Timothy OConnor, Agent Causation

1

• Timothy OConnor, Agent Causation
• The first task that TO sets himself is to show
  how the concept of agent causation can be
  understood as a distinct species (from event
  causation) of the primitive idea, which Ill term
  causal production, underlying realist or
  non-Humean conceptions of event causation. (258)
• That is a mouthful. Here is what TO is going to
  do
• 1. Articulate a certain conception of causality
  (or causal production). This conception is
  going to be realist and non-Humean. Well
  see what this means later on.
• 2. Show how the notion of agent causation is
  recognizable as a species of the kind of
  causality articulated in (1).

2

• Lets start with the Humean conception of
  causality to which TO is opposed. Humean
  refers to David Hume.
• On the Humean conception, the essential component
  of the concept of causality is the idea of a
  regularity in nature. On this conception, the
  truth of the claim that C caused E depends on a
  regular connection in nature between events
  like C and events like E.
• There are two ways of making this more precise.
• (a) C caused E it is a law of nature
  that whenever events like C occur events like E
  follow
• - law-like connection between C and E
• (b) C caused E had C not happened
  then E would not have happened
• - counterfactual connection between C and E

3

• TO calls this Humean conception of causality a
  reductive account. He thinks it is an attempt
  to reduce the notion of causality to the notions
  of law-like connection or counterfactual
  connection. By calling the opposing, anti-Humean
  view realist, he also shows that he views the
  Humean view as somehow skeptical or
  anti-realist about causality.
• TO points out that if the Humean conception of
  causality is correct, then the notion of agent
  causation is incoherent.
• Let A be an agent, and consider what the claim
  that A caused E would have to imply on the two
  different versions of the Humean view
• A caused E it is a law of nature
  that whenever events like A occur events like E
  follow
• A caused E if A had not
  occurred then E would not have occurred.
• Since A is an agent, the statements to the right
  of the arrows dont make any sense.
• So, if the Humean conception is right, then the
  notion of agent causation is incoherent (since
  claims about agent causation would have
  incoherent consequences.)

4

• What TO wants to show is that there is a
  different conception of causality, the realist
  or anti-Humean conception, on which the notion
  of agent causation is not incoherent.
• So what is the anti-Humean conception of
  causality?
• TO ...the core element of the concept of
  causality is a primitive notion of the
  production or bringing about of an effect.
• This anti-Humean view sees no connection between
  the concepts of causality and the concept of a
  regularity in nature. Rather, it understands
  causality in terms of the idea of an effect being
  produced or brought about by an object.
• Anti-Humean conception
• C caused E E was brought about
  by an object occurring in C

5

• When we apply the anti-Humean conception of
  causality to agent causation, TO thinks that no
  incoherency results.
• A caused E E was brought about by
  A
• TOs next step is to articulate an account of
  causation that makes use of this primitive notion
  of production or bringing about.
• To do this, TO first introduces the idea of a
  causal power.
• An objects causal powers determine the effects
  that the object will produce when placed in
  different circumstances. This allows us to think
  of causal powers as functions from circumstances
  to effects.
• Causal power function from circumstances to
  effects

6

• Example a stick of dynamite. A stick of
  dynamite has the causal power to produce an
  explosion when placed in certain circumstances.
  In circumstances in which the fuse is lit and
  there is plenty of oxygen you will get an
  explosion.
• circumstances effect
• lit fuse explosion
• sufficient oxygen
• lit fuse no explosion
• no oxygen
• unlit fuse no explosion
• sufficient oxygen

7

• The causal powers possessed by an object are
  determined by that objects properties.
• A stick of dynamite has certain chemical and
  physical properties. Because of these
  properties, the dynamite has a causal power that
  determines that it will produce an explosion when
  its fuse is lit and there is sufficient oxygen in
  the air.
• A picture of event-causation is emerging. When
  an event C causes another event E the following
  is the case
• - There is an object O which has certain
  properties that endow it with the causal power to
  produce E in certain circumstances.
• - C is event which contains an object O in
  circumstances in which its causal power to
  produce E is activated.

8

• So far weve been considering the anti-Humean
  conception of event causation. How does agent
  causation fit into this account of causality?
• Two kinds of causal powers
• Event-causal powers an object has a
  event-causal power to produce an effect E in
  certain circumstances if it is such that when the
  object is in those circumstances it necessarily
  produces E
• Agent-causal powers an agent has an
  agent-causal power to produce an effect E in
  certain circumstances if the agent is such that
  when she is in those circumstances it is possible
  for the agent to produce E
• Both notions invoke the primitive notion of
  something producing or bringing about an effect.
  The difference is that when an object exercises
  one of its event-causal powers in certain
  circumstances it must produce the effect that it
  produces. On the other hand, when an agent
  exercises one of her agent-causal powers in
  certain circumstances, she may or may not produce
  the effect.
• Hopefully you can see how this is going to help
  with the problem of free will.

9

• This anti-Humean, agent-causation account of
  causality gives us a picture of what happens when
  an agent performs an action freely.
• Agent A has a certain property or properties that
  endow her with the agent-causal power to fire a
  gun when placed in the right circumstances (shes
  holding the gun, there is no one forcing her
  finger on the trigger, etc.).
• A finds herself in the relevant circumstances in
  which her agent-causal power is activated. The
  activation of her agent-causal power makes it
  possible for her to produce a firing of the gun.
  It does not necessitate that she produces the
  firing. Her agent-causal power allows her to
  bring about the firing of the gun, but it does
  not necessitate that she will.
• A fires the gun. But since it was not
  necessitated by her the activation of her
  agent-causal power to fire the gun, it is still
  the case that she could have refrained from
  firing the gun. Hence, her action was performed
  freely.
• TOs main goal is just to show that there is a
  coherent notion of agent-causation, that allows
  for free action, which can be seen as a variety
  of a more general kind of causality.

10

• This description of what happens when an agent
  acts freely does not precisely capture TOs view.
  The problem is that on this account, what the
  agent produces is the action of firing the gun.
  This is not how TO thinks of it. TO thinks that
  the effect produced by the agent is an intention
  or volition - a mental event that proceeds to
  cause the firing of the gun.
• ac ec
• agent volition firing of the gun
• decision
• TO calls the event of the agents causing the
  volition a decision.

11

• Furthermore, TO thinks that we should call the
  action the whole sequence of events plus the
  agent at the beginning plus the existent causal
  relations.
• ac ec
• agent volition firing of the gun
• action
• TO In the case of an observable bodily movement
  such as waving my hand, my action consists of the
  causal relation I bear to the coming-to-be of the
  state of determinate intention volition to wave
  my hand, plus the sequence of events that flow
  from that decision.(266)
• This may seem like an odd way of thinking of the
  action. Wouldnt we normally say that the action
  is just the firing of the gun (or the waving of
  the hand)? It is odd to say that the action
  itself includes the intention or volition that
  caused it.

12

• TO thinks that the alternative to his way of
  individuating the action is wedded to the Humean
  conception of causality, which sees an action as
  part of a sequence of events. The intention is
  one event, it causes another event - the firing
  of the gun. The action is the firing of the gun,
  and it is caused by the previous intention.
• If we adopt the anti-Humean conception of
  causality we will view things differently.
  Instead of seeing sequences of events, we see
  objects and agents producing effects. On this
  conception, the natural way to view an action is
  as something produced by an agent. The notion
  of an action includes the agent, and it includes
  the fact of being produced by the agent. The
  agent and the agents production are part of the
  action.
• TO uses this way of individuating actions to
  respond to a dilemma raised by Davidson - but I
  dont want to talk about Davidsons dilemma. I
  leave this as an exercise for you how does
  individuating actions in the way TO proposes
  solve Davidsons dilemma?

13

• Lets now look at some objections and responses
  to this agent-causation account.
• The first objection that well consider appears
  on p. 268. TO quotes C.D. Broad
• Broad How can an event possibly be determined
  to happen at a certain date if its total cause
  contained no factor to which the notion of a date
  has any application? And how can the notion of a
  date have any application to anything that is not
  an event?(268)
• Suppose that a volition comes-to-be at a certain
  time t, and it is caused by the agent.
• agent volition
• happens at t

14

• Why does the volition occur at t, and not at some
  other time?
• Usually when we ask this kind of question about
  an event we answer it by citing the previous
  event that caused it. Since that previous event
  occurred at a time immediately prior to when its
  effect occurred, we can understand why the effect
  occurred when it did.
• But if the volition is caused by an agent and not
  by a previous event we cannot give this kind of
  explanation. The agent is not something that
  occurs at some time. Just citing the agent as
  cause of the volition seems to leave it
  completely mysterious why the volition occurred
  when it did.
• TOs response the volition occurred at t because
  it was caused to occur at t by the agent.
• TO Consider, for example, my consideration a
  while ago concerning whether to continue working
  on this paper for another hour or to stop and do
  something else. After a brief moment of
  consideration, I formed the intention (at time t,
  say) to continue working. According to the
  agency theory, we may suppose that at t I
  possessed the power to choose to continue working
  or to choose to stop, where this is understood as
  the capacity to cause either of these mental
  occurrences. And, in fact, the capacity was
  exercised at t in a particular way (in choosing
  to continue working.)(269)

15

• agent volition
• happens at t
• The agent caused the volition at t. - this
  explains why the volition occurred at t.

16

• This response seems sensible, but it immediately
  gives rise to another question
• Why did the agents causing of the volition
  occur when it did?
• Using TOs terminology, another way to put this
  is to ask
• Why did the agents decision occur when it did
  (i.e. at t)?
• And there is another closely related question
  that one might ask
• Why did the agent cause the volition that she
  did cause?
• or, equivalently
• Why did the agent make the decision that she
  made?

17

• Lets take the second question first
• Why did the agent make the decision that she
  made?
• Lets suppose that at a given time an agent
  causes the intention to X. Why did the agent
  make this decision?
• TO points out that there will be reasons -
  desires and beliefs - that support X. For
  example, the agent wants to ?, and believes that
  by doing X she can accomplish ?. These are her
  reasons for causing the intention to X.
• agent intention to X
• This also answers the first question, because the
  desire and belief are present at some particular
  time, and so they explain why the agent caused
  the intention to X when she did.

desire to ? belief that by Xing, she can ?
18

• Why did the agent decide to X?
• - because she desired to ? and believed that by
  Xing she could ?
• Why did the agent decide to X at t?
• - because at t she desired to ? and believed
  that by Xing she could ?
• The idea is that at t the agent found herself
  with this desire and belief. These are the
  circumstances that activate her agent-causal
  power to X.

19

• 2 objections
• To explain why the agent caused the intention to
  X we have to explain why she caused this
  intention and not any other intention. Just
  citing the reasons for X fails to explain why she
  didnt cause the intention to Y, or the intention
  to Z, or etc. and if we dont know why she didnt
  cause these other intentions, then we dont have
  an explanation for why she caused the intention
  to X.
• Citing the agents reasons for causing the
  intention to X renders the appeal to
  agent-causation superfluous. The presence of the
  belief and desire on their own explain the
  coming-to-be of the intention to X. In
  explaining this intention, there is no additional
  need to bring in the fact that the agent caused
  the intention.

20

• Lets start with the first objection. We are to
  imagine the following scenario
• agent intention to X
• intention to Y
• The objection is that merely citing the reasons
  for X wont explain why the agent decided to X
  because the agent also had reasons for Y, and so
  we dont know why the agent didnt decide to Y on
  the basis of her reasons for Y. This means that
  merely citing the reasons for X wont explain why
  the agent decided to X - we also have to cite
  reasons that show why the agent didnt decide to
  Y.

reasons for X
reasons for Y
21

• There is the threat of regress. It looks as
  though in order to explain why the agent chose X
  instead of Y on the basis of her reasons for X we
  have to cite some reasons she had for acting on
  her reasons for X rather than her reasons for Y.
  In other words, we have to explain why the agent
  decided to act on the reasons for X rather than
  the reasons for Y. This is another decision that
  needs to be explained by appealing to the agents
  reasons, and now the regress has started. Let
  X reasons for X
• agent
• Now the question is why did the agent cause the
  intention to act on X? To answer this, we have
  to appeal to further reasons - reasons for or
  against choosing the reasons for X. But then
  well need an explanation for why the agent
  favored the reasons for the reasons for X - and
  so on, into the night.

X
Y
intention to act on X
22

• TOs response is to reject one of the assumptions
  on which the objection is based.
• The objection makes use of two assumptions
• 1. Citing the agents reasons for X does not
  explain why the agent didnt choose to Y or to Z
  or etc.
• 2. Unless the explanation can explain why the
  agent didnt choose to Y or to Z or etc. it
  cannot explain why the agent chose to X.
• TO accepts the first assumption but rejects the
  second.
• In simply assuming that an explanation of the
  action will cite all the salient psychological
  features of the agent at the time of his
  decision, Kane is clearly presuming that there is
  only one type of adequate explanation of a
  choice, the type that explains why only that
  choice could have been at that point in the
  agents psychological history. But that is
  unsupported.(276)

23

• Compare
• The wall collapsed into a pile of rubble because
  the stick of dynamite exploded.
• The question is does this successfully explain
  why the wall collapsed into a pile of rubble? Do
  we also have to show why the wall didnt do
  something else, e.g. fall over flat, remain
  standing, split in half, etc.?
• If the 2nd assumption is correct, merely citing
  the explosion of the dynamite is not enough to
  explain why the wall collapsed into a pile of
  rubble. We would also have to explain why the
  wall didnt do anything else. TOs point is that
  this is an unreasonable demand on successful
  explanation.

24

• Recall the second objection
• Citing the agents reasons for causing the
  intention to X renders the appeal to
  agent-causation superfluous. The presence of the
  belief and desire on their own explain the
  coming-to-be of the intention to X. In
  explaining this intention, there is no additional
  need to bring in the fact that the agent caused
  the intention.
• TOs view
• agent intention to X
• desire and belief
• Objection
• desire and belief intention to X

explains
intention to X
explains
25

• agent intention to X
• To allow for freedom and responsibility the
  explanation of the intention to x which appeals
  only to the belief and desire would have to be a
  non-causal explanation. If the belief and desire
  cause the intention, then the agent is not
  responsible for the action that that intention
  ends up causing.
• TOs response ..the agents free exercise of
  his causal capacity provides a necessary link
  between reason and action, without which the
  reason could not in any significant way explain
  the action. It allows us to claim that the
  reason had an influence on the production of the
  decision, while not causing it. Were we to
  remove the element of causal production of
  decision altogether, and simply claim that the
  decision was uncaused, then noting the fact that
  the agent had a reason that motivated acting in
  that way would not suffice to explain it
  ....(277)

desire belief
desire belief cannot cause intention to X
26

• Randolph Clarke Toward a Credible Agent-Causal
  Account of Free Will
• The problem of rational action
• RC ...if an agents action is not caused by her
  having certain reasons for action, then it is
  unclear how she can be said to have acted on
  those reasons and how her action can be said to
  be rational (and rationally explicable).(286)
• On OConnors account, an action is not caused by
  the agents reasons. Remember reasons
  desire belief
• For OConnor, the agent is the one who causes the
  action (or rather, the intention which then
  causes the action). RC thinks this shows that
  for OConnor, no actions are rational.

27

• RC ...if an agents action is not caused by her
  having certain reasons for action, then it is
  unclear how she can be said to have acted on
  those reasons and how her action can be said to
  be rational (and rationally explicable).(286)
• As stated, this is the claim
• action is not caused by reasons R
• This is equivalent to
• the agent acted on reasons R the action
  was caused
• by R
• An action is rational if and only if in
  performing the action the agent acted on reasons.
  

implies
the agent did not act on R
implies
28

• The obvious question for RC is why believe these
  claims, i.e. why believe that
• acting rationally
• TO would undoubtedly deny this claim. Why does
  RC think that it is so obvious?
• An example will be helpful. Suppose an agent has
  a number of different reasons for the same
  action.
• agent
• Suppose R1, R2, and R3 are all reasons for
  doing X. Suppose the agent does X, and she does
  X for reasons R1 - not for R2 or R3. How do we
  explain this notion of doing something for a
  particular reason?

implies
being caused to act by ones reasons
R2
R3
R1
29

• Heres an example to flesh this out
• Suppose Claire voted for Ralph Nader in the last
  election. Before she voted, she had two reasons
  for voting for Nader
• Reason 1 she wanted to reduce the power of
  corporations in American politics
• Reason 2 she wanted to annoy her parents, who
  are Democrats
• When Claire actually voted for Nader, she voted
  for Reason 1 - i.e., the reason on which she
  acted was her desire to reduce corporate power.
• How should we make sense of this notion of the
  reason on which Claire acted?
• RC thinks the only way to make sense of this is
  to say that Reason 1, and not Reason 2, is the
  reason that caused Claire to vote for Nader.
  This is not something that TO can say.

30

• So lets suppose RC is right and that in order
  for an action to be rational it has to be caused
  by the agents reasons. How do we allow for
  rational action on an agent-causal view of
  action?
• RCs main idea
• ... what an agent directly causes, when she acts
  with free will, is her acting on (or for) certain
  of her reasons rather than on others ...(290)
• On TOs view, what the agent causes is an action
  (or, rather, an intention which then causes the
  action).
• agent intention action
• RC thinks that agents cause actions for reasons
• agent action for a reason

31

• Remember that for RC, acting for a reason is
  being caused to act by a reason. So his view is
  that what the agent causes is that her reasons
  cause her action.
• agent
• Claire

reason action
vote for Nader
desire to reduce corporate power
32

• There is another wrinkle in Clarkes account.
  The kind of event-causation that happens between
  reason and action is indeterministic or
  probabilistic. Prior to the agent causing the
  action, each of her reasons has some probability
  of event-causing one of various actions.

R1
R2
R3
30
30
40
agent
action1
action2
action3
33

• The probabilistic nature of the event-causal
  relation between reasons and actions is crucial
  to Clarkes account. Otherwise, if this relation
  were necessitating - i.e. if the presence of a
  reason necessitated the occurrence of the action
  - then there would be no role for agent-causation
  to play and no room for freedom and
  responsibility.

R1
R2
R3
0
0
100
agent
action1
action2
action3
34

• There is another wrinkle.
• Clarke ...what an agent directly causes, when
  she acts with free will, is her acting on (or
  for) certain of her reasons rather than on
  others, and her acting for reasons ordered in a
  particular way by weight, importance or
  significance as the reasons for which she
  performs that action.(290)
• Suppose that in a given situation I have a number
  of different reasons for performing different
  actions, and for any one of these actions I have
  more than one reason for performing it.

R2
R3, R4, R5
R1
4
agent
1
25
30
40
action 1
action 2
action 3
35

• agent

R4 R5 R3
action 3
Clarke Pam attends a lecture on Mapplethorpe,
say, primarily because she is interested in his
work and secondarily because she knows the
speaker. She might also have some desire to
accompany a second friend to an interesting movie
that is showing at the same time. But she causes
her acting on the first set of reasons, and on a
particular ordering of them, instead. What she
directly causes is her attending the lecture
primarily because of Mapplethorpes work and
secondarily because of her friendship with the
lecturer.(290)
36

• Review
• 1. Chisholms dilemma argument for a
  libertarian, agent-causal account of human
  responsibility and freedom.
• 2. Humean vs. Anti-Humean causation
• 3. OConnors account of agent-causation in
  terms of agent-causal powers
• 4. OConnor on the relation between reasons and
  actions
• 5. Clarke on rational action action for a
  reason action caused by a reason
• 6. Clarkes version of agent-causation
• Next week van Inwagen, An Argument for
  Incompatibilism