23.10.2007
1) Consider actions that are done in order to bring about some
change in the world; let us call some such action A, and the change
that the agent wants to bring about C. In order to consider taking
that action at all, the agent must have good reasons to think that
taking action A is instrumental in bringing about C with
some probability; in other words, the agent must be certain that he has
a good chance to bring about C by doing A. In addition,
the agent must be confident to be able to perform A. In cases
where it is normally so that doing A will bring about C,
but there is no chance for a given agent to actually do A,
it would not be reasonable for that agent to count A among the
available options.
Given that I decide to take an action A, in order to bring
about C, I have a bit of knowledge about the future (namely, I
know that C will happen). Since the grounds of this knowledge are in
my intention to perform A, this is called 'knowledge in
intention'.
This use of the term 'knowledge' assumes of course that it makes
sense to talk of knowledge about the future at all. This might be
disputed from a point of view about the future where the future is
entirely open, and therefore there is nothing to know about it - that
is, as long as an event is in the future, it is not possible to have
knowledge about it. For the sake of this discussion, let us
take it that there may be knowledge about future events. (If someone were to
hold a view where there can be no knowledge about the future, but still
wanted to make the points here, he should still be able to do so, although
obviously not in terms of knowledge.) In the case of knowledge in intention,
an agent knows that C will happen as long as she has decided to
perform A, i.e. as long as she intends to do A, and has good
reason to believe that it is in her power to do so. In the actual run of
events, impediments may of course come up which make her fail, or
which prevent C to happen in spite of her doing A. These are
not cases of knowledge in intention, for they are obviously not cases of
knowledge. However, if the agent succeeds in bringing about C via
doing A, I think we are justified in saying that the agent
knew that C would happen in the future, and this knowledge
was knowledge in intention.
There are other ways of knowing about future events: for instance
predictions from causal laws, or logical truths. The latter are
perhaps the least interesting, but even the former play a much
smaller role in daily life than knowledge in intention does. We rely
in many of our activities on knowledge about events that haven't
happened yet, but of which we know that they will happen, because
we are going to bring them about ourselves. (There is also inferred
knowledge of things that will probably come about because of other people's
intentions to make them happen. Obviously, though, there is a much higher
uncertainty in getting a grip on others's intentions than knowing one's own
ones.) This is part of our ability to project, which I have started to
discuss in a previous
post.
2) Before discussing a few aspects of this, let us consider an
example. You are on a business trip somewhere and your primary means of
communication with your office at headquarters is by phone. Assume you
have set up a meeting with a few co-workers to discuss a new project,
but whether it makes sense to do the meeting at all depends partially on
the results of your activities on your business trip. Let's say the
meeting is scheduled for the afternoon, it is still morning, and you
know you can cancel the meeting by just ringing one of the other
participants at headquarters and ask them to do so. So in this example,
the action A in question is ringing someone at headquarters, the
intended change C is a cancellation of the meeting, and if you
have just decided to make the call, you have knowledge in intention
about that cancellation. (Let us assume there is a high probability that
you will reach someone, and that they will do as asked and indeed call
off the meeting, etc., that is, let us assume, it really is
unproblematically up to you whether the meeting takes place or not.)
Now consider the importance of timing your action: it is still
morning, and that means your action actually can have the intended
effect; if, by contrast, it was already evening, the meeting would
have taken place already, and there would be no more point in making the
call. However, the two criteria which I have listed in 1) still
hold: taking action A brings about C (that is, as a
general rule, it's just that this rule doesn't apply in this case), and
you are perfectly able to perform A. What is different is that the
meeting has already taken place, and therefore cannot be prevented
anymore.
3) Perhaps we could build this sensitivity to timing into the
formulation of the first criterion (i.e. specify that doing A
brings C about, but only if C is a future event). But that
may not be the best idea. In fact, the time structure of things doesn't
necessarily play a role. To see this, let us modify our example a bit:
suppose you don't know when the meeting is to take place - you only know
that there will be a meeting at some time, but you don't know when. It
may be this morning already, or this afternoon, or perhaps only
tomorrow. Note that the two criteria do still hold. You can bring about
the cancellation, and you are certainly in a position to do so. You
would fail, of course, in case the meeting has already taken place. But
you can't know whether that's the case without trying.
Apparently, this kind of situation is in one crucial way different from
the more clear-cut example which we started with: there is no other
knowledge besides your knowledge in intention. In the example as it
was at the beginning, there was an independent knowledge about the meeting:
it was fixed when it was to take place, and you would know the time at which
it was to happen, in addition to your knowledge from your intention to
cancel it.
The second version of the example is built in a way that there is no
knowledge except your knowledge in intention. Still, the reasoning is
the same: you have to decide whether you are able to cancel the meeting, but
since you cannot know whether it has already happened, that possibility can
play no role for preferring either option. The uncertainty on your part
(about whether the meeting has already happened) is not a sufficient reason
to think that you are not able to cancel the meeting. If it hasn't happened
yet, you are perfectly able to do so, and you would do it that very same way,
by calling someone at headquarters and ask them to cancel, which they will
certainly do. If the meeting already took place, then of course you won't
be able to cancel it anymore, but you cannot assume that. Thus, as far as
your decision-making is concerned, you can cancel the meeting by making
the call, and there is nothing to prevent you to do so.
4) It seems, then, that the unavailability of independent knowledge
(i.e. knowledge of C independent of your intention to bring it about
by doing A) is a third criterion that plays a role in deciding
whether to perform A. There is no case where it still makes sense to
perform A if not all three criteria are fulfilled.
Take the meeting cancellation again. If there is independent knowledge
about the meeting having been held already, there is no point in taking
action A. Likewise, if you think your call would cause the meeting
to be canceled, but you know that you won't have a chance to reach a phone
until the evening, when it is too late (imagine you are traveling an
undeveloped countryside with little communication infrastructure), you don't
consider A an option anymore. Finally, if you think you could make
the call in time, but you could not be certain the meeting would actually
be stopped (perhaps you have an antagonist in the office who just sits there
and intends to make damn sure the meeting takes place when he learns you
want to cancel it, which he will), the same would apply. Only if you can
assume that you will be able to call, that the call will have the desired
result, and you don't know anything that indicates that C has
already taken place - then you would consider A as a reasonable course
of action.
5) Interestingly enough, this trias of criteria also seems to hold
if someone sets out to bring about some past event. I'm not going to
continue on this line here, but see Michael Dummett's famous article
"Bringing about the past"[1]. Much of this post is just an elaboration
of one of Dummett's points.
6) In order to regard a possible course of action as a
reasonable option at all, as we have seen, it is necessary for the agent to
assume that there is no independent knowledge (independent, that is, of the
agent's own intention to perform the action) about the outcome. I won't
even start to discuss the connections between the concepts of knowledge and
reality here (and also we should keep in mind what I remarked earlier,
in 1), about the question whether what I have called 'knowledge in
intention' could be strictly knowledge, since it is about the future,
of which one could arguably have no knowledge at all). But so much can be
said that there seems to be an assumption of 'openness' (whether it is
objective under-determination or just the impossibility of knowing about any
future states of affair) in any projection that we make in order to evaluate
possible courses of action. (Note that this 'openness', which is often
attributed to the future, has nothing to do with any 'fixedness' of the past
or present. It is an aspect of the future that has nothing really to do at
all with a supposed contrast between the future and the present. It rather
has to do with the space of options for potential actions.)
__
[1] Michael Dummett, "Bringing about the past", in: Truth and
other Enigmas, Harvard University Press 1978, 333-350.
