfurt
thinks the willing addict is responsible for taking drugs, despite his lack of
control. This intuition can be strengthened by supposing that the willing
addict has no awareness of his addiction and wholeheartedly chooses to
take drugs for reasons that have nothing to do with their irresistibility.
Chandra Sripada adds further support by offering a structurally anal-
ogous case that abstracts away from the particularities of addiction. His
example is of a “willing exploiter” who strongly desires to watch exploita-
tive kinds of pornography.

… suppose [his] desires and the actions they issue in are deeply ex-
pressive of his self. This man has a narcissistic kind of self-love at his
core. He is attracted to the idea that he is in a position of dominance
over others and the exploitiveness of the pornographic material is thus
exactly what he finds so deeply gratifying. The person thus stands
strongly in favor of his desires to view exploitive images and wouldn’t
change a thing. In envisioning this case, we are to keep all other rele-
vant aspects of the Willing Addict case the same. In particular,
the attitudes of this person’s self, via their role in deliberation and the
-21-
formation of practical judgments, provide motivational support in fa-
vor of his viewing the images. Additionally, the desires to view the
images are sufficiently powerful in their own right that, though he
doesn’t and wouldn’t ever try to resist these desires, were he to try, he
would fail.
19

The willing exploiter certainly seems to be morally responsible for his
viewing of the pornographic images despite the fact that he wouldn’t ever
be able to resist his desires to do so if he were to try to, and so control,
again, seems to be irrelevant to our willingness to ascribe responsibility.
3. Deep Self Theory
These cases also point towards a new kind of criterion that has nothing to
do with alternative possibilities or control. The willing exploiter seems re-
sponsible for what he does because it’s something he embraces, and he
does so precisely because he embraces it. This insight is key to the devel-
opment of a new kind of criterion for morally responsible agency.
Recall that the Actual Sequence model claims that alternative possibili-
ties are not relevant to moral responsibility and that we should instead fo-
cus on the actual sequence of events that leads up to the agent’s action.
But this falls short of a determinate theory of moral responsibility, since it
does not yet tell us which aspects of the action’s causal sequence are rele-
vant to determining whether or not the agent is responsible. As the willing
exploiter case illustrates, one natural way of filling in the story is by look-
ing towards whether an agent does what she in fact “stands in favor of”
doing. When agents stand in favor of the action they undertake, it seems
reasonable to hold that they express something about what those agents
are like.

19
Sripada (2017): 802-803.
-22-
Views that feature these sorts of criteria for moral responsibility are
variously referred to as “Self-Disclosure,” “True Self,” “Real Self,” or
“Deep Self” views. The most common title currently at use in the litera-
ture is “Deep Self,” and Deep Self views easily represent the most influen-
tial strand of Actual Sequence Compatibilism.
20
On a Deep Self view, what
matters for moral responsibility is that the agent acts in accordance with
what she really wants to do, where “really wanting” always involves
some further mental state beyond merely having a first-order desire. The
necessity of locating some further mental state comes from the fact that an
agent can act on one of her first-order desires without thereby standing
behind it.
Cases of compulsion illustrate this point well. Compulsion seems best
described as involving an agent being moved to ϕ against her will by a
rogue first-order desire or urge to ϕ that overpowers her identification
with some other course of action. Insofar as agents should not be held
morally responsible for their compulsive actions, there is reason to locate
the criterion for moral responsibility in the presence of some further kind
of mental state. And so Deep Self theorists each make some sort of demar-
cation within agential psychology that explains how only some subset of
an agent’s motivational states can produce actions for which an agent can,
in principle, be praiseworthy or blameworthy. An agent can only be held

20
One source of potential confusion here is that views that posit that control over self-
disclosing agential capacities as a condition on moral responsibility are occasionally re-
ferred to as Deep Self views despite not being Actual Sequence Views. Agnieszka Ja-
worska refers to these views as “Broad Identificationist Views” and Actual Sequence
Deep Self Views as “Narrow Identificationist Views” (Jaworska [2017]). It is unclear,
however, what precisely demarcates these Broad Identificationist Views from Classical
Compatibilist views that posit that the relevant sense of alternative possibilities to
ground moral responsibility has to do with counterfactual conditionals like “if one had
chosen/valued/endorsed acting differently she would have acted differently” (or their
New Dispositionalist equivalents). Here I will instead adopt the more popular taxonomy
that considers Deep Self views to be a proper subset of Actual Sequence Views.
-23-
morally responsible when her action is motivated in the right sort of way
because only then is her action produced in such a way that it can “speak
for” her as an agent.
Deep Self theorists generally proceed by proposing a special kind of
mental state that motivational states must align with in order for an agent
to be morally responsible. Many different Deep Self views have been put
forward, but the views that have been most prominent in the literature are
those that privilege either valuing, planning, caring, endorsing, or some
combination of these states. I will refer to these as “deep self mental
states.”
21
These special mental states are said to “mesh” or “align” with the
agent’s motivational states such that the motivational states “flow from”
the values, plans, cares, or endorsements.
While overly metaphorical language is often used to describe this rela-
tionship, there has been recent interest amongst Deep Self theorists in get-
ting clearer about what the relationship of “meshing” might amount to.
Many theorists seem to think about the relation as being causal: an agent
is attributionally-responsible for ϕ-ing if the motivational state that causes
the agent to ϕ is itself caused in part by the agent’s deep self mental states.
It’s worth noting that the deep self mental state need not make the differ-

21
Views that put forth other candidate deep self mental states include Susan Wolf’s “sane
Deep Self view” on which deep self mental states must meet further “sanity” require-
ments (Wolf [1987]); David Velleman’s view, on which deep self mental states are desires
to act in accordance with reasons (Velleman [1992]); and coherentist views on which deep
self mental states are those that bear special relationships to the agent’s other mental
states either by being relatively unopposed by other states (Arpaly and Schroeder[1999])
or by being narratively coherent (Matheson [2018]). While departing from traditional
Deep Self views in significant ways, other views sometimes said to “strike deep self
themes,” appear in Scanlon (1998), Arpaly (2003), Smith (2005, 2008), Sher (2009), and
Buss (2012). (Sripada [2016] offers the latter list with the caveat that these views may fail
to count as “Deep Self” views on many uses of the phrase. In Chapter 5 I follow Talbert
[2016] in referring to the views of Scanlon, Smith, and Sher as “New Attributionist”
views.)
-24-
ence between the agent acting as she does and her refraining, since deep
self mental states might also function by causally overdetermining the
agent’s course of action. One problem with broadly causal views, first not-
ed by Neil Levy, is that it seems possible that an effective first-order desire
and resultant action could be caused by a deep self mental state while in-
tuitively the deep self mental state is not expressed through the action.
22

Chandra Sripada gives the following example, which uses the proposed
deep self mental state of caring:

Suppose Jimmy’s son has gone missing in Afghanistan. He cares for
his son so much that he ruminates continuously, and this in turn gives
him a severe headache for which he must take an aspirin. Standard
theories of causation would say that Jimmy’s caring for his son causes
his taking an aspirin— very roughly there is a chain of causal depend-
ence that links the two. Jimmy’s taking the aspirin, however, does not
express his caring for his son.
23

An alternate understanding of the expression relation is what Sripada
and Shoemaker call a “content harmony” relation.
24
On this view, an agent
is attributionally-responsible for her action only if the motivational state
that she acts on is congruent with the content of the deep self mental state
in some sense. On Sripada’s understanding, the congruence amounts to
the motivational state’s being characteristically disposed to be produced
by the deep self mental state. For example, if I judge my mother’s health to
be of value to me, I might be characteristically disposed to be motivated to
take her to the doctor. A content harmony requirement might be added to
a causal requirement, or it may be thought to be a competing explanation

22
Levy (2011).
23
Sripada (2006): 1216.
24
See Sripada (2006), Shoemaker (2012, 2015b).
-25-
for what the expression relation is. For example, one possible view is that
an agent is attributionally-responsible for ϕ-ing iff she endorses her desire
to ϕ, whether or not that endorsement itself has any causal bearing on her
ϕ-ing.
3.1 The “Deep” in “Deep Self”
Given this description of the commitments of Deep Self views I have out-
lined, one might wonder what cause there is for adopting the language of
“deep self.” For all I have said, these views merely target some special sub-
set of mental states and propose that these states, rather than others, due
to the fact that they ensure agential identification, grant an agent’s actions
the ability to speak for that agent. But what sort of additional commit-
ments are taken on by adopting the language of the “deep self” and what
role do they play in the view?
Answers to this question by leading Deep Self theorists are extremely
varied. For example, David Shoemaker writes that

the ‘deep’ in ‘deep self’ simply refers to the psychic element’s place in
an agential structure as the ultimate psychological source of various
‘surface’ attitudes subject to its governance.
25

So, for example, cares are deeper than ordinary first-order desires since,
for example, caring about your family is the source of a desire to take your
daughter to soccer practice. The meaning of “deep” here does not imply
any sort of strong metaphysical commitment to the Self. On the other end
of the spectrum, Chandra Sripada thinks talk of deep selves commits him
to the existence of “fundamental conative states that robustly and globally
shape action,” the existence of which he takes to be a substantive claim
about actual human psychology.
26

27
Deep selves, for Sripada, are presum-

25
Shoemaker (2015b): 43.
26
Sripada (Unpublished Manuscript): 15.
-26-
ably ‘deep’ because on his view they play a crucial role in helping to ex-
plain a wide variety of agential phenomena including but not limited to:
moral responsibility, normative reasons for action, happiness,
and weakness of will. A similar but, in theory, distinct idea is the thought
that all of an agent’s mental states of the kind that are proposed to play
the role of deep self mental states together form some sort of whole which
either constitutes or provides us with some particularly important insight
into the agent’s Self.
I take these latter two conceptions to be contingent features of the set
of views generally recognized to belong to the Deep Self family of views.
Each view does need some story to tell about what privileges actions that
relate to deep self mental states such that they are the ones on the basis of
which we are permitted to hold an agent responsible. However, the ver-
sions of this story on which the deep self mental states together play a
foundational role in the core of an agent’s conative personality or are to-
gether constitutive of the agent’s Self only represent a couple of the op-
tions for fleshing out this story, among many other possibilities.
Further complicating these issues, as Lippert-Rasmussen points out,
people tend to conflate two different connotations of the phrase “deep
self.” On one conception, deep self mental states have special authority for
the agent, and on another, deep self mental states have more to do with
authenticity. As he puts it, on authenticity conceptions of the Deep Self, a
person’s Deep Self

27
While it does not seem to me that any such broad sweeping claims about human psy-
chology are required for proponents of Deep Self, Sripada thinks there is much less cause
for empirically-driven skepticism about the existence of such deep selves than what
many philosophers have been led to believe. According to Sripada, while certain seg-
ments of social psychology have been very influential as a source of data for philoso-
phers, data from neuroscience, human behavioral genetics, and personality psychology is
all fairly friendly to the idea of robust deep self psychology.

-27-

… is the person’s deepest and most genuine commitments and de-
sires…deep, idiosyncratic longings and repressed desires are strong
candidates [for deep self states] on [this] account.
28
(20).

This conception of the deep self is, to my mind, not relevant to ques-
tions of moral responsibility. Confusion in the literature between the two
senses of “Deep Self” is presumably part of what leads Nomy Arpaly to
her particularly damning accusation of Deep Self views. She has us imag-
ine a woman, Lynn, who discovers she is a lesbian but would much rather
have not come to such a discovery and does not want to be motivated by
such desires. Arpaly continues,

If Lynn were to go to her favorite college professor for help, she would
likely be told that she should try to accept herself for who she is, re-
frain from attempts to suppress her true self, and so on. If, on the other
hand, she were to read the moral psychology literature and believe its
claims, she would probably conclude that she was right and her homo-
sexual desires are not truly her own. For ‘Lynn’ and ‘homosexual de-
sires’, we could substitute ‘Victorian lady’ and ‘any sexual desire’,
‘nice Jewish boy’ and ‘hostility toward parents’, ‘severe perfectionist’
and ‘desire to get some rest’, ‘the young E.T.A. Hoffman’ and ‘desire to
be a writer’, or any of various characters from various novels and their
adulterous loves. In all these cases, the agent who dismisses these de-
sires as reasons for action and treats them as “outlaw desires” is likely
to feel that they are not really his.
29

28
Lippert-Rasmussen (2003): 20.
29
Arpaly (2003), 16.
-28-
But it need not be any part of a Deep Self view to hold that Lynn’s lesbian
desires are not an authentic part of who she is or that she should resist
them. A Deep Self view should merely say that if she were to engage in a
sexual act with a woman without in some sense valuing/ endorsing/ plan-
ning on/ caring about doing so, her action would be compulsive or lacking
in agential authorization in such a way that would undermine her being
an apt candidate for moral responsibility. It is perfectly consistent to addi-
tionally hold that Lynne ought to embrace her lesbian desires as being an
authentic part of her identity. Deep Self theorists ought to be clearer in re-
jecting the relevance of the authenticity conception of the Deep Self and
instead understand deep self mental states as those that have the authority
to speak for the agent; the mesh of deep self mental states with effective
motivation needn’t be understood to be anything over and above a condi-
tion for ownership over one’s action in the sense relevant for moral re-
sponsibility.
It is, in a way, unfortunate that the “Deep Self” name is the one that
has stuck, as it tends to evoke thoughts of a quite ambitious project to lo-
cate a central, fundamental, all-important seat of agency within the sea of
an agent’s mental states. The aims of a Deep Self theorist in practice are
generally a good deal more modest, (although as I’ve highlighted, they
vary quite a bit). But because of these confusions, it is difficult to say of
any particular view, including the one I advance in the rest of this disser-
tation, whether or not it ought to count as a Deep Self view.
4. Attributional-Responsibility
Deep Self theorists have historically been attracted to a particular kind of
notion of moral responsibility: attributional-responsibility. In “Two Faces
of Responsibility” Gary Watson first proposed that different parties to the
responsibility debate seemed to be implicitly committed to different ideas
about what sorts of responses were justified on the basis of the require-
-29-
ments they proposed.
30
He identified two aspects of responsibility: at-
tributability and accountability. The recent literature has expanded to
count answerability as a potentially distinct face of responsibility. This
chart very roughly explains the general differences in the concepts in a
way that is meant to be broad and inclusive:

There are few points of agreement in the literature about how we
should understand the relationship between these notions of responsibil-
ity. Some see the three as competing accounts, some think they represent
wholly distinct facets of responsibility, others call for their unification, and
others see one or more facet as a necessary condition on another.
31

Rather than debate these points at this level of abstraction, my strategy
instead will be to develop what I hope will be an attractive substantive
theory of attributional-responsibility. On my view, attributional-
responsibility is its own full-fledged form of moral responsibility, and ac-
countability-responsibility is its own entirely distinct form. So one may be
accountability-responsible without being attributionally-responsible and

30
Watson (1996).
31
See Watson (1996), Fischer and Tognazzini (2011), Strabbing (2011), Shoemaker (2011,
2013, 2015), Smith (2012), Talbert (2012), King (2014), Wolf (2015), and Zheng (2016) for
discussion.
If an agent meets the responsibility requirements, then we can…
Attributability judge, perceive, or otherwise react to her wrongdoing
as expressing a personal fault
Answerability demand justification of behavior that is prima facie
wrong
Accountability confront her on the basis of her wrongdoing, often
with the aim of demanding recompense or sanctions of
some form

-30-
vice versa. In this dissertation I gradually advance claims that narrow in
on this particular picture, although the earlier parts of the dissertation are
meant to be compatible with other views of the overall picture. In Chap-
ters 2 and 3, I defend my own picture of the requirements for attributabil-
ity, which is, in principle, compatible with other understandings of how
these responsibility concepts work in relation to one another. This in-
cludes the view that attributability’s main purpose is to serve as a neces-
sary condition for accountability-responsibility. In Chapter 4, I advance an
account of blame’s content that lays the foundation for understanding at-
tributional-responsibility as its own full-fledged form of responsibility. In
Chapter 5, I argue for the adoption of accountability-responsibility as a
wholly distinct form of responsibility. The structure is such that at any
point, the reader may get off the boat while still being able to, in theory,
accept the claims of the previous chapters.
In this chapter, I consider various Deep Self views as views of attribut-
ability in order to consider them as contenders to the view of attributabil-
ity I develop. On a Deep Self view of attributional-responsibility, an ac-
tion’s being appropriately related to deep self mental states is what allows
us to move from an evaluation of the moral quality of an action to an
evaluation of the moral quality of the agent on its basis. The main goal, as
I see it, is to find a filter that separates the sorts of acts that an agent can-
not truthfully claim come from ‘outside themselves’ from the ones that
stem from mere neurological noise, because this is the class of acts for
which agents may be blamed on the basis of what their actions say about
them as agents, making them appropriate target of aretaic assessments.
Even given an understanding of responsibility on which attributional-
responsibility constitutes its own full-fledged form of responsibility, there
are still additional conditions beyond the agential requirements put forth
in Deep Self theories that must hold for an agent to be blameworthy. Fur-
ther conditions, including the moral status of the act, and perhaps epis-
-31-
temic conditions on the agent
32
, are needed in order to know what any giv-
en action says about the agent.
When an agent stands in the proper agential relationship to her action
such that it opens her up in principle to being morally responsible for the
act (or what it reveals about her), I will describe such an agent as being
“attributionally-responsible” for her action. It is consistent with my un-
derstanding of the term that an agent may be attributionally-responsible
for a morally neutral action. So when I speak of a Deep Self view giving
sufficient conditions for attributional-responsibility, I do not mean to say
these are meant to be sufficient conditions for an agent being blamewor-
thy for any particular action.
5. B-Tradition versus H-Tradition
With this background in place, we are now in a position to examine the
first central question that this dissertation aims to answer: what are the
deep self mental states that should be required for an agent to count as at-
tributionally-responsible, and how should we mediate disputes between
alternative accounts?
In advancing my positive view in this dissertation, I will make the
methodological assumption that being responsible is metaphysically prior
to holding responsible. That is, in order to find out when it is appropriate
to hold someone morally responsible, we first need to know whether or
not the person meets the specifiable metaphysical conditions for actually
being responsible. David Shoemaker calls this the B-Tradition, which
stands in contrast to the H-Tradition, according to which holding respon-
sible takes priority.
33

Adherents of the H-Tradition tend to hold the view, which has been
quite pervasive in recent years, that moral responsibility ought to be ana-

32
I say more about just what I take these additional constraints to amount to in Chapter
4, §2.
33
Shoemaker (2015b): 20.
-32-
lyzed in terms of the responses (usually conceived of as reactive attitudes)
that are fitting in holding each other responsible. Fittingness is often taken
to be a sui generis primitive relation. Defenders of the B-Tradition rely on
intuitions about whether or not certain responses are appropriate as well,
but take these only to provide a defeasible epistemic guide to the condi-
tions of responsibility.
34
Adherents of the B-Tradition take seriously the
need to additionally locate some further explanation as to why the particu-
lar conditions that an agent must meet in order for it to be appropriate to
respond in certain ways must hold rather than some other conditions.
Providing an argument for the metaphysical priority of being respon-
sible over holding responsible is outside the scope of this dissertation, but
I do think it makes methodological sense to proceed as though the B-
Tradition is true until we have exhausted its possibilities. Gideon Rosen
provides a strongly worded defense of this method:

The Fittingness view is a theory of last resort. We should adopt it only
if we have tried and failed to analyze appropriateness or to assimilate
it to a relation studied elsewhere under another name….A theory of
responsibility aims to articulate the conditions under which blame is
appropriate, and then to explain why those conditions are as they are.
And the trouble is that the Fittingness View would furnish grounds for
abject pessimism about this project.
35

Defenders of H-Theory might nevertheless echo P.F. Strawson’s influen-
tial statement that the project of finding metaphysical conditions of re-
sponsibility has already failed or is bound to fail.
36
I hope, however, to of-

34
See also McKenna (2012), which advances the view that neither being nor holding re-
sponsible is more fundamental than the other.
35
Rosen (2015): 71.
36
Strawson (1962).
-33-
fer an attractive picture of the conditions for responsibility grounded in
the metaphysics of agency that might be taken as evidence to the contrary.
That said, while my arguments for my view will in some places as-
sume the B-Tradition, the view I advance is, with some modifications, it-
self compatible with the H-Tradition.
6. Internality’s Role in Deep Self Views
How do Deep Self theorists mediate disputes between alternative ac-
counts of the relevant deep self mental states? Although not every Deep
Self theorist is explicit about how they answer this question, most argue
that the tokens of only one particular type of mental state or another are
invariably “internal.”
37
A mental state is internal iff the agent is identified
with the state in such a way that it cannot legitimately be taken to be a
mere occurrence that does not belong to the agent since it is an “alien”
force.
38
I follow Agnieszka Jaworska here in distinguishing internality in
this ontological sense from subjective active identification that is based on
whether the agent perceives aspects of her psychology as being her own.
These senses are perhaps not wholly unrelated, however, as non-self-
deceptive subjective identification might be able to provide us with defea-
sible evidence of internality. While there is a possible view on which the
ontological category of internal mental states with which an agent can
rightly be identified amounts to nothing more than the states with which
the agent takes herself to be identified with, such a view would require an
argument.
The concept of an internal state is usually given a gloss as the kind of
state from which an agent cannot be alienated. Whether explicitly or im-
plicitly, something like this idea seems to play some role in explaining the

37
Internality may play a less central role in how these disputes are mediated when the
subject is approached from the H-Tradition.
38
Jaworska (2007): 531.
-34-
proposed authority of the particular kind of deep self mental states on
every major Deep Self view.
On the endorsing view, as put forth by Harry Frankfurt, the relevant
deep self mental states are higher-order volitions.
39
On this view, agents
have the ability to influence their actions via the formation of second-
order desires, which are desires about what the agent wants to desire to
do. Second-order volitions are desires not just about which desires an
agent endorses having but about which one of these desires the agent
wants to actually act on at a given moment in time. So an agent’s action is
attributable iff it is caused by a desire to ϕ that meshes with the agent’s
further desire to act on the desire to ϕ. Frankfurt seems to understand the
expression relationship that needs to hold between second-order volitions
and first-order desires in semi-causal terms. Either the first-order desire is
not by itself sufficient to motivate the agent and she needs the ‘push’ con-
ferred to it from her second-order volition, or her second-order volition to
act on a desire to ϕ accompanies a desire to ϕ that is already sufficient to
motivate her to action, and so her endorsement amounts to over-
determining or at least “okay-ing” the fact that she will be led to action by
such a desire.
An agent’s second-order volitions, for Frankfurt, have the authority to
speak for the agent because they are the output of an endorsement pro-
cess, the goal of which is to confer the status of internality on first-order
desires. In forming a desire to act on one of her first-order desires, an
agent identifies herself with her first-order desire because, for Frankfurt,
the process of endorsement is the process of identification and a state is in-
ternal iff the agent identifies with it.
40

39
Frankfurt (1971).
40
One much-discussed serious problem for Frankfurt is that it seems arbitrary that se-
cond-order desires, rather than say third or fourth-order desires have special agential au-
thority. The way Frankfurt thinks of the role of internality in the theory is part of what
generates the problem. If first-order states are granted the authority to speak for the
-35-
On the valuing view, the conception of agential architecture is quite
different.
41
On this model, agential psychology is divided between valuing
and mere desiring parts, and each has its own ability to motivate the
agent. The relevant deep self mental states are evaluative, although im-
portantly they do not consist merely in the pure cognitive judgment that
some course of action is best, but rather in the agent’s setting ends for her-
self. An agent is attributionally-responsible for her action iff what she does
is controlled by her evaluative system, which prescribes the overall best
course of action. She is not attributionally-responsible when what she
does is controlled by mere desires that do not flow from what she truly
values. Supporters of the valuing view argue that valuing states are invar-
iably internal by explicitly pointing to the following evidence that no
agent can truly be alienated from her values: when an agent comes to re-
pudiate one of her values it is always from the perspective of some contra-
ry value, and so the initial valuing state will fails to exist for her as a value
for her. In this way, an agent’s own values are guaranteed to always be in-
ternal since she cannot be alienated from them.

agent due to the fact that a second-order process can confer such authority, we might
think that the second-order states involved in the process need to get their authority from
a similar sort of even higher-order process. In order to solve this problem, Frankfurt later
concluded that the sequence must terminate in some sort of special kind of state or pro-
cess of identification, like a decision, that is invariably internal. See Frankfurt (1987,
1992). These views face a larger worry, however, in that by guaranteeing internality
through a special sort of identification state they fail to be reductionist views. Since I take
part of the motivation for identifying deep self mental states to be to give a reductionist
story of the conditions for attributional-responsibility, in this dissertation I will largely
draw on Frankfurt’s earlier second-order volition view. (While I do not address it head
on, many of my comments on the commitmental aspects of caring views apply to Frank-
furt’s even later view, on which identification amounts to passive commitment [Frank-
furt (2006)]).
41
See Watson (1975), Mitchell-Yellin (2014, 2015).
-36-
On the planning view, as proposed by Michael Bratman, the proposed
deep self mental states are personal self-governing policies about how one
will act in various circumstances.
42
In many cases these plans are set by
what the agent values, but in cases of normative silence in which agents
take there to be no distinct best course, they commit to personal policies
that apply also to similar situations in the future. An agent’s action is at-
tributable to her iff it is in line with her plans, and in this way plans confer
internality on motivational states that are instrumental or realizer desires
of these plans. Bratman proposes that plans are invariably internal for
agents like us because they partially constitute our diachronic agency. We
as agents cannot be alienated from our plans, not just because we set
them, but because they tie us together as coherent agents over time.
Proponents of the caring view offer a notably different picture on
which “identification is, for the most part, a passive process, garnering its
authority for self-determination from one’s nexus of cares.”
43
For Shoe-
maker (2003) these caring states are conceptual frames for clusters of emo-
tional dispositions that respond to the whims and woes of one’s cared for
object. For example, if an agent cares about the Phillies, she will experi-
ence anxiety over a potential loss, joy at a win, and despair if they don’t
make it to the playoffs. Sripada goes one step further and proposes that
cares are sui generis kinds of mental states with distinctive functionally
specifiable motivational, commitmental, cognitive, and emotional pro-
files.
44
In addition to a suite of emotional responses, if an agent cares about
X she will be intrinsically motivated to perform actions that promote the
achievement of X, be disposed to form judgments that cast X in a favora-
ble light, and will want to go on caring about X. An agent is attributional-
ly-responsible for her action iff, during the operation of the action-directed

42
See Bratman (2003).
43
Shoemaker (2003).
44
Sripada (2016).
-37-
psychological mechanisms that are involved in the etiology of the action,
her care exerts motivational influences (of sufficient strength) in favor of
acting as she does. Like plans on the planning view, cares are proposed to
play a crucial role in constituting diachronic agency via their commit-
mental aspect, which in turn helps to explain the fact that an agent cannot
be alienated from her cares. Since the motivational strength of cares stems
from the very thing that constitutes the agent as an agent over time, she
cannot be alienated from cares: they make up who she is as an agent.
One further option on offer for those who are worried that the other
accounts fail to establish a necessary condition on moral responsibility is
to hold an ecumenical, or disjunctive view. David Shoemaker currently
defends a view on which an agent either has to act in accordance with her
cares or with her values (or both) in order for her to be attributionally-
responsible for her action.
45
If the fact that they are both thought to be in-
variably internal is what makes cares and values good candidates to act as
deep self mental states, then the role of internality in an ecumenical view
is clear.
46
In principle, any combination of deep self mental state types
could be combined to form an ecumenical view just as long as the ways in
which the states are taken to confer internality on first-order motivational
states are compatible sorts of explanations and both ways of conferring in-
ternality on resultant actions can coexist.
7. Against Skepticism about Internality
I have shown how every major Deep Self view either explicitly or implicit-
ly relies on a concept of internality to explain why the actions that appro-
priately mesh with deep self mental states have the authority to speak for
agents that they do. But these views rely on support from intuitive under-

45
Shoemaker (2015a, 2015b).
46
Although Shoemaker’s own theoretical orientation, especially his endorsement of the
H-Tradition, on the face of it seems to de-emphasize the role of internality in Deep Self
theory.
-38-
standings of internal states as basically those from which one cannot be al-
ienated. But what would it mean to be alienated in the relevant sense in-
voked by each of these views?
If we understand internality in terms of alienation, and alienation in
terms of the idea that one’s mental state does not belong to oneself in
some way without further explication, this feeds directly into a skeptical
worry. The skeptical worry goes something like this: one reason we have
for thinking that it is possible for one’s desire to not belong to oneself in
some sense is that people report experiencing a feeling that one’s desire
somehow does not belong to oneself. But, if such reports are all we have to
go on, this may be quite shaky grounding for a theory of attributional-
responsibility since we can easily provide error theories that explain why
people report that their desires are alien.
Terence Penelhum expresses this line of criticism particularly forceful-
ly. He says, regarding an agent’s expression of the fact that his motivating
desire does not truly belong to him, that it is a

form of moral trickery…[that] involves an extension of the notion of
non-identification with one's own desires and behavior from the level
of harmless and even mildly illuminating metaphor to that of gross
literal false-hood. To say harmlessly that one is governed by a desire
that is not one's own is to utter a metaphor the literal translation of
which is that one is governed by a desire that one does not want to be
governed by. To say that the desire is not one's own and mean this lit-
erally is to say something obviously false: for the desire is operative
and therefore exists, and is not someone else's. This obvious falsehood
can be given the appearance of respectability with the aid of philo-
sophical theories about the division of the soul's faculties; and it is a
falsehood we are sometimes willing to swallow about others as well as
-39-
about ourselves, as in the Gallic concept of the crime passionel. But we
all know better.
47

Arpaly and Schroeder’s diagnosis is somewhat different:

When people find that they have not been as rational, sane, prudent or
moral as expected, they may experience…the cause of their misbehav-
ior as an alien intrusion
48
….In a culture such as our own, glorifying de-
cisiveness, self-control and ‘follow-through,’ and with a tendency to
medicalize failures of such traits, many agents will instinctively reject
evidence of themselves as straightforwardly akratic, as having simply
chosen poorly when they knew better. Instead, in some (and, it seems,
a growing number) of circumstances, they experience their failure as
apparently incomprehensible, an ugly intrusion upon their lives, and
the psychological cause of this failure seems an unpleasant intruder.
49

However, as both quoted passages suggest, internality skeptics have a
shared concern that such feelings of alienation are illusory, fabricated, or
the result of self-deception. Skeptics tend to base their accusations on two
factors. First, this way of speaking could easily be used as a fancy way to
excuse. Second, it seems obvious that all of a person’s desires are her own,
since they don’t belong to anyone else.
Frankfurt has a response to both of these worries, however.
50
He starts
by noticing that it is not obvious, except in a fairly trivial sense, that all of

47
Penelhum (1971): 670.
48
See also Buss (2012). Buss takes it that when some people speak of alienation, what they
really mean is that they act with a lack of a willing attitude. But, as she puts it, “just as
autonomous agency is compatible with stupidity and thoughtlessness, so too it is com-
patible with ambivalence, regret, disappointment, frustration, and self-criticism.”
49
Arpaly and Schroeder (1999): 383.
50
Frankfurt (1988): 61-62.
-40-
our desires belong to us since they do not belong to anyone else. We only
attribute some of the events in the history of a person’s body to that per-
son in a strict sense; some of them are mere happenings, such as getting
lurched forward on a bus or experiencing a bodily twitch. Just as it would
be unfair to say that because such behavior is not attributable to anyone
else, it must be attributable to the agent, so too is it unfair to make this in-
ference in regards to desires. Of course it does not decisively prove that
one may be alienated from one’s own desire, but the evidence in favor of
motivational alienation is not dissimilar to the kinds of evidence we have
of bodily-alienation. Allowing that a person can disclaim certain motiva-
tions as external is only as much of an opportunity for moral evasion as al-
lowing that a person can disclaim certain movements of her body as ex-
ternal is. And yet, we do not regularly take this as reason to be skeptical of
bodily twitches.
The difference here could be explained by the fact that motivational ex-
ternality may be fairly rare, such that some people almost never experi-
ence their motivating desires to do things like drink a beer or check their
ovens as being external.
51
Noting this fact can help give us a good explana-
tion for why many feel that speaking of external desires would be tanta-
mount to making up an excuse for one’s action. If those people who do
not experience externality were to speak about any of their desires in such
a way, they would be merely making an excuse for their behavior. The
tendency to extrapolate from personal experience in this regard is very
common, and bears similarity to public reaction to many psychological
conditions before (and sometimes even after) they are validated by sci-

51
This idea may not be particularly compatible with all Traditional Deep Self views,
though, since failures to meet the high bars they set for self-governing action are seem-
ingly commonplace. I return to this point again in Chapter 2, §5, once the positive view I
advance is on the table. The view I advance involves quite minimal conditions for non-
alienation and so is particularly well suited to give this response to the internality skep-
tic.
-41-
ence. Those with clinical depression are thought to be merely lazy, and
non-verbal autistic people are thought to be simply being difficult by peo-
ple who do not experience similar psychologies themselves. The ways in
which these accusations err seems only explicable by realizing that they
come about in part because the accusers extrapolate from what would be
going on in their own psychologies if they were to display similar behav-
ior. We should be wary to not let this kind of “intuition” color philosophi-
cal thinking about internality.
While I think this shifts the burden of proof onto internality skeptics, it
would be helpful for Deep Self theorists to have a more substantive, less
metaphorical account of internality. In the rest of this chapter, I aim to
provide one.
8. Internality is Approval
Consider the following case:

Exhausted Elsie: Elsie is extremely tired, but is desperately trying to
stay up to continue an important conversation with her friend. Despite
maximal effort to remain awake, she lies down because she knows she
is about to fall asleep. Elsie is not culpable for the cause of her exhaus-
tion, let’s suppose—it has just gotten very late. Is Elsie attributionally-
responsible for lying down to fall asleep, such that in principle it
would be appropriate to blame her on the basis of traits displayed by
her behavior?

According to every Deep Self view, Elsie is not attributionally-responsible.
What Deep Self views have in common in the way they intuitively explain
why an agent like Elsie is not responsible is that they show how even
though an agent like Elsie might be motivated to lie down, she does not ap-
prove of doing so. This lack of approval, I contend, is what licenses us to
say that her motivation acts counter to her, or ‘alienates’ her from her ac-
-42-
tion. In Chapter 2 I give an account of what I take approving to consist in,
but for now I just mean to point to an intuitive sense of approving: when
considering the options for what to do at t there is at some level some-
thing that the agent likes about or finds worthwhile about the prospect of
ϕ-ing at t.
52

If this intuitive notion of approving to some extent of an action rather
than merely being motivated to perform it is what makes the difference
between cases in which we are willing to grant that an agent’s action is
caused by an process that bears the mark of internality and ones in which
we are not, then we have located a common feature of any plausible can-
didate deep self mental state. Whether the deep self mental states are pro-
posed to be endorsements, valuings, plans, or cares, or some disjunction
of these, they succeed in guaranteeing agents’ resultant actions will be in-
ternal by guaranteeing that the agent will approve of her action. If I am
right about this, this means we can locate a common necessary condition
for attributional-responsibility shared by each major Deep Self theory.
9. Approval as a Common Necessary Condition for Attributional-
Responsibility
9.1 Approving is Necessary on the Endorsing View
It is perhaps easiest to see how approving of one’s course of action is a
necessary condition on attributional-responsibility on Frankfurt’s en-
dorsement view. Second-order volitions are meant to secure the fact that
the agent is not only motivated to act in the way that she does but that she
is personally invested in that particular course of action. This aspect of the
endorsement view comes out particularly clearly in Frankfurt’s discussion
of the contrast between wantons and full-fledged agents who form se-
cond-order volitions.

52
This should not be taken to imply that the agent necessarily takes it to be the best or
even a good course of action, just that she likes it.
-43-
A wanton, in Frankfurt’s sense, is someone who lacks second-order vo-
litions and so fails to take an interest in her will whatsoever.
53
The wanton
lets her strongest motivational states win out and move her to action irre-
spective of any opinion she might have about the matter. A wanton, ac-
cording to Frankfurt, lacks the capacity for self-reflective concern and thus
acts out of “mindless indifference to the enterprise of evaluating their own
desires and motives.”
54
Why, on Frankfurt’s view, are we meant to think
that the wanton’s actions aren’t attributable to her in the relevant sense?
For the wanton,

…it makes no difference to him whether his craving or his aversion
gets the upper hand. He has no stake in the conflict between them and
so…he can neither win nor lose the struggle in which he is engaged.
55

This means that when an agent is attributionally-responsible for her action
it is at least partially because she “has a stake” in the outcome of the con-
flict among the economy of her desires. Having a stake in the conflict be-
tween first-order desires competing to become an effective desire seems to
amount to having an opinion on the outcome. In other words, the agent
needs to approve of being motivated to act in the way that she does.
9.2 Approving is Necessary on the Valuing View
Approving is necessary for attributional-responsibility on the valuing
view as well, although a mistaken picture of the contrast between valuing
and desiring (in terms of interpretation of valuing deep self views) may
make this idea seem somewhat obscure. There is a picture of human agen-
cy that pits what an agent wants to do against what she thinks would be
best to do, conceiving of the two things as wholly separate. On this view it

53
Frankfurt (1988): 16.
54
Frankfurt (1988): 19.
55
Frankfurt (1988): 89.
-44-
is nice when an agent is motivated to do what she thinks is best, but this is
either accidental or caused by the agent bringing her motivations in line
with what is best; it is not that there is any motivational force to her judg-
ment that a certain course of action is best. Given this sort of picture, it
would be hard to see how approving of one’s course of action would be a
necessary condition on attributional-responsibility on the valuing view.
Valuing, however, is often held to have some more intimate connection
with motivation. And once this is granted, it is easier to see the connection
with approval.
This picture can be further specified in a number of different ways. For
example, on one view put forth by David Lewis, valuing X consists in de-
siring to desire X. Value, for Lewis, just is what a person would be dis-
posed to desire to desire in certain ideal circumstances.
56
If an agent acts in
accordance with her values, and valuing is given Lewis’s analysis, the
connection to the agent’s approving of her course of action is clear: the
agent who values ϕ-ing has a stake in wanting to be moved to ϕ. It’s inter-
esting to note that Lewis seems to take the intuitive connection between
valuing and approving to be strong enough to support an account where
valuing essentially just is a certain kind of approving.
While adherents of the valuing view need not be Lewisians about valu-
ing,
57
in order to make the view that valuing is connected up in the right
sort of way with agency in the sense that could reasonably ground attribu-
tional-responsibility, they do posit some sort of strong connection between
valuing and motivation via the fact that an agent approves of doing what
she takes to be the best thing to do.

56
Lewis (1989).
57
Given that coupling the Lewisian account of valuing with the valuing Deep Self view
would make the account of agency hierarchical, proponents of the view, like Gary Wat-
son, who criticize the hierarchical nature of the endorsing view might even have special
reason not to adopt it.
-45-
To act on one’s valuing state in the sense that defenders of the valuing
view conceive of it is never to merely act in accordance with what one co-
incidentally believes to be good. Rather, valuing is thought to have some-
thing to do with agency, and thus to have an essential connection to moti-
vation. Gary Watson’s characterization of valuing makes this connection
clear:

Now, to be sure, since to value is also to want, one’s valuational and
motivational systems must to a large extent overlap. If, in appropriate
circumstances, one were never inclined to action by some alleged eval-
uation, the claim that that was indeed one’s evaluation would be dis-
confirmed. Thus one’s valuational system must have some (considera-
ble) grip upon one’s motivational system.

So the notion of evaluation here is in an important way personal; an
agent’s values issue from a faculty that has a “grip” on her motivations. If
the thought “it’s the right thing to do” is meant to have a grip on motiva-
tion, it must be because the second thought, “and I approve of doing the
right thing,” is also present in some form. Whether the second thought is a
matter of the meaning of rightness, a truth about human nature, or a
standing disposition that happens to be present in agents like us (or some-
thing else), the fact that the agent approves of acting as she does because it
is right seems baked into the story.
Watson explains that the sort of motivational power exerted by valu-
ing is special because we are concerned to bring about the satisfaction of
desired ends for some reason that goes beyond the fact that acting allevi-
ates the suffering of having the unsatisfied desire. For an agent to value ϕ-
ing is for her not just to desire to ϕ but to set ϕ-ing as an end for herself.
And so an agent must not only be motivated to ϕ, but also actually ap-
prove of ϕ-ing for some reason. As Watson puts it,

-46-
Now, it must be admitted, any desire may provide the basis for reason
insofar as non-satisfaction of the desire causes suffering and hinders
the pursuit of ends of the agent. But it is important to notice that the
reason generated in this way by a desire is a reason for getting rid of the
desire, and one may get rid of a desire either by satisfying it or by
eliminating it in some other manner (by tranquilizers, or cold show-
ers). Hence this kind of reason differs importantly from the reasons
based upon the evaluation of the activities or states of affairs in ques-
tion. For, in the former case, attaining the object of desire is simply a
means of eliminating discomfort or agitation, whereas in the latter case
that attainment is the end itself. Normally, in the pursuit of the objects
of our wants we are not attempting chiefly to relieve ourselves. We
aim to satisfy, not just eliminate, desire.
58

And so, on the valuing view, valuing is necessary for attributional-
responsibility precisely because it guarantees that the agent’s effective de-
sire becomes effective because she approves of her course of action. And
so, proponents of the valuing view, too, should hold that approving is a
necessary condition on attributional-responsibility.
9.3 Approving is Necessary on the Planning View
According to the planning view, an agent is attributionally-responsible iff
she acts in accordance with her policy about how to act in such a situa-
tion.
59
Her policy-setting may be governed by her values in many cases,

58
Watson (1975): 210-211.
59
In the article I draw from here, Bratman specifically brackets off questions of responsi-
bility, focusing his discussion on identification alone: “…I want to see if we can, instead,
describe without independent appeal to judgments of responsibility—a fairly unified
phenomenon that is plausibly seen as the target of such talk of identification” (Bratman
[1996]: 2) His picture might just as easily be considered as a contending deep self account
of attributional-responsibility, however, and, with this caveat, I will proceed as though it
were put forth as one.
-47-
and in those cases the same considerations I raised in the last section
should lead to the conclusion that she must to some degree approve of her
course of action. But in cases that are normatively underdetermined, she
forms or acts on a previously determined policy that she just decides to
treat as reason-giving. If agents in these cases just follow personal policies
that are not governed by anything as strong as all-things-considered
judgments about what would be best, it might be far from clear that
agents who act in accordance with these policies need approve of their ac-
tions.
However, following Velleman, Bratman acknowledges the possibility
of a case in which an agent forms a plan in such a detached way that the
action she takes when she fails to act in accordance with it would still be
attributable to her. This provides a reason to supplement the story about
what must obtain in these sorts of cases for the agent to be attributionally-
responsible. Bratman supplements his account by adding that the agent
who ϕs must be satisfied with her decision to treat her desire to ϕ as rea-
son-giving. If an agent meets this condition, it seems to me that she would
have to approve of at least something about it.
Interestingly, for Bratman, the agent needs to be satisfied with the de-
cision to treat the desire as reason-giving not just at the time of the deci-
sion but also at the time of action. This is evident in the following passage:

In “The Importance of What We Care About,” Frankfurt emphasizes
that one can decide to care about something and yet “when the chips
are down” fail to care about it. Perhaps, similarly, I might decide to
treat my desire, say, to seek a reconciliation with an old acquaintance
as reason-giving and yet, when the chips are down, find myself unable
to treat it this way. I might find that, despite my decision, and despite
the fact that I am satisfied with that decision, I do not care enough
-48-
about reconciliation. In such a case it seems that I have not fully suc-
ceeded in identifying with my desire for reconciliation.
60

This seems right, since the agent’s approval at the time of action seems to
intuitively be what matters.
Now, Bratman understands satisfaction with a policy not in terms of
the presence of a particular attitude, but rather, in terms of the alignment
and integration of the policy with the agent’s other policies: “One is satis-
fied with such a decision when one’s will is, in the relevant ways, not di-
vided: the decision to treat as reason-giving does not conflict with other
standing decisions and policies about which desires to treat as reason-
giving.” But notice that this analysis of satisfaction only makes sense as
an analysis of satisfaction when we think of the sum of the agent’s other
policies as providing a guide to what the agent generally approves of do-
ing. Again, here, agential approval of some form seems necessary on the
view.
9.4 Approving is Necessary on the Caring View
Several aspects of the caring view might be thought to implicate approval.
First, among “joy” and “elevation,” “approval” is explicitly listed by
Sripada as one of the positively valenced emotions that agents are dis-
posed to experience when they act in ways that advance their cares.
61
If
the emotional aspects of caring states are meant to take center stage on the
view, it seems plausible that there must be some element of an approving
emotion toward one’s action in order for the agent to count as caring in
the relevant sense.
62
There might be a sense of caring on which happiness

60
Bratman (1999): 202.
61
Sripada (2016): 8.
62
Is approval an emotion? This may be more or less plausible depending on one’s ac-
count of what emotions are. As I will explain in more detail in Chapter 2, I take approval,
instead, to be a function of an agent’s motivational profile.
-49-
towards acting in ways that advance the cared-for object without approval
is sufficient for caring, as it seems possible in some sense to “care about
something despite oneself.” For example, you might get a small twinge of
pleasure from someone being called out on her bad grammar in an inter-
net comment thread, once you’ve already renounced that practice as being
classist. In one sense, it might seem appropriate to say that you still care
about grammar, despite yourself. But it seems implausible that cares in
this sense would be good candidates for deep self mental states. Even a
wholly unwilling addict might get some pleasure from drinking, but a
theory that gives the result that an unwilling addict is responsible since
she cares about drinking alcohol seems counter to the aims of a Deep Self
view.
In addition, the evaluative judgment aspect of the caring view further
emphasizes the importance of the agent’s approval of her action. If an
agent is attributionally-responsible for her action that furthers the end X,
her action will be suitably related to the fact that she is disposed to form
judgments that cast X in a favorable light. Recall again the earlier conclu-
sion, regarding the valuing view, that the relevant sorts of evaluative
judgments that can ground an agent’s attributional-responsibility must be
related to her agency via the fact that she approves of those actions on the
basis of her values. These same considerations apply to the evaluative as-
pect of the caring view as well. Sripada writes that the valuing view
“places all elements of the relevant class of evaluative judgments within
one’s deep self.” In contrast, “the care-based view allows that many eval-
uative judgments don’t bear any connection to the deep self, namely those
that don’t bear the right dispositional tie to one’s cares.”
63
In this way, the
caring view is even more explicit about the fact that acting in accordance
with evaluative judgments usually implicates internality because agents
act on their judgments because it matters to them to act in accordance with

63
Sripada (2016): 1210.
-50-
what’s right. This sense of mattering, it seems to me, is fundamentally tied
to approving.
Finally, the motivational and commitmental aspects of caring seem to
implicate approval. For Sripada, if an agent cares about X she is intrinsi-
cally motivated to perform actions that promote the achievement of X. She
also will want to continue caring about X. Together, this makes it the case
that when ϕ-ing promotes the achievement of X, and X is something the
agent cares about, the agent has an intrinsic desire with a positively va-
lenced higher-order attitude towards being motivated by it. Although the
consideration of the promotion of the achievement of X may not outweigh
other factors in a given case, the agent still would seem to have to approve
of being motivated to ϕ in at least a minimal or pro tanto sense for this to
be the case.
It’s important that the desire to promote the achievement of X is intrin-
sic, because this rules out cases where the agent might only be motivated
to ϕ due to wanting to quell an external urge to promote the achievement
of X, as I discussed in the section on the valuing view. If an agent acts on a
desire to promote the achievement of X to quell such a desire it would
seem strange to describe the agent as caring about X, and it would equally
seem strange to describe the agent as approving of her action. Shoemaker
makes a related point in discussing how his caring view differentiates
non-attributable actions of unwilling addicts from attributable weak-
willed actions:

For the unwilling addict, it matters greatly that his desires for the drug
are satisfied or eliminated—it may not matter how. For the merely
weak willed, however, the primary object of one’s care-dependent de-
sire is not the mere satisfaction (or elimination) of some other non-
care-derived desire.
64

64
Shoemaker (2003): 103.
-51-

Again here, the contrast seems best explained via the concept of approval.
The weak-willed agent, unlike the unwilling addict, actually approves of
something about acting as she does that goes above and beyond the alle-
viation of an urge.
And so, given each major Deep Self view, it is a necessary condition on
attributional-responsibility that the agent approves to some degree of her
action.

-52-
Chapter 2: A Sufficient Condition for Attributional-Responsibility
65

1. Introduction
Chapter 1 surveyed some of the advantages of the Deep Self approach to
attributional-responsibility. I showed that the appeal of each major kind of
Deep Self view currently on offer lies in the fact that when an agent’s ac-
tion meshes with the proposed deep self mental states, her action is guar-
anteed to be caused by a process from which she is not alienated. I sug-
gested that this lack of alienation amounts to the agent’s approving of her
action in some sense, and argued that proponents of each Deep Self view
have reason to hold that approving of one’s action to some degree is a
necessary condition on attributional-responsibility.
In this chapter I argue that the notion of approval is not only necessary
for attributional-responsibility, but also sufficient. First, I show that
each major Deep Self view contains additional elements beyond securing
the agent’s approval of her action that are unnecessary for attributional-
responsibility. None of the deep self mental states: valuing, caring, en-
dorsing, or planning, as conceived of by proponents of traditional Deep
Self theories, are necessary for attributional-responsibility. Some argue
that attributional-responsibility might be secured by more than one men-
tal state kind, and I carry this move to its logical conclusion. Instead of at-
tempting to identify a kind of mental state or process that is invariably in-
ternal, I suggest that the way to proceed is to give an analysis of internali-
ty itself and hold that agents are attributionally-responsible for any in-
stance of an action caused by an internal process. I analyze internality in

65
This chapter greatly benefitted from comments on earlier drafts by Eugene Chislenko,
Nathan Robert Howard, Renee Jorgensen Bolinger, as well as discussion by the partici-
pants of the fourth New Orleans Workshop on Agency and Responsibility, Temple Uni-
versity’s Freedom and Responsibility Fall 2017 Seminar, and USC’s Spring 2016 Disserta-
tion Seminar.
-53-
terms of minimal approval, so an agent is attributionally-responsible iff
her action is caused by a process the presence of which ensures that she
minimally approves of acting on the motivational state that she in fact
does.
To minimally approve of ϕ-ing, I suggest, is to have a hypothetical,
partial desire to act on a motivation to ϕ for some further aim other than
to get rid of the desire to ϕ. I motivate each aspect of this analysis in turn,
and then turn to some advantages the view has over traditional Deep Self
views. My view is comparatively agnostic regarding the processes in
volved in the production of action, and as such, it is better able to stand up
to some of the most deeply entrenched problems for the Deep Self ap-
proach.
2. Unnecessary Components of Deep Self Views
As I argued in Chapter 1, each major Deep Self view is consistent with the
idea that approving in some sense is a necessary condition on attribution-
al-responsibility. However, proponents of each major Deep Self view take
there to be additional elements to their account that are necessary for at-
tributional-responsibility. This, I argue, is false. Each Deep Self view suc-
ceeds at locating a sufficient condition for attributional-responsibility just
to the extent that it secures the fact that in each instance of intuitively at-
tributable agency, the agent will approve to some degree of her action. Se-
curing the fact that the agent approves of her action in some minimal
sense is both necessary and sufficient for attributional-responsibility.
The first part of this argument will consist of showing that each major
Deep Self view contains elements that are unnecessary for attributional-
responsibility. In this section I show how the valuing, caring, and plan-
ning views contain unnecessary elements. I hold off on addressing the
unnecessary aspects of the endorsing view until §4 and §5 of this chapter,
as this part of this argument serves to set up my positive view.
-54-
2.1 Valuing is Not Necessary for Attributional-Responsibility
If we think of the valuing view as being supported primarily by the idea
that the mark of agency comes from an agent’s acting on her consideration
of what she takes to be best to do, it is fairly easy to generate counterex-
amples. There are many actions for which an agent is attributionally-
responsible but does not judge her course of action to be all-things-
considered best. For example, in cases of normative silencing, an agent
cannot form an all-things-considered judgment about what it would be
best to do since there is no best option. Consider instead the view on
which the agent must consider the action to be among her best options.
Even on this view, there are counterexamples in which an agent may act
in a way that she is attributionally-responsible for where she does not take
her action even to be one of the all-things-considered best things to do but
nevertheless embraces it on some level. Gary Watson, the original propo-
nent of the valuing view, now endorses this objection and has come to see
his earlier view as being too rationalistic. As he puts it,

When it comes right down to it, I might fully ‘embrace’ a course of ac-
tion I do not judge best; it may not be thought best, but is fun, or thrill-
ing; one loves doing it, and it’s too bad it’s not also the best thing to do,
but one goes for it without compunction.
66

Cases in which people act out of love that they themselves take to be
wholly irrational seem to be commonplace as well as particularly compel-
ling examples. People who are in love wholeheartedly act in ways that are
not among the options they consider all-things-considered best, nor do
they sometimes even consider them good. On a fairly regular basis people
return knowingly to undeserving exes, begrudgingly do unreasonable fa-
vors for loved ones, and support family members in ways that go beyond

66
Watson (2004): 168.
-55-
the realm of reason. And they seem to do so with enough agency that we
rarely think of these agents as being excused from responsibility for their
actions as a result of their reason being undermined.
But, as David Shoemaker notes, despite their intuitive appeal, these
cases are liable to be recast in terms of value by proponents of valuing
views.
67
For example, Angela Smith argues that in these sorts of cases,
agents merely speak in terms of ‘having no reasons’ to do as they do as a
façon de parler.
68
When it comes down to it, agents really do judge what
they are doing out of love to be best given their reasons, such as reasons of
shared history, being entangled with someone in such a way, or the like.
When an agent has the explicit thought that she ought not go back to her
ex but does so anyway, she uses ‘ought’ only in the inverted commas
sense. The thought she has, valuing view adherents allege, must really be
something like everyone would counsel me not to do this, but still I find it most
valuable.
Shoemaker appeals to cases of well-planned revenge as another exam-
ple of actions an agent may be attributionally-responsible for despite un-
dertaking them independently of what she values. During the plotting of
revenge an agent may come to realize that her doing so is morally wrong
and prudentially disastrous, and even that there is nothing valuable what-
soever in so-acting yet nevertheless carry out her plot. And yet, agents are
paradigmatically attributionally-responsible for acting on such desires for
revenge. While these seem like perfect cases for making the point that val-
uing is not necessary for attributional-responsibility, valuing theorists
could argue that such agents are impossible. In the ordinary case, agents
who act like this, contrary to appearances, do value retributivism either in
the abstract or in the particular case, and they are either wholly self-
deceived about their anti-retributivist values or else the values are merely

67
Shoemaker (2015a): 129-134.
68
Smith (2012).
-56-
aspirational: such agents only desire to be such that they value anti-
retributivism. Otherwise, perhaps the agents are suffering from the sort of
condition that should undermine their being held attributionally-
responsible.
It seems to me, frankly, implausible that every single case of acting out
of love and every single case of acting out of revenge are realistically re-
cast in such a way. One worry is that the intuitions that these cases are
well described in terms of value are theory-laden and influenced by inde-
pendent considerations that favor motivational judgment internalism, the
view that there is a necessary connection between moral judgment and
motivation. Motivational judgment internalism is only plausible when it
includes the caveat that agents are motivated by their values except in
cases of volitional disorder. The concern is that if finding motivational
judgment internalism plausible influences intuitions about the cases in
which an agent acts out of love or revenge, some sense of what differenti-
ates volitional disorder from normal cases might already be smuggled in
to the picture, when that is the very thing that is meant to be in question.
Part of the force of these revenge and love cases, and our resistance to
recast all of them in terms of values, is that they are realistic, common-
place, and seem to differ phenomenologically from the way they are de-
scribed by proponents of the valuing view.
69
Perhaps the best case against

69
Mere hypothetical cases are more fairly subject to be recast. To take a case that has been
offered in support of the view that valuing is not necessary for attributional-
responsibility that I do not think will be successful in convincing any valuing theorists,
Shoemaker gives a case of an artist and philosopher who cares only about being the type
of person who lives beyond the realm of justification such that when he judges some-
thing to be best for him he instead acts on a perverse desire to do the very opposite.
When he acts in this way, according to Shoemaker, he clearly doesn’t think his action is
all-things-considered best, but he certainly seems to be attributionally-responsible for
what he does. In order to make the case resistant to the recasting problem, Shoemaker
specifies explicitly that this agent doesn’t desire to live beyond the realm of justification
because he values it, but I think this is unlikely to convince any valuing view proponents.
-57-
the valuing view is a real life example Shoemaker provides in which his
own volitionally necessitated will came apart entirely from what he
judged to be all-things-considered best. He describes a time in which he
encountered an injured mouse in his apartment, and knew that the best
thing to do would be to take a hammer and kill the mouse instantly, sav-
ing the creature from a more protracted and painful death that would be
caused by simply letting it out into the wild. But as he stood over the
mouse with his hammer, he found he couldn’t bring himself to do it, and
chose to let the mouse free instead. This is a case on which he judged it
clearly best to take one course of action but couldn’t bring himself to act
accordingly in such a way that nevertheless revealed something about
what he is like as an agent. It seems very implausible to say that he was
self-deceived about his view that putting the mouse out of its misery
would be best; he is surely a very reasonable guy when considering what
one ought to do. It makes most sense to suppose that he was instead moti-
vated by some combination of squeamishness and contorted mercy, nei-
ther of which operated via encouraging him to think it was among his all-
things-considered best options to let the mouse go free. And yet, despite
the fact that he acted completely contrary to reason, he certainly seems at-
tributionally-responsible for his action. And so it does not seem that con-

Valuing view proponents will describe such a man as engaging in paradoxical decision
making. When he performs one of these contrary-to-judgment actions, the valuing theo-
rist will argue, it must surely be that he takes himself to have justificatory reason to do
so; he can even provide the reason: he does it because it is the opposite of what he judges
best! In coming to this realization, the undertaking of this opposite action becomes what
the agent judges it best to do, but now should he do the opposite of that? Since this kind
of case is not common, there is no reason to doubt the claims of the valuing view adher-
ent that this sort of case is paradoxical.
-58-
sidering a course of action to be among one’s best options is necessary for
attributional-responsibility.
70

2.2 Neither Planning nor Caring is Necessary for Attributional-
Responsibility
I reject caring and planning accounts of deep self mental states as being
necessary for attributional-responsibility for reasons that apply to both,
and so in this section I treat them in tandem.
Both caring and planning views posit that in order for agents to be at-
tributionally-responsible they must act from a motivation that is in some
way importantly tied to their past and/or future motivational architec-
tures. As I discussed in Chapter 1, this aspect of those views is usually
touted as an advantage. If agents are constituted over time by plans or
cares, this offers a neat explanation as to why compulsive behavior that
does not mesh with one’s plans or cares is alienating: it literally stems
from outside of the agent. While diachronic coherence in one’s action ar-
guably is a marker of agency par excellence, it is much less clear that acting
in accordance with this kind of diachronic scaffolding is necessary for at-
tributional-responsibility.
David Shoemaker, arguing against the necessity of Bratman’s planning
states, gives an example of a racist who prudently decides “to never let his

70
Silverstein (2017) gives a compelling case that practical reasoning, understood as rea-
soning about what to do and normative reasoning, understood as reasoning about what
one ought to do, are non-identical. An agent can settle the question of whether or not she
should dine out for lunch while deciding to forestall reasoning about whether or not she
will in fact do so until tomorrow. In this case normative reasoning has finished long be-
fore practical reasoning has even begun. While this by itself does not foreclose the possi-
bility that attributable agents must act in accordance with reason, it does drive a further
wedge into the supposed tight connection between valuing and ordinary attributable ac-
tion.
-59-
hatred structure his will.”
71
When he witnesses a KKK event though, one
day, he says, “woo hoo!” In this case the man pretty clearly has not decid-
ed to treat his racist desires as giving him a reason for action, but never-
theless appears to be attributionally-responsible. Shoemaker explains that
this man is attributionally-responsible because, despite the fact that he
judged it best not to let his racism show, he is a racist; he has a longstand-
ing hatred for Black people and these attitudes reveal what he is generally
like as a person.
But consider a modification to the case that seems to tell against the
caring view as well. Suppose the man who lets out a “woo hoo!” upon
seeing a KKK event has never had a prior racist thought or attitude and
never has one again. He is merely caught up in the racist fervor of the
KKK demonstration and is temporarily overcome. Nevertheless, the man
in this situation seems just as attributionally-responsible for his reaction.
This indicates that it is the man’s approval of his action in the moment, ra-
ther than across time, that matters for attributional-responsibility. Suppos-
ing we were confidently assured of the fact that this was and would al-
ways be his sole racist action, this might have some effect on the sorts of
attitudes it would appropriate to direct towards him compared to the
thoroughgoing active racist, but it does not affect the fact that his current
act is attributable to him.
It’s possible that mental states like caring or planning may help illumi-
nate aspects of an agent’s personality that are elusive when considering
only time-slice properties, but I think we should be skeptical of any claims
that such understanding is required for attributional-responsibility and
think to do so would be to conflate two different senses of the Deep Self.
72

Agents who are swept up in moments of anger, passion, or fleeting inter-
est can still act in ways that they stand behind; sometimes they even do so

71
Shoemaker (2005a): 121.
72
See Chapter 1, §3.1.
-60-
wholeheartedly. Thus, there is reason to think that even if the planning
and caring views point to mental states that guarantee that an agent ap-
proves of her behavior in the right sense for attributional-responsibility,
acting in accordance with these mental states cannot be necessary for at-
tributional-responsibility.
Another, albeit less conclusive, reason to be suspicious that caring or
planning states are necessary for attributional-responsibility is that they
come with ontological baggage. Both Bratman’s planning view and Sripa-
da’s caring view involve countenancing the existence of new kinds of
mental states that are said to play a crucial role in agential architecture.
Considerations of parsimony put the burden on these theorists to prove
that adding new mental state kinds to our ontology is absolutely neces-
sary. This concern is amplified by the fact that, at least outside of the do-
main of thinking about responsibility, it is intuitively plausible that plans
and cares might be constructed out of beliefs and standing desires with
particular sorts of content.
73
Parsimony considerations are always weighed
against the explanatory power of views, and both Bratman and Sripada
offer arguments elsewhere about the global roles of caring/planning states
on agent architecture. These comprehensive pictures of agency could each
be given dissertation-level treatments, and so I will certainly not attempt
to discredit them here. I merely mean to mark the fact that, without ante-

73
It may be worth considering the viability of a planning or caring view of attributional-
responsibility coupled with the view that planning or caring states are in fact constructed
out of simpler component parts. Shoemaker (2003) in some ways presents an example of
such a view, and so is not subject to the same criticism I pose in the section for Sripada’s
and Bratman’s views. In thinking about the viability of demarcating deep self mental
states via their content rather than their kind, it may be worth considering the view put
forth in (Velleman [1992]), on which agents effective motivation must be suitably related
to a standing desire with a particular content: the desire to act in accordance with rea-
sons. Depending on how one construes Frankfurt’s second-order volitions, his endorse-
ment account may also be construed in this way.
-61-
cedently agreeing with the arguments that these theoretical posits are nec-
essary to explain other aspects of human agency, it is more difficult to get
on board with the planning or caring view than it would be to accept a
view that makes do with mental states that are more or less universally
agreed to exist. There is reason to wonder whether we really need to posit
a special kind of mental state to do the work of supplying a criterion for at-
tributional-responsibility.
3. From Type-Disjunctive to Token-Disjunctive Views
Let’s take stock. I have now argued that each traditional Deep Self view
provides a deep self mental state that is sufficient but not necessary for at-
tributional-responsibility. Each traditional Deep Self view succeeds insofar
as it articulates a way that an agent may approve of her action such that
she is not alienated from it. But the additional elements of each view that
go above and beyond the fact that the agent approves of her action are su-
perfluous; as long as the agent approves of her action and this is related to
why she acts, it doesn’t seem to matter much how she comes to do so.
One way of proceeding in light of this would be to hold a disjunctive
view on which an attributionally-responsible agent’s action may be
caused by any of the several types of mental state that implicate that the
she will approve of her resultant action. As mentioned in Chapter 1, §6,
David Shoemaker’s ecumenical view is a disjunctive view that is also mo-
tivated by the thought that there is no single type of deep self mental state
that is necessary for attributional-responsibility. On the ecumenical view,
an agent is attributionally-responsible for her action iff her effective moti-
vation aligns either with her cares or her values. If endorsing and plan-
ning views also provide sufficient conditions for attributional-
responsibility, one way to capture this would be to hold a disjunctive view
on which an agent is attributionally-responsible for her action iff her effec-
tive motivation properly aligns with her cares or her values or her en-
dorsement or her plans.
-62-
Recall, though, the distinction from Chapter 1 between the B-Tradition
and the H-Tradition. On the H-Tradition, holding responsible is meta-
physically prior to being responsible and to be analyzed in terms of the re-
sponses (usually conceived of as reactive attitudes) that are fitting in hold-
ing one another responsible. Since Shoemaker follows the H-Tradition, his
ecumenical view is allegedly supported by the fact that agents’ actions
that mesh with either cares or evaluative commitments both seem to make
fitting certain paradigm aretaic sentiments in response such as admiration
and disdain. In Chapter 1, §5, I expressed my commitment to the B-
Tradition, according to which being responsible takes priority over hold-
ing responsible. So, while the responsibility responses that appear to be
fitting may provide a defeasible epistemic guide to the conditions of re-
sponsibility, I have argued that we also need to locate some further reason
why those particular criteria must hold rather than some other criteria.
Given a methodological commitment to find some deeper reason relat-
ed to the structure of attributable agency in virtue of which the proposed
conditions for attributional-responsibility are in fact the correct conditions,
the disjunctive view occupies a somewhat precarious position. Given the
B-Tradition, any disjunctive view invites the further question: in virtue of
what should we count the kinds of mental states the disjunctivist posits,
and only the kinds of mental states she posits as constituting sources of the
deep self? Suppose she answers this question by identifying some proper-
ty of these kinds of mental states, F, which justifies their inclusion as states
that can imbue first-order motivational states with the authority to speak
for the agent. If the fact that states of that type have F is the reason they
can speak for the agent, the disjunctivist deep self theorist should admit
that a state of any kind, just so long as it has F, can imbue an agent’s first-
order motivational states with the authority to speak for her.
To maintain her position in light of this challenge, the disjunctivist
would have an additional argumentative burden not shared by traditional
Deep Self theorists. While it is open to traditional Deep Self theorists to
-63-
claim that there is something uniquely special about endorsing, valuing,
planning, or caring that makes that kind of state the only kind that can
imbue first-order motivational states with the authority to speak for the
agent, the disjunctive Deep Self theorist would have to admit that there is
some property shared by each kind of Deep Self state that has that author-
ity and then furthermore argue that no other state could in principle have
that property.
To preserve the advantages of the disjunctive view without having to
meet that challenge, one could instead hold a disjunctive view of the fol-
lowing form:

Type-Disjunctive Attributional Responsibility: An agent is attribu-
tionally-responsible for ϕ-ing iff she acts on a motivational state that
meshes in the relevant way with a further mental state type that has
some property, F, by virtue of its being a token of that type.

But if F is a feature most fundamentally not of mental state types but of
all the token mental states that fall under a type (for example, if F is a fea-
ture of every caring state), then this invites a further question: what is the
relevance of mental state types to the theory? If the type of mental state
plays no role in the theory except that of being a good predictor of wheth-
er the token will have F, then not including other mental state tokens that
have F despite not being of a type that guarantees that all its members
have F would be arbitrary.
A better alternative is to adopt what I’ll call token-disjunctivism. Ac-
cording to token-disjunctivism, an agent is attributionally-responsible for
any action caused by a motivational state that properly meshes with any
further mental state or collection of states that have F. These mental states
may be tokens of types whose tokens invariably have F, but the types of
which they are tokens need not have only tokens that have F.

-64-
Token-Disjunctive Attributional-Responsibility: An agent is attribu-
tionally-responsible for ϕ-ing iff she acts on a motivational state that
aligns in the relevant way with a further mental state token (or group
of mental state tokens together) that has some property, F.

What should we fill in for F? Since, as I have argued, each Deep Self
view seems to provide a successful criterion for attributional-
responsibility only to the degree that it ensures that the agent approves of
her action, and that no additional elements are necessary, I suggest that
for F we fill in “the fact that the agent approves to some degree of her ac-
tion.” I am proposing that the fact that an agent’s action is brought about
by a process that is non-coincidentally related to the fact that she approves
of her action is both necessary and sufficient for attributional-
responsibility.
In what follows I give an account of what it means for an agent to min-
imally approve of her action, and argue that a token-disjunctive account of
attributional-responsibility based on minimal approval has a number of
advantages over traditional Deep Self views.
74

4. Partial Identification
I am going to build up an account of what it takes for an agent to approve
of her action in the sense that I believe should ground an account of at-
tributional-responsibility. Just as on Harry Frankfurt’s endorsing view, I
think the agential criterion for attributional-responsibility has to do with

74
I argued in Chapter 1 that the “Deep Self” label is ill-defined, and that the positive view
I advance may or may not count as a Deep Self view depending on the criteria that are
prioritized in classification. However, I will adopt the convention of referring to my view
as a “Deep Self view”. That said, I don’t put much stock in whether or not the Deep Self
label rightly fits the view I put forth, and I only label it as such to bring out the fact that it
clearly shares some of the advantages, both dialectical and substantive, of traditional
Deep Self views.
-65-
the structure of one’s will and is built out of desires. Although the view I
will put forward is quite different from Frankfurt’s, I arrive at it by noting
the ways in which it contrasts with views like Frankfurt’s.
The first way in which I think an account of attributional-responsibility
needs to diverge from Frankfurt’s view is that it ought to locate the crite-
rion for partial identification rather than complete identification. In Chap-
ter 1 I hinted at the view that identification with a course of action need
only be partial in order for the agent to be attributionally responsible for
it, since it seems that agents need only approve of their course of action in
some way, or to some degree. This idea stands in contrast to most tradi-
tional Deep Self views. For example, recall that on Frankfurt’s view, in or-
der to be attributionally-responsible for ϕ-ing, an agent must reflect on her
first order desires and come to desire most to act on her desire to ϕ. Cru-
cial to this view seems to be something like the idea that an agent must
completely stand in favor of coming to act as she does. But views like this
seem unable to respect the intuitive judgment that agents can be attribu-
tionally-responsible for actions that are the result of weakness of will.
75
In
Chapter 3 I will argue that the view I put forth in this chapter is the most
promising way for deep self theorists to handle weakness of will cases.
For now I just want to show how weakness of will cases intuitively move
us away from the view that complete identification is necessary for attrib-
utional-responsibility to the view that merely partial identification is suffi-
cient.

75
This objection has been raised, in various forms, by Vihvelin (1994), Haji (1998), Haji
(2002), Fischer (2010), Fischer (2012), McKenna (2011), McKenna and van Schoelandt
(2015), and Strabbing (2016). A complication that should be noted here is that not all
Deep Self views are always presented as views of the conditions for responsibility, and
some theorists propose these sorts of conditions merely as views of self-governance, in-
tentional action, autonomy, or agency par excellence. This criticism should be taken to ap-
ply only to the (many) Deep Self theorists who do take acting in accordance with one’s
Deep Self to be a criterion for responsibility.
-66-
Consider the following case:

Murderous Max: Max strongly desires to go out on a killing spree this
morning because he hates people and there is almost nothing he likes
more in the world than shooting them—in fact he thinks of himself as
having made an art of it, perfecting his technique more and more with
each kill. There is one thing, however, that he cares about even more:
he is extremely committed to his morning workout routine. As much
as he wants to go out on a killing spree, he also realizes that if he does
that, he’ll have to forego his morning workout ritual. He knows that if
he misses even one morning of working out, he’ll probably fall off of
his routine, and he’ll thus sacrifice the progress he hopes to be making.
So, after considering and giving some post-reflective weight to his op-
tions to act on both desires, he decides that acting on his desire to work
out is what he most wants to do, it aligns best with his values, and is
consistent with the plans he has set for himself. However, his desire to
hone his murderous craft by going on that killing spree ends up just
being so intense that he caves from lack of willpower and goes out and
does the deed.

Max’s killing spree certainly seems to reveal something about what he is
like such that his action speaks for him for the purposes of attributional-
responsibility; he is clearly blameworthy despite the fact that he endorses,
in the senses relevant for most Deep Self accounts, going to the gym at the
time of action. On Frankfurt’s view, Max forms a second-order volition to
go to the gym but some other first-order desire becomes his effective mo-
tivation. Max does seem to identify with his desire to go on a killing spree
in a way that seems relevant. But if Frankfurt’s account were the right ac-
count of identification, we would have to say that he does not and, as a re-
sult, is not attributionally-responsible for his killing spree. But this seems
clearly to be the wrong result.
-67-
Traditional Deep Self views suffer from this problem because they are
specially designed to show how acting on desires that agents themselves
do not see as most favorable undermine agency in such a way as to make
such actions non-attributable, as in the case of compulsion. But weakness
of will is often described as the failure to act in accordance with what the
agent finds to be the most favorable course of action, and yet intuitively
we think weak-willed actions are attributable. These views tend to lack the
resources to differentiate compulsion from weakness, so in exempting
compulsive action they overextend to exempt weak-willed action.
The lesson we should draw from this is that while it makes sense to
say that in compulsive cases agents are wholly alienated from their ac-
tions, it seems that there are cases, like the case of Murderous Max, in
which an attributable agent doesn’t stand behind his action as being the
thing to do and yet still endorses it, in at least a partial way. While most
Deep Self views give accounts of what would be required for an agent to
fully stand behind her action, an agent does not need to wholeheartedly
endorse her course of action in order to be responsible for it. If lack of
identification with one’s action n is to be a relevant consideration in ex-
empting an agent from attributional-responsibility, we’ll need to under-
stand this in terms of total lack of identification with one’s action, rather
than less-than-complete identification with one’s action.
76

In order to see the difference between what I’ll call complete and mere-
ly partial endorsement, it will be easiest first to look at a case in which an
agent feels herself being pulled in more than one direction by her first-

76
This is not to deny the possibility that degree of identification with one’s action may
play some further role in determining features of blameworthiness. It seems at least prima
facie intuitive that wholehearted embrace of morally wrong action could be cause for ex-
tra/more intense blame or blame of a special kind of character, and my proposal should
not be taken as being incompatible with this. I am merely suggesting that we shift from a
conception in which anything less than wholehearted identification falls short of attribut-
ability to one in which merely partial identification makes the cut, so to speak.
-68-
order desires and then explicitly deliberates about what to do at the se-
cond-order. Consider the following case:

Three Desire Theresa: Theresa is currently at work and has three first-
order desires, each with the potential to pull her in a different direc-
tion: she wants to complete the assignment she has been given by her
boss, she wants to spend her time writing some thank you emails to
some relatives who are waiting to hear from her, and she wants to slap
her boss. She then considers each of these desires and considers which
of them she most wants to act on. It’s not that she is uncertain about
what she ought to do: she knows she should complete her assignment;
but she nevertheless is unsure prior to deliberating about which of her
desires she most wants to act on. Upon reflection she gives some
weight to the possibility of acting on her desire to get her work as-
signment done and some weight to acting on her desire to send her
thank you emails. There’s a part of her that wants to act on her inclina-
tion to do what she’s supposed to do, and there’s a part of her that
would really prefer to act on her inclination to make sure her relatives
hear from her today. Acting on her urge to slap her boss, she realizes,
isn’t even a contender for what she should do right now; even though
she is struck by the raw urge to do it, she would never really want to
do this. She’s not even angry with her; it’s just an occasional urge that
pops into her mind. She decides, in the end, to work on her assign-
ment.

Three possible outcomes in this scenario lead to three different levels of
potential alienation. First, there is the outcome in which what she actually
does is her work assignment. She feels no sense of alienation in this case,
given that she acts on the desire of hers that she most wants to act on; she
stands behind her action.
-69-
Next, there is the outcome in which, when she opens up her work as-
signment, she shifts over to her email and instead starts writing those
thank you notes. It’s possible this may feel somewhat alienating to her, as
she is not, in the end, motivated by the desire that she decided she most
wanted to act on. However, acting on a desire to write those thank you
emails is in line with at least something that she wanted for herself. She
may have decided in the end that what she wanted for herself more was to
act on her desire to work on her assignment, but nevertheless she did
want to some degree to act on her desire to write the thank-you emails.
Furthermore, let’s make the reasonable stipulation that it was no mere co-
incidence that she gave some weight to her desire to write thank you notes
and the fact that she actually wrote thank you notes. The elements of her
psychology that led her to give some post-reflective weight to her desire
to act on her desire to write thank you notes were also involved in causing
her to actually write the thank you notes. This act, I want to argue, is
therefore not wholly alien to her since she at least partially endorses it.
Finally, there is the outcome in which she, despite never seriously con-
sidering it as a contender for the motivation she should act on, slaps her
boss. If this happens, something seems to have gone seriously awry. Even
though she had the first-order motivation to slap her boss, she experiences
complete alienation since she acts on a motivation wholly outside of what
was even in contention for what she wanted for herself to do upon reflec-
tion. In this case she is moved to action by a bit of her psychology that
stands entirely outside of the complex psychological dispositions that af-
fect choices about which motivations she would want herself to act on. Be-
ing moved to action when one would fail to even partially endorse its mo-
tive is, I contend, what makes actions that are compulsive in this way
stand outside of one’s agency, and thus fail to be able to speak for the
agent. So, in order for an action to be able to speak for an agent it must be
suitably related to the motivations the agent would at least give some se-
cond-order weight to in deciding on which motivation to act.
-70-
4.1 Partial Identification and Desire Individuation
While I have still not yet provided a complete analysis of minimal ap-
proval, since its foundation is the notion of partially endorsing a first or-
der desire, and this partial endorsement is a special kind of desire, it is
worth pausing to examine the notion of ‘desire’ at play here. I have ar-
gued that Theresa is attributionally-responsible for sending thank you
emails because she gives some weight to acting on her desire to do so. In
other words, she has a higher-order desire of some strength to act on her
desire to send the emails.
Given certain conceptions of desire, though, one might think that
whenever it is the case that Theresa wants to send the emails it is already
the case that she has a desire to act on her desire to send the emails. For
example, if ‘desire’ is given a simplistic dispositional analysis, Theresa’s
desire to send the emails just amounts to a disposition to send the emails.
Since the only way to send the emails is by acting on her desire to send the
emails, a desire to act on a desire to send the emails would seem to
amount to nothing more than a disposition to send the emails. Thus, the
fact that she gives some weight to her first-order desire can be nothing
over and above her first-order desire itself.
However, there is reason to reject the simple dispositional analysis as
inadequate. As I argue in “Depression’s Threat to Self-Governance,” in
order to make sense of certain aspects of the phenomenology of agency,
desires need to be individuated at least in part by their propositional con-
tents.
77
The case of melancholic depression gives us reason to think that
desires to ϕ and desires to act on desires to ϕ have distinct existences. This
insight comes from thinking about cases in which an agent with melan-
cholic depression wants to act on a desire to get out of bed but lacks the
corresponding desire to get out of bed. The fact that getting out of bed is a
satisfaction condition for the depressed agent’s second-order endorsement

77
Gorman (Unpublished Manuscript).
-71-
is no guarantee that she already possesses a first-order desire to get out of
bed. Likewise, the fact that acting on a desire to send emails is the satisfac-
tion condition of Theresa’s first-order desire to send emails is no guaran-
tee that she already possesses a second-order desire of some strength to
act on her desire to send emails.
78

That said, aside from this stipulation, I will not aim to mediate the dis-
pute over the nature of desire here. In fact, my usage of ‘desire’ bears
more similarity to the tradition of using the term ‘desire’ to refer to any
motivational state whatsoever. When I use the term ‘desire’ in an example
to refer to the each of the states that together make up a particular agent’s
partial endorsement of a course of action, the reader may substitute in her
favorite motivational state for the term ‘desire,’ each time it is used, just so
long as it is the kind of state that is individuated in part by its proposi-
tional content.
79

5. Hypothetical Versus Explicit Endorsement
Besides requiring only partial rather than complete identification, I also
think the kind of endorsement required for the type of minimal approval
that matters to attributional-responsibility differs in another respect from

78
Other views of desire may also be difficult to hold in conjunction with the view that
partial endorsement matters for attributional-responsibility, even if they are not outright
ruled out like the simplistic dispositional analysis. For example, consider the view that a
desire to ϕ consists in nothing more than a judgment that one has some pro tanto reason
to ϕ, coupled with an account of reasons on which pro tanto reasons are very easy to come
by. If I take myself to have a pro tanto reason to act on a desire to eat my car, say because
it has nutrients in it, such a view would describe me as having a weak desire to act on my
desire to eat my car. If I then acted on an urge to eat my car, it is hard to see how the fact
that I very weakly desired to act on my desire would make me attributionally-
responsible for attempting to eat it. So views on which desires are this cheap to come by
will not be particularly good candidates to pair with a partial-endorsement based view.
79
Including but not limited to: intentions and evaluative judgments with motivational
components.
-72-
the Frankfurt’s traditional endorsement view. The traditional endorse-
ment view posits that it is the actual act of endorsement that makes the
agent’s resultant action belong to her; the act of taking a stand results in
ownership of one’s motivation and its results.
80
However, not every case
of attributable action involves the amount of intra-psychological reflection
of Theresa or Max. In fact, we rarely go through the explicit process of tak-
ing a stand on which of our desires to act on in the course of deciding how
to act. Yet, we are still able to say with some degree of confidence whether
or not our actions align with the motivations we would have wanted for
ourselves to act on if we did consider which motivations to act on. And
this is what seems to make the difference between whether or not we
identify at all with our actions. It is the fact that what the agent to some
degree would endorse and what she in fact does align with one another, I
propose, rather than any actual act of endorsement, that makes it the case
that an action is suitably related to an agent’s deep self such that it can
speak for her. Since what’s important here are the psychological disposi-
tions and not the endorsement itself, the actual act of endorsement can be
merely hypothetical. So the view, so far, is this. An agent’s action can
speak for her iff her act is suitably related to the fact that she would at
least partially endorse (give some weight to) the desire that she acts on, if
she were to consider which of her motivations she wanted to act on.
We frequently do not explicitly deliberate about or give additional
weight to our competing first-order motivations en route to acting, and so
views that require explicit endorsement predict that we will regularly fail
to identify with the springs of our actions. However, our actions are al-
most always caused by the kinds of motivations to which we would give
at least some weight if we were to deliberate, and in the rare cases in which
a person’s action fails to be produced by such a motivation, we generally

80
This is crucial to David Velleman’s understanding of Frankfurt’s view. See Velleman
(1992).
-73-
attribute the result to some sort of dysfunction. In moving from explicit
endorsement to hypothetical endorsement, the view entails that relatively
few actions fail to meet the requirements of attributability. I think this
constitutes a point in favor of the hypothetical view for a couple of rea-
sons.
First, it can help explain why some objectors to traditional Deep Self
views are tempted to say that people who claim to be alienated from their
actions must be making excuses. Recall that objectors to traditional Deep
Self views sometimes object that they can too easily imagine situations in
which they themselves might not go through the proposed attributability-
granting processes and even, perhaps in hindsight, feel some sense of al-
ienation from their actions, while they nevertheless maintain strong intui-
tions that they are attributionally-responsible for those actions. But a hy-
pothetical endorsement view explains why complete alienation of the kind
that undermines attributability is rare enough that it is entirely possible
that these objectors have almost never met the conditions for it.
81

Another benefit is that it can preserve the connection between felt al-
ienation and actual alienation. While alienation in the sense that is rele-
vant to attributability is not just a feeling, it is frequently accompanied by
a feeling. One (perhaps non-essential) common feature of felt alienation is
that it involves a sense that the process by which one comes to act is non-
typical. The hypothetical partial endorsement view, since it holds that
there are so few actions that fail to be attributable to agents, is compatible
with the fact that our interpersonal relationships function in ways that as-
sume that our actions generally issue from us in a way that speaks for us.
If having many of our actions fail to speak for us was very common, the
feelings that accompany any particular action failing to speak for us might

81
Or, perhaps more accurately, most objectors have never experienced externality in the
productions of actions that matter much. The scratching of itches at inopportune times
might be a fairly common action that is often external.
-74-
feel less prototypically “alien” since it would be a more regular occur-
rence. And so in further restricting the number of actions the account
deems alien, the hypothetical view better preserves the connection be-
tween felt alienation and alienation in the sense of failing to identify with
one’s action.
Despite these considerable benefits, some may protest that hypothet-
ical endorsement is a poor substitute for actual endorsement. Why should
we care about what motivations agents would endorse rather than what
they actually do?
In an analogous discussion about hypothetical consent in law and ap-
plied ethics, David Enoch makes the point that the question of whether or
not hypothetical consent is a poor substitute for actual consent must be
answered by paying careful attention to whether or not hypothetical con-
sent ever has the same kind of normative significance as actual consent. In
order to answer that question, he says, we would need to know why actu-
al consent matters and whether or not hypothetical consent can supply
those same resources.
82
In the case of consent, Enoch argues that actual
consent is sometimes important for reasons of sovereignty and sometimes
for reasons of non-alienation. Hypothetical consent can answer to the lat-
ter but not the former concern.
Something very similar seems to me to be true of endorsement. There
are certain ways of thinking about actual endorsement that cast it as grant-
ing motivations with special agential power. For example, David Vel-
leman conceives of Frankfurt’s second-order volitions as providing a
compatibilist response to agent-causal libertarianism; requiring an en-
dorsement process on his interpretation is an attempt to “put the agent
back into the picture” of the causal story of action.
83
If this is the aim, then
it’s clear hypothetical endorsement won’t do. But, in the next couple of

82
Enoch (2017).
83
Velleman (1992).
-75-
sections I will offer some cases in which non-attributable agents seem to
make choices for themselves about what to do via processes that are very
similar to the processes of attributable agents. This casts doubt on the idea
that what goes awry in non-attributable actions is that the agent is missing
from the picture in some way.
I have argued that what is really at issue is the fact that non-
attributable agents are alienated from the motivational states that move
them into action. And here, it is the coordinated symmetry between who
the agent is and what she does and the fact that her action comes about
due to a mechanism that plays a part in ensuring that coordinated sym-
metry that ensures non-alienation. Think about it this way: we’re not
frightened by the notion of failing to actively pick which motivation to act
on and acting anyway—we do so all the time when we let ourselves run
on autopilot. What’s frightening is the prospect of being autopiloted in di-
rections that have nothing to do with our own perspective of our inter-
ests—the motivations that we would give weight to were we to reflect.
84

6. Approving as Endorsement with a Further Aim than Elimination
A hypothetical-partial-endorsement-based view provides an appealing
explanation of why certain paradigm cases of non-attributable action are
non-attributable, while, unlike many traditional Deep Self views, does not
overextend its reach to classify actions performed out of weakness of will
or ordinary actions undertaken sans deliberation as non-attributable. Ac-
tions that are caused by rogue motivations that stand wholly outside of

84
What, though, would we make of being perpetually on autopilot in such a way? This is
a bit of a frightening prospect, but it might be frightening for reasons that have little to do
with motivational alienation. For example, we might just value having phenomenal con-
sciousness of our agential processes for its own sake, or because it helps us clarify our
self-perception or gain insight that leads to a sense of narrative coherence. Another con-
founding factor here is that we might need to not be on autopilot to form dispositions to
partially endorse acting on certain desires.
-76-
the complex of psychological states that inform what agents want for
themselves count as non-attributable. For example, straightforwardly
compulsive actions, which the agent wholly disavows, will count as non-
attributable. Frankfurt’s unwilling addict, understood as someone who
would give no weight to his desire to take drugs at the time of action, is
another perfect example of a person whose action does not speak for him,
according to this view.
Such a view also gives the verdict that willing addicts, even those who
are merely partially willing at the time of action, are responsible for their
taking drugs. That is, an agent who comes to takes a drug due to a chain
of mental states that ensure that she, at the time of action, would give
some post-reflective weight to her desire to take drugs, acts in such a way
that is self-disclosing for the purposes of attributional-responsibility.
85

When we fill in the story in a certain kind of way, this seems to be the
right result. For example, imagine a woman whose conflicting first-order
motivations are the motivation to take heroin and the motivation to tend
to her child, and then imagine her being such that if she were to deliberate
about what to do, she would give some weight to each motivation. “On
the one hand” she might think to herself, “I want to act on my desire to
tend to my child, but on the other hand I do really love heroin a lot and
maybe the fun of taking heroin is more important to me than my child’s
wellbeing…well, I guess my child’s wellbeing edges out my desire for
heroin, so I guess that’s what I’ll do?” In this case, if the woman ends up
taking heroin due to her desire to have fun, despite the outcome of her ac-
tual or hypothetical deliberation, it seems that she is attributionally-
responsible for her action.

85
Other factors may differentiate the wholly and partially willing in terms of the nature
of the response that will be appropriate. For example, if the partially willing addict’s tak-
ing drugs in this case meets the remaining criteria for blameworthiness, this kind of
agent may be less blameworthy than a wholly willing addict, but she is nevertheless an
appropriate target of blame.
-77-
However, consider another kind of case that falls somewhere in be-
tween the paradigmatic cases of the willing and unwilling addict: the be-
grudgingly willing addict. Michael Bratman describes the case as involv-
ing an addict who, “since [he] is confident that his desire for drugs will
soon so overpower him as to prevent him from acting intentionally, and
since the struggle to remain drug-free is extremely painful… decides to
cease resisting his desire, and to take the steps necessary for satisfying it.”
“To be sure,” Bratman writes, “he would rather not perform an act of
drug-taking. Nonetheless, given his options, he would rather perform this
particular drug-taking act.”
86
Such an addict does minimally endorse act-
ing on his desire to take drugs, and so would count as attributionally-
responsible for his action on the view currently under consideration, yet
this doesn’t seem like the right result.
To elaborate, the view apparently entails that if the agent were to
wholeheartedly resist by not endorsing a desire to take the drug despite
inevitably having his will being taken over by the urge, his resultant ac-
tion would not be attributable. However, if he were to wisely recognize
that failure would be the inevitable outcome of his resistance and get it
over with more quickly by acting on the desire to take the drug to get rid
of the urge faster, his action would be attributable. But it seems that the
wrong thing makes the difference in these two cases; the action’s connec-
tion to the agent shouldn’t be determined by whether or not the agent de-
cides to give in to an overwhelming impulse, but rather, by whether or not
his desire is the kind of thing he’d wanted to be motivated by in the first
place. The begrudging addict’s action is still caused purely by non-
agential neural activity, no matter how long the agent refrains from acting
on it. He may feel just as alienated from his action as the unwilling addict,
and would be right to feel this way since he feels compelled to endorse

86
Bratman (1996).
-78-
acting on a desire he would not want to act on were it not for the feeling of
horrifying inevitability.
87

People whose psychological make-up has this structure are not limited
to hypothetical addicts. Recent research into Tourette syndrome reveals
that the initiation of ticcing behavior often comes about in a similar way.
88

Once thought to be akin to involuntary twitches like muscle spasms, Tou-
rettic tics are now acknowledged to be intentional movements consciously
undertaken in response to premonitory urges, which are experienced by
agents as alien. More than 90% of people with Tourette syndrome report
that their tics are “voluntary” in the sense that they believe they take an
active (though subservient) role in the action’s coming about.
89
The re-
ported phenomenology echoes the neurophysiological findings. Ordinary
action involves signals being sent from an agent’s frontal lobe, the site of
considered judgments about what to do, to the motor cortex, the initiator
of action. But in the brains of people with Tourette syndrome, disordered
neural connections between the basal ganglia and motor cortex cause er-
rant signals to encourage the initiation of simple body movements (in-
cluding the utterance of words) that build up as a sort of mental pressure
over time, making the sufferer more and more uncomfortable until she
chooses to act on the urge. People with Tourette syndrome can, with some

87
Some theorists believe that all addiction functions similarly to the way it’s described in
these cases, that, in a sense, there are no truly unwilling addicts since addiction affects
one’s judgment about what one should do in this kind of way. See, for example, Buss
(2012). Most Deep Self theorists, by way of contrast, take it as given that compulsion of-
ten involves an agent’s deep self mental states being overpowered by a rogue urge. But
this dispute needn’t be settled here. It seems that our view should tell us that irrespective
of the prevalence of cases of its type, if there were a case of addiction in which an unap-
proved desire simply overpowered one’s approved desires, the agent’s action would
count as non-attributable.
88
Schroeder (2005) argues for the importance of the integration of this research into our
theorizing about responsibility.
89
Leckman et. al (1999).
-79-
difficulty, use their judgment to postpone the discharging of these urges,
but ultimately usually choose to give in to the urges as a way of alleviat-
ing the pressure.
Compare the person who utters slurs for the pure joy of harming mi-
norities to the person with Tourette syndrome who utters slurs purely to
relieve the unbearable pressure of a strong premonitory urge to tic. The
former, unlike the latter, seems to be agentially involved in the right sort
of way with the motivation of her action such that her action represents
something about her, while the latter’s action does not seem to be repre-
sentative at all. According to the kinds of endorsement views under con-
sideration so far, however, most agents’ tics would be attributable, since it
seems that people with Tourette syndrome do give weight to the desire to
act on their urges, which comes from their desires to rid themselves of the
uncomfortable urges.
Something seems to be going awry in the psychology of the begrudg-
ing addict and the person with Tourette syndrome such that the relation-
ships they have to their effective motivating desires do not grant the de-
sires the usual authority to speak for them. Here is what I think goes
awry. There are oftentimes two functions of acting on a desire: acting on a
desire to ϕ brings it about so that you are ϕ-ing, but it also in many cases
gets rid of your desire to ϕ. For example, if you want to go to the library,
and then you successfully act on this desire, you will be at the library and
thus no longer have an occurrent desire to go to the library. In most cases
in which we endorse a desire to ϕ we do so because we want the satisfac-
tion of the desire to ϕ, not merely because we want to no longer want to ϕ.
In cases in which we endorse acting on a desire because we approve of do-
ing so, we do so because we have some aim that acting on our desire satis-
fies other than its elimination through our action. When you ‘endorse’ giv-
ing in to such an urge due solely to its power, the force the urge exerts on
you is no less purely mechanistic than when it overpowers your endorse-
ment. To distinguish, I will call the kinds of endorsements that are not
-80-
merely due to wanting to rid oneself of one’s motivation with no further
aim “approvals.” So, it is approval rather than just any kind of endorse-
ment that is relevant to attributability.
With all the pieces in place now, we can formulate an account of what
it takes to “minimally approve”. To “minimally approve” of acting in a
certain way is to approve of it in just (at least) a hypothetical and partial
sense.

Minimal Approval: An agent minimally approves of ϕing-at-t iff at t
she is such that if she were to reflect on her desire to ϕ at t, she would
want to some degree to act on it with some further aim in doing so
other than merely eliminating the desire.

One more kind of case that sits between the willing and unwilling addict
cases is also worth discussing: cases in which a person appears to act in
order to eliminate their desire and does so to further some aim she has in
eliminating it. Consider the following case:

Thoughtful Tourette Sufferer: Tom has a Tourettic urge to say the
word “dyke,” which he knows is a slur that targets lesbians. Tom has
absolutely no animus against lesbians and, in fact, he aims to minimize
the harms his frequent utterances cause for lesbians who hear them
and do not know about his condition. He is currently in the room with
one lesbian, and knows that another will additionally enter the room
in 5 minutes. He feels an overwhelming urge to utter the slur, one that,
with some great effort, he would be able to postpone, but only for
about 5-6 minutes. He decides that he prefers to act on the urge sooner
rather than later, so as to only harm one person rather than two, and
does so.

-81-
Is Tom attributionally-responsible for uttering the slur? Our intuitions
pull in both directions. Tom seems praiseworthy for avoiding harming an
additional person by uttering the slur now rather than later, which seems
to indicate that he is attributionally-responsible. But Tom does not seem
blameworthy for harming the one person, since his utterance was due on-
ly to a need to eliminate a rogue urge. The Minimal Approval view can
explain this by showing how he meets the conditions for attributional-
responsibility under one description of his ϕ-ing, and fails to meet them
under another.

Tom seems praiseworthy for uttering the slur now rather than later be-
cause the actual sequence of mental states involved in the production
of Tom’s action guarantees that the agent at t is such that if he were to
reflect on his desire to “utter the slur now rather than later” at t, he
WOULD want to some degree to act on it with some further aim in do-
ing so than merely eliminating it.

Tom does not seem blameworthy for uttering the slur full-stop because
the actual sequence of mental states involved in the production of
Tom’s action guarantees that the agent at t is such that if he were to re-
flect on his desire to “utter the slur rather than not utter the slur” at t,
he would NOT want to some degree to act on it with some further aim
in doing so than merely eliminating it.

While it is controversial whether we ought to individuate actions and de-
sires this finely, and thus whether or not our concepts of praise and blame
ought to be sensitive to these distinctions, I think the intuitive reactions to
these kinds of cases tell in favor of doing so. The Minimal Approval View
can help to make sense of our intuitions that initially seemed contradicto-
ry.
-82-
7. Minimal Approval and the Mechanism of Action
I have argued that minimally approving, in the sense I have articulated,
should play a central role in attributional-responsibility, but I have not yet
specified just what the relationship is between minimal approval and the
causal chain that leads the agent to act. The fact that the agent would min-
imally approve of ϕ-ing and the fact that she in fact ϕs, at minimum,
should not be wholly coincidental. Consider the following case:

Coincidental Compulsion: Carrie has a compulsion to do a jumping
jack at 3pm every day. She has absolutely no ability to stop herself
from acting on her first-order urge to do it, but she almost always hates
that she does it. It often interferes with whatever else she actually
wants to be doing at 3pm, but her first-order urge takes over and she
does the jumping jack anyway. Today, as 3pm rolls around Carrie is
feeling a bit low energy, and thinks to herself that doing a jumping
jack might actually be nice for helping her wake up a bit, and so she
desires to some degree to act on her desire to do the jumping jack for a
reason other than just to get rid of the urge. However, she will be in a
work meeting at that time, and all things considered she definitely
would not prefer to act on the urge, since it will be embarrassing and
difficult to explain to her colleagues. At 3pm the urge overtakes her
considered opinion about what she wants to do entirely via the exact
same mechanism it does every day (one that is not sensitive to any in-
formation about what she approves of doing), and she does the jump-
ing jack.

Carrie, it seems, is not attributionally-responsible for her jumping jack.
Even though she did minimally approve of doing the jumping jack, this
had absolutely nothing to do with what caused her to act. Minimal ap-
-83-
proval seems to matter only when it is non-coincidental that the agent acts
in accordance with what she minimal approves of doing.
But, given the fact that minimal approval is hypothetical, we can’t ex-
plain the connection by telling a simple causal story on which the approv-
al causes the action. It might be the case that an attributionally-responsible
agent explicitly reflects, approves of acting on one of her desires and this
approval itself brings her to act. But this is not the only possibility on
which her approval and the causal chain leading her to act might be non-
coincidentally related. The process leading to her action, or some part of it,
might cause her to be such that she would minimally approve. Or she
might have a mental state that causes both her minimal approval and her
action. The common principle shared by these relations is that for some
particular chain of mental states leading to action, C, which leads the
agent to ϕ, if the agent’s ϕ-ing is caused by C, then the agent minimally
approves of ϕ-ing.
But now consider the following case:

Conditional Fallacy Frankie: Frankie is addicted to alcohol and would
give anything to stop drinking, but despite his other plans for himself
he is often overtaken by an extremely powerful urge to drink. One
day, while feeling hopeless about his powerlessness against his urge
he decides to think hard about whether or not he should just form and
act on a desire in order to act on it before it overpowers him. Ultimate-
ly, he does. Given what I have been arguing, it seems he should not be
attributionally-responsible. His thinking leads him to act and so is part
of the mental states making up the causal chain that leads to his action.
But, now imagine that this same thinking also leads him to put his
head down on some objects on his desk, one of which he hits hard
enough that it stimulates his brain to have certain mental states, mak-
-84-
ing it the case that at the moment of action he minimally approves of
acting on his desire to drink alcohol.
90

In this case, Frankie’s causal chain of mental states leading to his action
does make it the case that he minimally approves of acting on his urge to
drink. But, intuitively, this is not the kind of connection that would make
him attributionally-responsible for his resultant action.
91
Instead I think
the relevant connection is not that the causal chain of mental states lead-
ing to the agent’s action actually makes it the case that the agent minimal-
ly approves, but rather that the causal chain of mental states disposes the
agent to minimally approve in the vast majority of cases (all non-fluky
cases).
92

93
The thought here is that it is not any particular counterfactual
that matters for attributional-responsibility, but rather the fact that the
agent has the actual mental states that dispose her to approve, whatever
they may be. We’re not really interested in what happens in any particular
counterfactual world, but rather, that generally speaking, in (most of) the

90
I call this case “Conditional Fallacy Frankie” because the problem here may be an ap-
plication of some version of the Conditional Fallacy. For discussion of the Conditional
Fallacy in general, see Bonevac, Dever, and Sosa (2006).
91
Less science-fiction-y cases can also be generated, although it is less clear that they pose
a genuine threat to the view as stated. For example, a person who never reflects on her
motivations might undergo an organic personality change if she were to reflect on her
first-order desires. It is not wholly clear that these post-personality-change hypothetical
approvals are not what matters for attributional-responsibility. But if this seems prob-
lematic enough to provide reason to think that the view should say that certain aspects of
agent’s personalities ought to be held fixed, this gives additional reason to move to a dis-
positional view.
92
This means that there will also be inverse Conditional Fallacy Frankie cases on which
the agent does not actually minimally approve due to some fluke caused by the causal
chain leading to action.
93
The reader is invited to insert her own favorite theory of the metaphysics of disposi-
tions here.
-85-
worlds in which the agent considers this particular choice situation, she
approves to some degree of her desire. In ruling out certain outlier worlds,
moving to a dispositional account ensures that there is actually some fea-
ture of the agent’s psychology that plays the relevant role here, rather than
a mere fluke of the environment.
One final tweak must be made before an adequate account can be for-
mulated. It seems important to attributional-responsibility that the mental
states involved in the causal chain of action, as well as in the agent’s min-
imal approval are really her own. By that I mean something quite modest,
just that they are not inserted, say by an evil scientist or the like.
94
There is
an extensive literature here, and I am aware that making this concession
opens the door for potential incompatibilist arguments to get a foothold.
95

However, I reluctantly adopt a modest ownership condition since it seems
to me a trade-off worth making.
96
Although the account loses theoretical
simplicity, it gains the ability to make good on some of our strongest intui-
tions about attributability.
Putting all of these pieces together, we can now formulate the Minimal
Approval view:

The Minimal Approval View of Attributional-Responsibility: An agent
is attributionally-responsible for ϕing-at-t iff the actual sequence of
mental states involved in the production of her action is non-
implanted and together with her other mental states makes it the case

94
See Matheson (2018) for a recent discussion of Manipulation Cases as they pose coun-
terexamples to Deep Self views.
95
The most famous of these argument, Derk Pereboom’s Four Case Argument aims to
show that there is no relevant difference on which to base an ownership condition be-
tween a case in which an agent’s deep self mental states are implanted and when they are
acquired in the normal fashion, given the truth of determinism (Pereboom 2001).
96
For a recent defense of adopting a modest ownership condition to protect against ma-
nipulation cases, see McKenna (2016).
-86-
that at t, if she were to reflect on her desire to ϕ at t, she would be suf-
ficiently likely to want to act on her desire to ϕ at t, with some further
aim in doing so other than merely eliminating this desire.
97

8. Advantages of the View
8.1 Solving Frankfurt’s Infinite Hierarchy Problems
Though the idea that higher-order desires have something to do with our
agential identities is intuitive, the fact that higher-order accounts like
Frankfurt’s endorsement account have problems dealing with infinitely-
ascending orders is often assumed to make higher-order accounts non-
starters. Watson first frames his valuing view as a view that can avoid the
problems of infinite hierarchies caused by hierarchical account. Frankfurt
himself has been aware of these sorts of issues since his first article dis-
cussing the view, and has modified his view numerous times in order to
try to better handle the problems. In this section I will show how two of
the most commonly discussed problems of infinite hierarchies are caused
in part by contingent features of hierarchical views that Frankfurt’s ac-
count has but that the Minimal Approval view lacks.
In order to see how the first problem of infinite hierarchies arises, first
consider a case in which there is a conflict at the second-order about
which first-order desire to endorse. Consider first an agent deciding be-
tween going to the gym and doing some grading who then acts on the de-
sire to go to the gym because she forms a second-order volition that is
constituted by a desire to have her desire to go to the gym be the one to
move her to action. But it seems very possible to have an agent who is not
only first-order conflicted but also second-order conflicted; this agent not

97
“ϕ-ing” should be understood here as standing for an action, and the account should be
taken to cover only attributional-responsibility for actions. Attributional-responsibility for
omissions and for consequences is, on my view, derivative on attributional-responsibility
for actions. I explore some of these issues in Chapter 5.
-87-
only wouldn’t be sure whether to go to the gym or get some grading done,
but additionally, she wouldn’t be sure whether to form the desire to act on
her desire to go to the gym or to form the desire to act on her desire to get
some grading done. Since the way Frankfurt seems to imagine that con-
flicts among desires of the same order are resolved is by moving to a yet
higher order of reflection, we would then have to look at the agent’s third-
order desires to be moved by second-order desires to be moved by first-
order desires. If these desires were also in conflict, we would have to look
at the agent’s fourth-order desires, etc. This process seems like it could too
easily be such that it never terminates in wholehearted endorsement and
so we might worry that this is just a bad candidate description for how the
process that leads to autonomous action works in agents like us. Further-
more, it seems as though an agent who never decisively reaches a level at
which she is wholly unconflicted can still act in ways that she should still
be held attributionally-responsible for.
But we are now in a position to see how this kind of problem of infinite
hierarchies is related to contingent features of a higher-order desire ac-
count. One feature of Frankfurt’s view is that it seems to imply that
wholeheartedness is necessary for an agent’s process to issue in action.
This is what makes conflict at the n level a problem for understanding how
the agent could come to act, and forces us to examine level n+1 to find the
wholehearted endorsement. If we instead admit, as the Minimal Approval
view does, that we may act despite having not settled conflicts among de-
sires at the second-order, we are left without reason to worry about third-
order desires as a way of mediating conflict in order to act. This is not
possible on Frankfurt’s view because according to Frankfurt the reason se-
cond-order desires matter is in large part because they help settle conflicts
between first-order desires leading the agent to make a decision that is-
sues in action.
The second, and more troubling, problem of infinite hierarchies is the
one Watson speaks most directly to in “Free Agency” when he writes,
-88-
“Can’t one be a wanton, so to speak, with respect to one’s second-order
desires and volitions?”
98
Watson’s concern is that just as an agent can be
alienated from her first-order desire in the sense that she has not endorsed
it, she can be alienated from her second-order desires as well. Since first-
order desires on Frankfurt’s view do not have the authority to speak for
the agent unless they go through a process involving second-order de-
sires, it seems as though there is no reason to suspect that second-order
desires wouldn’t also have to go through a higher-order process of their
own to be granted such authority. Ever-higher levels would need to be
appealed to in order to generate the appropriate authority, making it im-
possible for any desire to have the authority to speak for the agent. What
we should conclude from this, Watson thinks, is that second-order voli-
tions do not really have the authority to speak for an agent since they do
not guarantee that the agent is not alienated from them.
99
Even if an agent
forms a full-fledged unconflicted second-order volition, we have no rea-
son to privilege the authority of that state over a third-order volition to
not want to act in accordance with the desire that aligns with her second-
order volition. Since second-order status does not automatically grant se-
cond-order volitions any special authority, it seems they have no more
claim to represent the agent than any other candidate mental state.
But on Minimal Approval non-alienation is already established by bare
reflective endorsement of any strength. It is a contingent feature of a hier-
archical view that we should be concerned with determining the most de-
cisive or most authoritative endorsement. On the Minimal Approval
view, even if the agent has a wholehearted third-order desire that repudi-
ates the second-order desire, as long as the second-order desire exists,
then this is not enough to establish full alienation since the agent’s action

98
Watson, (2004): 28.
99
Velleman (1992) argues that this problem is more general and is a problem for Wat-
son’s own view as well.
-89-
would still come about in a way that is related to something that the agent
would want for herself upon reflection (unlike when she is motivated by a
rogue first-order desire). While higher-order desires may reveal further
truths about us, attributional-responsibility is not about revealing the
deepest truths about ourselves, but rather, it is about meeting a very min-
imal set of ownership conditions of our actions ensuring that their source
is not completely alien to our interests. To be attributionally-responsible
for our actions we must only be invested in them in some reflective way
that goes beyond a bare desire. Once this reflectivity is established, there
is no need to worry about ever-ascending orders of reflection.

8.2 Capacity Without Process
We tend to think that human agents’ actions can put them on the hook for
what they have done in a way that non-agents’, like non-human animals’
behavior does not. We think that the kinds of moral responsibility re-
sponses that are appropriate for persons go above and beyond the ways it
is reasonable to respond to a dog. Planning, endorsing, and valuing ver-
sions of Deep Self views generally aim to be able to make this distinction
by positing some special kind of capacity that agents actively exercise
when they undertake actions for which they are attributionally-
responsible. For example, on the endorsing view, an agent exercises her
capacity to choose which action to initiate by picking amongst her first-
order desires.
But this feature of these views is also problematic, because the idea
that we as agents go through some sort of special mental process each
time we act in an attributable way is implausible.
100
The problem is that
Frankfurtian descriptions of agents selecting which first-order desire to act
upon, or Bratmanian descriptions of agents simultaneously acting and
making choices about how to settle conflicts amongst desires in the future,

100
See, for example, Arpaly (2002, 2006), Smith (2005), and Buss (2012).
-90-
just do not seem to be what we ordinarily do as agents in everyday life.
Our conduct for which we are often rightly held morally responsible is
sometimes spontaneous, initiated by subconscious motivation, out of
character, or brought about in a fit of emotion. Furthermore, as the results
of numerous social psychology studies appear to show us, we sometimes
lack reflective access to some of the motivational influences on our actions,
perhaps in ways that would implausibly preclude us from being attribu-
tionally-responsible for a large range of actions given the conditions of
these more agentially demanding Deep Self views.
On the Minimal Approval view, though, action brought about by sub-
conscious processes can still meet the requirements for attributability. The
process that causes action needs to make it so that the action is in line with
what the agent to some degree would want for herself, which requires the
agent to be the sort of creature who has the capacity to form higher-order
desires. But that capacity need not be exercised in the form of actual re-
flection, thus avoiding the charge that traditional Deep Self views face that
in ruling out animal action, they rule out too much.
While the Minimal Approval view has this advantage, the minimalist
nature of the view might seem to run a different risk, namely, being una-
ble to plausibly explain why agents can be attributionally-responsible in a
way that most non-human animals cannot. The Minimal Approval view is
relatively silent about the nature of action-production and so even a crea-
ture with relatively instinctual or mechanical sorts of action production
could, in theory, be eligible for attributional-responsibility. However,
there is good reason to think that most non-human animals in fact do not
have the capacity to form higher-order desires since they do not have the
capacity for higher-order thought, and so would not be candidates for at-
tributional-responsibility according to the Minimal Approval view.
While it is not wholly uncontroversial, the idea that humans are
unique among animals in being able to have higher-order thoughts has a
rich history of support in the literature on consciousness. Some argue that
-91-
higher-order thoughts require possession of an I-concept in such a way
that the thinker can understand themself as a self in a way that involves
complicated linguistic capacities that lower animals do not possess.
101

Daniel Dennett argues that the ability to say which mental state one is in is
more fundamental than having a higher-order thought, and provides the
basis of having a higher-order thought. Since animals cannot say which
mental state they are in, according to Dennett, they do not have the capac-
ity for higher-order thought.
102
Others question whether or not lower ani-
mals are capable of possessing mental-state concepts at all
103
; preliminary
studies seem to indicate that they are not.
104

So a significant advantage of the Minimal Approval view is that it can
preserve the distinction between the way agents and non-agents reveal
themselves through action due to their having special capacities, while
simultaneously allowing that agents may not actually engage in such me-
thodical deliberative processes when they act in ways that they can be
held attributionally-responsible for.
8.3 Criterion Operates Independently from the Type of Mental State
that Causes Action
A further advantage of the Minimal Approval view is that, unlike some
Deep Self accounts, it is compatible with attributable actions being caused
by mental states of any type, just so long as they meet the requirement of
being appropriately related to hypothetical approval states. This agnosti-
cism about what the actual process of action production looks like is an
advantage for at least two reasons.
First, unlike several Deep Self views, accepting the view does not re-
quire accepting controversial positions in moral psychology. For example,

101
See Quine (1995), Bermúdez (2003), and Bennett (1964, 1966, 1988).
102
See Dennett (1991).
103
See Davidson (1984, 1985), and Bermúdez (2003).
104
See, for example, Povinelli and Vonk (2004).
-92-
the valuing view as advocated by Gary Watson assumes at least some
form of motivational judgment internalism, and the caring views of
Chandra Sripada and David Shoemaker rely on accepting that there exist
complex states or dispositions that we can identify as caring states, which
are distinct from mere desires and play a central role in action production.
In contrast, the Minimal Approval view is compatible with each of these
pictures of action production, but its proponents can remain agnostic
about which sorts of states have the ability to motivate. It is even con-
sistent with extremely minimal theories of action production including
simple forms of Humean psychology. This might make the view attractive
to those who are averse to more traditional Deep Self views due to the
more complicated systems of action-production that they posit.
But there is a further advantage to the fact that the Minimal Approval
view does not posit that any particular kind of mental state must be in-
volved in the causal chain in order for an action to be attributable: certain
mental state types may sometimes produce attributable action and some-
times produce non-attributable action. In order to illustrate this point, I
want to focus on a class of actions to which Deep Self theorists have per-
haps paid insufficient attention: actions in which agents act directly “out
of” emotions. Emotions are often thought to be partly constituted by mo-
tivational states or, at the very least, they are generally thought to have
some unmediated influence on motivation. This accords with the com-
mon-sense ideas that we can “strike someone out of anger” or “hide out of
embarrassment.” Such actions often characteristically do not align with
our plans, values, cares, or second-order volitions concerning what we
think the most preferable thing to do is in a given situation. Imagine, for
example, an anti-retributivist who nevertheless is swept up in a wave of
vengeful anger, or an ethically non-monogamous person who has disa-
vowed the appropriateness of jealousy being nevertheless moved to action
by it. While these actions fail to align with Deep Self states, they neverthe-
less seem to be the sorts of things for which one can be attributionally-
-93-
responsible.
On the other hand, if a theory were to hold that a person is attribution-
ally-responsible for any action done out of emotion, it would not be viable.
Actions caused by psychological and neurological disorders that we intui-
tively tend to think exculpate an agent from attributional-responsibility
can, it seems, cause action by impacting the agent’s emotional state such
that she “acts out of” a given emotion. If this is right, a theory of attribu-
tional-responsibility should have a way of distinguishing between cases of
acting out of an emotion that involve one’s agency in the right sort of way
and ones that circumvent agency.
One lesser-known disorder illustrates the importance of drawing such
a distinction. Misophonia is a neurobehavioral syndrome in which certain
ordinary human-produced repetitive sounds, such as the sounds of others
chewing, sniffling, or clearing their throats, trigger reactions of anger, dis-
gust, and fear in otherwise psychologically healthy individuals.
105
While
research on misophonia is in its infancy, it is hypothesized that the cause
of such reactions is extra-connectivity between a set of emotional pro-
cessing centers of the brain and the anterior insular cortex, the site of in-
teroception (the ability to sense what is happening to one’s own body) in
the brain.
106
Due to this over-connectivity, ordinary sounds cause these
sufferers to react emotionally as if these innocuous sounds are threats, set-
ting off fight-or-flight reactions. When misophonia sufferers are in “fight”
mode, their anger is not just an expression of being overwhelmed, but ra-
ther, tends to take the form of a directed expression of anger and disgust
towards the source of the offending sound. To be clear, this is not just an-
ger towards the person for making a sound that they know bothers the
sufferer, as anger can be just as strong towards those making sounds who
do not realize their sounds are upsetting to the sufferer. Crucially, at the

105
See Braut et. al (2018) for a cross-disciplinary review of the research on misophonia.
106
See Kumar et. al (2017) , Edelstein et. al. (2017)
-94-
very same moment in time that she acts out of anger, a person with miso-
phonia is able to acknowledge that it makes no sense to be angry and that,
for example, making sounds while chewing is entirely innocuous. Due to
these irrational, embarrassing, and inescapable responses, people with
misophonia often live increasingly reclusive lives as the disorder pro-
gresses in order to try to avoid both sounds and accidentally lashing out at
those who they know have done nothing wrong.
Consider the following pair of cases:

Manners Mary: Manners Mary was taught as a child to always chew
with her mouth closed and greatly appreciated the value of the lesson.
Following her parents, she grew up believing that a decline in manners
in society was the root of much evil and that it is deplorable that some
people chew with their mouths open. In her adulthood, she has come
to see this as a bit overblown, but she has retained the sense that it’s
bad form to chew with an open mouth as well as an accompanying
sense of disgust when she sees others behaving with such poor man-
ners. At an important dinner party she notices her fellow guests chew-
ing with open mouths, and thinks to herself that someone ought to tell
them to stop, and that perhaps if no one else does, she should be the
one to do so. However, she knows that these guests would only be of-
fended and would not change their ways if she were to mention their
behavior, and so decides that this would probably be a bad time to say
something. However, she fails to hold her tongue, gets increasingly
angry, and yells out “chew with your mouth closed!” despite knowing
that everyone will only be offended and not change their ways.

Misophonia Mary: Misophonia Mary does not care about manners in
the slightest and makes no effort to chew with her mouth closed.
However, she suffers from misophonia, which makes her inexplicably
angry when she hears people making chewing sounds. Even when in
-95-
the throes of an episode of misophonia she recognizes that there is
nothing wrong or bad in any way about eating with one’s mouth open,
yet due to errant signals in her brain that trigger a fight-or-flight reac-
tion, Mary feels compelled to flee or else lash out at those making the
sounds. With nowhere to flee to at a dinner party, out of anger Mary
yells out “chew with your mouth closed!” despite knowing that every-
one will only be offended and not change their ways.

Intuitively, it seems we should hold Manners Mary attributionally-
responsible but not Misophonia Mary, though they are both most directly
motivated by their anger. This is some indication that our view should be
consistent with the fact that acting out of anger is neither sufficient for at-
tributional-responsibility nor disqualifying for it.
But because both agents’ actions are motivated by anger and not suita-
bly related to their plans, endorsements, judgments, or possibly even
cares, traditional Deep Self views will have a difficult time explaining why
Manners Mary’s action is attributable and thus licenses a different re-
sponse than Misophonia Mary’s.
107

The Minimal Approval, by contrast, is well-suited to explain the con-
trast. If we were, at the time of action, to ask Manners Mary to consider
her motivation to yell out at the guests chewing with their mouths open
she would give some weight to that option. After all, she thinks it is some-
what important that such ill-mannered behavior not go wholly ignored.
But it does not seem appropriate to hold Misophonia Mary attributionally-
responsible for her yelling, and the Minimal Approval view shows how
her motivation stands outside of her agency. If we were to ask her wheth-

107
It might be argued that Manners Mary’s action is related to her judgments in a way
that Misophonia Mary’s action is not. However, it is clear that Manners Mary does not
take her action to be among the best options. She might take herself to have a pro tanto
reason to act as she does, but acting in accordance with one’s pro tanto reason is not suffi-
cient for attributional-responsibility. See Chapter 3, §4.2 for further discussion.
-96-
er or not she would like to be motivated to some degree by the desire to
tell the dinner guests to stop chewing she would say, even in a moment of
her anger, that she has no desire to be so moved. The only reason she
might give any weight to the desire to lash out would be to relieve the
psychological pressure of not saying anything, caused by her involuntary
fight-or-flight reaction, and thus she would fail to meet the conditions of
the Minimal Approval view.
This pair of cases helps illustrate the fact that the Minimal Approval
view can hold that agents who act out of emotions are often attributional-
ly-responsible for their actions while leaving room for the possibility that
emotional motivation may factor prominently in action caused by non-
agential neurological activity for which we should not hold agents attribu-
tionally-responsible. The set of cases in which agents act out of emotions
helps illustrate the broader point that in having a set of criteria for at-
tributability that does not require any particular mental state type to fea-
ture in the action-causing sequence it has better flexibility for handling
some of the nuances of attributability and neurological dysfunction. This
gives the Minimal Approval view yet another significant advantage over
traditional Deep Self views.
-97-
Chapter 3: How Should Deep Self Theorists Account for
Weakness of Will?
108

1. Introduction
In the first two chapters of this dissertation I articulated and defended a
new view of attributional-responsibility, the Minimal Approval view.
Along the way I illustrated some of its advantages over more traditional
Deep Self views. One such advantage mentioned but not yet explored in
depth is the view’s ability to account for the attributional-responsibility of
weak-willed agents. In this chapter I zoom in on this advantage of the
view and argue that the Minimal Approval view is uniquely well-suited
among Deep Self views to account for the difference between weak-willed
actions, which are attributable to agents, and compulsive actions, which
are not.
I begin by showing how deeply entrenched the weakness of will prob-
lem is for Deep Self theorists. Traditional Deep Self views lack the re-
sources to adequately distinguish compulsion from weakness of will,
which leads to their wrongly classifying certain attributable weak-willed
actions as non-attributable. Most current solutions involve implausible
bullet-biting, or else cede dialectical ground to control-based theorists. I
suggest that Deep Self theorists instead must adopt an understanding of
the self as having multiple strands that are competing yet individually
have the power to speak for the person qua agent. The Minimal Approval
view, I argue, is the only realization of this picture that appropriately sep-
arates non-attributable compulsive action from attributable weak-willed
action. Thus the Minimal Approval view has a distinctive advantage over

108
This chapter benefitted from discussion by the participants at a 2018 Eastern APA
symposium session.
-98-
competing Deep Self views. I conclude by addressing a family of fairness
concerns for the Minimal Approval view’s treatment of these cases.

2. The Weakness of Will Problem
Consider the following case:

Sam’s Exam: Suppose Sam knows that she should be studying for her
final exam, even though there’s a party going on that night that she re-
ally wants to go to. She judges it would be best for her to study for her
exam, endorses her desire to study for her exam, plans to study for her
exam etc., and yet somehow she ends up going to the party anyway.
Sam does not act on what she judges it best to do, or in line with what
she plans to do, or on the desire she higher-order endorses. But com-
mon sense tells us that Sam’s action is still self-expressive and so she
nevertheless ought to be attributionally-responsible for her action.

This poses a problem for traditional Deep Self theorists because, according
to their views, agents are not attributionally-responsible for their actions
unless their actions are aligned with their planning, endorsing, valuing, or
caring states. Traditional Deep Self views thus face an important objection:
they counter-intuitively hold that we are not responsible for weak-willed
acts, and so fail to provide a necessary condition for attributability. Per-
haps, then, the bounds of the Deep Self could just be drawn differently so
as to cast a wider net around the class of attributable actions?
The problem, however, is more deeply entrenched than this. Deep Self
views are custom-made to show how acting on desires that the agent her-
self does not see as most favorable undermine agency in such a way as to
make such actions non-attributable, as in the case of compulsion. But
weakness of will is usually described as the failure to act in accordance
with what one finds to be the most favorable course of action, and yet we
-99-
do intuitively think weak-willed actions are self-expressive.
109
One of the
motivations for Deep Self views is making sense of the claim that acting
attributably just involves doing what one really wants to do, where “really
wants” is qualified in such a way as to bracket off compulsive desires,
which Deep Self theorists see as having the power to coerce agents ‘from
the inside,’ so to speak. But they do this by bracketing off motivations that
are counter to what agents judge best, endorse, or plan for, which also in-
escapably seems to bracket off weak-willed actions. So traditional Deep
Self views not only incorrectly classify weak-willed actions as non-
attributable, they also seem to do so almost by design.
110

109
I should note at the outset that I am operating under the assumption that compulsion
is primarily a conative or volitional phenomenon rather than a merely cognitive one. It is
possible that some things that we colloquially call ‘compulsions’ may instead be instanc-
es of acting in accordance with pathologically acquired beliefs about the world. For ex-
ample, an agent might come to believe that her house would burn down if she didn’t
check that her oven was turned off 18 times. Suppose she does so instead of getting to a
meeting on time. On a view like the Minimal Approval view, we could easily explain
how she would be exempt from blameworthiness by showing how even though her ov-
en-checking is attributable to her, given the background conditions, which include her
epistemic state, she shouldn’t be blameworthy for doing what she reasonably judged to
be the best course of action. If this account of why compulsion exempts agents from
blame could be made plausible, there are two ways it might be bolstered to at least ad-
dress the distinction between compulsion and weakness of will. One possibility is that it
could be argued that weakness of will is a truly volitional occurrence while compulsion is
cognitive. Another route would be to argue that both compulsion and weakness are cog-
nitive phenomena and the distinction lies in differences in the beliefs, or the manner of
their acquisition. Even if some compulsive cases do function in this this way, I am deeply
skeptical that all compulsive cases function this way. If that’s right, there is a still a seri-
ous problem to solve here. I am grateful to Sarah Buss for pressing me to articulate this
assumption.
110
This problem has been articulated many times, notably in Vihvelin (1994), Haji (1998),
Haji (2002), Fischer (2010), Fischer (2012), McKenna (2011), McKenna and van Schoelandt
(2015), Strabbing (2016), and McKenna (forthcoming), and is often considered by many to
be a knock-down objection to Deep Self views, despite Deep Self theorists having tradi-
-100-
Take, for example, Bratman’s planning view. On Bratman’s view an
agent’s act is only attributable if it meshes with her self-governing policies
and plans about which desires of hers to act on. So if an agent makes a
self-governing policy to always act on her desires to stay home and study
before an exam, but then when the time comes is unable to get herself to
act on her self-governing policy, her action will be non-attributable. But
Bratman is also inclined to describe weakness of will in this way, as an
agent’s inability to get herself to act on her own self-governing policies. In
short, weakness of will is usually defined as a failure of self-governance,
and self-governance is usually what is required for attributable action.
This makes the fact that weak-willed actions do seem blameworthy very
difficult to accommodate on Deep Self views. This is a very serious prob-
lem for these theories since weak-willed cases appear to many as cases in
which it is particularly appropriate to hold agents responsible; in fact,
Gideon Rosen has even argued that agents are blameworthy for their ac-
tions only when they are weak-willed.
111

We do, intuitively, want to say that compulsive cases are cases of non-
attributable actions, and Deep Self views are able to give a richly explana-
tory account of why agents are not responsible for compulsion. When act-
ing on compulsive desires, agents’ standpoints are overpowered such that
their resultant actions do not express anything about what it is they really
want to do. Yet the attraction of this proposal is severely undercut by the

tionally relatively little to say about it. For example, Fischer writes of Frankfurt’s view,
“The problem of weakness of the will is, in my view, a decisive problem for Frankfurt’s
approach. Somewhat surprisingly, it has not received nearly as much attention as the so-
called ‘regress’ problem. (Indeed, I am not aware of any discussion of the relationship be-
tween his account of acting freely and the problem of weakness of will by Frankfurt)”
[Fischer (2010)].
111
Rosen (2014) argues that agents could only be responsible for weak-willed actions be-
cause only weak-willed agents meet the high epistemic standards for knowing wrongdo-
ing that he takes to be a prerequisite for blameworthiness.
-101-
fact that weak-willed cases have these same features yet intuitively these
actions are attributable. It seems there must be some way of distinguishing
compulsion from weakness of will as well as what makes the latter but not
the former attributable, but traditional Deep Self views seemingly have no
resources available to make such a distinction. Since it seems the distinc-
tion between compulsion and weakness of will must be found elsewhere,
we may wonder if we should look elsewhere, too, for an explanation of
what makes agents non-blameworthy for compulsive actions.
3. Current Solutions and Their Problems
It is crucial that Deep Self theorists find some solution to this problem. In
this section I will discuss existing solutions to the weakness of will prob-
lem, but I want to be clear that I will only be addressing them as resources
for a Deep Self theorist to tell compulsion apart from weakness of will and
also to correctly classify weak actions as attributable and compulsive ac-
tions as non-attributable. While non-Deep Self accounts of responsibility
such as control-based and reasons-responsive accounts will also have to
differentiate compulsion and weakness of will, these cases do not neces-
sarily pose any special problem for them, as the distinction might fall out
as a result of the account’s calibration of the degree of control or reasons-
responsiveness required for responsibility. I will ultimately be arguing
that this way of dividing up the cases is misguided, and hope to provide a
compelling alternative picture, but giving a positive argument for my
view over the solutions these non-Deep Self accounts posit is outside the
scope of this chapter. Instead I will focus on the in-house debate over how
to handle these cases that exists within the broadly Deep Self family of
views. And so in discussing the options currently on the table, I will be
analyzing them only as they are compatible with the aims of Deep Self
theory.
-102-
3.1 Mismatched Accounts
One possible way out of the puzzle, though it has not been much dis-
cussed in the literature, would be for the Deep Self theorist to deny that it
is the very same kind of mental state that weak-willed agents fail to act in
accordance with that also grounds ascriptions of attributional-
responsibility. For example, a Frankfurtian could say that weakness of will
consists in an agent acting contrary to what she judges best, while at-
tributable action does not require acting in accordance with what one
judges best, but rather, with the desires one endorses, since Frankfurt leaves
open the possibility that agents may endorse desires arationally. So it is
possible for Sam to judge it best that she should stay home and study,
while endorsing her desire to go to the party. This at least shows that it is
not impossible for a Deep Self view to account for some weak-willed ac-
tions being attributable.
But unless it is plausible that in every case of weakness of will the
agent endorses her course of action in Frankfurt’s sense, the view still fac-
es a weakness of will problem. All we need is one case to regenerate the
problem that the view fails to identify a necessary condition for attributa-
ble action. A familiar case from Huckleberry Finn is helpfully illustra-
tive.
112
Huck Finn befriends a slave named Jim and helps him escape from
slavery. While on a raft being used for the escape, Huck is plagued by
what he refers to as “conscience.” He believes, as do other white people in
his society, that helping a slave escape amounts to stealing, and that steal-
ing is morally wrong. He judges that the moral wrongness of this action
outweighs the demands of loyalty to one’s friends, and never considers
the idea of doubting what his society has told him is right. Though he re-
solves to turn Jim in because he judges it to be the right thing to do, he
nevertheless finds himself psychologically unable to follow through on

112
This case first appears in Bennett (1974) and Adams (1985), and is discussed in regards
to Deep Self views in Arpaly (2002).
-103-
this resolve, chastising himself for remaining a “bad boy.”
113
While this
case has a non-standard feature of examples of weakness of will cases,
since what Huck ends up doing is the right thing rather than the wrong
thing, this is still clearly a case of weakness of will. It also seems that Huck
is attributionally-responsible for helping Jim escape from slavery, and po-
tentially even praiseworthy.
Having the mismatch of a judgment-based conception of weakness of
will and an endorsement-based conception of attributability will not help
with cases like Huck’s. On the judgment-based conception of weakness of
will, Huck judges it best to turn Jim in but ends up helping him escape
nevertheless. So we correctly count Huck as weak-willed. But Frankfurt’s
traditional endorsement view won’t lead to the result that Huck is attribu-
tionally-responsible, because it does not seem plausible to suppose that
Huck in Frankfurt’s sense endorses his first-order desire to help Jim es-
cape. While it’s plausible that Huck may endorse a more general desire to
help his friends, or desire to some degree to help Jim, he clearly does not
form a second-order volition to act on his desire to help Jim escape. In fact,
he desperately wants to be moved by his desire to do what he takes to be
the right thing, so this looks like a case in which the desire he most wants
to be moved by is his first-order desire to turn Jim in. So we have a case on
which the mismatched view under consideration still incorrectly predicts
that a weak-willed action will not be attributable, when intuitively it is.
Other mismatched accounts will be even less plausible, since it’s quite
clear that there are weak willed actions that agents are attributionally-
responsible for in which they neither judge the course of action they un-
dertake to be best nor have they planned or committed to act in accord-
ance with it.

113
Arpaly (2002): 75.
-104-
3.2. Skepticism About Weakness of Will
While experiences of weakness of will seem to be very common, there is a
long history of skepticism about their existence. As Gary Watson points
out there are at least two different ways to be a skeptic about weakness of
will.
114

Traditional Socratic weakness of will skepticism consists in the claim
that no one ever truly acts against her best judgment. While there is noth-
ing logically incompatible about such denial and a Deep Self approach to
attributability, denying the possibility of acting against one’s better judg-
ment is counter to one of the primary motivations of Deep Self theory.
One of the main things that the Deep Self is used to explain is the fact that
when an agent acts on a motivation external to her standpoint, like the
unwilling addict does, she is not attributable for such action. So coupling
such an approach with the idea that a person never acts in ways that con-
flict with what she judges best would be an unusual move. Socratic skep-
tics usually argue that all of our acts are what we at the time of action real-
ly judge to be the best courses of action. Otherwise, they allege, we
wouldn’t act in such ways. Coupled with Watson’s view, the claim that
we always do what we judge to be best leads to the conclusion that all ac-
tion is attributable action. Frankfurt’s view sits strangely with Socratic
skepticism as well. One of the most interesting parts of Frankfurt’s view is
that it allows for some actions to count as attributable that are not judged
best, but there just are no such actions if weakness of will skepticism is
true.
A more natural kind of skepticism for the Deep Self theorist to adopt is
skepticism that anyone ever freely acts in conflict with what she most
wants to do.
115
For the Deep Self theorist this would amount to biting the

114
Watson (1977).
115
Gary Watson considers and argues for such a view in Watson (1977). However, his ar-
gument only speaks to views on which freedom is understood as being related to control,
so it does not provide an argument for favoring skepticism over the view I ultimately de-
-105-
bullet and accepting the consequence of her theory that weak-willed ac-
tions for which agents seem to be attributionally-responsible are really
compulsions for which they are not attributionally-responsible, or else are
misdescribed. In the absence of a compelling argument for such skepti-
cism, the Deep Self theorist opens herself up to a charge that such skepti-
cism is ad hoc. But even more worryingly, such a view would be highly re-
visionary. While this in and of itself is not a knock-down argument, the
view would have some very difficult consequences to accept. To illustrate,
consider again the case of Murderous Max from Chapter 2 (reprinted be-
low):

Murderous Max: Max strongly desires to go out on a killing spree this
morning because he hates people and there is almost nothing he likes
more in the world than shooting them—in fact he thinks of himself as
having made an art of it, perfecting his technique more and more with
each kill. There is one thing, however, that he cares about even more:
he is extremely committed to his morning workout routine. As much
as he wants to go out on a killing spree, he also realizes that if he does
that, he’ll have to forego his morning workout ritual. He knows that if
he misses even one morning of working out, he’ll probably fall off of
his routine, and he’ll thus sacrifice the progress he hopes to be making.
So, after considering and giving some weight to each option, he de-
cides that acting on his desire to work out is what he most wants to do,
it aligns best with his values, and is consistent with the plans he has set
for himself. However, his desire to hone his murderous craft by going
on that killing spree ends up just being so intense that he caves from
lack of willpower and goes out and does the deed.

fend in this chapter. For critiques of the argument in Watson (1977), see Watson (1999),
Strabbing (2016), and Watson (Forthcoming).
-106-
Is this case either misdescribed or else Max is not attributionally responsi-
ble for his killing spree, as the skeptic would have it? The characterization
of his mental process seems normal enough (despite, of course, the strange
content of his thought process). And it seems very difficult to believe that
his action in this case is not self-disclosing for the purposes of attribution-
al-responsibility. The fact is, he does desire to a great extent to act on his
desire to kill people, and it doesn’t seem right to excuse him for his action
just because he wanted to act on another motivation more.
3.3. Attributability for Failure to Exercise a Deep Self Capacity
In a recent paper, Jada Twedt Strabbing suggests that, in response to the
weakness of will problem, Deep Self views ought to allow that in addition
to exercises of attributability-relevant capacities, failures of attributability-
relevant capacities also are self-expressive for the purposes of attribution-
al-responsibility. This is Strabbing’s principle, which she says can be
adopted by any Deep Self theorist:

Having the Capacity (HC) Principle: An agent is attributionally re-
sponsible for an action A if and only if 1) A results from the exercise of
his attributability-relevant capacity to do A or 2) A results from the
failure to exercise his attributability-relevant capacity to avoid doing
A.
116

Now, just as objectively wrong actions are not by themselves sufficient
for attributional-blameworthiness, failures to act may constitute wrongdo-
ing but are not by themselves sufficient for attributional-blameworthiness.
What Deep Self views add to wrongdoing to generate attributional-
blameworthiness is some form of agential authorization or assent that
shows us that the agent’s action is not only wrong but also expresses
something about what she is like through that action. So to take seriously

116
Strabbing (2016): 14.
-107-
the fact that Strabbing offers her principle as a Deep Self principle, the
most charitable interpretation of this principle as it applies to weak-willed
action is to understand failures to exercise and act on attributability-
relevant capacities as expressing implicit assent to either the omissions or
to the alternate courses of action undertaken.
When the HC Principle gets applied to a particular Deep Self view,
several ambiguities arise, and no interpretation of the view ends up feasi-
ble. I will illustrate these problems by applying the HC Principle to Frank-
furt’s endorsement view after I briefly preview the structure of the argu-
ment. The first ambiguity arises in interpreting what is meant by an “ex-
ercise of an attributability-relevant capacity.” On one interpretation the
“exercise” is the formation of the deep self mental state, and on another it
is an agent’s getting herself to act in accordance with that deep self mental
state. Given the first interpretation, another ambiguity arises: is the attrib-
utionally-responsible agent the one who fails to form a deep self mental
state to do something that would prevent her from acting impermissibly if
she were to act in accordance with it? If so, the principle does not give the
result that weak-willed agents are attributionally-responsible. Or is the at-
tributionally-responsible agent the one who fails to form a deep self men-
tal state that will actually be effective in getting her to avoid acting im-
permissibly? If so, I’ll argue, the view mischaracterizes the character flaw
revealed by the weak-willed agent. This leads to an interpretation on
which the attributionally-responsible agent is the agent who fails to get
herself to act in accordance with her deep self mental state given that she
had the ability to. But on this interpretation, I’ll argue, the view cedes
crucial dialectical ground and opens itself up to an especially difficult in-
compatibilist challenge.
Let’s see what these interpretations of the HC Principle look like when
applied to Frankfurt’s view. Given the first interpretation, an agent “exer-
cises her attributability-relevant capacity” by forming a second-order voli-
tion. If interpreted this way, we get the following view:
-108-

Frankfurt-HC1: An agent is attributionally responsible for an action A
if and only if 1) A results from the agent’s forming a second-order voli-
tion to act on a first-order desire to A or 2) A results from the failure to
form a second-order volition to act on a first-order desire to avoid do-
ing A.

On this view, the only sorts of omissions for which a person can be attrib-
utionally-responsible are those that result from failing to form second-
order volitions.
But unless we adopt a Mismatched Accounts strategy, it is natural for
Frankfurt to think that weak-willed agents do form second-order volitions
to avoid doing A but just don’t act on them.
117
Sam, for example, forms a
second-order volition to act on her desire to study, which, if it were suc-
cessful, would lead to her avoiding doing A (in this case, going to the par-
ty), she just isn’t motivated to action by it. On this interpretation there is
no way to distinguish weak-willed actions from compulsive ones, and
both seem non-attributable. Sam does form a second-order desire to act on
a first-order desire to avoid doing A, namely, she forms a second-order
desire to act on her desire to study, a way of avoiding going to the party.

117
An alternate, but misguided interpretation of Frankfurt’s “second-order volition”
would have it that second-order volitions directly cause actions when unimpeded. On
this view, we would still need to understand why weak willed urges don’t count as im-
pediments while compulsive urges do; this just moves the problem to a slightly different
location. I call this approach misguided because, as I argue in “Depression’s Threat to
Self-Governance,” melancholic depression’s impact on the will seems best described by a
disconnect between an agent’s second-order volition and her action even in the absence
of a countervailing first-order urge. See Gorman (Unpublished Manuscript) for further
discussion both of this point and of the two competing conceptions of second-order voli-
tions.
-109-
And so, on Frankfurt-HC1, since she does not fail to form such a second-
order volition, it seems Sam is not attributionally-responsible.
There is something that Sam fails to do, though: she fails to form a se-
cond-order volition to study that actually gets her to avoid going to the
party. Even though the agent might believe that forming a second-order
volition to study will get her to avoid going to the party, she is wrong.
This leads us to the second possible interpretation. Could we instead in-
terpret Frankfurt-HC1 as delimiting weak-willed actions as the set of ac-
tions that involve an agential failure to form a second-order volition to do
something that actually would make the agent avoid doing A, independ-
ent of her beliefs? Cases like the following one show why such a view
would be unappealing:

Secret Spinning Desire: Suppose Sam believes, incorrectly, that the on-
ly second-order volition she could form that would get her to avoid
acting on her urge to go to the party is the second-order volition to act
on her desire to study. However, unbeknownst to her, doing so would
be wholly ineffective. The only second-order volition she could form
that would get her to avoid going to the party would be a second-
order volition to act on a desire to spin around on her desk chair one
time and then start studying. But Sam never considers acting on such a
desire. Instead, she forms the second-order volition to act on a desire to
study, which is ineffective, and she ends up going to the party instead.

While I think this interpretation of the view would correctly hold that Sam
is attributionally-responsible for going to the party, it wholly mischarac-
terizes the character-flaw that weak-willed action expresses. It’s not be-
cause she failed to realize that acting on a desire to spin around on her
desk chair and then study that she is attributionally-responsible. This sort
of epistemic flaw neither reflects the fact that an agent implicitly assents to
the course of action (a crucial part of Deep Self explanations) nor does it
-110-
seem to track our common-sense understanding of what it is about agents
like Sam that is objectionable.
So, more charitably, I think we should interpret Strabbing’s proposal
as applied to Frankfurt’s view in a third way:

Frankfurt-HC2: An agent is attributionally responsible for an action A
if and only if 1) A results from the agent’s getting herself to act on a
desire to A that she second-order endorses or 2) A results from the
failure to get herself to act on a first-order desire to avoid doing A that
she second-order endorses.

The problem with this suggestion is that in some ways, just like weak-
willed actions, compulsive actions might seem like failures to exercise and
act upon general attributability-relevant capacities. Compulsive actions
would thus count as attributable on this view when they are in fact non-
attributable.
Instead, the view would need to be coupled with an argument that
compulsive actions are not failures in the relevant sense to act in accord-
ance with one’s second-order volitions but rather, cases in which the agent
lacks the ability to act in accordance with her second-order volition. This
echoes a suggestion recently offered by Michael McKenna that the appro-
priate response to the problem of weakness of will for theorists like Frank-
furt and Watson is to reintroduce talk of ability. For example, building on
some textual support for this position in the article where Watson first
proposed the valuing view, McKenna suggests that valuing theorists
ought to adopt the following modified principle:

-111-
We act freely just in case we are able to act in accord with what we
value, as issuing from our evaluative system; we act unfreely just in
case we are unable to act in accord with what we value.
118

McKenna has in mind here the kind of freedom relevant to attributional-
responsibility, and so we can derive the following principle:

An agent is attributionally-responsible for her action just in case she is
able to act in accord with what she values.

The idea here is that the valuing theorist could account for the fact that
agents are attributionally-responsible for weak-willed actions despite the
fact that such agents do not act in accordance with what they most value,
because they have the (unexercised) ability to act in accordance with what
they most value.
Something similar seems to be what Strabbing really has in mind when
she writes, for example,

But why is an action attributable to an agent just in virtue of resulting
from his attributability-relevant capacity, even if that capacity is not
exercised? On my view, the answer is this: when an action results from
the agent’s attributability-relevant capacity, he has control over the fact
that he performs it. An agent clearly has control over the fact that he
performs action A when A results from his exercising his attributabil-
ity-relevant capacity to perform A. Yet an agent also has control over
the fact that he performs A when A results from his failing to exercise
his attributability-relevant capacity to avoid A, and this is precisely be-
cause he could have avoided A.
119

118
McKenna (Forthcoming): 5.
119
Strabbing (2016): 22-23.
-112-

So weak-willed action on this view is self-expressive because, for example,
Sam’s going to the party when she possesses the relevant ability to stay
home and study instead is tantamount to her implicit assent to the alterna-
tive course of action.
The fact that Strabbing invokes the notion of self-control within the
context of attributability is notable, since focusing on an agent’s volitional
control is usually seen as a competing approach to a focus on attributability
and the Deep Self. For example, Al Mele argues that the fact that we are
not responsible for actions stemming from manias, compulsions, and ad-
diction is better explained by the fact that we lack volitional control over
such mechanisms, and so talk of the Deep Self seems unnecessary.
120
Giv-
en that it is presented as a friendly modification for Deep Self theorists,
Strabbing’s idea, instead, is that volitional control itself is not sufficient for
responsibility, but rather, whether or not an agent exercises volitional con-
trol over her Deep Self-relevant capacities reveals something about what
she is like. But it’s just not clear what work the Deep Self is actually doing
in this picture, or if, instead, this is really just the ability view in disguise.
Ability, it seems, is really the heavy lifter on this interpretation of the ac-
count.
McKenna agrees. He argues that taking on board the notion of ability
to distinguish weakness of will from compulsion amounts to giving up on
the dialectical aims of Deep Self theory. As he puts it,

Relying upon the notion of ability does not fit well with the strict aims
of a [Deep] Self view and the attempt to treat compromises to free
agency as external impediments to acting freely. If the weak willed
non-addict’s desire to take the drug is after all external to her, then it
appears not to be the case that externality rather than internality is

120
Mele (1992, 1995).
-113-
what explains self-determination or lack of it. Something else is doing
the work—an ability. But now, if it is merely an ability to resist a de-
sire, what’s it matter if it is internal or external to the agent? What mat-
ters is whether her ability affords her sufficient control over it to be
free with respect to whether she acts on it.
121

Dialectically, this view just cedes too much important ground to the abil-
ity theorist to really be attractive to a Deep Self theorist. In Chapter 1, §2.1,
I explained how the Deep Self view’s unique advantage over Classical
Compatibilism is that it is able to bracket off incompatibilist concerns
about the relevant kind of ability required for responsibility by shifting
the conversation entirely away from talk about ability altogether. Any
principle that reintroduces ability is subject to incompatibilist interpreta-
tion. On Strabbing’s, weak-willed action is attributable to the agent only
when she could have acted on her attributability-relevant capacity but
didn’t. But the incompatibilist will say that no agent can act otherwise
from how she actually acts so long as she is determined. So, given the
truth of determinism, no one is attributionally-responsible for any weak-
willed actions. The burden is then on the compatibilist to prove that her
compatibilist-friendly conception of ability is instead what differentiates
failures to exercise a capacity from instances of entirely lacking the capaci-
ty. Although moving past this stalemate may not be an impossible task, it
is a task that Deep Self views were designed to circumvent.
The great advantage of Deep Self views is the fact that they move be-
yond this stalemate by attempting to show how ability to do otherwise is
irrelevant to questions of responsibility and by replacing this with another
condition that is meant to be unquestionably compatibilist-friendly, name-
ly, internality. This sort of move is leveraged by Frankfurt’s arguments
against the Principle of Alternative Possibilities. By analyzing internality

121
McKenna (Forthcoming): 5.
-114-
in part in terms of ability, we lose part of the motivation for shifting to a
focus on internality in the first place. And if a Deep Self theorist finds the-
se arguments compelling, she should be wary of reintroducing talk of abil-
ity.
122

Furthermore, some of the charges against compatibilist conceptions of
ability are sharpened further by considering them in the context of a Deep
Self view. In particular, the charge that compatibilist notions of ability fall
short of explaining how an agent’s actions/omissions are “up to her.” For
example, combining the view with Michael Smith’s account of ability we
get the view that failures to exercise capacities are cases in which an agent
does not actually act on the desire she endorses when there is a raft of
counterfactuals in which she does act in accordance with the desire she en-
dorses.
123
This thin notion of ability is open to the following charge, which
is well articulated by Pamela Hieronymi:

122
Making the distinction normative rather than metaphysical does not help matters
much. In both weak-willed and compulsive cases it is already true that the agent has, for
example, endorsed her first-order desire to do the right thing. So to ask whether or not it
is fair to expect the agent to exercise her attributionally-relevant capacity is tantamount
to asking whether or not it is fair to expect the person to resist her desire to do the wrong
thing. But attributionally-relevant capacities drop out of the explanation altogether here,
as the answer to this question is just an answer about what amount of control over our
desires we think is reasonable to expect of people to have, and the explanation makes no
essential reference to anything about attributionally-relevant capacities. At the very least,
the notion of the Deep Self would need to be quite different from how it is ordinarily un-
derstood to play any more significant role in this explanation. The boundaries of the
Deep Self would need to be set by normative facts about when it is reasonable to take an
act to be self-disclosing, rather than by facts that indicate whether or not the agent in fact
(in at least some way) stands behind her action. See Chapter 5, §4.2 for further discussion
of this point.
123
See Smith (2003) for a development of this view.
-115-
Suppose I have a heart attack. It may well be that, in a host of similar
possible worlds, I do not have a heart attack. Further, the fact that I do
not have a heart attack in those worlds may be explained by the under-
lying structure of my cardio-vascular system. Thus, I have the capaci-
ty, in Smith’s sense, to have not suffered the heart attack. The truth of
this claim does nothing to show that it was up to me whether I had a
heart attack…
124

In the context of a Deep Self view, this charge is particularly troublesome.
If it’s not up to you, nor does it have anything to do with what you really
want, whether or not you act on the desire you endorse, how can whether
or not you act on the desire you endorse determine whether or not your
action is self-disclosing?
Even setting aside all of the dialectical problems with the reintroduc-
tion of ability, it’s just not clear that the agent does assent to her action by
failing to bring herself to act in accordance with her values/ endorsements/
commitments when she has the ability to. In order for this to be true, we
would need an argument that this minimal kind of assent is the relevant
kind of assent. One could make an argument that this notion of assent
seems too minimal to do the justificatory work it’s meant to do. Instead,
my criticism comes from a slightly different angle: if we need to make the
notion of assent much weaker than it’s ordinarily understood in tradition-
al Deep Self theories, we may as well just identify the kind of mental state
that constitutes this form of assent, and make that mental state itself the
criterion for attributability.
4. Adopting a Mosaic Conception of the Deep Self
The only truly feasible strategy for a Deep Self theorist to respond to the
weakness of will problem is to alter the way she conceives of deep self

124
Hieronymi (2007): 16.
-116-
mental states such that it makes sense to say that weak-willed agents do
identify with the actions they undertake in a way that makes these actions
self-disclosing. In order to understand the way in which an agent can as-
sent to a course of action that is counter to what she takes to be her most
favorable course of action, the traditional Deep Self theorist will have to
alter her conception to allow that both the favored option and the weakly-
willed option would, if undertaken, have the power to speak for the agent.
So, for example, both Sam’s desire to stay home and study, were she to act
on it, and her desire to go to the party would both count as being part of
her Deep Self. The idea behind this approach is to in some way weaken
the pre-requisites for agential self-expression so that an agent does express
something about who she is directly via her weak-willed action.
Chandra Sripada calls the view of the Deep Self that corresponds to
this alternate picture on which more than one mental state and their re-
sultant actions can speak for the agent a “mosaic” conception. Mosaic con-
ceptions of the Deep Self permit conflicts and tensions within the Deep
Self since, as Sripada writes, “conflict can and often does extend all the
way to our very practical foundations.”
125
He contrasts this with “homog-
enous” conceptions of the Deep Self, according to which the Deep Self can
contain no conflicts, and all apparent conflicts are merely illusory. On
such views, for an action to issue from a person’s Deep Self is for the agent
to authorize an action, and an agent can only authorize an action by
uniquely and decisively picking it out in some way. I have shown that
most traditional homogenous Deep Self theories are unable to accommo-
date weak-willed actions as attributable actions because weak-willed ac-
tions are caused by motivations that conflict with the motivations the
agent has authorized herself to act on. This suggests that the way to solve
the weakness of will problem for a Deep Self theorist is to adopt a mosaic

125
Sripada (2016): 24.
-117-
picture of the Deep Self and correspondingly, a weakened criterion for at-
tributability.
4.1. The Minimal Approval View as a Mosaic Deep Self View that Can
Solve the Weakness of Will Problem
Adopting the Minimal Approval account solves the weakness of will
problem because it explains the way in which even the weak-willed agent
may assent to her course of action: via minimally approving of her course
of action. It is able to solve the problem precisely because it is a mosaic
view: more than one course of action can be assented to since conflict is
permitted at the agential foundations that are relevant for responsibility-
ascriptions.
Recall that the Minimal Approval account just requires the following
for attributional-responsibility:

The Minimal Approval View of Attributional-Responsibility: An agent
is attributionally-responsible for ϕing-at-t iff the actual sequence of
mental states involved in the production of her action is non-
implanted and together with her other mental states makes it the case
that at t, if she were to reflect on her desire to ϕ at t, she would be suf-
ficiently likely to want to act on her desire to ϕ at t, with some further
aim in doing so other than merely eliminating this desire.
126

The account accommodates the fact that weak-willed agents have less-
than-complete identification with the courses of action they undertake,
and are nevertheless attributionally-responsible for them. For example, it
makes sense to think that Murderous Max’s action comes about in part

126
“ϕ-ing” should be understood here as standing for an action, and the account should
be taken to cover only attributional-responsibility for actions. Attributional-responsibility
for omissions and for consequences is, on my view, derivative on attributional-
responsibility for actions. I explore some of these issues in Chapter 5.
-118-
due to the fact that he wants to some degree to act on his desire to go on a
killing spree, and that this desire is not just to relieve some urge but is ra-
ther related to his love of the art of murder. Therefore, his action, though
weak-willed, is clearly attributable.
It might be helpful at this point to see how the Minimal Approval view
comes apart from traditional ability-based views in terms of what it posits
about the difference between weakness and compulsion. Now it may be,
as a matter of empirical fact, that it is often harder to get oneself not to act
on a desire that has no correspondence to what you endorse, making it
such that you have less control over acting on such desires, and so there
may be fairly substantial overlap in the extensions of the two theories. The
difference is that, on my view, the difficulty one has in resisting a piece of
behavior is not essential to what makes something non-attributable, nor to
what makes something a compulsion. An agent may be such that she
could easily have avoided the compulsive behavior if she tried, but simply
due to absent-mindedness forgot to. She would still not be responsible for
her compulsive action because it does not express anything about who she
is. On the other hand, Sam, whether she has the capacity to resist her
weak-willed desire to go to the party or not, still does something she is at-
tributionally-responsible for because her doing so is related to the fact that
she would want to act on this desire due to some aim, and this tells us
something about what Sam is like. I think is precisely what a Deep Self
theorist should want to say about the cases, not only because it is exten-
sionally adequate, but also because it gives the right kind of explanation
for why weak-willed but not compulsive agents are attributionally-
responsible.
4.2 Can Other Deep Self Views Adopt a Mosaic Conception to Solve the
Weakness of Will Problem?
The reason the Minimal Approval view is able to solve the weakness of
will problem while Frankfurt’s endorsement view is not is that it does not
-119-
require a homogenous conception of the Deep Self and so can offer a
weaker criterion for attributability. This raises the question: can any other
Deep Self view follow suit and adopt a criterion that is compatible with a
mosaic notion of the Deep Self in order to solve the weakness of will prob-
lem?
Not every Deep Self makes sense in mosaic form. Bratman’s planning
view, for example, seems inextricably bound up in a homogenous concep-
tion of the Deep Self since the agent’s deciding decisively on a certain
unique course of action to prioritize is meant to be precisely what author-
izes attributional-responsibility for that action. Bratman emphasizes the
role of cementing agential coherence in planning agency, and adopting a
mosaic conception of planning would certainly forgo these appeals of the
view. However, even if a version of the planning view did allow for an
agent to make conflicting plans, it is not feasible to think that in all at-
tributable weak-willed cases the agent must have a plan to act in the way
she does, nor need her action be related to any of her larger plans. In fact
it’s quite a natural thought that part of the nature of weak willed actions is
the very fact that they deviate entirely from our plans for ourselves.
In Chapter 2, I argued that each traditional Deep Self view proposes a
criterion that contains additional elements that are unnecessary for at-
tributability. Mosaic versions of valuing and caring views, I’ll now argue,
are subject to a parallel criticism in their handling of weakness of will cas-
es. In both cases weakening the criterion to accommodate a mosaic Deep
Self conception does not weaken it enough to capture all cases of attribut-
able weak-willed action.
Strabbing considers and then, for this very reason, dismisses a mosaic
version of the valuing view on which rather than requiring that agents act
on what they most value in order to be attributionally-responsible for their
actions, agents just need to act in accordance with something they (pro tan-
to) value at all. So according to this view, in Sam’s case, we could say that
while she values studying for her exam the most, she does value going to
-120-
the party as well to some degree. But there are cases, as Strabbing points
out, in which a weak-willed agent does not even pro tanto value the course
of action she takes, yet she still seems attributionally-responsible for her
action.
127

Watson’s own examples of a mother overcome with a sudden urge to
drown her bawling baby in the bathtub, and a squash player with a desire
to smash his opponent in the face with his racquet can be used to illustrate
here.
128
There are cases in which a person desires something without valu-
ing it at all, yet we appropriately take a person’s acting on such a desire to
say something about what she is like. Agents who act on weak-willed de-
sires to do such things provide counterexamples to the claim that weaken-
ing the criterion of the valuing view in this way can solve the weakness of
will problem.

127
Another problem with this view might be that it seems in other respects too expan-
sive—it gives the result that agents who are intuitively compulsive who nevertheless see
something of value in taking a drug, etc. are in fact attributionally-responsible. This point
is not crucial to my argument in this chapter since there is, I take it, already sufficient rea-
son to reject this view. But it is worth mentioning since a similar though somewhat less
serious worry exists in regards to the Minimal Approval account. Consider a case in
which an alcoholic desires to some small degree to act on her desire to drink, not just be-
cause she wants to get rid of the desire to drink, where her second-order desire is not ex-
plained by any sort of special epistemic circumstances, and where acting on a desire to
drink is morally problematic. My view may seem unreasonably austere in classifying
such agents as attributionally-responsible and blameworthy. While I won’t fully develop
it here, if this result is unpalatable, one possible route is to say that attributability is grad-
able such that the alcoholic’s action is less attributable than the action of an agent who
wanted most to satisfy her desire to drink (where such desire to satisfy is not explained
by wanting to rid herself of a desire to drink. (In theory, a mosaic value theorist could
make some sort of analogous move to circumvent this issue.) A natural idea might be
that the gradability comes from the strength of the desire to drink relative to her other
desires to some degree satisfy first-order desires. I hope to explore this issue further in
future work.
128
Watson (1987).
-121-
This is an application of the charge against the valuing view I made in
Chapter 2. It is just a fact that we are attributionally-responsible for some
things that we don’t value at all, and there are many weak-willed cases
that illustrate this lesson particularly well. As Watson himself now
acknowledges, the valuing account is just too rationalistic; this is a prob-
lem for accounting for the fact that even some non-weak-willed actions
are attributable in the absence of valuing.
129
In such cases an agent whole-
heartedly decides to do something that she does not judge to be good, and
intuitively is attributionally-responsible for her action.
The caring view is a better candidate for solving the weakness of will
problem, and in fact, it is usually advanced in an explicitly mosaic form,
as a view that is therefore well-positioned to handle cases of weakness of
will.
130
As with the valuing view, though, a somewhat parallel criticism to
the one I advanced against caring views in Chapter 2 extends to the view’s
treatment of weak-willed cases.
The issue is what Strabbing calls the problem of weak-willed whims.
131

The idea is that oftentimes the desire an agent acts on in a weakness of
will case is a whimsical desire that intuitively is not related to what the
agent cares about. Nevertheless, intuitively, the agent is responsible. Cer-
tainly some cases that we are tempted to describe as weak-willed whims
may truly be instances of very minor compulsions, and the line here may
not always be clear. Whereas attributional-responsibility for acting on
whims may seem inconsequential in many cases, in weak-willed cases act-
ing on these whims is in many cases what makes the agent fail to do what
she believes she ought to do. Supposing she is right about what she ought
to do, if her responsibility for failing to do what she ought to do is deriva-
tive on her responsibility for acting as she in fact does, then it matters very

129
Watson (1987).
130
See, for example, Shoemaker (2003), and Sripada (2016).
131
Strabbing (2016): 16.
-122-
much whether or not she is attributionally-responsible for her whimsical-
ly-motivated action.
132

Suppose Sam has decided to stay home and study for her exam and is
overtaken by a sudden whim to pace around her dorm room instead. It
seems strained to say that Sam cares about pacing around the room or that
her desire to do so would be suitably related to any sorts of distinctive car-
ing states she has. But it is likely that Sam’s desire to pace around the
room is not just a random fluke either. In fact, it may be that whimsical
desires always stem from subconscious intrinsic desires. Arpaly and
Schroeder argue that “in each case [involving someone acting on a whim]
it is easy enough to imagine credible intrinsic desires that each person
might have such that the person’s whim is instrumental toward, or a real-
izer of, the content of the intrinsic desires.” It seems plausible that there
will always be similar stories to tell about whimsical desires even if the
agents themselves don’t always have access to the explanations.
133

If whimsical desires are always instrumental or realizer desires of sub-
conscious intrinsic desires, it is easy enough to see how a weak-willed
agent who acts on a whimsical desire would satisfy the requirements for
Minimal Approval for these actions. The fact that she would act on her
whimsical desire is related to the fact that if she were to reflect she would
have a desire (not necessarily consciously held) to act on a motivational
state that would further her intrinsic desire (her further aim).
But dialectically speaking, care theorists cannot appeal to the presence
of a mere intrinsic desire to show that the agent is attributionally-
responsible, because they are at pains to show that caring states are not
reducible to mere intrinsic desires, but rather, involve a complex set of

132
I think it makes the most sense for mosaic Deep Self theorists to consider responsibility
the actions that are omitted in weak-willed cases as derivative from responsibility for act-
ing on the motivation the agent in fact acts on. I develop this idea more fully in Chapter
5.
133
Arpaly and Schroeder (2013): 10.
-123-
dispositions. Chandra Sripada does take caring states to be partially con-
stituted by intrinsic desires, and so he may tell a similar story to show
how the presence of these intrinsic desires is evidence that the agents
whimsical desires are suitably related to her caring states. However, it is
just not clear that the sorts of intrinsic desires for which whimsical desires
are instrumental towards or realizer desires of need always be related to
what she cares about in Sripada’s sense.
It is plausible that Sam’s pacing around the room is caused by a sub-
conscious fear of failure that is suitably related to the fact that she cares
about being a good student. But it is equally plausible that her pacing is
due to a spontaneous and fleeting intrinsic desire to not think so hard; she
is moved to act on a desire that realizes an intrinsic desire she has no long-
term or emotional investment in whatsoever. This echoes the objections to
the caring view I advanced in Chapter 2.
It is interesting to note that David Shoemaker, who also advances a
care theory, seems to openly embrace the fact that according to his view
agents are not attributionally-responsible for acting on whims.
134
The costs
of this move are mitigated by the fact that, according to his picture, agents
who act on whimsical desires may be candidates for answerability-
responsibility and/or accountability-responsibility. These forms of respon-
sibility, for Shoemaker, are serious forms of moral responsibility that do
not require agents’ actions to be attributable (at least not in the kind of
sense required by attributional-responsibility). For care theorists who take
attributability to be necessary for the most central kinds of moral respon-
sibility, however, denying that agents who act on weak-willed whims are
attributionally-responsible comes at a much higher cost.
This shows that there is a distinctive advantage to adopting the Mini-
mal Approval View. While the main innovation in responding to the
weakness of will problem is to lower the criterion for attributability such

134
See Shoemaker (2015b): 113.
-124-
that conflicting strands of the self can both speak for the agent, the Mini-
mal Approval view is the most promising implementation of this strategy.
5. Responses to Fairness Objections
Despite the advantages of the Minimal Approval view’s handling of the
weakness of will problem, it does have some features that may seem unin-
tuitive due to worries about the fairness of demarcating compulsion and
weakness in the way that I have suggested we should. In this section my
aim is to respond to these concerns.
First, some may object that on my view we let people off too easily for
their compulsions. I am committed to the claim that it can be the case that
a person is not attributionally-responsible for her action even if she could
have done otherwise had she resisted. On my view, compulsions in theory
need not even be particularly difficult to resist to make a person exempt
from attributional-responsibility for the resultant actions. To the extent
that this strikes some people as implausible, I take it this is motivated by a
concern about fairness.
I have several lines of response to such worries. First, there is much
contested ground over which questions about free will and moral respon-
sibility should be answered in the domain of metaphysics and which
should be answered in the domain of first-order ethical theory, but I take
there to be a methodological problem with raising concerns about fairness
at least with this part of the picture. This is not to say that questions of
fairness need not enter consideration at all, just that if they do, they do so
at a different level of generality. We might reasonably ask the question,
given considerations of fairness, what is the appropriate basis on which to judge
people to be attributionally-responsible for their actions? Once we have decided
that the answer to that question is that it is appropriate to hold people at-
tributionally-responsible for their actions just in case they reveal some-
thing about what they are like as agents, and commit to a Deep Self view,
our further question is now: what does it take for someone to reveal something
-125-
about what she is like as an agent through her action? Our criterion for answer-
ing that question should be based on how well the proposed state or pro-
cess corresponds to the proper notion of self-disclosure that is relevant to
praise and blame. Reintroducing concerns about fairness in answering this
question seems to admit a certain kind of defeat for the metaphysical in-
quiry of the B-Tradition. We sought to discover the conditions that tell us
when an act actually is self-disclosing, not just when it would be best for us
to think of someone’s act as self-disclosing given the consequences of do-
ing so as they bear on considerations of fairness.
Setting aside these methodological concerns, it is not even clear that
considerations of fairness would make us favor an ability-based theory of
when people are exempt from attributional-responsibility for compulsive
action over the view I have put forward.
There are at least two different concerns about fairness that might be
raised in regards to my proposal. First, one might think the victims of
moral transgressions that are caused by compulsive agents who could
have easily resisted their compulsive actions have a right to blame the
people who wronged them. Moral wrongdoing could have been prevent-
ed easily, and so, given considerations of fairness, blame seems warrant-
ed.
135
Second, given that on my view agents are attributionally-responsible
for weak-willed actions that are very difficult to resist, it seems unfair to
exempt similarly situated compulsive agents who would not have the
same difficulty resisting were they to try. One thing to note is that the
view is a view about attributional-responsibility, not just blameworthi-
ness, so presumably the inverse fairness concerns could equally be raised
for praiseworthiness. These concerns might be thought in a way to balance

135
I actually agree that such victims might deserve an apology, but do not think they
have the right to blame compulsive agents. I develop a brief sketch of a sense of respon-
sibility that might make the former but not the latter response appropriate in Chapter 5,
§5.
-126-
each other out. But given that susceptibility to blame may be more bur-
densome than susceptibility to praise is a benefit, noting this may do little
to quell worries.
But isn’t it also unfair that compulsive agents are expected to shoulder
the burden of resisting external urges? If compulsive urges are in fact ex-
ternal, why should they be the responsibility of the agent to have to man-
age them? One could try to somehow balance this consideration of fair-
ness against those raised against the proposal, but I think to make fairness
the sole criterion for the theory would be to make a methodological mis-
take. It would amount to reducing blame and responsibility practices to
mere burdens rather than acknowledging them as practices that are inex-
tricably bound up in a context that makes such practices apt.
If remaining concerns linger, the following is at least dialectically open
to the Minimal Approval theorist. She may admit that even if concerns
about fairness do not influence ascriptions of attributability, they may in-
form the background conditions of the agent’s action such that they influ-
ence blaming practices, thus affecting the way in which it is appropriate to
interpret and respond to the meaning of the agent’s action. The Minimal
Approval theorist may even accept that while it plays no role in assign-
ment of attributional-responsibility, difficulty resisting countervailing mo-
tives may have a role to play in distinguishing between whether or not,
for example, an agent reveals through her action an overtly malicious trait
or a mere lack of moral fortitude. In this way the Minimal Approval theo-
rist can accommodate the datum that degrees of difficulty resisting does
seem to play some role in our assessment of agent blameworthiness with-
out ceding the crucial point that control plays no role in setting the
bounds of attributability.

-127-
Chapter 4: The Finely Individuated View of Blame’s Content
1. Introduction
My central goal in this dissertation so far has been to advance an account
of the conditions of attributability, the Minimal Approval view. I have ar-
gued in favor of an account that locates attributability in a set of agential
conditions, and accordingly, my focus has primarily been on the attitudes
of the responsible party. In this chapter I illuminate some of the core fea-
tures of blame itself, and so my focus is instead on the attitudes of the
blamer, rather than the person blamed.
Recall that the view I articulate in the first three chapters is meant to
be, in theory, compatible with a wide range of stories about the route from
attributability to blameworthiness and blame, and about the relationship
between attributional-responsibility and other kinds/faces of moral re-
sponsibility. In this chapter I will develop one possible story about the re-
lationship between attributability and blame that I think ought to be
adopted, and in Chapter 5 I will say more about what I take the relation-
ship between attributional-responsibility and accountability-responsibility
to be. But it should be noted that Chapters 1-3 are, in a way, self-standing.
One may accept the claims of the first three chapters without also accept-
ing the claims I will make in this chapter or the next.
That said, an accompanying account of blame can make a particular
account of its preconditions more or less plausible. In particular, the Min-
imal Approval view leaves us with a view on which the “self” that is im-
plicated in attributional-responsibility may be both arational and fragmen-
tary. This is a less robust conception of the self or will than might be re-
quired for certain accounts of blame. Does the possibly arational and
fragmentary nature of the self that accompanies the Minimal Approval
picture threaten our ability to tell a compelling story about blame? My
main goal in this chapter will be to argue that it does not.
-128-
Developing a complete theory of the nature of blame would require
providing answers to two different questions. First, there is the question
of what kinds of mental states blame consists in. Candidate answers to
this question in the current literature include judgments, emotions, de-
sires, and dispositions to communicate or protest. Second, there is the
question of what the content of these blaming attitudes are. For example,
suppose blame amounts to targeted resentment: does B blaming A mean
just that B resents the fact that A ϕ-ed? The judgment that A’s ϕ-ing reason-
ably implicates? A’s quality of will as exhibited by her ϕ-ing? A’s charac-
ter as exhibited by her ϕ-ing?
These two questions may not be wholly independent, but since I will
not be able to offer a complete account of the nature of blame in this short
chapter, I focus primarily on the latter question, which speaks more di-
rectly to the challenges that arise for articulating a theory of blame to go
along with the Minimal Approval view. My aim is to sketch a theory of
blame’s content that is attractive in its own right, but also one which is
particularly well suited to be accepted in conjunction with the Minimal
Approval view.
2. From Attributability to Blameworthiness
Once it is determined that an agent is attributionally-responsible for ϕ-ing,
what else must be added to this fact in order for it to be appropriate to
blame the agent on the basis of her ϕ-ing? If the kind of blame in question
is moral blame, then one thing that must be added is that the agent does
something that is in some sense morally objectionable. I take the relevant
sense of an agent doing something morally objectionable to be that she
does something that she morally ought not to have done relative to her
non-moral beliefs at the time of action.
136
Although it is controversial, I

136
I say her “non-moral beliefs” because I believe that unlike other sorts of ignorance,
moral ignorance does not exempt, but I will not here be able to give a defense of the fact
-129-
take it that this way of conceiving of the sense in which a blameworthy
agent does something morally wrong avoids the need to posit an addi-
tional epistemic requirement on blameworthiness
137
or an additional re-
quirement to rule out cases in which an agent acts under severe duress. It
is not that these factors are irrelevant to an agent’s blameworthiness, but
rather that they are factors that play a role in normative ethical theorizing
about wrongdoing itself. There is no need to offer additional criteria here
beyond the fact that the agent attributably does something morally objec-
tionable for blameworthiness because I take it that it is just not morally
wrong in the sense relevant to blameworthiness to tell a lie due to your
child being threatened, or to give someone a glass of poison when you be-
lieve that it is gin.
138

A proponent of the Minimal Approval view need not be beholden to
this picture of the relationship between moral wrongness and blamewor-
thiness, however; there are interesting and largely underexplored issues
regarding which of these issues ought to fall under the domain of theoriz-
ing about moral wrongness and which should fall under the domain of

that moral ignorance is special in this way. For recent discussion of this issue in the litera-
ture see the papers collected in Robichaud and Wielend (2017).
137
A belief-relative notion of moral wrongness subsumes several kinds of proposed epis-
temic conditions on blameworthiness including but not limited to Susan Wolf’s “sanity
condition.” See Wolf (1987).
138
Many think we need to leave room in the picture for the possibility of morally wrong
action committed by agents who are excused from blameworthiness due to duress in or-
der to explain intuitions about cases of, say, murder or assault under duress. Although I
won’t argue for it here, I suspect these intuitions can be explained instead by a combina-
tion of uncertainty about the degree of duress that makes typically wrong actions become
permissible, and sufficient attention to the importance of the gradability of moral wrong-
ness.
-130-
theorizing about blameworthiness.
139
While these are interesting issues in
their own right, I will largely bracket them here.
Aside from the fact that the agent does something morally objectiona-
ble, my proposal is that there are no additional requirements for being an
appropriate target of blame over and above the fact that the agent’s action
is attributable to her. She need not act with ill-will, she need not possess
any (further) kind of free will, nor need she possess any special sort of
communicative or normative-competence.
3. Robust Traits and Implicit Judgments: Between Scylla and Charybdis
But blame is not merely the recognition that an attributable moral wrong-
doing has occurred. It is important to respect the fact that, as George Sher
points out, it is central to our conception of blame that blame is fundamen-
tally “a reaction to a person on the basis of the wrongness of what he has done”
in which we take “wrong acts to…reflect badly on the agents who perform
them”(7). Even a moral responsibility skeptic might allow that there are
morally wrong acts and that we are justified in reacting to the fact that

139
To give just one example, Taylor (2003) argues that hierarchical accounts of autono-
mous agency ought to be rejected given that they generally fail to account for cases of se-
vere duress since they wrongly predict that agents are still autonomous when they act
under duress. The Minimal Approval view does in a certain sense propose conditions for
“autonomous agency” but the sense of autonomy relevant to attributional-responsibility
is taken to be rather minimal. (There is a good case to be made that the fact that acting
under duress is not compatible with autonomy in its various stronger senses invoked
outside of discussions of responsibility, including medical and political contexts.) While I
take it that it is important that the complete specification of blameworthiness make room
somewhere to explain the fact that agents are not blameworthy in cases of severe duress,
it is not obvious to me that this must be done via showing how duress undercuts at-
tributable agency, given just how minimal the sense of attributable agency required for
the Minimal Approval view is. But views that posit that less minimal agential conditions
and thus stronger senses of autonomy are required for attributability may seem more ob-
jectionable if they relegate exemptions for duress to the normative domain instead of ex-
plaining them via their accounts of autonomous agency.
-131-
such actions have occurred. But this seems to fall short of genuine blame,
which goes above and beyond a mere judgment that someone has com-
mitted a wrongdoing or the admonishment of such an act. Dispassionately
telling a murderer that what she did was wrong and that she ought not do
so again in the future falls short of blaming her on a fundamental level.
When we blame her we, in some important respect, react to her on the ba-
sis of her action. The central question an account of blame needs to an-
swer, therefore, is what blaming adds over and above a judgment that
someone has acted wrongly that somehow relates her wrongdoing to a re-
action to the wrongdoer herself.
However, two straightforward ways of implicating the agent herself in
the content of blaming attitudes, via robust traits or implicit judgments,
are not available to proponents of the Minimal Approval view.
According to various traditional Deep Self views, an agent’s act is at-
tributable iff it is caused by a mental state that has an especially tight con-
nection to an agent’s practical standpoint or character. For example, Mi-
chael Bratman posits that agents’ actions must align with their planning
states, and their planning states, when taken all together, jointly constitute
an agent’s diachronic practical identity. On certain readings of Frankfurt’s
theory, such as on David Velleman’s interpretation, second-order volitions
play the role of being functionally identical to the agent herself such that
blaming attitudes directed at the initiation of an attributable action just are
blaming attitudes directed at the agent herself. Blame’s sting, on such
views, comes from the fact that one’s attributable acts express one’s deep-
est commitments and so criticism of an agent’s attributable action im-
pugns the core of her being. Sometimes this idea is coupled with the idea
that an agent’s diachronic commitments or values make up her character
traits, and so when we blame an agent due to her action, we are really
blaming her for having certain morally problematic traits, which are ex-
pressed through her action. Call this the Robust Trait view.
-132-
But given a simple correlation between attributability and attribution-
al-blame, the Minimal Approval theorist cannot avail herself of the Robust
Trait view. If an agent’s act is attributable, according to the Minimal Ap-
proval theorist, all we know is that some part of her self stands behind it,
not that who she most deeply is stands behind it. Given the way the Min-
imal Approval view handles weakness of will, it could be possible for a
person to be generally kind, and even to most strongly endorse doing the
kind thing in every scenario, but still be blameworthy for acting unkindly
in a one-off weak-willed scenario. For her to be blameworthy, doing that
unkind thing must be something of which she approved to some minimal
degree, but unkindness needn’t be a part of any larger or more defining
feature of her will.
On another view, one that is meant to be compatible with more mini-
mal conceptions of attributability, the content of blame is not the character
of the agent, but rather, the meaning of the agent’s action, which is in part
a function of the agent’s position with regard to the person doing the
blaming. As T. M. Scanlon puts it, the meaning of an action for a person is
“the significance that person has reason to assign to it, given the reasons
for which it was performed and the person’s relation to the agent.”
140
On
Angela Smith’s version of the view, which she calls the Rational Relations
view, the content of blame is the judgment of the agent taken to be implic-
it (by the blamer) in her so acting.
141
It is because Smith takes it that at-
tributable actions can reasonably be taken to reveal the judgment of the
agent that agents are answerable, or can be called upon to provide justifi-
cation for their actions. But according to the Minimal Approval view,
agents do not need to take themselves to have normative reasons to per-
form the attributable actions they perform; they may approve of them for

140
Scanlon (2008): 54.
141
Smith (2005): 17.
-133-
no reason at all. So it would be unreasonable for blamers to assume that
agents take their attributable actions to be justifiable.
4. The Ledger View
Given that the Minimal Approval view neither requires attributable ac-
tions to be expressions of robust character traits nor claims that attributa-
ble actions make it the case that others can reasonably assume that they re-
flect in some way on the judgments of the agent, there is reason to adopt a
view on which the contents of blame are taken to be much more minimal.
One candidate view is the Ledger View of blame. Ledger views hold that
what blame adds to the judgment that someone has done something
wrong is that the wrong act itself adds a ‘negative mark’ to the wrongdo-
er’s ‘moral record.’ The content of blame then, is the wrongdoing itself as
it bears on the blamed person’s overall record. A theory of attributability, then,
would give the conditions for when an agent’s action does/ does not re-
flect on her moral record.
The insight of the Ledger view that is worth preserving is that it is the
fact that the agent (attributably) committed the wrongdoing itself that re-
flects poorly on the person blamed rather than the fact that the agent’s
traits more generally align with the propensity to commit similar wrong-
doings. What should it matter to the victim of a horrific crime if the perpe-
trator was acting in a way that, while she endorsed it in the moment, does
not reflect her more general character? The Minimal Approval view can
explain why the fact that an agent acts “out of character” sometimes
seems to explain why she is exempt from blame due to the fact that acts
that she does not minimally approve of are oftentimes out of character.
But it is not some further requirement on blame that actions need be relat-
ed to past or future traits or dispositions. The Ledger view’s focus on the
attributable act’s mark against the person’s character itself better reflects
this.
-134-
Nevertheless, the Ledger view, in the ways it is usually defended, suf-
fers from a couple of serious problems. First, in focusing on a person’s
overall ‘score’ it overemphasizes the degree to which the blamed person’s
wrongdoing is blameworthy due to diminishing her moral standing in
general. When a person with a generally exemplary history of moral be-
havior commits a wrongdoing, she is still blameworthy, though her score-
card may still be much better than average. The Ledger view leaves us
with unanswered questions about why such people should be blamewor-
thy for failing to achieve overall moral perfection.
Second, our practices of holding one another responsible as they actu-
ally exist often seem quite removed from the practice of moral grading
and accounting posited by the Ledger view. As a result, this sort of view
can run the risk of distorting our interpersonally engaged social practices
to make it seem, as Gary Watson puts it, “as though in blaming we were
mainly moral clerks, recording moral faults... from a detached and aus-
terely ‘objective’ standpoint.”
142
Furthermore, even if this did provide an
apt description of our actual processes, I doubt that a system of demerits
and point-scoring could really be the institution that many of us seek to so
fiercely defend.
A suitable account of blame’s content to accompany the Minimal Ap-
proval view ought to borrow from the Ledger view the idea that an agent
may be blameworthy on the basis of evidence from an individual wrong-
doing without it necessarily revealing a larger character flaw, while aim-
ing to avoid these pitfalls of the view.
5. The Finely Individuated Trait View
5.1 A Paradigm Shift for Thinking About Aretaic Traits
In order to avoid the unappealing consequence of thinking in terms of a
person’s overall ‘score’ being the focus of appropriate blame directed at a

142
Watson (2004): 226–227.
-135-
person, it will be helpful to reconsider the relevance of aretaic traits. On
the orthodox view of aretaic traits, traits are robust character dispositions
that influence agents across a variety of circumstances. However, recent
evidence from social psychology threatens to show that robust traits as
philosophers have often conceived of them simply do not exist. The Situa-
tionist Challenge, as it’s been called, argues that our morally relevant
traits are highly contextual and influenced by morally trivial situational
factors. Since a single agent usually has evaluatively inconsistent disposi-
tions triggered by these various contextual factors, it is very rarely apt to
ascribe traits like “viciousness” or “kindness” to agents.
143

To take a commonly cited study, participants were 84% more likely to
help a woman pick up her papers if they found a dime in a phone-booth
just prior to the papers scattering. One natural conclusion, given a wealth
of similar data across other studies, is that it’s the dime finding and not
the good will that leads participants to help the woman.
144
A possible re-
sponse to this concern is to hold that we ought not utilize the concept of
traits at all in our moral practices. But does it mean that we ought not
blame a study participant for failing to help the woman (supposing it is
morally wrong not to help in the scenario)? It seems that we still can co-
herently blame the participant because she does reveal something about
her character; she reveals that she is the kind of person who wouldn’t help
a woman whose papers are scattering, at least in the case in which she has
not first found this dime. Nothing in the Situationist critique tells against
the fact that it is appropriate to have a blaming response to the fact that
someone attributively acts unkindly in a particular situation. Even if an at-
tributable morally wrong action is aptly described as being influenced by

143
See Doris (2002, 2015).
144
Although see Earp and Trafimow (2015) for the alleged replication crisis for social psy-
chology, which may cast doubt on some of the data used to bolster criticism of the exist-
ence of robust character traits.
-136-
a morally trivial situational factor, and is thus not correctly described as
stemming from a more general trait like cruelty or dishonesty, that doesn’t
mean we shouldn’t respond to the more finely individuated trait that is
evidenced just from the agent’s attributably performing that action alone.
In order to say that we blame agents for their traits in these sorts of
scenarios, we would need to consider a paradigm shift in thinking about
the metaphysics of aretaic traits. Ordinarily, more robust traits are taken
to be fundamental and to consist in dispositions to perform token actions
of a certain type. For example, viciousness is taken to consist in a collec-
tion of dispositions to, say, steal candy from a baby, ruthlessly punch
someone, con someone out of out of money, etc. Instead, I think we
should think of the paradigm examples of traits as being more finely indi-
viduated, such as: being the kind of person who would on at least one particular
occasion steal candy from a baby. Arguably, we can still appropriately use
terms like “vicious” to point to patterns of more finely individuated traits
that are similar or co-occur due to a common cause, and we can do so
without countenancing the existence of viciousness as having an inde-
pendent existence or as having the power to shape behavior more globally
across an agent’s psychology.
The orthodox way of thinking about aretaic traits obscures the fact that
we do learn something about an agent’s moral character when she at-
tributably commits a moral wrong even on a view like the Minimal Ap-
proval view with very minimal conditions for attributability. Namely, we
learn that she is the kind of person who would attributably ϕ, where ϕ-ing
is morally wrong, in the kind of circumstances in which she in fact ϕs. The
proper content of blaming attitudes, I want to suggest, are these very fine-
ly individuated aretaic traits, individuated roughly as finely as actions
themselves. While this may seem too fine to individuate traits compared
to the way they are often conceived of in the philosophical literature, I be-
lieve we do in fact blame people on the basis of quite finely individuated
traits all the time. For example, consider the following paradigmatic blam-
-137-
ing statements: “I can’t believe you are the kind of person who would do
that!” or “I didn’t think, when I first started dating her, that she was the
kind of woman who would ever say something like that to me!” While the-
se statements are focused on the character of the person blamed, their fo-
cus is quite narrow. The man who blames his wife for cheating on him
may have an interest in whether or not her act is part of a more general
propensity to be unfaithful, but it would be absurd to insist that he should
relinquish his blame entirely if he were to be presented with conclusive
evidence that it was a one-time thing.
In individuating traits this finely it might seem that the view runs the
risk of collapsing into the view that blame’s content is just the morally
wrong action itself. But if we allow that blaming involves attitudes be-
yond mere judgments about or that something has occurred, the role of
finely individuated traits becomes less obscure.
For example, on George Sher’s account, blame centrally involves a de-
sire not only for the bad action to not have occurred, but also for the agent
to have been different, although the evidence that the agent has a trait that
we blame her for is fully supplied by her attributable act itself. In blaming
it seems we want not just for the horrible thing to not have been said, but
for the agent to not have had something in her psychology that she would
countenance at all that would lead to her saying such a thing. As Sher puts
it, blame, in its most characteristic form, often seems bound up in the frus-
tration of a desire that leads “not to the generalized frustration that we
feel when we get stuck in traffic or botch a plumbing repair, but rather to
bad feelings that are directed specifically at the wrongdoer or bad person
himself.” To see why this is not surprising, Sher continues,

we need only remind ourselves of the peculiarly close connection be-
tween that person and what is wanted. When we have an unsatisfiable
desire to escape a traffic jam or fix a broken drain, we may indeed
want other people to act in certain ways…but we want this only be-
-138-
cause it would produce a farther result that does not essentially in-
volve them. By contrast, when we have an unsatisfiable desire that
someone…not have a bad character, our desire is directed at that per-
son not merely in the superficial sense that we want something that he
could bring about, nor yet in the somewhat deeper sense that we want
something that we cannot fully describe without mentioning him, but
rather in the deepest sense that we want him to have exercised his own
decision-making capacities in a certain way.
145

This kind of view makes sense of the fact that something about the agent
herself is the content of blaming attitudes, rather than the act itself.
Consider also a view on which some sort of negative behavioral re-
sponse to an agent is made appropriate due to the fact that the agent is
blameworthy. For example, blame may consist in part in the blamer’s al-
tered patterns of attention towards the blamed person and/or the revoca-
tion of charitable interpretations of the blamed person’s behavior more
generally. The licensing of these sorts of responses seems to be a feature of
the blamed agent revealing something about what she is like through her
action rather than of wrong actions themselves.

5.2 Why Blame Can Sometimes be Emotional and Impair Relationships
Recall that one criticism of the Ledger view is that it seems to advance a
picture of blame as a detached objective assignment of a demerit, and this
threatens to distort the emotional and interpersonal dimensions of blame.
While simply attending to the phenomenon of blame in real life gives us
plenty of evidence that blaming someone can be infused with emotion and
can play some role in modifying or even ending relationships, there is a
further question as to whether either of these features might be essential to
blame. But even if neither of these aspects plays an essential role in a theo-

145
Sher (2005): 105.
-139-
ry of blame, there ought to be a coherent story to tell about why these fea-
tures are often present, even if they are secondary to blame itself. The
Finely Individuated Trait view is well suited to explain these connections.
Some philosophers who aim to give an account of blame solely in
terms of fitting reactive emotions draw a sharp distinction between agent-
focused anger or resentment on the one hand and generalized frustration
on the other. According to these views blameworthy agents are defined by
the fact that they are the appropriate targets of fitting resentment. But no-
tice that the Finely Individuated Trait view is well-positioned to both help
distinguish between frustration and resentment as well as explain why re-
sentment may often be warranted when an agent is blameworthy. Anger
in its most general form tends to be a reaction to the threat of something
taken to be valuable. It is plausible that people take it to be valuable that
people who are important to them (and perhaps people in general) not
behave in certain ways. When a person attributably acts in one of these
disvalued ways, she is in a way both the person who threatened the thing
taken to be valuable as well as, in a way, the lost value itself. By locating
what is blameworthy as an aspect of the blamed agent’s self in some
sense, we can better see why agential anger, or resentment, is often pre-
sent in an episode of blaming rather than mere frustration.
In many cases these sorts of valued social norms that prohibit people
from acting in certain ways are part of the implicit contracts that make up
the bounds of our interpersonal relationships. I think we should accept the
Strawsonian idea that we attach great importance to what the actions of
others reveal about their attitudes towards us. We all have normative ex-
pectations of one another that sustain our social practices, it would seem.
We rely on the fact that most of our friends are not the kind of people who
will intentionally harm us, our wives will not laugh in our faces, and our
students will not throw paper airplanes at us while we’re lecturing.
In some but not all of these cases, our relationships are conditional on
not seeing a person in such a way and, in revealing what they are like, the
-140-
person being blamed impairs her relationship with those affected. The
frustration of the affected persons’ expectations may be grounds for with-
drawal from the relationship on their part. We think we know what kind
of people we are dealing with when we form certain kinds of bonds with
them, and when we come to learn, through their actions, that they are not
the sorts of people we took them to be, we may come to question our con-
nections with them. We needn’t agree with T. M. Scanlon’s idea that blam-
ing consists in part in taking one’s relationship to be (at least partially)
impaired by the blamed person’s action to see why via the process of
blaming people often come to see that there are impairments to their rela-
tionships.
146
Notice that our reactions in these sorts of cases are better en-
capsulated by statements like “I just can’t be friends with someone who
would say such a thing to me” than “you’ve reached 10 demerits so your
moral scorecard is too low for me to respect you as a fellow moral agent.”
The particularities of the situation matter and relate to sometimes highly
specific relationship-customized norms and expectations about partici-
pants’ characters.
5. The Time-Slice Property Objection
On the Finely Individuated trait view of blame, the contents of blame in-
volve a property of an agent that is time-specific. This leads to a couple
features of the view, which, on the face of it, might seem puzzling.
First off, even if a blamer knows that the blamed person’s action re-
veals that she has the property of being the kind of person who would ϕ given
circumstances C, the blamer might be in a position to know that the blamed
person will never again be in circumstances C. Despite the fact that after
the act the blamed person would still have the property of being such that
she would act in an immoral way in circumstances C, we might wonder

146
See Scanlon (2008, 2013).
-141-
why we should care about that when that trait has no bearing on the per-
son’s current or future actions.
Secondly, the blamed person might be immediately regretful such that
she changes after the act so that she no longer does have the trait of being
such that she would attributably ϕ in circumstances C, say because she
comes to realize that you will blame her for it. Since she no longer ap-
proves of her action, we may wonder why it would be permissible to
blame her on the basis of a time-slice property she no longer possesses.
Another way to think about the worry, specific to the Minimal Approval
View, is this. It seems that at t2 an agent can be alienated from the desire
she approved of at t1, but it is only in virtue of not being alienated from
her desire that she is blameworthy for it. Why then, should we not take
the kind of alienation she has at t2 as disqualifying her from blame at t2?
But in both cases, I want to give an initially somewhat flatfooted re-
sponse. The action is related to an agent, even one who has changed, be-
cause she is still the one who performed the action. The content of the
blame is not time-indexed; it just contains the details about the circum-
stances within its content and specifies that this time-particular trait is true
of the agent. The agent now is numerically identical, if not qualitatively
identical to the agent who performed the action, and so it is related to her
by being a part of her agential history. The affected parties still learn
something about the blamed person and it’s still perfectly reasonable in
many cases for someone to take learning that someone is the kind of per-
son who could ever be capable of doing what they’ve done to be grounds
for taking a blaming stance towards that person. Worrying that the
blamed agent is alienated from her action is misplaced, because we know
that if her action is attributable then at the time of action she was not al-
ienated from it, and so she met the ownership conditions for it. It might
make sense to speak about some kind of alienation that the agent now has
from her past action, but this is not the kind of externality-grounding al-
ienation that is relevant to responsibility. The function of inquiries about
-142-
alienation is just to find out if the agent meets the minimum conditions for
ownership over her action such that it is expressive at the time of action.
However, this might seem to make the view subject to an opposite
worry. It seems that the fact that an agent has changed for the better after
the fact does and ought to in some cases modify our stance towards her. If
an agent who does something she is blameworthy for becomes such that
she no longer would do such a thing, apologizes, and does restitution, it
seems that forgiveness becomes appropriate. On the view I have been ar-
guing for, though, given that she is still numerically identical to the per-
son with the blameworthy trait, it might seem that the view would predict
that instead, blame continues to be warranted.
But the view I have put forth should not be confused with the view
that once a morally problematic trait is revealed, blame is the one set re-
sponse that is all-things-considered best. I have only argued that the con-
tents of blaming attitudes are finely-individuated morally problematic
traits as revealed by agent’s actions. Depending on what kind of attitudes
these are, other factors may influence their appropriateness in a given cir-
cumstance.
For example, take the view that warranted blame involves the appro-
priateness of a certain kind of directed attention. One possible scenario is
that the aptness of blame gives you pro tanto reason to focus on the
wrongdoer through a certain lens, but, given her repentance, you have
more reason to focus your attention elsewhere. Another possibility is that
it may be equally permissible to blame or to withdraw one’s blame, since
the appropriateness conditions for focusing one’s attention plausibly in-
volve lots of cases in which it’s permissible to go in any one of several dif-
ferent ways.
Similarly, an emotion-based view of blame can, arguably, make sense
of cases on which, for example, there is reason to be mad at a person on
the basis of their having exemplified a certain trait, but there is also coun-
tervailing reason to feel some other emotion that is mutually-exclusive
-143-
with remaining angry. In general, it should be emphasized that describing
the content of blaming attitudes does not yet settle an ethics of blame.
I admit, however, that it is hard to see how countervailing considera-
tions of an agent’s post-action transformation could act to mitigate blame
on the basis of finely individuated traits given a view on which blame
consists merely in beliefs or judgments of some sort. On a crude version of
the view on which blaming just is the judgment that someone has a moral-
ly objectionable trait, and traits attach to persons across time, then blame-
worthy people seem inexorably doomed to lives on which it is once and
forever all-things-considered appropriate for others to blame them. It is
possible that more sophisticated accounts of the judgment view, when
paired with the Finely Individuated Trait view, might be able to concoct
ways to avoid this problem. However, it may be more promising to pair
the Finely Individuated trait view with a different account of the kinds of
constitutive attitudes of blaming.
6. The Minority Report objection: Pre-Blame
In the Stephen Spielberg film Minority Report, fortune-tellers known as
pre-cogs are able to apprehend would-be criminals by foreseeing that they
will commit a crime. The idea that someone ought to be punished on the
basis of a crime without actually committing the crime (yet) is, on the face
of it, unappealing. Certain compatibilist views have been accused of hav-
ing the unintuitive consequence of allowing, at least in theory, for the pos-
sibility of morally sanctioned pre-punishment.
While the Finely Individuated Trait view is not a theory of the ethics of
punishment, it does face an analogous problem: what I’ll call the problem
of “pre-blame.” Suppose that neuroscientists were able to conclusively
show that someone is the kind of person who would do some morally
wrong thing, say Ψ-ing, in a set of circumstances, but it is a set of circum-
stances that she is not yet in, or perhaps will never be in. According to the
Finely Individuated Trait view, since such a brain scan or other neurologi-
-144-
cal test seems to reveal the relevant information about the person, it seems
that she would be attributionally-blameworthy for being the kind of per-
son who would Ψ in a particular set of circumstances, despite the fact that
she has not actually Ψ-ed. This seems counterintuitive, presumably be-
cause we tend to think that it is only through a person’s actual action that
we can rightly blame her. While my strategy to respond involves biting
the bullet, I want to show that focusing on several possible factors that
drive its underlying intuition can diffuse the force of the counterintuitive
consequence.
6.1 Actual Action as Usual Evidence
In real life as we know it, the only completely decisive evidence we are
ever given that someone has such a finely individuated trait comes via
their action. Given the current state of science, it is not possible to defini-
tively prove that someone is the kind of person who would ϕ in some
maximally specified set of circumstances. That means that the only time
we have conclusive evidence that someone is the kind of person who
would ϕ in circumstances C is when we know that that person attributive-
ly attempts to ϕ in circumstances C. Given that these things always over-
lap in the actual world, it’s not surprising that we would come to the con-
clusion that it is only appropriate to blame someone when she actually ϕs.
We have deeply embedded norms, perhaps supported by principles of
morality, about not assuming the worst about someone and not acting as
though we know how they will act before we have conclusive evidence of
it. Despite the fact that we would have conclusive evidence if we were able
to scan people’s brains, according to the objection, it’s possible that our in-
tuitions are nevertheless influenced by the fact that we are so used to the
only conclusive evidence of the relevant kind of trait being an actual ac-
tion.
We don’t know what it would be like to live in a world in which we
could tell exactly what someone would do before they have done it, and
-145-
there is just a general strangeness in imagining the scenario which may
end up pervading all of our intuitions about it. Imagine we reject the view
that it is appropriate to blame someone for their finely individuated trait
in favor of a view on which her action must actually occur. Whatever kind
of attitude we take blame to consist in, it is strange to think that in a world
in which we knew exactly what was going to happen we would be able to
entirely withhold that attitude until the action took place. Once we know
that someone will commit a horrific murder we will resent them, make
certain judgments about them, desire that they are otherwise, potentially
want them to suffer or be punished, have reason to modify our relation-
ship to them, etc. These cases in which we know what someone will do be-
fore they actually do it just seem to lead to seemingly unpalatable conse-
quences about blame no matter whether we hold that people in such a sit-
uation can be blamed before their actions or only afterwards. This is rea-
son not take a counterintuitive consequence of what a view posits about
pre-blame to be a decisive objection.
6.2 Victim Relation and the Standing to Blame
Take another case in which the Finely Individuated Trait view seems to
predict that it is appropriate to blame someone that might seem unintui-
tive.

Trapped Tara: Tara wakes up one morning and declares that she is go-
ing to murder someone, and you have every reason to believe that, if
given the opportunity, she would follow through on her declaration.
Little does Tara know, though, that while she was asleep last night her
bedroom was air-lifted to a deserted island with her in it, so no mur-
dering will be possible. Escaping from the island, let’s suppose, is also
impossible.

-146-
In this case you are given good evidence that Tara has the finely individu-
ated traits of being such that she would murder someone in many circum-
stances in which she has people available to murder. However, you also
know that she will never be in those circumstances. The Finely Individu-
ated Trait view says that blaming Tara would be apt, but this may seem
somewhat unintuitive. After all, Tara didn’t actually do anything wrong,
and she hasn’t harmed anyone.
But I think our reaction is not so much that it’s unfair to blame Tara,
but rather that it’s not so clear why we should bother with blame in this
case. In particular, part of the intuition that blaming Tara is inappropriate
seems to derive from the fact that it seems that some of the standard ways
of outwardly expressing blame would be inappropriate in a case in which
no wrongdoing was committed. But it is standard to draw a prima facie
distinction between the permissibility of blame and the permissibility of
outwardly expressing blame or confronting the blamed person. The latter
might take into account additional moral reasons one might have to ex-
press or refrain from expressing blame, and it also might take into account
the standing of the blamer.
147

It is controversial just what conditions give a person the standing to
express blame, but three commonly cited factors thought to undermine
one’s standing to blame are complicity, hypocrisy, and meddling.
148
One
suggestion as to what might unify these disparate seeming conditions is
that standing is at least in part a function of the blamer’s relationship to
the victim of the wrongdoing.
149
When the blamer is complicit, she is par-
tially at fault for the victim’s situation; when the blamer is hypocritical,

147
Although, one might also need certain standing to appropriately blame in the unex-
pressed sense. This may seem more or less plausible depending on the account given of
blaming attitudes and their contents. It is less controversial that standing matters for
outward expressions of blame than for blame simpliciter.
148
See Coates and Tognazzini (2013).
149
See, for example, Bell (2013).
-147-
she cannot honestly align herself with the victim; and when she is med-
dling she lacks the requisite connection to the victim.
If there’s no possibility of Tara actually committing the wrongdoing,
there is no real victim. Even if the content of blame strictly speaking
doesn’t require a victim, merely the existence of a trait, the permissibility
of expressing it might be mitigated by the absence of a victim. It might be
inappropriate to outwardly direct blame at Tara when there is no one she
has actually harmed. To a lesser degree, the appropriateness of expressing
blame might be mitigated in a similar way in pre-blame cases. While there
is a future victim, there is no current victim, and that might alter the ap-
propriateness of expressing blame. Since it can be difficult to untangle in-
tuitions about the appropriateness of blame from the appropriateness of
expressing blame, this is one further reason that we should give pause to
putting a lot of stock in the seeming counterintuitiveness of appropriate
pre-blame.
6.3 Stubborn Incompatibilist Intuitions
The intuition that the permissibility of pre-blame is unacceptable might al-
so stem in part from sticky incompatibilist intuitions, which, methodolog-
ically speaking, we shouldn’t permit at this stage to provide a knock-
down objection to the view. Libertarians will hold that it isn’t really possi-
ble to have 100% certainty about what someone will do, even in a world
that has access to the most advanced possible scientific discoveries. In an
indeterministic world, therefore, anytime someone predicts the future, the
prediction can only ever be a confident guess. If a fortuneteller predicts
that a person will act in an immoral way, it is, however unlikely, within
that person’s power to overcome the situationally-influenced factors and
change courses due to a pure exercise of the will. It’s possible that what
rubs some of us the wrong way about these scenarios in which a person is
pre-blamed is that deep down we think blaming someone on the basis of a
-148-
prediction is unfair since she might not act in the way she is predicted to
act. If so, then we have clearly failed to isolate the appropriate intuition.
Of course, an incompatibilist might use the fact that we seem unable to
move away from these kinds of intuitions that tell against the permissibil-
ity of pre-blame as evidence that we have unshakable incompatibilist intu-
itions. This could be bolstered into an argument that the compatibilist pro-
ject has gone awry at some point. However, it is part of the nature of the
dialectic that we do have stubborn intuitions that pull us in favor of liber-
tarianism, but that we also have stubborn intuitions favoring aspects of
compatibilism and skepticism as well. All three of the following state-
ments have intuitive appeal, and yet they are jointly incompatible: “We
are morally responsible” “Being morally responsible requires having mul-
tiple paths available at the moment of choice.” “Determinism is true, and
rules out responsible action.” The point I want to make here is just that in-
sofar as our intuitions about cases may stem from intuitive support for
one of these general ideas, we have reason to at least be cautious about us-
ing them to rule out views.
6.4 Denying the Existence of Resultant Moral Luck
I have, so far, acknowledged the fact that it is somewhat counterintuitive
that pre-blame would hypothetically be appropriate on the Finely Indi-
viduated Trait view, though I have argued that we ought to have a meas-
ured response to its counter-intuitiveness. But it should also be noted that
this same consequence of the view actually helps explain our intuitions
about another issue: resultant moral luck.
It seems that, in the actual world, it is appropriate to treat a successful
murderer and a murderer whose plan is thwarted by something wholly
outside of his control differently. This is puzzling, though, since the only
real difference in the two cases is something that has nothing to do with
-149-
any features of the two individuals.
150
But an adherent of the Finely Indi-
viduated Trait view could make use of what is known as the epistemic ar-
gument for denying the existence of resultant moral luck. In other words,
the Finely Individuated Trait view is consistent with and even helps bol-
ster a powerful argument that, despite appearances, it really is not appro-
priate to treat successful and (un)luckily unsuccessful murderers differ-
ently when all else is held equal.
As I have been arguing, the epistemic argument against resultant mor-
al luck holds that the reason we often treat the two cases differently is be-
cause in the real world we rarely know the strength of someone’s com-
mitment to undertaking a certain course of action unless we have evi-
dence from the fact that they actually went through with it.
151
As Dana
Nelkin explains,

Thus, rather than indicating our commitment to cases of resultant
moral luck, our differential treatment of successful and unsuccessful
murderers indicates our different epistemic situations with respect to
each. If we were in the unrealistic situation of knowing that both
agents had exactly the same intentions, the same strength of commit-
ment to their plans, and so on, then we would no longer be inclined to
treat them differently.
152

The Finely Individuated Trait view helps explain why we ought to treat
the two the same if we were in the same epistemic circumstances: what

150
Discussion of these issues in the contemporary literature tends to center on the treat-
ment of the issue provided in Williams and Nagel (1976), although discussion also ap-
pears earlier in Feinberg (1962). For one recent treatment of the issue as it pertains to
blameworthiness and praiseworthiness, see Hartman (2017).
151
See Richards (1986), Rescher (1993), Rosebury (1995), and Thomson (1993).
152
Nelkin (2013).
-150-
matters is the fact that the person has the quality of being such that she
would ϕ in circumstances C.
Countenancing moral luck is generally taken to be a problem of some
sort, and so it is meant to be counterintuitive that moral luck should make
a difference. Since something like the Finely Individuated Trait view is
needed to bolster a crucial argument that moral luck doesn’t make a dif-
ference, this should count in its favor. The same aspect of the view that
under one lens looked counterintuitive, also has strong intuitive appeal
when focusing on issues of resultant moral luck. Thus the problem of pre-
blame does not give us anywhere near conclusive reason to reject an ac-
count of blame’s content that seems as promising as the Finely Individu-
ated Trait view.

-151-
Chapter 5: The Case of Forgetting

1. Introduction
In the previous four chapters I developed the Minimal Approval account
of attributional-responsibility. I suggested that an agent can be attribu-
tionally-responsible for acting when the production of her action meets
fairly minimal ownership conditions. In Chapter 3 I showed how this ac-
count explains why agents are attributionally-responsible for weak-willed
actions: unlike in cases of compulsion, agents minimally approve of their
weak-willed actions. Given this explanation, it seems right to say that
weak-willed agents are also derivatively attributionally-responsible for
failing to do what they would have done were they to have been strong-
willed.
In this chapter I consider the prospects for extending a similar line of
thought to account for cases in which agents seem responsible for forget-
ting to do something they ought to have done. I argue that this move is
not as appealing as it might seem, and, in addition, I argue that similar at-
tempts to expand the notion of attributional-responsibility by other theo-
rists to account for these kinds of cases are misguided.
Instead, I argue that agents are responsible in a non-appraising, role-
responsibility sense for these sorts of forgettings, but they are not attribu-
tionally-responsible. This sets an important limit on the project of the dis-
sertation as a whole; while I argue that blaming practices that center
around finely individuated traits as expressed by actions of which agents
minimally approve play a crucial and significant role in our moral respon-
sibility practices, they are not exhaustive of our responsibility practices.

-152-
2. The Minimal Approval View and The Case of Forgetting
Consider the following case from George Sher’s book, Who Knew? Respon-
sibility Without Awareness:

Alessandra, a soccer mom, has gone to pick up her children at their el-
ementary school. As usual, Alessandra is accompanied by the family's
border collie, Bathsheba, who rides in the back of the van. Although it
is very hot, the pick-up has never taken long, so Alessandra leaves
Sheba in the van while she goes to gather her children. This time, how-
ever, Alessandra is greeted by a tangled tale of misbehavior, ill-
considered punishment, and administrative bungling which requires
several hours of indignant sorting out. During that time, Sheba lan-
guishes, forgotten, in the locked car. When Alessandra and her chil-
dren finally make it to the parking lot, they find Sheba unconscious
from heat prostration.
153

Assuming that Alessandra is a generally caring and thoughtful person
who loves Sheba, it is not easy to get clear on our intuitive responses to
such a case. On the one hand, we might feel bad for Alessandra for mak-
ing such an upsetting mistake, one that does not on the face of it seem to
reflect any sort of characteristically morally wrong personality traits. On
the other hand, she seems to have acted negligently—we think she should
still be held responsible in some sense for what happened. To bring out
this intuition, imagine that Alessandra were to offer no apology to her
children and family for what she had done.
154
This, I think, would seem

153
Sher (2009): 24.
154
Although, perhaps apology is warranted even in cases that involve no responsibility.
See, for example, Talbert (Forthcoming): 17, and Scanlon (2008): 150. Perhaps a similar in-
tuition could be brought about, though, by considering the case in which Alessandra
feels no remorse or special duty to comfort others affected as a result of her causal re-
sponsibility.
-153-
highly inappropriate. While her family might rightly be sensitive to the
fact that Alessandra herself might be suffering from the tragedy of the in-
cident, we might not think them out of line to expect her to make amends
of some sort. The verdict here is unclear, but there is at least a prima facie
case for Alessandra’s being responsible in some way.
What does the Minimal Approval view say about Alessandra’s neglect
of Sheba in this case as it pertains to her attributional-responsibility? It
might be thought that the Minimal Approval view actually has resources
that other views in the Deep Self family lack to show why Alessandra is,
after all, attributionally-responsible for leaving Sheba in the car. In Chap-
ter 3, I argued that agents could be attributionally-responsible for acting
out of weakness of will. When Sam acts out of weakness of will and ends
up going to a party rather than studying for her exam, she is attributional-
ly-responsible for going to the party. It makes sense in these cases to also
say that she is derivatively attributionally-responsible for not studying. If
being attributionally-responsible for what you actually do can render you
attributionally-responsible for what you fail to do as well, then we might
think we should conclude from the fact that Alessandra is (let’s stipulate)
attributionally-responsible for staying to talk to the school administrators,
that she is also attributionally-responsible for leaving Sheba in the car.
In order to evaluate whether or not this is right, though, we’ll need to
pay more attention to the principle that lets us move from Sam’s attribu-
tional-responsibility for her action to her attributional-responsibility for
her omission. One principle that might get us from Sam’s responsibility
for going to the party to Sam’s responsibility for failing to study is the fol-
lowing:

Modal Bridge Principle: If an agent, A, is attributionally-responsible
for ϕ-ing-at-t, then A is also attributionally-responsible for not ψ-ing,
where ψ-ing is anything else A could have done at t.

-154-
Adopting this principle would also give the result that Alessandra is re-
sponsible for not tending to Sheba, since, arguably, she could have tended
to Sheba rather than stayed to talk to the administrator. But notice that
adopting this principle reopens the issue that Deep Self views were par-
tially created to avoid. Given the fixity of the past and the truth of deter-
minism, Incompatibilists will argue, there is nothing that Alessandra
might have done at t other than what she actually did—stay and talk to
the administrator. Given the hard work of maneuvering around these
questions and arguing for their irrelevance in the case of attributionally-
responsible action, it would be unfortunate if they were simply to reap-
pear in the case of attributionally-responsible omission.
There is also a larger problem with this view. Consider the following
case:

Oblivious Ollie: Ollie is walking to work and meets the conditions of
Minimal Approval for walking to work. Unbeknownst to him, there is
a small child drowning in the river behind him, but he never turns
around and notices the child drowning. Is Ollie attributionally-
responsible for failing to turn around and help the child? It would
seem that he should not be, but given a plausible (compatibilist) con-
strual of the Modal Bridge principle, he is attributionally-responsible
for walking to work, and could have instead turned around and saved
the drowning child.

So perhaps, instead, we ought to adopt a principle like the following:

Evidence Bridge Principle: If an agent, A, is attributionally-responsible
for ϕ-ing-at-t, then A is also attributionally-responsible for not ψ-ing-
at-t where ψ-ing is any act that the agent has sufficient evidence is a
potential course of conduct for her.

-155-
This principle both correctly exempts Ollie from attributional-respons-
ibility, and, at least on the face of it, might even avoid the incompatibilist
worry. But consider the explanation it gives as to why Alessandra is re-
sponsible for leaving Sheba in the car. She is responsible for talking to the
school administrators when it is the case that she also had evidence (to
which she failed to attend) that she could go tend to Sheba to avoid catas-
trophe at that moment instead. But what role does the existence of the ev-
idence to which Alessandra failed to attend play in explaining why she is
responsible? Alessandra may fail to attend to the evidence due to a ran-
dom misfiring in her brain, and so the fact that this evidence existed
doesn’t tell us anything about what Alessandra is like agentially. It may be
true that it tells us that a surface-level normative fact is true about Ales-
sandra: there was evidence that an alternative course existed, evidence to
which she ought to have been responsive. But grounding the explanation
of an ascription of attributional-responsibility in a normative fact like this
would require accepting a sort of deep foundational asymmetry between
attributability for acts and omissions. Since attributability for actions on
the Minimal Approval view is determined by pure metaphysical condi-
tions of agency, it would be odd if first-order normative theorizing needed
to take place in order to determine the conditions for attributability for an
omission.
Instead, I think the Minimal Approval view should be coupled with a
principle that ties attributional-responsibility for omissions more closely
to the fact that the mechanism of the agent’s action relates to the fact that
the agent would approve to some degree of what she does do instead of
the omitted action. This will explain why Sam is attributionally-
responsible for failing to study, but will have the result that Alessandra is
not attributionally-responsible for failing to tend to Sheba. This is the
principle I think the Minimal Approval theorist ought to adopt:

Contrastive Approval Bridge Principle: If an agent, A, is attributional-
-156-
ly-responsible for ϕ-ing-at-t, then A is also attributionally-responsible
for not ψ-ing-at-t, when ψ-ing is an alternative considered in the
worlds in which A meets the conditions of minimal approval.
155

Sam is attributionally-responsible for not studying, not because she could
or should have studied, but because of a fact about her agency. Sam min-
imally approves of going to the party even though she knows that she
could be studying, and this is related to what leads her to go to the party.
Notice that, given this bridge principle, Sam’s responsibility for her omis-
sion actually has something to do with her approval. She’s responsible for
going to the party because she minimally approves of it, and she’s respon-
sible for not studying because she minimally approved of doing some-
thing else instead. Her responsibility for her omissions comes from the fact
that we know something about what she is like when she is faced with the

155
What grounds these bridge principles? One possibility is that they are derived from
principles that would explain why agents are responsible not just for their actions, but al-
so for the consequences of their actions. On this strategy, the fact that an agent does not ψ
instead when she ϕs is just a consequence of ϕ-ing like any other. Different principles of
this form correspond with the modal, normative, and contrastive approval bridge princi-
ples. For example:
Modal: If an agent, A, is attributionally-responsible for ϕ-ing-at-t, then A is also attributionally-
responsible for all of the consequences of ϕ-ing that she could have foreseen, (possibly including
the fact that A will not ψ instead).
Evidence: If an agent, A, is attributionally-responsible for ϕ-ing-at-t, then A is also attributional-
ly-responsible for all the consequences of ϕ-ing that she had enough evidence to foresee, (possible
including the fact that A will not ψ instead)
Normative: If an agent, A, is attributionally-responsible for ϕ-ing-at-t, then A is also attribu-
tionally-responsible for all of the consequences of ϕ-ing that she should have foreseen (possibly in-
cluding the fact that A will not ψ instead).
Contrastive Approval: If an agent, A, is attributionally-responsible for ϕ-ing-at-t, then A is also
attributionally-responsible for all of the consequences of ϕ-ing she foresees in the worlds in which
she reflects on which of her motivations to act on at t and approves of ϕ-ing, (possibly including
the fact that A will not ψ instead).
-157-
choice situation in which the option of studying features—there’s a part of
her that wants to go to the party anyway.
The same can’t be said of Alessandra. Her minimal approval of talking
with the administrator doesn’t tell us anything about what she cares about
with respect to Sheba, since considerations about Sheba aren’t among
those that would factor into her agential psychology whatsoever when she
so approves.
Note that Alessandra’s failure to think about Sheba’s being in the car is
plausibly one of the things we ought to hold fixed when assessing wheth-
er or not she has the disposition required for being attributionally-
responsible for talking to the administrator according to the Minimal Ap-
proval view. To see why this should be so, notice that people are plausibly
attributionally-responsible for ϕ-ing even if there are other options that
they do not remember that would silence their desires to ϕ were they to
come up with those options. If I had remembered that I had spaghetti in
the pantry I would never have given any weight to a desire to go to the
grocery store, but this doesn’t mean that I’m not attributionally-
responsible for going to the grocery store.
Alessandra never considers Sheba, and so there is no bridge that
shows that just because she is attributionally-responsibility for talking to
the administrator, she should be attributionally-responsible for forgetting
Sheba. This leaves us needing to look elsewhere to explain the intuition
that Alessandra is in some sense responsible for failing to tend to Sheba.
3. The Tracing Strategy
One possible way to show that Alessandra might, after all, be attribution-
ally-blameworthy for failing to tend to Sheba would be to use the tracing
strategy.
The tracing strategy is used to explain why agents are responsible for
certain acts that are not directly attributable to agents. According to pro-
ponents of the tracing strategy, an agent may also be attributionally-
-158-
blameworthy for acts that should have been and could have been prevent-
ed by an earlier act, when the agent is attributionally-blameworthy for the
prior omission. For example, an out of control drunk driver who hits an-
other car may not meet the attributability requirements for hitting the car,
but nevertheless be blameworthy. Given the tracing strategy, we can ex-
plain her attributional-blameworthiness for her behavior via the fact that
she is attributionally-blameworthy for her earlier acts of getting intoxicat-
ed and choosing to drive rather than catch an Uber to and from the bar.
Note that the agent must be blameworthy for the prior action in order for it
to explain why the agent is blameworthy for the current action. It cannot
merely be that the person made a choice in the past that lead to the bad
outcome. If a mother decides to pick her child up from school rather than
have him take the bus home, and on the way home the child gets injured
in a car accident due to no fault of the mother’s driving, she does not be-
come blameworthy just because she made the prior choice to pick him up
by car.
While some have voiced concerns with the viability of the tracing
strategy in general, for the sake of argument I’ll grant that the tracing
strategy, generally speaking, works, and provides a good explanation for
a broad range of cases of blameworthiness.
156
The question then is whether

156
Tracing strategies are more often invoked by reasons-responsiveness and control-
based theorists of responsibility. While the tracing strategy is in principle open to de-
fenders of the Deep Self view, defenders of Deep Self views sometimes couple their theo-
ries of attributability with defenses of the H-Tradition of responsibility and aim to defend
a Strawsonian quality-of-will thesis on which blameworthy agents’ actions must express
ill-will. It’s not clear that a tracing strategy is compatible with this sort of view. Although
I’ve developed the Minimal Approval account as B-Tradition account of responsibility,
interestingly, the hypothetical nature of the reflectiveness required for Minimal Approval
might make the tracing strategy somewhat less necessary in certain standard cases, and
so adopting it might help quality of will theorists avoid the tracing strategy. Another re-
cent set of objections to the tracing strategy concerns certain articulations of the strategy’s
compatibility with reasonable epistemic conditions on responsibility. For this line of ar-
-159-
or not a Minimal Approval theorist could use the strategy to explain the
intuition that Alessandra is attributionally-blameworthy.
I think such an explanation is bound to fail. There are things that Ales-
sandra could have done that would have made it so that she didn’t forget
Sheba in the car. She could have tied a string around her finger or set a
phone alarm reminding her that the dog was in the car, or she could have
avoided bringing the dog altogether. She failed to do those things, but in
order to make it the case that she is attributionally-blameworthy for leav-
ing Sheba, we would also need to know that she is attributionally-
blameworthy for those prior omissions. Given the reasonable assumptions
that she would be no more than a minute or two picking up her kids, that
she doesn’t have any general tendency to forget important things when
her plans get modified, and doesn’t have any particular tendency to forget
about Sheba, it’s just not clear that she would be attributionally-
blameworthy for not setting a phone alarm.
Using just our intuitive notion of attributional-blameworthiness, it is
just not clear that Alessandra does anything that reflects a bad trait prior
to her leaving Sheba in the car. And using the Minimal Approval view
combined with the Contrastive Approval Bridge Principle doesn’t seem to
give us the result that there is anything that she is attributionally-
blameworthy for either. In order for her to be derivatively attributionally-
blameworthy for leaving Sheba in the car due to a tracing explanation, it
would have to be the case that in every specification of the story that gen-
erates the intuition that Alessandra is responsible for leaving Sheba in the
car, she is also such that she is blameworthy for minimally approving of
what she was doing instead of setting a phone alarm, or blameworthy for
minimally approving of what she was doing instead of thinking about
ways to keep herself vigilant about Sheba’s wellbeing.

gumentation, see Vargas (2005), and Shabo (2015).

-160-
Furthermore, I don’t think tracing can give us the right kind of explana-
tion for our intuition that Alessandra is responsible in some sense. Take a
case in which Alessandra has complete knowledge and full awareness of
the fact that there is a 1/10,000 possibility that she will leave Sheba in the
car at some point. If Alessandra were to fail to set a phone alarm, given
the other demands of her busy life as a mother, I think few would blame
her for this. Nevertheless if she were to in fact leave Sheba in the car I
think this would still elicit our reaction that she owes some form of rec-
ompense to her family in light of leaving the dog in the car. Even when
she has done nothing wrong in the past, we still think she ought to be held
responsible in the present.
4. An Argument Against the Minimal Approval View as the Correct
View of Attributability?
Although I have presented the Minimal Approval view as a view that is
robust across a wider variety of different causal stories than traditional
Deep Self views, cases like Alessandra’s open up the possibility of the ob-
jection that it is not robust across a wide enough variety of causal stories.
One such challenge comes from proponents of a loosely associated set of
views sometimes called New Attributionist views.
157
In this section I raise
the possibility that the case of Alessandra could be leveraged into an ob-
jection that agents are in fact attributionally-responsible for an even
broader range of acts and omissions than what the Minimal Approval
view can account for. I consider George Sher’s and Angela Smith’s views
as offering competing views of attributional-responsibility that give the
result that Alessandra is attributionally-responsible for leaving Sheba in
the car, but argue that neither account succeeds at this task.

157
New Attributionists include George Sher, Angela Smith, Matt Talbert, and T. M.
Scanlon. For a discussion of this term and its application, see Talbert (2016). Matt Talbert,
however, explicitly takes the view that Alessandra is not responsible [see Talbert (Forth-
coming)].
-161-
4.1 Sher’s Agent-Constituting Mechanism View
On Sher’s account, agents need not approve of their motivations in any
sense in order to be attributionally-responsible for them. The criterion for
attributional-responsibility on his view is instead just that the action or
omission in some way stems from the constitutive features of the agent
qua the particular rational agent she is. He takes these features to include
not just the agent’s cares, values, and plans but also her general disposi-
tions and tendencies, as well as the relevant neurophysiological mecha-
nisms that cause such things.
158
In this way, he is able to hold that agents
may be attributionally-responsible for their actions and omissions even if
they are not endorsed by the agent in any way, so long as they are caused
by one of the kinds of mechanisms that are involved primarily in agency.
In Alessandra’s case, is the fact that she leaves Sheba in the car caused
by her constitutive agential features? According to Sher, it is. Things like
“her concern for her children, for example, or her tendency to focus in-
tensely on whatever issue is at hand” are just the sorts of things that make
her the agent that she is. And even though Alessandra would never ap-
prove or judge it best to leave Sheba in the Car, according to Sher, “…we
must locate the significance of Alessandra's failure to remember Sheba not
in what it reveals about her judgments about reasons, but rather in its be-
ing caused by the same psychophysical structure that sustains her ability
to make such judgments.”
159
And so, he thinks, we should conclude that
she is in fact attributionally-responsible for her omission.
I’d like to echo a general line of criticism taken up by Angela Smith
and others to Sher’s larger account, though.
160
The problem is that, while
he articulates a promising-sounding necessary condition for attributional-
responsibility, Sher fails to provide a sufficient condition, since a mere

158
Sher (2005): 122.
159
Sher (2009): 131.
160
See, for example, Smith (2008), and Mason (Provisionally Forthcoming).
-162-
causal connection between these kinds of structures and action is intui-
tively not sufficient to make an agent attributionally-responsible. Take the
following example: Imagine a case in which an agent has the quality that
she thinks really really hard about her decisions just about all the time.
But this same feature of her agency sometimes causes her brain, at other
times, to short out due to being in overdrive, making her collapse in place.
Suppose that she does not know that these incidences are related. One day
she collapses and falls into someone on the way down, injuring this per-
son. Pretty clearly, this agent is not attributionally-responsible for her fall-
ing, and yet her behavior is, it seems, caused by one of the features that
makes the agent who she is qua rational agent.
Perhaps Sher could tell a story about how this behavior is not caused
in the right sort of way by her tendencies, but it is unclear what that story
could amount to without changing the extension of the view pretty signif-
icantly. The broader lesson here is that the criterion for attributional-
responsibility can’t just be that the behavior is a side effect of the agent’s
rational-agent tendencies, as Sher would have it.
4.2 Smith’s Rational Relations View
In a series of articles, Angela Smith defends the Rational Relations view of
responsibility according to which, broadly speaking, if an agent’s behavior
displays a lack of rational concern, it is appropriate to hold that agent re-
sponsible for the behavior.
161
Her view is generally taken to be an account
of attributional-responsibility that can accommodate the fact that agents
like Alessandra are attributionally-responsible for things like forgetting.
However, I will show that depending on the interpretation of Smith’s
view, it either does not give the result that Alessandra is responsible, or is
not really a theory of attributional-responsibility.
I’ll start with the first horn of the dilemma. It is understandable that
Smith’s view would generally be taken to be able to give an account of

161
Smith (2005, 2008, 2012, 2015).
-163-
Alessandra’s attributional-responsibility, since Smith’s own paradigm
case of responsibility that she uses to motivate her view is very similar to
the Alessandra case, and Smith seems to explain responsibility in this case
as being grounded in a revealed lack of concern via an omission. This cer-
tainly sounds like attributability language. In Smith’s case, she forgets a
friend’s birthday and the friend takes her to be morally responsible for her
forgetting, as her dispositions and patterns of attention reasonably convey
to her friend a lack of care.
Smith describes the birthday-forgetting case in the following way:

I did not intend to hurt my friend’s feelings or even foresee that my
conduct would have this effect. I just forgot. It didn’t occur to me. I
failed to notice. And yet, despite the apparent involuntariness of this
failure, there was no doubt in either of our minds that I was, indeed,
responsible for it. Although my friend was quick to pardon my
thoughtlessness and to dismiss it as trivial and unimportant, the act of
pardoning itself is simply a way of renouncing certain critical respons-
es, which it is acknowledged, would, in principle, be justified.
162

But does Smith’s forgetting her friend’s birthday reveal an objectiona-
ble degree of concern about her friend? Certainly it might. Someone self-
absorbed or who undervalued her friendship might make this mistake
frequently precisely for the reason that she failed to cultivate the appro-
priate level of concern. But it is not obviously true that any such descrip-
tion is true of Smith in this case. It might be that perhaps a better friend
would focus her attention so intently on the birthdates of all her friends
with the special intent to ensure this never happens. But if Smith has no
special reason to worry that she’ll forget, behaving this way would seem

162
Smith (2005), 236.
-164-
to be supererogatory. It’s possible, it seems, to have an appropriate level
of concern for one’s friends without doing this.
Commenting on the preceding passage from Smith, Matt Talbert says
the following:

Though Smith doesn’t say that she is blameworthy here, it’s a natural
way to read the case and it’s easy to imagine a person in the position of
Smith’s friend responding to her with the negative attitudes involved
in moral blame. I take Smith’s view to be that if blame is appropriate in
this case, then what makes it appropriate is just that in forgetting her
friend’s birthday, Smith reveals something objectionable about her ori-
entation toward her friend.
163

Presumably Talbert’s reading of Smith would also apply to Alessandra’s
case. If Alessandra’s forgetting did display objectionable lack of concern
for Sheba, then this would be sufficient for appropriate blame. The ques-
tion then becomes, in Alessandra’s case: is blame, after all, appropriate in
the particular case due to objectionable lack of concern for Sheba?
I agree with Talbert that if it’s true that we can tell the story such that
Alessandra more generally cares a great deal about Sheba then we should
conclude, on an account like Smith’s, that leaving Sheba in the car does
not reveal anything objectionable about her orientation towards the dog,
and thus Alessandra is not attributionally-responsible for her omission.
And so we are left with cases on which it seems intuitive that Alessandra
is responsible in some sense, despite the fact that leaving Sheba does not
reveal anything objectionable about her level of concern for Sheba. Again,
the only way of avoiding this outcome for Smith would be to argue that
the very fact of Alessandra’s omission makes it the case that it is impossi-

163
Talbert (Forthcoming).
-165-
ble for her to have sufficiently cared about Sheba. But this seems implau-
sible.
As Talbert points out, the idea that anytime someone leaves a dog in
the car this shows an objectionable attitude of lack of care must be mistak-
en, because were an extraordinary circumstance to occur, even a high level
of concern for Sheba would not be enough to prevent forgetting her in the
car. But perhaps we could modify the view to something like, in the ab-
sence of extraordinary intervention (and presumably Alessandra’s discus-
sion with administrators should not count as extraordinary) if Alessandra
had properly cared for Sheba she wouldn’t have left her in the car. But, as
Talbert is I think right to point out,

Even under normal circumstances, people sometimes forget or fail to
notice things about which they care very much. More generally, I take
it that even under normal circumstances, what we notice, what we re-
member, what fails to occur to us, and so on, doesn’t necessarily indi-
cate what we value or how much we value it.
164

It might be that forgetting a birthday or forgetting a dog in the car general-
ly indicates that a person has an objectionable lack of concern, but this is
not sufficient evidence that a lack of concern is actually attributable to any
particular agent. This is why it’s crucial in an account of attributability to
have some condition that makes reference to the agent’s psychology, not
just the agential psychology that her action/omission makes us warranted
in believing might be the case. As a theory that can explain why Alessan-
dra’s responsibility for leaving Sheba must be grounded in some trait that
is attributable to Alessandra, Smith’s theory falls short.
This leads to the other horn of the dilemma, though. It seems to me
that Smith might be able to agree that Alessandra is responsible even in

164
Talbert (Forthcoming): 16.
-166-
the cases in which her forgetting doesn’t indicate an objectionable lack of
concern for Sheba. In fact, in various places it seems as though Smith con-
ceives of the connection between our tendencies and our evaluative judg-
ments to be that our tendencies merely give those around us prima facie ev-
idence of our evaluative judgments such that we are required to answer
for our tendencies—to either own up to the fact that they do reveal what
we are like in a given instance or, crucially, to explain how in the particu-
lar case they came about in some way that shows that they did not stem
from our lack of concern. Smith writes that:

If one judges some thing or person to be important or significant in
some way, this should (rationally) have an influence on one's tendency
to notice factors which pertain to the existence, welfare, or flourishing
of that thing or person. If this is so, then the fact that a person fails to
take note of such factors in certain circumstances is at least some indi-
cation that she does not accept this evaluative judgment.
165

I am not sure whether Smith means here that it’s reasonable for the
blamer to take it as some indication or it’s reasonable for us as philoso-
phers knowing the full specification of the story including the mental-
causal story of the potentially blamed person’s behavior to take it as indi-
cation that she does not accept the evaluative judgment. If we read Smith
on the former interpretation, though, then this gives her a way to preserve
the idea that Alessandra is responsible even if Talbert’s claims are all true.
In Alessandra’s case, her leaving Sheba in the car gives others a prima facie
reason to think she does not care sufficiently about Sheba, to which she
ought to respond. And the required response may involve apology and/or
other recompense for what happened that helps to make clear that she
does not have any lack of general care about Sheba or her family.

165
Smith (2005): 244.
-167-
But just because some fact is evidence that Alessandra doesn’t care
enough about Sheba, this doesn’t suffice to show that Alessandra actually
doesn’t care enough about Sheba. And so it seems that, whatever the mer-
its of this view, it is no argument against the Minimal Approval view as a
view of attributional-responsibility since it is possible no negative trait is,
in fact, attributable to Alessandra, (though it might not be unreasonable
for her family to initially attribute one to her.) It might be possible to use
the word “attributability” to refer not the set of actions that reveal some-
thing about what the agent is like such that the agent is properly subject to
praise and blame on the basis of those traits but instead to the set of be-
haviors for which it is reasonable to respond to as if they revealed some-
thing about what the agent was like due to the fact that they have a special
kind of significance for moral-social relations.
166
But doing so puts the
view out of conversation with the Minimal Approval view, which aims to
give the fully specified conditions on which a behavior does, after all, re-
flect a blameworthy trait of the agent. So, on this second horn, the reading
of Smith’s view may be more charitable, but it can’t be leveraged into an
objection that the Minimal Approval view misidentifies the bounds of at-
tributional-responsibility due to its handling of the Alessandra case.
5. Accountability Without Attributability
Let’s take stock. I have argued that while it seems that Alessandra is in
some sense responsible for leaving Sheba in the car, the most promising

166
For evidence that there is more than one sense in which the word “attributability”
tends to be used in this literature, see the discussion in the comments thread of this post
at the PEA Soup blog: http://peasoup.typepad.com/peasoup/2009/05/scanlon-on-moral-
responsibility-blame-part-1.html Here, T.M. Scanlon writes that the special relevance of
attitudes that are in principle judgment-sensitive comes from the fact that they are “par-
ticularly significant for our relations with each other (not because they are “attributable”
or “belong to” the agent in a sense in which other attitudes to do not).” “Or, to put the
same point differently,” he writes, “we tend to think of these attitudes as ‘attributable’ in
a way that others are not only because their content has this greater significance.”
-168-
bridge principle to explain attributional-responsibility for omissions on
the Minimal Approval view gives the outcome that she is not attribution-
ally-blameworthy. The tracing strategy also gives the result that Alessan-
dra is not attributionally-blameworthy even in cases in which she still
seems responsible. Alternative views of attributional-responsibility that
are able to return the result that Alessandra is attributionally-
blameworthy for her omission seem either to face serious problems of
their own or else not truly be views of attributional-responsibility in the
sense I have been concerned with.
One option would be to bite the bullet and conclude that Alessandra is,
after all, exempt from all moral responsibility for leaving Sheba in the car.
Sometimes theories should prompt us to give up our common sense intui-
tions about every day cases, and our intuitions in this case seem far from
clear. Some have defended the view that mistakes like Alessandra’s are
not the kinds of things for which a person can be morally responsible, and
this is not always seen as a difficult bullet to bite.
167
This line of thought
can be further supported by the fact that Alessandra ought to reassure her
family that her omission was not due to lack of concern for Sheba. Fur-
thermore, people may often be in poor epistemic circumstances in regards
to knowing whether they are in fact attributionally-responsible for their
mistakes, and so if Alessandra were to confidently proclaim her inno-
cence, given the stakes we might rightly find her to lack appropriate hu-
mility. Biting the bullet here is not obviously wrong, and is certainly open
to the Minimal Approval theorist.
But I don’t think the intuition that Alessandra is in some sense respon-
sible should be written off so easily. One reason is that there’s an im-
portant difference between an uncertain intuition and an ambivalent intui-
tion. And our intuitions about Alessandra seem to me to be of the latter

167
See, for example, Zimmerman (1986), and Talbert (Forthcoming).
-169-
type. It’s not just that we’re not sure if Alessandra’s case counts as a case
of something one can be morally responsible for, it’s that she does seem
responsible in one sense while not seeming responsible in another sense.
Following David Shoemaker’s approach to ambivalent intuitions about re-
sponsibility, I think we should take this ambivalence in our intuitions as
evidence of the fact that Alessandra’s omission fails to satisfy the condi-
tions for one type of responsibility but does satisfy the conditions for an-
other type.
168
In particular, I think that our intuitions reflect the fact that
this is not the kind of thing that does reveal something about what Ales-
sandra is like as a person, and so character-implicating responses of the
kind that are appropriate in cases of attributional-responsibility are inap-
propriate. And yet, it still seems that Alessandra owes something to her
family in light of her mistake. Even in the case in which her family is
completely assured of her lack of attributional-responsibility, it seems that
they would be justified in expecting and feeling entitled to an apology.
And it seems she owes it to them. This sense of owing recompense of
some sort plausibly implicates some form of responsibility response that
goes above and beyond the mere the appropriateness of agent-regret since
it is not just that she Alessandra ought to feel badly about what happened;
it seems that her family deserves something from her: at minimum, an
apology.
My proposal is that Alessandra is not attributionally-responsible but is
responsible in a non-appraising accountability sense. The relationship be-
tween attributional-responsibility and accountability-responsibility is fre-
quently thought to be that attributability is sufficient for mere appraisals
while an agent must meet the further requirements for accountability in
order to be subject to proper sanctioning responses of any kind. On this
picture, attributability is a necessary condition for accountability and in
order for an agent to be accountability-responsible an agent’s behavior

168
Shoemaker (2009, 2011, 2015).
-170-
must not only be attributable but she also must possess some further ca-
pacities such as normative-competency or even libertarian free-will. On
this sort of picture, if Alessandra is not attributionally-responsible for
leaving Sheba in the car, she is certainly not accountability-responsible for
doing so. I want to instead adopt an alternate picture on which attribu-
tional-responsibility and accountability-responsibility are wholly dis-
tinct.
169

I will not be able to give anywhere near a full account of accountabil-
ity-responsibility, but I do want to very briefly sketch one possible picture.
Accountability-responsibility, on this picture, is derived from role respon-
sibility.
170
Many of the roles we play in relationship to others, such as our

169
As Zheng (forthcoming) points out, this way of distinguishing concepts, while not as
popular in moral philosophy has a number of analogues in in political philosophy. Two
distinct concepts of responsibility, an appraising and non-appraising sense, have some-
times been distinguished by using different terminology. For example, Schmidtz and
Goodin (1998) distinguishes “blame responsibility” from “task responsibility” and Lavin
(2008) distinguishes “liberal” from “post-liberal” responsibility.
170
A somewhat similar picture is sketched in Zheng (2016, Forthcoming). In Zheng (2016)
the distinction between attributional-responsibility and accountability-responsibility is
made by grounding it in a “conceptual genealogy” meant to show how attributability
and accountability concepts arise from distinct sources of philosophical concern. Zheng
shares my general conception of attributional-responsibility as arising from a metaphysi-
cal question about how to make room on a naturalist picture of the world for our sense
that we are in some important way the authors of our own conduct such that our behav-
ior can open us up for appraisal-based praising and blaming responses. The concept of
accountability, in contrast, concerns issues of fairness and arises from strictly practical
normative concerns in moral and political philosophy. For Zheng, accountability-
responsibility derives from the fair division of labor. Individuals have duties to shoulder
certain social burdens that they are involved in causing because, as Zheng puts it, “when
a person’s action brings about some negative consequences for others, this generates a
social problem that simply cannot go unaddressed. These costs must be picked up some-
how and by someone, even if there is no bad intention or fault on the part of the person
involved, because there are victims who deserve redress. This means that under a fair
system of distributing burdens, it will often be appropriate for the person who performed
-171-
roles as caretakers, guardians, teachers, and bosses, generate certain kinds
of role-related responsibilities. As Sam Scheffler puts it, “...it is a familiar
fact that such ties are often seen as a source of special responsibilities. In-
deed, we would be hard pressed to find any type of human relationship to
which people have attached value or significance but which has never
been seen as generating such responsibilities.”
171
Role-related responsibili-
ties might be more or less moral depending on the particular role in ques-
tion.
It might be thought that all we mean when talking about role-
responsibilities is that we have some duties or obligations that obtain in
virtue of our relationships. However, I suspect the connection between
role-responsibilities and moral responsibility is tighter than that. It seems
to me that we are directly accountable to others for failing at our role-
related responsibilities, and when our responsibilities are moral, we are
generally straightforwardly morally accountable for doing so.
172
The ques-
tion of whether or not I am responsible in this sense completely circum-
vents the question of whether or not my action/omission expresses some-
thing morally bad about me. I might take responsibility for the welfare of
another person’s children while they are on my watch, but if one of these
children gets hurt on my watch this does not necessarily reflect something
bad about what I am like as a person, such that it would be appropriate to

the action to bear a large share of the costs: she can be asked to compensate for damages,
make reparations, or to change her practices to prevent future failures. But notice that
none of this requires an assessment of character, intentions, attitudes, or values—she can
justifiably be required to bear (at least some of) the costs of her behavior whether she per-
formed them out of malice, negligence, or sheer (non-culpable) ignorance or accident.”
Zheng (forthcoming) builds on this foundation and argues that accountability in this
sense is grounded in role-related responsibility.
171
Scheffler (1997): 190.
172
Similar ideas are developed in several different contexts in: Scheffler (1997), Miller
(2004), Cane (2016), Zheng (2017, forthcoming), and Lawford-Smith and Collins (2017).
-172-
blame me in any kind of appraising sense. I still owe my friend an apolo-
gy and special concern for the child’s recovery given that I am accountable
to her.
This is not to say that there is no set of excusing conditions for ac-
countability-responsibility, but these conditions do not come from failing
to meet certain metaphysical agential criteria. Rather, they might pertain
to the fairness of being held to have certain role-related responsibilities at a
given time and circumstance. For example, the fairness of the conditions
under which a person comes to take on a role-related responsibility may
be one relevant factor. As David Miller puts it when discussing a similar
concept of responsibility,
173
fair ascriptions of role-related accountability
may also

... rest on implicit norms concerning the capacities that human beings
can be expected to possess. Thus we do not hold people responsible for
the consequences of their actions in cases where those consequences
could only have been avoided by a superhuman display of strength.
Likewise we do not hold people outcome responsible in cases where
they are coerced into acting as they do, and in making these judgments
we rely on our intuitive sense of how much pressure a normal person
could be expected to resist. At the same time, diminished capacity does
not relieve a person of outcome responsibility. People cannot escape it
merely because through ignorance they failed to anticipate the results
of their actions. An unusually clumsy person can be held responsible
for the damage he causes as he blunders about the world
174

173
Miller speaks of “outcome responsibility” which roughly maps on to what I am calling
“accountability” and contrasts this with “moral responsibility” which he describes as
playing a similar role to what I have called “attributional-responsibility.”
174
Miller (2004): 245.
-173-
I suspect that in many cases when a person “takes responsibility,” it is in
this role-related accountability sense. In taking responsibility an agent
clarifies or reaffirms what she takes to be her reasonably assigned role-
related duties for which she is directly accountable even in the absence of
displaying any immediate morally objectionable traits in failing to meet
those duties.
Returning to the case of Alessandra, her role as Sheba’s caretaker
makes her liable for something bad happening to Sheba on her watch, and
this holds completely independently of whether or not her failure to do so
reflects a bad (even finely individuated) aretaic trait of hers. Alessandra,
while not attributionally-responsible for leaving Sheba in the car does owe
redress to her family members for her mistake. Her family would be
wrong to think that she has a bad trait, or to alter their perceptions of her
or their relationships to her in light of her mistake. Her family, however,
would be right to expect that Alessandra should apologize and make any
and all attempts to make it up to them. She should do these things not just
because she is the mother of a family that has just experienced a horrific
event but also because of the specific accountability to others affected that
she has, given that she had a role-based obligation to care for the dog at
that point in the day.
It should be noted that, were Alessandra to attributably fail to live up
to the demands of her role responsibility in her response to the situation,
she would then additionally be properly subject to appraisal-based attribu-
tional-blame. In this way, we can capture the full force of the intuition that
if Alessandra were to coldly fail to offer an apology, she would not only
be failing to deliver something that is owed to her family but she would
also be revealing a morally bad trait through her behavior.
Though this sketch of the relationship (or lack thereof) between attrib-
utional-responsibility and accountability-responsibility is admittedly in-
complete, I believe it represents a promising starting point for clearing
away conceptual confusion regarding ascriptions of moral responsibility,
-174-
and this is not limited to cases of forgetting. I’ll briefly mention three other
potentially fruitful applications of distinguishing the concepts in this way.
Responsibility for implicit bias may be helpfully thought of as involv-
ing accountability without attributability in cases in which an agent does
not minimally approve of her action that causes a harm associated with
acting in accordance with implicit bias. These agents nevertheless harm
others through their actions, and perhaps in doing so they fail in their
roles as citizens. As a result they owe something to the recipients of these
harms.
Similarly, white people might have role-related responsibilities as citi-
zens to help undo the effects of structural racism, even the effects that they
personally had no role in bringing about.
175
Even though, as an individual,
a particular white person might not have undertaken any attributable ac-
tions that would reveal objectionably racist traits, it might be that group-
membership generates role-related responsibilities to actively mitigate the
effects of structural racism. Omitting these actions, then, can make white
people accountably-responsible for harms to black people. Of course, once
awareness of this fact is raised for the white person’s consideration, she
can then become attributionally-blameworthy for actively ignoring it.
176

175
See Zheng (Forthcoming).
176
Avia Pasternak makes a similar point when she says that “group members are not
necessarily blame-responsible when their group acts badly (that would be determined in
light of their own behaviour). But a group's [collective moral responsibility] affects group
members in other ways: First, it generates feelings that are associated with moral wrong-
doing, such as 'we shame' and regret for what the group has done, and possibly also a
sense of personal shame for being associated with the group.

Moreover, the [collective
moral responsibility] of a group grounds certain task-responsibilities for its members,
such as the duty to change their group's practices and norms, or the duty to share the
burden of repairing the damage their group caused.

When group members end up hav-
ing these forward-looking duties, then a failure to comply with them would amount to a
personal moral failure” (Pasternak [2011]).

-175-
Finally, consider also the various agency-impairing conditions that I
have argued exempt agents from attributional-responsibility: from the
person with Tourette syndrome who utters slurs to the person with miso-
phonia who verbally lashes out every time someone chews gum. While
these agents are subject to forces that undermine their agency, this does
not give them license to harm people without explanation or apology, be-
cause this would be a display of a lack of regard for the role-related re-
sponsibilities they take on in relation to those their behavior affects.
Again, this sets an important limit on claims of being exempt from re-
sponsibility due to failures to meet the conditions of the Minimal Approv-
al account. While it is crucial for agents to respond to the behavior for
which they do meet these conditions and so are attributionally-
blameworthy, this does not exhaust the extent of what is required of their
humanity in terms of responding to the harms they have a duty to amelio-
rate, whether they come about by choice or chance.

-176-
Bibliography

Adams, Robert Merrihew (1985). Involuntary sins. Philosophical Review 94
(1):3-31.
Arpaly, Nomy (2002). Unprincipled Virtue: An Inquiry Into Moral Agency.
Oxford University Press.
Arpaly, Nomy & Schroeder, Timothy (1999). Praise, Blame and the Whole
Self. Philosophical Studies 93 (2):161-188.
Arpaly, Nomy & Schroeder, Timothy (2013). In Praise of Desire. OUP.
Bell, Macalester. “The Standing to Blame: A Critique.” In Justin Coates, D.
& A. Tognazzini, Neal (2013). Blame: Its Nature and Norms. Oxford Uni-
versity Press USA.
Bennett, Jonathan (1974). The Conscience of Huckleberry
Finn. Philosophy 49 (188):123-134.
Björnsson, Gunnar & Pereboom, Derk (forthcoming). Traditional and Ex-
perimental Approaches to Free Will and Moral Responsibility. In Jus-
tin Sytsma & Wesley Buckwalter (eds.), Companion to Experimental Phi-
losophy. Blackwell.
Bonevac, Daniel, Dever, Josh & David Sosa, and (2006). The conditional
fallacy. Philosophical Review 115 (3):273-316.
Bratman, Michael E. (1996). Identification, Decision, and Treating as a
Reason. Philosophical Topics 24 (2):1-18.
Bratman, Michael E. (1999). Faces of Intention: Selected Essays on Intention
and Agency. Cambridge University Press.
Bratman, Michael E. (2003). A desire of one's own. Journal of Philosophy 100
(5):221-42.
Braut, Jennifer J. (2018). “Investigating Misophonia: A Review of the Em-
pirical Literature, Clinical Implications, and a Research Agenda.” Fron-
tiers in Neuroscience. 12
Buss, Sarah (2012). Autonomous Action: Self-Determination in the Passive
Mode. Ethics 122 (4):647-691.
-177-
Buss, Sarah. (1994). “Autonomy Reconsidered.” Midwest Studies in Philoso-
phy 19(1), 95- 121.
Cane, Peter (2016). Role Responsibility. The Journal of Ethics 20 (1-3): 279-
298.
Clarke, Randolph (2003). Libertarian Accounts of Free Will. Oxford Universi-
ty Press USA.
Clarke, Randolph (2009). Dispositions, Abilities to Act, and Free Will: The
New Dispositionalism. Mind 118 (470):323-351.
Clarke, Randolph (forthcoming). Free Will and Abilities to Act. In Streit
um die Freiheit: Philosophische und theologische Beiträge. Paderborn:
Schoeningh/Brill.
Clarke, Randolph, McKenna, Michael & Smith, Angela M. (2015). The Na-
ture of Moral Responsibility: New Essays. Oxford University Press.
Coates, D. Justin & Tognazzini, Neal A. (2013). The Contours of Blame. In
D. Justin Coates & Neal A. Tognazzini (eds.), Blame: Its Nature and
Norms. Oxford University Press. pp. 3-26.
Dennett, Daniel. (1991). Consciousness Explained. London: Allen Lane.
Doris, John M. (2002). Lack of Character: Personality and Moral Behavior.
Cambridge University Press.
Doris, John M. (2015). Talking to Our Selves: Reflection, Ignorance, and Agen-
cy. Oxford University Press.
Earp, Brian D. & Trafimow, David (2015). Replication, falsification, and
the crisis of confidence in social psychology. Frontiers in Psychology 6.
Edelstein, Miren et al. “Misophonia: Physiological Investigations and Case
Descriptions.” Frontiers in Human Neuroscience 7 (2013), 296.
Enoch, David (2017). Hypothetical Consent and the Value of Autono-
my. Ethics 128 (1):6-36.
Fara, M. (2008). Masked Abilities and Compatibilism. Mind 117 (468):843-
865.
Feinberg, Joel (1962). Problematic responsibility in law and mor-
als. Philosophical Review 71 (3):340-351.
-178-
Fischer, John Martin (2010). The Frankfurt cases: The moral of the sto-
ries. Philosophical Review 119 (3):315-336.
Fischer, John Martin & Tognazzini, Neal A. (2011). The Physiognomy of
Responsibility. Philosophy and Phenomenological Research 82 (2):381-417.
Fischer, John Martin. (2012). “Responsibility and Autonomy: The Problem
of Mission Creep.” Philosophical Issues 22(1), 165-184.
Fischer, John Martin. (2012). “Semicompatibilism and Its Rivals. Journal of
Ethics 16 (2), 117-143.
Frankfurt, Harry (1987). Identification and Wholeheartedness. In Ferdi-
nand David Schoeman (ed.), Responsiblity, Character, and the
Emotions: New Essays in Moral Psychology. Cambridge University
Press
Frankfurt, Harry (1992). The Faintest Passion. Proceedings and Addresses of
the American Philosophical Association 66 (3):5-16.
Frankfurt, Harry G. (1969). Alternate Possibilities and Moral Responsibil-
ity. Journal of Philosophy 66 (23):829.
Frankfurt, Harry G. (1971). Freedom of the will and the concept of a per-
son. Journal of Philosophy 68 (1):5-20.
Frankfurt, Harry G. (1988). The Importance of What We Care About: Philo-
sophical Essays. Cambridge University Press.
Frankfurt, Harry G. (2006). Taking Ourselves Seriously & Getting It Right.
Stanford University Press.
Gorman, A.G. (Unpublished Manuscript.) “Depression’s Threat to Self-
Governance.” University of Southern California, Department of Phi-
losophy.
Haji, Ishtiyaque. (2002). “Compatibilist Views of Freedom and Responsi-
bility.” In Robert H. Kane (ed.), The Oxford Handbook of Free Will.
(Oxford: Oxford University Press.)
Haji, Ishtiyaque (1998). Moral Appraisability: Puzzles, Proposals, and Perplexi-
ties. Oxford University Press.
-179-
Haji, Ishtiyaque (2002). Compatibilist views of freedom and responsibility.
In Robert H. Kane (ed.), The Oxford Handbook of Free Will. Oxford
University Press
Haji, Ishtiyaque. (1998). Moral Appraisability: Puzzles, Proposals, and Perplex-
ities. (Oxford: Oxford University Press.)
Hartman, Robert J. (2017). In Defense of Moral Luck: Why Luck Often Affects
Praiseworthiness and Blameworthiness. Routledge.
Hieronymi, Pamela (2007). Rational capacity as a condition on blame. Phil-
osophical Books 48(2):109–123.
Jaworska, Agnieszka. (2015). "Identificationist Views," In Meghan Griffith,
Neil Levy, and Kevin Timpe (eds.), Routledge Companion to Free Will,
(Abingdon: Routledge).
King, Matt (2014). Two faces of desert. Philosophical Studies 169 (3):401-424.
Kumar, Sukhbinder et al. “The Brain Basis for Misophonia.” Current Biolo-
gy27.4 (2017): 527–533. PMC. Web. 14 Sept. 2017.
Lawford-Smith, Holly & Collins, Stephanie (2017). Responsibility for
states' actions: Normative issues at the intersection of collective agency
and state responsibility. Philosophy Compass 12 (11):e12456.
Lavin, C. (2008) The politics of responsibility. University of Illinois Press,
Champaign
Leckman, James F. and Donald J. Cohen. (1999). Tourette’s Syndrome—Tics,
Obsessions, Compulsions: Developmental Psychopathology and Clinical Care
(Hoboken: Wiley-Blackwell), 27.
Levy, Neil (2011). Expressing who we are: Moral responsibility and
awareness of our reasons for action. Analytic Philosophy 52 (4):243-261.
Lewis, David. “Dispositional Theories of Value II” in Smith, Michael, Da-
vid Lewis, and Mark Johnston. "Dispositional Theories of Val-
ue." Proceedings of the Aristotelian Society, Supplementary Volumes 63
(1989): 89-174.
Lippert-Rasmussen, Kasper (2003). Identification and responsibil-
ity. Ethical Theory and Moral Practice 6 (4):349-376.
-180-
Mason, Elinor, (Provisionally Forthcoming) “Taking Responsibility.” In
Oxford Studies in Agency and Responsibility Volume 6, ed. David Shoe-
maker, Oxford University Press.
Matheson, Benjamin (2018). The Threat from Manipulation Argu-
ments. American Philosophical Quarterly 55 (1): 37-50.
McKenna, Michael (2008). Compatibilism. Stanford Encyclopedia of Philoso-
phy.
McKenna, Michael (2008). Frankfurt's argument against alternative possi-
bilities: Looking beyond the examples. Noûs 42 (4): 770-793.
McKenna, Michael (2011). “Contemporary Compatibilism: Mesh Theories
and Reasons-Responsive Theories.” In R. Kane, ed., 2011, Oxford Hand-
book of Free Will, 2nd ed. (New York: Oxford University Press): 175-98.
McKenna, Michael (2012). Conversation & Responsibility. Oup Usa.
McKenna, Michael and Chad Van Schoelandt (2015). “Crossing a Mesh
Theory with a Reasons-Responsive Theory.” In A. Buckareff, C. Moya,
and S. Rosell, Agency and Responsibility (Palgrave Macmillan).
McKenna, Michael and Chad Van Schoelandt. (2015). “Crossing a Mesh
Theory with a Reasons-Responsive Theory.” In A. Buckareff, C. Moya,
and S. Rosell, Agency and Responsibility (Basingstoke: Palgrave Macmil-
lan).
McKenna, Michael. (2011). “Contemporary Compatibilism: Mesh Theories
and Reasons- Responsive Theories.” In R. Kane, ed., 2011, Oxford
Handbook of Free Will, 2
nd
ed. (New York: Oxford University Press), 175-
98.
McKenna, Michael (2016). A Modest Historical Theory of Moral Responsi-
bility. The Journal of Ethics 20 (1-3):83-105.
McKenna, Michael. (Forthcoming). “Watsonian Compatibilism.” In Oxford
Studies in Agency and Responsibility, Justin Coates and Neal Tognazzini,
eds. Vol. 5.
Mele, Alfred R. (1992). Springs of Action: Understanding Intentional Behavior.
Oxford University Press.
-181-
Mele, Alfred R. (1995). Autonomous Agents: From Self-Control to Autonomy.
Oxford University Press.
Mele, Alfred R. (1999). Kane, luck, and the significance of free
will. Philosophical Explorations 2 (2):96-104.
Mele, Alfred R. (2006). Free Will and Luck. Oxford University Press.
Miller, David (2004). Holding nations responsible. Ethics 114 (2):240-268.
Mitchell-Yellin, Benjamin (2014). In Defense of the Platonic Model: A Re-
ply to Buss. Ethics 124 (2):342-357.
Mitchell-Yellin, Benjamin (2015). The Platonic model: statement, clarifica-
tion and defense. Philosophical Explorations 18 (3):378-392.
Nelkin, Dana K. (2013). Moral Luck. Stanford Encyclopedia of Philosophy.
Pasternak, Avia (2011). The collective responsibility of democratic pub-
lics. Canadian Journal of Philosophy 41 (1):99-123.
Penelhum, Terence W. (1971). The importance of self-identity. Journal of
Philosophy 68 (October):667-78.
Pereboom, Derk (2005). Living Without Free Will. Cambridge University
Press.
Rescher, Nicholas, 1993, “Moral Luck”, in Moral Luck, D. Statman (ed.),
Albany: State University of New York Press.
Richards, Norvin, 1986, “Luck and Desert”, Mind, 65: 198–209; page refer-
ence is to the reprint in Statman 1993b.
Robichaud, Philip & Wieland, Jan Willem (eds.) (2017). Responsibility - The
Epistemic Condition. Oxford University Press.
Robin, Zheng (2016). Attributability, Accountability, and Implicit Bias. In
Jennifer Saul & Michael Brownstein (eds.), Implicit Bias and Philosophy,
Volume 2: Moral Responsibility, Structural Injustice, and Ethics. New York:
Oxford University Press. pp. 62-89.
Robin, Zheng (2016). Attributability, Accountability, and Implicit Bias. In
Jennifer Saul & Michael Brownstein (eds.), Implicit Bias and Philosophy,
Volume 2: Moral Responsibility, Structural Injustice, and Ethics. New York:
Oxford University Press. pp. 62-89.
-182-
Rosebury, Brian, 1995, “Moral Responsibility and Moral
Luck,” Philosophical Review, 104: 499–524.
Rosen “The Alethic Conception of Moral Responsibility” nature of moral
responsibility.
Rosen, Gideon(2004).Skepticism about moral responsibility. Philosophica l
Perspectives 18 (1):295–313.
Sartorio, Carolina (2016). A Partial Defense of the Actual-Sequence Model
of Freedom. The Journal of Ethics 20 (1-3):107-120.
Scanlon, T. M. (1998). What We Owe to Each Other. Belknap Press of Har-
vard University Press.
Scanlon, T.M. (2008). Moral Dimensions: Permissibility, Meaning, Blame,
Cambridge, MA: Harvard University Press.
Scanlon, T.M. (2013) “Interpreting Blame”, in Blame: Its Nature and Norms.
Ed. Coates & Tognazzini: 84–99.
Schmidtz D., and Goodin R.E. (1998) Social Welfare and Individual Re-
sponsibility: for and against. Cambridge University Press, Cambridge
Schroeder, Timothy & Arpaly, Nomy (1999). Alienation and externali-
ty. Canadian Journal of Philosophy 29 (3):371-387.
Schroeder, Timothy. (2005). Moral Responsibility and Tourette Syn-
drome. Philosophy and Phenomenological Research 71(1), 106–123.
Shabo, Seth (2015). More Trouble with Tracing. Erkenntnis 80 (5):987-1011.
Sher, George (2005). In Praise of Blame. Oup Usa.
Sher, George (2009). Who Knew?: Responsiblity Without Awareness. Oxford
University Press USA.
Shoemaker 2012 http://peasoup.typepad.com/peasoup/2012/06/survey-
says.html
Shoemaker 2015a “Ecumenical Attributability”
Shoemaker, David (2009). Responsibility and disability. Metaphilosophy 40
(3-4):438-461.
Shoemaker, David (2013). Qualities of will. Social Philosophy and Policy 30
(1-2):95-120.
-183-
Shoemaker, David (2015). Responsibility From the Margins. Oxford Univer-
sity Press.
Shoemaker, David W. (2003). Caring, identification, and agency. Ethics 114
(1):88-118.
Shoemaker, David. (2003). “Caring, Identification, and Agency.” Ethics 114
(1):88-118.
Shoemaker, David. (2011). “Attributability, Answerability, and Accounta-
bility: Toward a Wider Theory of Moral Responsibility.” Ethics 121(3),
602-632.
Shoemaker, David. (2015). “Ecumenical Attributability” in The Nature of
Moral Responsibility ed. Randolph Clarke, Michael McKenna, and An-
gela Smith. (Oxford: Oxford University Press.)
Silverstein, Matthew (2017). Ethics and Practical Reasoning. Ethics 127
(2):353 - 382.
Smith, Angela M. (2005). Responsibility for attitudes: Activity and passivi-
ty in mental life. Ethics 115 (2):236-271.
Smith, Angela M. (2008). Character, blameworthiness, and blame: com-
ments on George Sher’s In Praise of Blame. Philosophical Studies 137
(1):31-39.
Smith, Angela M. (2012). Attributability, Answerability, and Accountabil-
ity: In Defense of a Unified Account. Ethics 122(3):575-589.
Smith, Angela M. (2008). Control, responsibility, and moral assess-
ment. Philosophical Studies 138 (3):367 - 392.
Smith, Angela M. (2015). Responsibility as Answerability. Inquiry : An In-
terdisciplinary Journal of Philosophy 58 (2):99-126.
Smith, Michael (2003). Rational Capacities, or: How to Distinguish Reck-
lessness, Weakness, and Compulsion. In Sarah Stroud & Christine
Tappolet (eds.), Weakness of Will and Practical Irrationality. Oxford: Clar-
endon Press. pp. 17-38.
Sripada, Chandra. “At the Center of Agency, the Deep Self”
https://umich.app.box.com/s/vlknn6s4ggnc7tuefft2
-184-
Sripada, Chandra (2016). Self-expression: a deep self theory of moral re-
sponsibility. Philosophical Studies 173 (5):1203-1232.
Sripada, Chandra (2017). Frankfurt’s Unwilling and Willing Ad-
dicts. Mind 126 (503):781-815.
Strabbing, Jada Twedt (2016). Attributability, weakness of will, and the
importance of just having the capacity. Philosophical Studies 173 (2):289-
307.
Strabbing, Jada. Moral Responsibility: Attributability, Accountability, and Ca-
pacities. Dissertation. Princeton University. (2011).
Strawson, Peter F. (1962). Freedom and resentment. In Gary Watson
(ed.), Proceedings of the British Academy, Volume 48: 1962. Oup Oxford.
pp. 1-25.
Talbert, Matthew (2012). Accountability, Aliens, and Psychopaths: A Re-
ply to Shoemaker. Ethics 122 (3):562-574.
Talbert, Matthew (2016). Moral Responsibility: An Introduction. (Cambridge:
Polity).
Talbert, Matthew (Forthcoming). “Omission and Attribution Error” in D.
K. Nelkin and S. C. Rickless (eds), The Ethics and Law of Omissions
(OUP).
Taylor, James Stacey (2003). Autonomy, duress, and coercion. Social Phi-
losophy and Policy 20 (2):127-155.
Thomson, Judith Jarvis, 1993, “Morality and Bad Luck”, in Moral Luck, D.
Statman (ed.), Albany: State University of New York Press.
Timpe, Kevin; Griffith, Meghan & Levy, Neil (eds.) (2017). The Routledge
Companion to Free Will. New York: Routledge.
Tognazzini, Neal A. (2011). Understanding Source Incompatibil-
ism. Modern Schoolman 88 (1/2):73-88.
Vargas, Manuel (2005). The Trouble with Tracing. Midwest Studies in Phi-
losophy 29 (1):269-290.
-185-
Vargas, Manuel (2011). Revisionist Accounts of Free Will: Origins, Varie-
ties, and Challenges. In Robert Kane (ed.), Oxford Handbook on Free Will,
2nd Edition. Oxford University Press.
Velleman, David J. (1992). “What Happens When Someone
Acts?” Mind 101(403), 461- 481.
Velleman, J. David (1992). What Happens When Someone Acts? Mind 101
(403):461-481.
Vihvelin, Kadri (2004). Free Will Demystified: A Dispositional Ac-
count. Philosophical Topics 32 (1/2):427-450.
Vihvelin, Kadri (2011). How to think about the free will/determinism
problem. In Michael O'Rourke, Joseph Keim Campbell & Matthew H.
Slater (eds.), Carving Nature at its Joints: Natural Kinds in Metaphysics
and Science. MIT Press. pp. 314--340.
Vihvelin, Kadri (2013). Causes, Laws, and Free Will: Why Determinism
Doesn't Matter. Oup Usa.
Vihvelin, Kadri. (1994). “Are Drug Addicts Unfree?” In S. Luper-Foy and
C. Brown (eds.) Drugs, Morality and the Law (New York: Garland), 51-
78.
Watson, Gary (1975). Free agency. Journal of Philosophy 72 (April):205-20.
Watson, Gary (1996). Two Faces of Responsibility. Philosophical Topics 24
(2):227-248.
Watson, Gary (1999). Disordered Appetites: Addiction, Compulsion and
Dependence. In Jon Elster (ed.), Addiction: Entries and Exits. Russell
Sage Publications
Watson, Gary (2004). Agency and Answerability: Selected Essays. Oxford
University Press.
Williams, Bernard & Nagel, Thomas (1976). Moral Luck. Aristotelian Socie-
ty Supplementary Volume 50 (226):115 - 151.
Wolf, Susan (1987). Sanity and the Metaphysics of Responsibility. In Fer-
dinand David Schoeman (ed.), Responsibility, Character, and the Emo-
-186-
tions: New Essays in Moral Psychology. Cambridge University Press. pp.
46-62.
Wolf, Susan (1990). Freedom Within Reason. Oxford University Press.
Wolf, Susan (2015). Responsibility, Moral and Otherwise. Inquiry : An In-
terdisciplinary Journal of Philosophy 58 (2):127-142.
Zheng, Robin (forthcoming). What is My Role in Changing the System? A
New Model of Responsibility for Structural Injustice. Ethical Theory and
Moral Practice: 1-17.
Zimmerman, Michael J. (1986). Negligence and moral responsibil-
ity. Noûs 20 (2):199-218.

Abstract (if available)

Abstract In this dissertation I argue in favor of the Minimal Approval view, an original account of an agent’s moral responsibility for her actions, understood as the conditions that must be met so that an agent’s actions speak for her such that she can appropriately be blamed on their basis. My account shares a general theoretical orientation with Deep Self views, which hold that in order to be responsible an agent need not have had the ability to do otherwise, but rather, that an agent’s action must be caused by some deeper mental state that represents her agential identity. However, my view diverges from extant Deep Self views in several respects. For one, I argue that they tend to seriously over-generate exemptions, such that agents are exempt from responsibility even from ordinary weak-willed actions. Accordingly, much of my focus is on giving a more nuanced, and in certain senses narrower, explanation of genuine exemption cases. According to the Minimal Approval view, in order for an agent to be an apt target of blame on the basis of her action, the chain of mental events that leads her to action must sufficiently dispose her to be such that she would minimally approve of her effective motivation if she were to reflect. To minimally approve is to want just to some degree for one’s motivating desire to be satisfied for some further aim than merely to get rid of it.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Getting our act(s) together: the normativity of diachronic choice

PDF

The mental states first theory of promising

PDF

Responsibility and the emotional structure of relationships

PDF

The constitution of action

PDF

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

PDF

The virtue of reasonableness: on the normative epistemic implications of public reason liberalism

PDF

Rationality and the primacy of the occurrent

PDF

Self-reported and physiological responses to hate speech and criticisms of systemic social inequality: an investigation of response patterns and their mediation…

PDF

Responding to harm

PDF

Reasoning with degrees of belief

PDF

The causal account of liability to self-defensive harm reconsidered

PDF

Defending Rawls's Law of Peoples against cosmopolitan critiques

PDF

What "ought" ought to mean

PDF

A deontological explanation of accessibilism

PDF

Protest of the oppressed: a case study on Black female administrators’ experiences in K-12 education from student to leadership

PDF

The partial Good Samaritan states: China and Japan in the international relations of autocracy and democracy

PDF

The regime of religious pluralism: uncovering the cultural dimensions of American religious belonging

Asset Metadata

Creator Gorman, Amanda Grace (author), Gorman, August G (author)

Core Title The minimal approval view of attributional-responsibility

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Philosophy

Publication Date 10/19/2018

Defense Date 08/15/2018

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

Tag Addiction,Blame,compatibilism,compulsion,deep self,ethics,free will,Harry Frankfurt,misophonia,negligence,OAI-PMH Harvest,OCD,responsibility,Tourette Syndrome,weakness of will

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Quong, Jonathan (committee chair), Finlay, Stephen (committee member), Shoemaker, David (committee member), Watson, Gary (committee member)

Creator Email agorman@usc.edu,gorman.august@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-90441

Unique identifier UC11675749

Identifier etd-GormanAugu-6796.pdf (filename),usctheses-c89-90441 (legacy record id)

Legacy Identifier etd-GormanAugu-6796.pdf

Dmrecord 90441

Document Type Dissertation

Format application/pdf (imt)

Rights Gorman, Amanda Grace; Gorman, August G

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