Re: Epiphenomenalism and Zombies

From: Stevan Harnad (
Date: Mon Jun 10 2002 - 23:04:12 BST

> On Sun, 26 May 2002, Anonymous wrote:
> Hi,
> Something I planned to comment on in last years CM302 coursework was the
> consciousness and evolution issue, but unfortunately I submitted mostly
> waffle.. One thing I overlooked was the key objection, epiphenomenalism,
> but I now (finally..) have a couple of things to say about that.
> Firstly if epiphenomenalism were true then we behave as though we are
> zombies.

I'm not sure why you say that. Epipenomenalism just means that
consciousness (feeling) does not have any independent causal power:
iIt is just some sort of effect dangling inconsequentionally on the
functional stuff that is doing the real work (but doesn't "need" the
feelings per se).

We don't know whether Zombies are possible. A Zombie would be a TT-passing
robot that acted just as if it felt, but really didn't. We would have no way
to know whether this was so, even if it were possible, hence even less
way of knowing whether it is indeed possible. But either way, it is not
connected to the question of epiphenomenalism. Feelings could have,
or fail to have, independent causal power whether or not there can
be Zombies.

> But if so on what basis could I deny with certainty the claim
> that I might be a zombie?

The claim that you are a Zombie is the claim that you don't really have
feelings, you just act as if you did. Only you can know whether that's
false (for if it's true, you don't know anything at all: you are just a
system acting as if it knew). So (assuming you feel), you can be sure you
feel (indeed, as Descartes points out, you couldn't possibly doubt it).

What none of us can know with that kind of certainty is whether anyone
ELSE feels.

> Surely without the input of mental events
> there's nothing to cement this belief? And if I'm granted a zombie twin,
> it must also claim that it could not possibly be a zombie.

I'm afraid this is all beside the point. There is nothing to "cement."
If you feel, you feel; if you don't you don't. Feeling is already
"mental." No need for further mental stuff.

And I'm not sure why you introduce a Zombie twin: You mean you feel, but
the "twin" does not? Perhaps, but really only interesting if you are
natural and the "twin" is man-made (and then it's just the robot TT all
over again). A zombie bio-twin is pure sci-fi: Why should the same genes
and brain produce feelings in one case and none in the other?

> Also I see a particular difficulty in accounting for the statement,
> 'I exist / there is existence'. How could the claim be made without
> input from a mind that knows it to be true?

Descartes' Cogito was "I think, therefore I exist," not "I exist
therefore there is existence." (I don't even know what "there is
existence" might mean! Existence exists? Sounds odd, and at the least,

Descartes' Cogito is supposed to be that I cannot doubt that I am thinking
if/when I indeed thinking (for doubting itself would be thinking). So
I must exist.

But that's a bit mysterious. Here's a clearer version: "I feel. Therefore
it makes no sense to doubt that there is feeling going on (when I feel)."

(When I feel, it is always NOW, so I can't deny that feeling is going
on right NOW; perhaps also it is going on HERE, wherever here is --
but that is already sounding doubtable. Our spatial ideas could be all
wrong. And so could our ideas about what "I" am.)

So the Cogito really just says: When I am feeling something, it is
impossible for me to doubt that that feeling is happening then. So
whatever feelings are, they do exist. The way feelings feel, it is more
or less part of them that a feeler is doing the feeling, so maybe that
is what the "I" is -- but this may be splitting hairs a bit to finely.

I feel, therefore there is feeling going on now: "I" seem to be the one
doing the feeling, whatever that means (I know what it feels like, but
not what it means: maybe I'm waiting for cognitive science to tell me).

> Existence (leaving aside
> mathematical existence) cannot be logically deduced. For example,
> when questioned a zombie could not justifiably say, 'I must exist,
> otherwise I couldn't make this reply', as it can only assume that is
> making a reply.

Saying things and inferring (assuming) things is no problem. A computer
can do that. What makes it a Zombie is that it DOESN'T feel.

By definition, if there can be a Zombie, then, although it states the
Cogito, its statement is false, because although it says "I am feeling,
therefore feeling is going on," that is not true: No feeling is going

So no "existential" conclusions are to be drawn from these false
statements (false precisely because we are assuming it's really a Zombie
who is saying them).

> How could it respond to the follow-up, 'and how did you
> know for sure that you were making a reply?' So surely a zombie couldn't
> claim to be sure about existence and neither could we if epiphenomenalism
> were true.

Chris, I'm afraid this is missing the mark. You have to distinguish what
the Zombie says from what is true. It says "I feel," but, on the premise
that it really is a Zombie, that statement has to be false. And
therefore no Cartesian conclusions can be drawn from it (either by you
or by the Zombie, which may as well be a rolling stone).

Nor does this have anything to do with epiphenomenalism, one way or the

These are tricky concepts, puzzled over by human minds for centuries:
Don't expect to get them all sorted out the first time (or the second,
or third...)!

> And finally the argument that evolution won't favour a conscious entity
> over a behaviourally identical non-conscious entity doesn't support by
> itself the claim that consciousness couldn't affect evolution, as this
> claim would only be true if epiphenomenalism were true.

The suggestions that evolution can't mind-read is just another reason
for believing epiphenomenalism.

It's good to keep in mind the alternatives, when you speak of
epiphenomenalism. They include (1) telekinetic dualism (mind over
matter; feeling as an extra causal force in the universe) and (2) various
forms of functionalism (which try to show that feelings have some
functional/causal role to play: I suggest that if the role is not
telekinetic, then it is doomed to be epiphenomenal, i.e., no causal
role at all; there is no other alternative).

Your statement above seems a little scrambled. Evolution is simply the
functional consequences of whatever helps survival and reproduction. To
show that epiphenomenlaism is false, you have to show what causal role
feelings themselves could play in helping survival and reproduction. Be
careful to make sure that it is feelings themselves, and not just some
unconscious function with which they happen to be mysteriously correlated,
that helps survival/reproduction (because otherwise the feelings become
epiphenomenal after all). And be careful also not to introduce telekinetic
dualism (because there is no evidence for it, and it would probably not
be compatible with the other laws of physics to introduce an independent
mental "force" alongside the known ones [gravity, electromagnetism,
atomic forces, etc.]).

> If conscious x
> and nonconscious y are identical individuals with some new behaviour,
> and both already exist, then neither is favoured. But it might be that
> x has a higher probability of being born in the first place, if the
> adaptation required to produce it is smaller (due to some functionality
> of consciousness) than the adaptation required to produce y.

> Chris Mcintosh

Yes, but the trick is to say exactly what that "some functionality of
consciousness (feeling)" might actually be; otherwise this is just

But bravo, Chris, for continuing to think about this so long after marks
have any causal role to play...

Cheers, Stevan

This archive was generated by hypermail 2.1.4 : Tue Sep 24 2002 - 18:37:31 BST