Back to CQ
Back to Archives' Table of Contents
The Following are comments by Gary Schouborg and
Ami Thomasen on the post that became A Defense of Emergent Downward Causation.
Gary Schouborg writes:
Teed, you argue that drinking a cup of hemlock caused both Soc's death and
Xantippe's becoming a widow because both were the same event. You contrast this
with the two separate events of "the cup's being empty" and Soc's death.
In the sense in which you seem to be using 'event', the cup's being empty is a
fact, not an event. The corresponding event would be Soc's emptying the cup or the
cup being emptied. Under the historical conditions, Soc's emptying the cup caused
his death. We can also say Soc's death caused X's becoming a widow, whereas we
cannot say her becoming a widow caused his death. Therefore, it is unclear in what
sense these two events are the same. We could say that there is a complex event
which includes two cause-effect events of Soc's death and X's becoming a widow.
This might require you to tweak but not abandon your distinction between practical
and metaphysical causality.
Based on your identification of Soc's death and X's becoming a widow, you
conclude that "we could have saved Socrates' life by having him divorce Xantippe".
Assuming you can answer my preceding objections, you have here to deal with the
fact that divorcing one's wife does not make her a widow. Therefore, your
conclusion is a non sequitur flowing from simple equivocation rather than from the
more interesting confusion between practical and metaphysical causality. This
detail is probably easily repaired, allowing you to move on to your distinction.
Your distinction between practical and metaphysical causality seems to come
down to, more simply, the distinction between a partial but useful explanation and
a complete one, respectively. Among partial explanations, you can then distinguish
between the weak sense of epiphenomenal as irrelevant or not salient vs. causal as
relevant or salient.
Your criteria for usefulness -- (1) the factor that changed most recently; and
(2) the factor over which we have the most control -- though promising, are
inadequate. For example, I can imagine coming into a room where the gas has been
left on, lighting a match, and later (having miraculously survived) saying that the
house blew up because I had inadvertently left the gas on. The criterion here is
*salience*. In some case, what has changed most recently would be salient, in some
cases not. This also applies to control. For example, when I say that thousands
drowned from a tidal wave, I don't mean that the tidal wave was the factor those
people had the most chance of getting control of. Control may or may not be
You refer to the metaphysical cause of an event as "everything in the universe
that was responsible for that event taking place, whether anyone knew about it, or
was able to have any control of it. A metaphysical cause, unlike a practical cause,
cannot be described with a single sentence." But what about, "God is the uncaused
cause of everything"? Of course, this doesn't really undermine the substance of
your position, but it does suggest you should not tie your point too closely to
sentences, or if you do, tweak your formulation slightly.
Though I don't believe anything substantive in your thesis depends on it, I
suggest you tidy up your language about laws causing things to happen. Laws don't
cause, they explain. No law caused the Belfast explosion, lighting the match did.
If we want to explain why or how lighting the match did, then we can appeal to
certain laws. Events proceed according to certain laws, but are not caused by them,
just as we drive along certain roads but our driving is not caused by them. (You
could, of course, go the Aristotelian route and distinguish between formal and
efficient cause; but your whole analysis here seems to be about efficient causes,
so that formal causes might be epiphenomenal here.)
The criterion of salience allows you to distinguish the conditions under which
physical events are appropriately appealed to as causes and the conditions under
which mental events are. Physical as well as mental events can be epiphenomenal (in
the weak sense), when they are not relevant or salient. Thus, "Under physical
descriptions, physical attributes are genuinely causal, and mental attributes are
epiphenomenal. But under mental descriptions, physical attributes are epiphenomenal
and mental attributes are genuinely causal."
You object to Kim's argument that if we say that a mental event M causes
another mental event M*, and that both are realized in physical events P and P*,
then P can cause P* all by itself and appeal to M and M* is unnecessary; therefore
M and M* are epiphenomenal. You conclude that if Kim is correct, "mental
descriptions are somewhere between being shallow and being outright falsehoods."
First, the conclusion of shallowness / falsehood does not follow. If M is
physically realized by P, there is nothing to require us to make P the reality and
M the epiphenomenon. An idealist can claim M is the reality and P is mentally
realized by M.
Secondly, the dilemma is resolved by holding that M is related to P as a
relationship among members of a set P is related to the members themselves. This
gives ontological priority to P, out of which M emerges as a relationship.
Similarly, you and I are correspondents, where you and I have ontological priority
over the relationship of correspondence. Therefore, if we assume that a belief is a
relationship among certain brain states / processes, then we can say that "M
causes M*" reflects that fact that P causes P*. When I believe something (M), there
is operating a certain brain state (P). To say that M causes M* reflects the fact
that P causes P*. However, this does not make M and M* epiphenomenal, because
communication with others and with myself can, in the face of our current ignorance
of Ps, be done only in terms of M. Talk of Ms is a higher level of discourse than
talk of Ps. The situation is analogous to my being able to write this to you even
though I know nothing about the actual physical events going on in my computer. We
can say the application software is causing certain changes in the computer which
show up on the monitor screen, but the application software is nothing more than
relationships among physical events. It's not as if the computer were in a
particular time and place and the application software were somewhere else.
Ami Thomasen writes:
First, about events:
You talk about an event as having many attributes that will be picked out
differently in different descriptions of that event, and that each may or may not
play a role in the "responsibility nexus" for a given event. It's an unusual
picture of events, ontologically, since events are usually taken to consist in an
object's coming to have (or lack) a particular property at a time, which would mean
there's only one property (or attribute) per event. You obviously have a more
robust conception of an event, where one and the same event can have many different
attributes (mental and physical, or drinking hemlock and emptying the cup).
As an aside, I'm not sure I'd want to say that, e.g., Socrates' death and
Xantippe's becoming a widow *are* the same event. If you aren't already familiar
with Alvin Goldman's arguments for the fine individuation of events (under which
these would count as two events), you might find them of interest--they're in
Chapter 1 of his A Theory of Human Action. And he argues, in part, that we
shouldn't identify such events since they have different properties.
I take it that generally, when folks speak about different descriptions of the
same event, they assume that these descriptions pick out the same property and
object. (E.g. Davidson seems to assume that mental and physical events are one and
the same, differing only at the level of descriptions.) But that wouldn't leave the
room you want to allow different properties (attributes) involved in an event to
have different (causal vs epiphenomenal) roles in a particular cause-effect
Then, about mental causation:
But suppose we start from some such robust conception of events, and allow that
one and the same event can have both mental and physical attributes (I presume
that, along the lines of Kim's example, we are supposing that these are separate
attributes, the mental not being reducible to the physical). Then, I take it your
idea is to say that the mental attributes may be efficacious with regard to the
mental effects (the physical there being epiphenomenal), and the physical are
efficacious with respect to the physical (the mental there being epiphenomenal).
Here's where I see the tricky parts arising. What justifies us in thinking that
the physical attributes of the event are ever (all) epiphenomenal? Seems like they
wouldn't be--like, without the physical aspects of the first event, the mental
event wouldn't occur (we can test this by drastically changing or removing the
relevant physical basis). I agree with your final example that *some* physical
aspects may be irrelevant (as in the difference between P and Q). But, at least as
long as we allow the mental to depend on the physical, it seems like whatever
mental attribute of an effect there is, there will be a physical attribute of the
cause that *is* part of the "nexus of responsibility". And if so, then we once
again face the threat that that physical aspect will get/deserve all the causal
credit, leaving those mental attributes of the causing event epiphenomenal after
all, even with respect to other mental events. Maybe this is a challenge that can
be met (I hope it can), but I think that's where the tricky part lies.
One way out of this is mine--to distinguish causation from determination and
provide the physical attributes with a role in determination but not causation as
such (which goes at the level of the mental), but that's another story (and one you
Without having to make such a distinction, a possible avenue for reply would be
to try to say that *both* the mental and physical attributes are necessary for
causing a mental attribute, giving the mental some work to do. Of course this would
need to be argued for, we'd need to know why the mental attribute is also required.
But your original example works in parallel with this kind of solution, because,
while drinking the hemlock [P] is (and being married [M] isn't) in the
responsibility nexus for Socrates' death [P*], drinking the hemlock [P] *and* being
married [M] are *both* in the responsibility nexus for Xantippe's becoming a widow
I'm somewhat in sympathy with your points about the physical, but I'm not sure
how much. I think that each of the physical atributes of any system could be
changed without changing its mental qualities. We could, after all, in principle
replace all of a brains components with silicon, and mentally it could remain the
same. Or a more dramatic example, there is real evidence that various mental
functions are actually performed by different parts of the brain at different
times in a person's life. (Sometimes within the space of a few months). Of course
the brain will always possess physical characteristics of some sort, but given
that no one physical characteristic is inevitable, I think that is a good sign
that each of those physical characteristics is epiphenomenal.
But-- Even if all of the above is true, I don't agree that the fact that a
*particular* physical characteristic could be replaced with a different one shows
that the physical is not (at least in part) causally responsible. If a tree blows
over in a storm, it is true that it could've fallen in just the same way if it was
pushed real hard, hit by a truck, or blown by a rather differrent gust, but none of
that shows that the wind that did push it over wasn't causally responsible (was >
merely "epiphenomenal") for that tree's falling. Similarly, my dining room table
is held up by its four wooden legs; it could equally well be held up by crates or
metal posts, but that doesn't mean that its legs are "epiphenomenal" with respect
to its standing. Mental characteristics may be *generically* (causally?) dependent
on merely some physical event > of a relevant type, but that does not mean that
they are not dependent > on the physical, or that the physical characteristics are
"epiphenomenal" with respect to them.