You are on page 1of 25

Final draft. Published in Philosophical Studies 139 (2008): pp. 73-89.

The Inconceivability of Zombies

Robert Kirk

ABSTRACT
If zombies were conceivable in the sense relevant to the ‘conceivability argument’
against physicalism, a certain epiphenomenalistic conception of consciousness – the ‘e-
qualia story’ – would also be conceivable. But (it is argued) the e-qualia story is not
conceivable because it involves a contradiction. The non-physical ‘e-qualia’ supposedly
involved could not perform cognitive processing, which would therefore have to be
performed by physical processes; and these could not put anyone into ‘epistemic
contact’ with e-qualia, contrary to the e-qualia story. Interactionism does not enable
zombists to escape these conclusions.

I. INTRODUCTION
Zombies would be like us in all physical respects, but without phenomenal
consciousness. It is widely agreed that if zombies are possible, physicalism is false. A
much debated argument for the possibility of zombies starts from the claim that they are
conceivable, then urges that whatever is conceivable is possible. Many physicalists
agree that zombies are conceivable – even in a strong sense – but disagree that
conceivability entails possibility. Whatever may be the correct view on that last point, I
think all are wrong about the conceivability of zombies. I will argue that zombies are
not conceivable in any sense strong enough for the conceivability argument.1
There are plenty of objections in the literature to the conceivability of
zombies.2 But the idea is so alluring that those who think zombies are conceivable tend
to feel there must be something wrong with the objections; the zombie idea may be
problematic (they say) but surely it is not actually incoherent. I will argue that, on the
contrary, it is indeed incoherent, involving a grossly distorted conception of phenomenal
consciousness. To counter its appeal there needs to be a way to bring out the wrongness

1
of that conception – an approach which has strong intuitive impact, but does not just pit
one lot of intuitions against another. That is what I will try to provide.
My approach centers on a particular conception of consciousness, the ‘e-
qualia story’, to be explained shortly. This is a variety of epiphenomenalism: the view
that although all physical events are physically caused, consciousness depends on non-
physical ‘qualia’ which are physically caused but have no physical effects. The
argument of this paper has two stages. The first is to argue that the e-qualia story is not
conceivable (in the relevant sense) because it is contradictory. The second is to argue
that if zombies are conceivable, so is the e-qualia story. In outline:

(A) The e-qualia story is not conceivable.


(B) If zombies were conceivable, the e-qualia story would be conceivable.
Therefore zombies are not conceivable.3

(A) is defended in sections III-IX; (B) in sections X and XI.

II. ‘ZOMBIES’ AND ‘CONCEIVABLE’


The idea of zombies suggests itself as soon as one accepts the causal closure of the
physical. If every physical effect has a physical cause, all human behavior is explicable
in physical terms. But then how does consciousness – phenomenal consciousness, the
sort involved in there being ‘something it is like’ to have experiences – fit into the story?
Apparently it can only be a causally inert by-product, and epiphenomenalism or
parallelism must hold. In that case, as G. F. Stout argued,

it ought to be quite credible that the constitution and course of nature would be
otherwise just the same as it is if there were not and never had been any experiencing
individuals.4

What Stout envisaged is a ‘zombie twin’ of our world: a physical duplicate of the actual
world on the assumption that the physical world is closed under causation – so that
everything physical goes on just the same – but without phenomenal consciousness.
Zombies must be understood to be complete physical systems in the sense that all
effects in them are produced physically. They can be defined as follows: creatures

2
without phenomenal consciousness but physically just like us on the assumption that the
actual physical world is causally closed.5
Once the zombie idea has been explained it is almost impossible to resist the
thought that such things are at least conceivable. Most philosophers would agree they
are conceivable in some sense, but that sense is often too broad for the purposes of the
conceivability argument. For our purposes it is enough to say that a proposition or
situation counts as inconceivable if it can be known a priori to be false; otherwise it is
conceivable.

III. THE E-QUALIA STORY


Does the possibility of zombies entail epiphenomenalism? Some have argued that it
does; but that seems to be a mistake.6 Why shouldn’t zombies be possible even if the
actual world is interactionistic? Later I shall defend a different claim: that the
conceivability of zombies entails the conceivability of the e-qualia story, a particular
version of epiphenomenalism. This story consists of theses (E1)-(E5) below. Although
close to epiphenomenalism as usually understood, it does not aim to be a fair reflection
of epiphenomenalists’ views, nor of current views about qualia.

(E1) The world is partly physical and its physical component is closed under causation:
every physical effect has a physical cause.7
(E2) Human beings are physical systems related to a special kind of non-physical
properties, ‘e-qualia’. E-qualia make human beings phenomenally conscious.
(E3) E-qualia are wholly caused by physical processes but inert: they have no effects
either on the physical world or among themselves.
(E4) Human beings consist of nothing but functioning bodies and their related e-qualia.
(E5) Human beings are able to do such things as notice, attend to, think about, compare,
and (on occasions) remember their e-qualia.

I take it that (E1) and (E4) are clear enough for our purposes, given (E2), and will
explain (E2) now, and say more about (E3) and (E5) in the next two sections.
The notion of e-qualia is significantly different from at least one current
notion of qualia. David Chalmers defines qualia, or ‘phenomenal qualities’, as “those
properties of mental states that type those states by what it is like to have them”.8 It
seems the existence of qualia in that sense could be accepted even by physicalists, since

3
the definition does not entail that these properties must be non-physical. Nor, unlike the
e-qualia conception, does it entail that they could be stripped off without affecting the
physical world. To say qualia in that sense exist amounts to no more than saying we are
phenomenally conscious. Clause (E2) of the e-qualia story is crucially different: it
emphasizes that what makes us conscious is our relation to these special non-physical
properties.
Although perhaps not many philosophers would endorse the position
outlined by (E1)-(E5), many would assume it is at least conceivable. The argument in
the following sections aims to show that the e-qualia story is not conceivable in the
relevant sense because (E1)-(E4) are incompatible with (E5). If there were e-qualia
satisfying (E1)-(E4), the epistemic relations envisaged by (E5) would be ruled out. It is
notorious that epiphenomenalism gets into trouble over our epistemic relations with
qualia; I aim to show that this difficulty is terminal.

IV. E-QUALIA, EPISTEMIC CONTACT, AND COGNITIVE PROCESSING


By (E3), e-qualia are caused by physical processes but inert. E-qualists will
acknowledge that the qualities of our conscious experiences (which is what e-qualia are
supposed to be) seem to have effects on our behavior, but insist it is only apparent
causation, not real. When a thorn in my finger hurts and makes me wince, most of us
would say the harm to my finger caused pain, which in turn caused the wince. E-
qualists, in contrast, will say that one physical event (stimulation of pain-receptors)
caused both a pain e-quale and another physical event (wincing). It seems that the pain
caused the wincing (e-qualists would say) but in fact what caused it was the physical
event, which also caused the pain. To see that this account of conscious experience is
not just strange but inconceivable, we must first note a consequence of the inertness of
e-qualia.
Like all varieties of epiphenomenalism, the e-qualia story is an attempt to
characterize the metaphysics of a world whose inhabitants are phenomenally conscious
just as we are: that is, in the sense that there is something it is like for them to perceive
the world around them, and to have sensations and other experiences. We are able to do
such things as notice, attend to, think about, compare, and on occasions remember the
qualities of our experiences. So the e-qualia story must ensure that the inhabitants of a
world satisfying its conditions – an E-world – engage in those activities too. And since
in an E-world the qualities of its inhabitants’ experiences are provided by e-qualia (by

4
(E2)), its inhabitants must be able to do such things as notice, attend to, think about,
compare, and remember their e-qualia: for short, they must be in epistemic contact with
their e-qualia.9 Hence (E5).
Epistemic contact in that sense involves cognitive processes such as
conceptualization and the storing and retrieving of information, which in turn involve
the causation of changes and the persistence of unconscious items. The case of
remembering illustrates both points. If information is stored about an event, that must
leave traces which can have effects later, when the subject is recalling or otherwise
using the stored information; but these traces are not normally conscious. Other forms
of epistemic contact, such as thinking about, attending to, or comparing items, in turn
depend on information being stored and retrieved. Epistemic contact also involves
causation and unconscious persistence, because it depends on conceptualization, which
requires more or less persisting cognitive structures contributing causally to the
subject’s ability to group things together.10
But e-qualia are inert, and so cannot themselves engage in the cognitive
activities which would be necessary to put E-worlders into epistemic contact with their
e-qualia. Also, they do not persist through time (at least not unconsciously) and for that
reason too cannot perform the cognitive functions necessary for epistemic contact. The
point is crucial. By (E4), human beings consist of nothing but bodies and their
associated e-qualia. It follows that in an E-world, the cognitive functions in question
must be performed by physical processes. You might suggest that non-physical items
other than e-qualia could perform those functions; but by (E4) there are no such items.
The fact that the cognitive work in an E-world must be done by physical
processes presents a fatal difficulty for the e-qualia story.

V. HOW IS EPISTEMIC CONTACT POSSIBLE?


According to the e-qualia story, physical processes cause a temporally extended
complex of e-qualia constituting an individual’s stream of consciousness. At any given
time the individual is supposed to be in epistemic contact with whichever e-qualia are
occurring at that time. But what ensures there is such contact? To get an idea of the
difficulty we can start by imagining an E-world w that becomes ‘zombified’: at a certain
time it loses all its e-qualia. Since by (E1) the physical component of an E-world is
closed under causation, the absence of e-qualia leaves the physical component of w
functioning as before; the difference is that its inhabitants are zombies. Suppose, then,

5
that a single temporally extended complex of e-qualia, ψ, is introduced into this zombie
world. ψ is the kind of e-qualia complex that every E-worlder is supposed to be in
epistemic contact with, but in this example it is not caused by physical processes but has
come into existence spontaneously. Might some E-worlder nevertheless be in epistemic
contact with ψ: think about, attend to, compare, or remember some of the e-qualia in it?
11

No. ψ is an e-qualia complex suitable for one individual; so at most one E-


worlder could be in epistemic contact with it. However, by definition there is nothing to
connect ψ with one particular E-worlder rather than with any other. Therefore nothing
could put any particular E-worlder into epistemic contact with ψ. Even if it somehow
constituted a stream of consciousness all by itself, it would be one whose constitutive
experiences no one could attend to, think about, or remember. This would be so even if
w had only one inhabitant. Nor would it matter how many such e-qualia complexes were
introduced into the zombie world w, nor how numerous its population was. The mere
co-existence of e-qualia complexes and zombies would not result in anyone’s being in
epistemic contact with those complexes.
However, the e-qualia story does not envisage mere co-existence. Each
complex of e-qualia is supposed to be caused by some of the physical (presumably
neural) processes in a human-like body; and that might at first appear to ensure that
those processes were in epistemic contact with it. Causation alone would not be enough,
however. Assuming physical processes can cause non-physical items in the first place
(as we must for argument’s sake) there is no reason a priori why the laws of nature in w
should not be such that, for any arbitrary type of e-qualia complex, the neural processes
in a given E-worlder’s body caused an e-qualia complex of just that type. Suppose for
example that I had a counterpart K in w. There is no a priori reason why processes in K’s
body should not cause an e-qualia complex of a type that might have been associated
with an Inuit whose life included dangerous encounters with polar bears. I am not an
Inuit, have comparatively slight experience of snow and ice, and have never been close
to a polar bear. So even if the e-qualia story were true of me, very few, if any, of my
neural processes could have put me into epistemic contact with the e-qualia in such a
complex. Since the neural processes in my counterpart’s body mirror my own, they
would not put anyone into epistemic contact with that particular e-qualia complex.
An obvious reply would be that the laws of nature in an E-world ensure that
the relevant neural processes in its inhabitants’ bodies cause only such sequences of e-

6
qualia as stand in an appropriate isomorphic relation to (features of) those processes.12 If
I am attending to a sequence of clock chimes, for example, then my cognitive
processing of the chimes must be temporally and otherwise correlated with the e-qualia
they supposedly cause. E-qualists might now seem to have a satisfactory response to the
original question. Shortly I will argue that that response does not work; but we need to
be as clear as possible about the difficulty for e-qualists, and the resources they have for
dealing with it.

VI. THE PROBLEM


It is agreed on all sides that we can refer to our experiences. But if referring requires the
referent to have an effect – however indirect – on the referrer, then because e-qualia
would be inert, no one could refer to them. For those who accept a causal account of
reference, the e-qualia story is mistaken for that reason alone. However, the same
thought also leads epiphenomenalists to reject the causal account of reference, so
presupposing it when arguing against the e-qualia story would be question-begging. My
argument does not depend on a causal account; nor is the problem I am focusing on the
same as the problem epiphenomenalists have over referring to experiences.
The following question helps to clarify the problem:

(Q) How, in an E-world, could physical processes in an individual body contribute to


anyone’s being in the relevant sort of epistemic contact with e-qualia: what could hook
them up, epistemically?

I will argue that e-qualists cannot give a satisfactory answer to this question. The
structure of an E-world would prevent its inhabitants from being in epistemic contact
with ‘their’ e-qualia because the cognitive processing essential for the relevant sort of
epistemic contact would have to be performed by physical processes epistemically
insulated from all e-qualia. That general point is based on the following counter-
example to the suggestion that if e-qualia are caused by and isomorphic to the relevant
physical processes, then there is epistemic contact.

VII. MY CRANIAL CURRENTS


Suppose that ours is an E-world, and that by some quirk in the prevailing laws of nature,
those of our brain processes which allegedly cause e-qualia also induce minute patterns

7
of electrical activity which are in relevant respects isomorphic to them – but have no
effects on them. For each sensory modality these cranial currents mirror the relevant
brain activity just as it is supposed to be mirrored by e-qualia. The patterns formed by
currents induced by activity in my visual cortex, for example, would serve as a
continuous record of my visual experiences – hence (as the e-qualia story implies) of
my visual e-qualia.
As it happens, no one actually observes these induced currents or knows
anything about them. It follows that their being caused by and isomorphic to some of
my brain processes is not enough to put anyone into epistemic contact with them. From
the point of view of my epistemic history they might as well not be there; they could
cease or start up again, or change character, without affecting the epistemic situation in
the slightest. So they are a counter-example to the suggestion that for such processes to
cause or be isomorphic to something would be enough to put someone into epistemic
contact with it.
That result applies directly to e-qualia. Even if my brain processes cause or
are isomorphic to my e-qualia, that cannot put me into epistemic contact with them. My
e-qualia, like my cranial currents, could cease or start up again, or change, without
affecting the epistemic situation. They are so thoroughly insulated epistemically that
they might as well not be there at all. So far, then, e-qualists cannot answer question
(Q).
But if causation and isomorphism will not do the trick, what else can e-
qualists appeal to? It must be something additional or different. But by definition there
is nothing in an E-world but its physical component and its e-qualia, so the only
available factors are:
(a) the intrinsic properties of the physical component;
(b) the intrinsic properties of e-qualia;
(c) the ways in which natural necessity might relate those properties so as to constitute
subjects in epistemic contact with their e-qualia.
None of these can help e-qualists.
(a) If the intrinsic properties of my physical cognitive processes could put me into
epistemic contact with anything, they could do it for my cranial currents; which they
don’t. Nor could any changes to them make a relevant difference, since the argument
depends only on the broad features of an E-world without reference to any physical
details.

8
(b) Keep in mind that the cranial currents case is a counter-example to the general
suggestion that if certain suitable physical cognitive processes cause or are isomorphic
to something – anything – which has no effects on them, then they are enough to put
someone into epistemic contact with it. It is a counter-example regardless of the intrinsic
properties of the item supposedly caused by and isomorphic to the physical processes in
question. So it remains a counter-example if that item happens to consist of e-qualia,
and regardless of what the intrinsic properties of those e-qualia may be. Their intrinsic
properties could be whatever you please (provided they remained inert); I should still
not be able to notice, think about, attend to, remember, or compare them.
If e-qualists are going to be able to escape the cranial currents argument,
therefore, it must be on the basis of (c) the ways in which the physical and non-physical
components of an E-world are related (presumably by natural necessity). But if
causation by, and isomorphism to, physical cognitive processes are not enough for
epistemic contact, what else might e-qualists appeal to? The only other suggestion I
know of is that the counterfactual dependence of particular e-qualia on particular
preceding physical events might be enough. But that will not work because the cranial
currents case remains a counter-example. If ours were an E-world where the physical
processes which caused e-qualia also caused isomorphic patterns of cranial currents,
there would be the same counterfactual dependence between those currents and
subsequent physical events as the present suggestion envisages between e-qualia and
those physical events. Since I would remain out of epistemic touch with my cranial
currents, the same must go for my e-qualia.
The cranial currents case is more than a counter-example to the two
suggestions just discussed. It highlights features of an E-world which ensure that e-
qualia are epistemically insulated from all cognitive processes. Both e-qualia and cranial
currents are (we imagined) caused by the same brain processes, while neither has any
effects on them. Those two cases have the same essential structure; the only differences
are between the intrinsic properties of e-qualia and those of cranial currents. But we saw
two paragraphs back that the details of those intrinsic properties cannot affect the
argument; so the differences between e-qualia and cranial currents are not relevant. It
follows that if any feature of an E-world could make our brain processes put us into
epistemic contact with our e-qualia, it would equally make them put us into epistemic
contact with our cranial currents. But – given the scenario sketched earlier – nothing

9
could put us into epistemic contact with our cranial currents; therefore nothing could put
us into epistemic contact with our e-qualia. (See also Objection 3 in the next section.)
We have considered all the resources by means of which e-qualists might
have been able to escape the cranial currents counter-example. It turns out that they
cannot do the job. It is the overall structure of the e-qualia story, not the details of the
realm of e-qualia, still less the details of the physical realm, that would prevent the
physical processes in an E-world from putting anyone into epistemic contact with e-
qualia. I conclude that, quite generally, clauses (E1)-(E4) of the e-qualia story are
inconsistent with (E5).
E-qualists will be keen to press objections. After summarizing the argument
so far, I will reinforce it by discussing those I have come across.

VIII. THE ARGUMENT SO FAR


1. E-qualia are inert (E3).
2. Being in epistemic contact with one’s e-qualia involves activities such as noticing,
attending to, thinking about, remembering, and comparing them ((E5) and section IV).
3. These activities involve causation and the persistence of unconscious items and
structures, and therefore (in an E-world) can be performed only by physical processes
(IV).
4. That raises the question (Q) of how physical processes could put anyone into
epistemic contact with e-qualia (V and VI).
5. One suggestion is that the causation of e-qualia by, and their isomorphism to, certain
physical processes might combine to put E-worlders into epistemic contact with their e-
qualia (VI, VII).
6. The example of the patterns of electric currents induced in my brain by the relevant
physical processes is a counter-example (VIII).
7. It is also a counter-example to the idea that the intrinsic properties of e-qualia or of
the physical component of an E-world might provide for epistemic contact, and to the
suggestion that counterfactual dependence of particular e-qualia on particular preceding
physical events might do the trick (VIII).
8. The same example helps to bring out the fact that it is the overall structure of the e-
qualia model of consciousness which prevents physical processes from putting anyone
into epistemic contact with e-qualia (VIII).

10
9. Thus no one in an E-world could be in epistemic contact with e-qualia in the sense of
clause (E5) of the e-qualia story, and the e-qualia story involves a contradiction: it is
inconceivable in the relevant sense (VIII).

IX. OBJECTIONS
Objection 1: ‘It looks as if the argument presupposes a causal account of aboutness and
reference. If so, you’re after all just pitting your intuitions against those of your
opponents – and begging the question.’
Reply 1: That’s a misunderstanding. The argument doesn’t presuppose or in any way
depend on a causal account of aboutness. It depends on the fact that e-qualia cannot do
cognitive processing and are epistemically insulated from the physical processes which
do such processing. Certainly, part of the trouble is that e-qualia have no effects (and
that mere counterfactual dependence would not be enough). But another is that they
cannot provide for the persisting unconscious structures involved in cognitive
processing. The upshot is that the relevant cognitive processing would have to be done
by physical processes; while the cranial currents example helps to show how the
structure of the e-qualia model of consciousness is guaranteed to prevent such processes
from being able to put an individual into epistemic contact with e-qualia. Reflection on
the reasoning might incline one to favor a causal account of aboutness and reference;
but the argument doesn’t depend on a causal account.

Objection 2: ‘Your presentation reflects a tendentious view of the subject of experience.


You imply it is the functioning body or brain: a conception in which e-qualia are add-on
extras. From that perspective the notion of e-qualia can seem mysterious. But for e-
qualists a different perspective is more natural, according to which the primary locus of
the subject is in the stream of consciousness itself. Far from the subject being
constituted by physical processes, those processes are in a way peripheral.’
Reply 2: Suppose for argument’s sake that the e-qualia caused by physical processes in
an individual’s body did form a stream of consciousness – even a subject. By the
argument in section VII, those e-qualia’s inertness would still ensure that they and the
putative subject were epistemically insulated from all cognitive processes. No subject
could think about, notice, attend to, or remember items in that stream of consciousness:
no one could be in epistemic contact with them in the relevant sense. So this objection
does nothing to undermine the argument based on the cranial currents case.13

11
Objection 3: ‘You overlook a crucial possibility: we might stand in an epistemic relation
of acquaintance to e-qualia. That would break the analogy between e-qualia and cranial
currents. Its mere possibility blocks the inference from cranial currents to e-qualia,
rendering the cranial currents case irrelevant and demolishing the whole argument of
section VII. When e-qualia are caused by bodily processes, the acquaintance relation
puts the subject into epistemic contact with them, within the stream of consciousness.
Beneath the surface of consciousness there are various kinds of information processing.
By virtue of the counterfactual dependence of these underlying processes on e-qualia,
and the causal impact they have back on the stream of consciousness, the subject is after
all in a position to notice, compare, and remember e-qualia.’14
Reply 3: Keep in mind that what e-qualists have to explain is how, in an E-world,
physical processes in an individual body could contribute to the relevant sort of
epistemic contact with e-qualia: that is question (Q). I will argue that the suggested
notion of ‘acquaintance’ goes no way towards answering that question.
Evidently the relata in the acquaintance relation are supposed to be e-qualia
on the one hand and ‘the subject’ on the other. But something must underlie this
relation: must make the difference between its holding and its not holding. What could
that something be? We know the only resources available are: (a) intrinsic properties of
the physical component; (b) intrinsic properties of e-qualia; (c) the ways in which
natural necessity might relate those properties. None of these can provide for the
relevant kind of epistemic contact.
(a) is ruled out directly by the cranial currents example. Since the intrinsic
properties of the physical component of an E-world could not put me into epistemic
contact with my cranial currents, they could not put me into epistemic contact with
anything else that was related to my physical cognitive processes only by being caused
by and isomorphic to them.
The objection seems to envisage that (b) what underlies the relation is an
intrinsic property of e-qualia. But this thought was anticipated and dismissed in section
VII. My cranial currents are a counter-example to the suggestion that if something –
regardless of its nature – were caused by and isomorphic to some of my physical
processes (while not affecting them), then these would put me into epistemic contact
with that something in the relevant sense (section VII). It follows, as we saw, that the
intrinsic properties of the item in question make no difference to the argument.

12
Whatever they may be, they are locked up beyond the epistemic reach of my cognitive
processing. So if what underlies the acquaintance relation is an intrinsic property of e-
qualia, then both that property and the subject are marooned within that realm, and
cannot put me into any relevant kind of epistemic contact with e-qualia.
It is suggested that (c) acquaintance operates in combination with the
counterfactual dependence of underlying physical processes on relations among e-
qualia. This dependence, together with the effects of those physical processes on the
stream of consciousness, is supposed to put the subject ‘in a position to notice, compare,
and remember e-qualia’. However, I argued in section VII that counterfactual
dependence could not provide for that kind of epistemic contact; and we have just seen
that the suggested relation of acquaintance does not block that argument.

Objection 4: ‘You’re construing the special epistemic access that e-qualists claim we
have to our qualia on the model of familiar kinds of epistemic access such as sense
perception. But according to e-qualists, our epistemic access to qualia is private in this
sense: to have them is to know them. The accessibility of electric currents is just
observability: they are observable by anyone suitably equipped. So the cranial currents
example doesn’t work because patterns of electric currents are nothing like what e-
qualia are supposed to be.’
Reply 4: The argument shows there is no way anyone could ‘have’ e-qualia in the
relevant sense: no one could be their subject and know them. E-qualia are certainly very
different from electric currents, and our relation to them would indeed be unlike our
relation to such things. But the argument exploits a fact which holds regardless of how
different those relations and their relata may be in other respects: that the cognitive
activities involved in epistemic contact cannot be performed by e-qualia and must be
performed by physical processes (section IV) – which cannot be epistemically linked
with e-qualia (section VII).

Objection 5: ‘E-qualists need only point out that a different conception of e-qualia
might be devised, according to which they were causally active.’
Reply 5: That is not in the e-qualia story; and I am arguing only that the e-qualia story
itself, not some variant of it, is inconceivable. (See also X below.)

13
I conclude that the conclusion stands. E-worlders’ physical processes cannot put anyone
into epistemic contact with their e-qualia in the sense of (E5), so that (E5) is
inconsistent with the rest of the e-qualia story. That completes my case for thesis (A): in
the relevant sense, the e-qualia story is inconceivable. I now have to show (B): that the
conceivability of zombies would entail the conceivability of the e-qualia story. The
argument has two phases, set out in the next two sections.

X. CONCEIVABILITY OF ZOMBIES WOULD ENTAIL CONCEIVABILITY OF


INERT CONSCIFICATION
The claim is not that the conceivability of zombies would entail that the e-qualia story
was true,15 only that it was conceivable. So I have to show that the conceivability of
zombies would entail that the e-qualia story cannot be known a priori to be false (see III
above). The broad idea is simple: if zombies were conceivable, then conceivably what
started off as a zombie twin of our world could be transformed into a world where the e-
qualia story held. I expect many readers will concede that straight off, but it is vital to
see that some apparent escape routes are blocked; this needs some care.
Not all zombists are dualists: some are physicalists, some idealists, some
panpsychists. However, as I will explain shortly, all zombists are committed at least to
the conceivability of dualism: the view that the world consists of a physical component
and a non-physical component which are ‘separate existences’ in the sense that neither
entails the existence of the other; and the non-physical component makes us conscious.
(The non-physical component might consist of our minds, or non-physical qualia, or a
single cosmic center of consciousness. For convenience I will refer to the non-physical
component in the singular.)
Here is why zombists are committed to the conceivability of dualism. Given
the definition of zombies (section II), the conceivability of zombies entails the
conceivability of a purely physical world that is closed under causation, and whose
inhabitants are not only behaviorally like us, but physically just as physicalists suppose
we are – yet not conscious. I assume it is at least conceivable that we are in physical
respects just as physicalists suppose we are. So, since zombies would lack something we
have, what we have and they would lack must be non-physical. This non-physical
component of our world, which made us conscious, would be logically independent of
its physical component.16 That is dualism. So if zombies are conceivable, so is dualism.

14
I will now argue that zombist dualism commits you to the conceivability of
something very close to the e-qualia story: ‘inert conscification’. (In the next section I
will go on to argue that the conceivability of inert conscification entails that of the e-
qualia story itself.)
We can take it that dualism is either interactionist, epiphenomenalist, or
parallelist. Now, epiphenomenalists believe the physical component of our world is a
causally closed system, while the non-physical component is caused by the physical
component but inert. Just as their position entails that, conceivably, the extinction of this
inert non-physical component would transform our world into a world of zombies, so it
entails that, conceivably, the genesis of a suitable inert non-physical item in a zombie
twin of our world would result in the existence of a (phenomenally) conscious human-
like population.17 Thus epiphenomenalists are committed to the conceivability of inert
conscification in the following sense:

(C) A causally inert non-physical item ψ could be associated with a zombie twin z of our
world so as to transform it into a world z* whose inhabitants enjoyed our kind of
phenomenal consciousness, including epistemic contact. ψ is caused by the physical
component of z*.

Parallelists too, although they hold there is no causal action between the physical and
the non-physical, are committed to the conceivability of (C). For they must hold that
causal relations are contingent, so that even if a does not in fact cause b, it must be
conceivable that it should do so.
Zombist interactionists, however, may think they can escape commitment to
(C), first because they reject the causal closure of the physical; second because they
insist that the non-physical component of our world is causally active (ert?). I will argue
that, even so, they must concede that (C) is conceivable. Before defining zombist
interactionism I will refine the definition of a ‘zombie twin’ world. A world z is a
zombie twin of ours just in case:

z is a purely physical, causally closed system; z is physically as like the actual world as
possible (with physical causes substituted for non-physical ones where necessary); the
human-like inhabitants of z lack phenomenal consciousness.

15
Zombist interactionism can now be defined:

(ZI) Zombies are conceivable and the actual world consists of a physical component and
a non-physical component. The latter includes (or consists of) something ψ* such that:
(i) ψ* makes us phenomenally conscious and keeps us in epistemic contact with our
experiences; (ii) ψ* has effects on the physical component of the world and is affected
by it.

I will argue that (ZI) entails that conceivably our world as thus characterized – I will
refer to it as i – could be transformed into a world like z*; hence that (C) is conceivable.
What forces zombist interactionists into this position is their peculiar conception of
consciousness. For (ZI) entails the following three propositions.
(1) Conceivably the laws of nature governing i (the world according to (ZI)) could
change at a certain time so that from that time on: (i) no non-physical items in i had
effects; (ii) whatever had been directly or indirectly caused by non-physical items was
instead caused by physical items, so that every physical effect now had a physical cause;
(iii) all non-physical items other than ψ* ceased to exist, so that ψ* was the only non-
physical item.

(2) If the changes described in (1) are conceivable, then it is also conceivable that after
the changes ψ*, in spite of being inert, should continue to make the inhabitants of i
conscious and keep them in epistemic contact with their experiences.

(3) Given (1) and (2), (C) is conceivable.

If (ZI) entails (1), (2), and (3), obviously it entails that (C) is conceivable. I will argue
that (ZI) does indeed entail those premisses.

Premiss (1). Condition (i) is unproblematic. Causation is contingent, so zombists cannot


deny that conceivably anything that actually has effects – including ψ* – might have
existed without having effects.18
(ZI) also entails that condition (ii) is straightforwardly conceivable together
with (i), at least for the case of physical effects. For (ZI) has it that zombies are
conceivable; and in a zombie world all physical events in i would be caused physically.

16
(One result would be that our successors in i behaved exactly like us. Some
interactionists might deny that physical events could cause human-like behavior, but
they could not be zombists.)
As to non-physical effects, the question is whether there are any which could
not conceivably be produced physically; and interactionists can have no a priori
objections to that. So (ZI) entails that conceivably conditions (i) and (ii) are jointly
satisfiable.
(iii): since (ZI), being a variety of dualism, entails that the non-physical
component of reality is logically independent of the physical component, it entails that
that all non-physical items in i might conceivably cease to exist; a fortiori that
conceivably all non-physical items other than ψ* could cease to exist, leaving ψ* (now
inert) as the only surviving non-physical item. (In that case, notice, if ψ* ceased to exist
the result would be a zombie twin of our world.) Clearly there is no inconsistency
between that and conditions (i) and (ii); so (ZI) entails (1).

Premiss (2). Evidently, one consequence of the changes envisaged in conditions (i) and
(ii) of premiss (1) would be that the physical component of i was closed under
causation. Also, because all those kinds of physical items that had previously been
caused non-physically would continue to be caused (though physically), i would remain
physically similar to what it had been; in fact the physical component of i would be
exactly like a zombie twin of our world. The question now is whether (ZI) entails that
conceivably ψ* would continue to make the inhabitants of i conscious and keep them in
epistemic touch with their experiences, or whether the changes would absolutely19
prevent that.
Consider, then, what differences the changes would make to i. They are that
(in i after the changes) (i) no non-physical items have effects; (ii) physical items have all
the kinds of effects that were originally produced by non-physical items; (iii) the only
non-physical item is ψ*. (i) and (ii) are crucial, of course, since they stop i from being
an interactionist world and put it on the way to being like z*. But do they absolutely
prevent ψ* from contributing to making i’s inhabitants conscious? If so, that must be on
account either of (a) what ψ* and any other non-physical items were doing before the
changes – what they caused, what functions they performed – or/and of (b) what those
non-physical items were: their nature.

17
(a): what causes what is contingent, at least for zombists. The laws of nature
may preclude the physical causation of items supposedly caused by non-physical items;
but zombist interactionists cannot deny it is conceivable that those things should be
done physically. So (a) appears to be ruled out.
(b): since the changes leave ψ* untouched, they cannot affect its nature. So
we can dismiss (b) too.
Since ψ* includes whatever ‘makes us conscious and keeps us in epistemic
touch with our experiences’, it is hard to see how the third difference – that after the
changes there are no non-physical items other than ψ* – could prevent ψ* from
continuing to make the inhabitants of i conscious.
Thus there is at least a strong case for the claim that (ZI) entails the
conceivability of premiss (2). However, zombist interactionists might think I have not
done them justice. They might urge that it is not by merely nomic necessity that what
provides for consciousness is causally efficacious: they might claim it is a priori
necessary; so that even if ψ* continued to exist after the changes, its inertness would
prevent its continuing to make anyone conscious.20 Now, I agree with the widespread
view that whatever makes us conscious cannot be inert. But I will argue that zombists
are committed to a conception of consciousness which entails that the contrary is at least
conceivable. (Interactionist zombists might endorse the correct view about
consciousness and causality; but if I am right that would make them inconsistent: their
zombism commits them to an incorrect view as well. In that case, appealing to the
correct view cannot protect their incorrect view from the arguments of sections III-IX.)
Like most of us, zombists claim to know they are conscious. Unlike some of
us, they cannot justify this claim by reference to observation of physical facts such as
behavior – because they think zombies would be physically indistinguishable from us, at
least superficially. This means that interactionist zombists must think their knowledge
that they are conscious, hence their knowledge that ψ* exists, comes from the fact that
they have conscious experiences themselves, not from knowing any physical facts. A
consequence is that they cannot consistently claim it is a priori that ψ* must have
effects in order to make us conscious. If we can know we are conscious by actually
having conscious experiences, then we can know it without also knowing whether or not
consciousness has effects. So zombist interactionists cannot maintain it is inconceivable
that ψ* should make us conscious in spite of being inert. They may point out that we
observe what we take to be effects of our being conscious. But we cannot observe that

18
they are effects of ψ*: that is part of a theory which, for them, might conceivably be
mistaken. (Consistently with our experience, those effects and ψ* might be joint effects
of some common cause, for example.)
Zombist interactionists might raise another objection. They might accept that
epistemic contact requires consciousness to have effects, but maintain that even if ψ*
continued to make our successors conscious, its lack of causal efficacy would prevent it
from continuing to sustain epistemic contact.21 I find it hard to make sense of that
suggestion. Significantly, Chalmers has said there is “not even a conceptual possibility”
that a subject should have a red experience “without any epistemic contact with it”.22
And surely he is right about that. After all, we are discussing phenomenal
consciousness, the idea that there is something it is like for the subject. If there is
nothing it is like, the subject is not phenomenally conscious. Now, could there be
anything it was like for a subject who had phenomenally conscious experiences but
could not notice, attend to, or compare them? How could that be different from what it
was like to be completely unconscious: like nothing? Zombists typically base their
claims about what is conceivable on what is imaginable. But can we even imagine the
situation described? When I imagine myself being phenomenally conscious, I
automatically imagine myself having experiences that I can notice, attend to, remember,
or compare. When I try to imagine not being able to do those things, imagination fails: I
cannot take the first step of an argument for the conceivability of the situation described.
Nor does that seem to be a psychological defect. It seems more like a symptom of the
fact that being in epistemic contact with one’s conscious experiences is part of what it is
to have them. Absent any argument to the contrary, I conclude it is not conceivable that
our successors in i should be conscious without being in epistemic contact with their
experiences. If that is right, the objection fails.23

Premiss (3). We noted that after the changes, the physical component of i would be
closed under causation and resemble a zombie twin of our world. The difference
between i and a zombie world is of course ψ*, which, though caused by something
physical (or at least conceivably so caused24) is inert. As we have just seen, ψ* would
make i’s inhabitants conscious and ensure they were in epistemic contact with their
experiences. A look at (C) confirms that if we take ψ* = ψ, those features of i after the
changes are the defining features of z*. Hence if (ZI) commits its exponents to (1) and
(2), it commits them to the conceivability of (C).

19
Earlier I argued that all zombists are committed to the conceivability of
dualism, and that epiphenomenalists and parallelists are committed to the conceivability
of inert conscification as defined by (C). I have just argued that interactionist zombists
are committed to (1), (2), and (3), and conclude they too are committed to the
conceivability of (C). Given that all dualists belong to one or other of those classes, it
follows that all zombists are committed to the conceivability of inert conscification, or
(C).

XI. CONCEIVABILITY OF INERT CONSCIFICATION WOULD ENTAIL


CONCEIVABILITY OF THE E-QUALIA STORY
It is easy to see that the definitions of a ‘zombie twin’ world and of (C) ensure that the
conceivability of (C) entails the conceivability of a world z* satisfying the following
conditions, which mirror (E1)-(E5):
(Z1) z* is partly physical and its physical component is closed under causation: every
physical effect has a physical cause.
(Z2) The human-like organisms in z* are related to a special kind of non-physical item
ψ. ψ makes them phenomenally conscious.
(Z3) ψ is wholly caused by physical processes but inert: it has no effects either on the
physical world or internally.
(Z4) The human-like inhabitants of z* consist of nothing but functioning bodies and ψ.
(Z5) The human-like inhabitants of z* are able to notice, attend to, think about,
compare, and (on occasions) remember the experiences provided for by ψ.
So the conceivability of (C) entails the conceivability of (Z1)-(Z5). But that
entails the conceivability of (E1)-(E5). For there are only three differences between
(Z1)-(Z5) and (E1)-(E5): “z” occurs in place of “the world”; “human-like inhabitants” in
place of “human beings”; “a special kind of non-physical item (or items) ψ” in place of
“e-qualia”. The first two are not significant, since they arise from the fact that while
(E1)-(E5) are a story about how the actual world and its human inhabitants might
conceivably have been, (Z1)-(Z5) are about a special kind of world and its human-like
inhabitants – which is also how the actual world and its human inhabitants might have
been on the assumption that zombies are conceivable. Obviously those differences do
not prevent the conceivability of (Z1)-(Z5) from entailing the conceivability of (E1)-
(E5).

20
The third difference may appear troublesome. What compels zombists to
concede that conceivably a world satisfying (Z1)-(Z5) might have e-qualia as its special
non-physical component ψ, rather than something different? There is no real difficulty
here. By (C), ψ makes z*’s inhabitants conscious and thereby ensures that their
experiences have, in Chalmers’s words, “those properties of mental states that type
those states by what it is like to have them” (see section III). That means ψ underlies
their qualia in a broad, neutral sense of ‘qualia’. By (C), ψ is also inert and caused by
physical items, so the same must go for these qualia. Moreover, there is nothing in i
other than its physical component and ψ. But then these qualia satisfy those parts of the
e-qualia story which define qualia (clauses (E2), (E3), and (E4)) and must be counted as
e-qualia.
So the conceivability of inert conscification, or (C), entails the conceivability
of (E1)-(E5). We saw in the last section that the conceivability of zombies entails that of
(C). So the conceivability of zombies entails that of (E1)-(E5), which gives us:

(B) If zombies were conceivable, the e-qualia story would be conceivable.

But we already have:

(A) The e-qualia story is not conceivable.

By contraposition, zombies are not conceivable.25

_____________________

NOTES

1 For the view that the zombie possibility entails the falsity of physicalism see e.g.
David Chalmers, The Conscious Mind (New York: Oxford University Press, 1996); R.
Kirk, Zombies v. Materialists, Aristotelian Society Proceedings, supplementary vol. 48
(1974): 135-152 and Zombies and Consciousness (Oxford University Press, 2005):

21
7-23. For the conceivability argument see e.g. Saul Kripke, Naming and Necessity,
(Cambridge: Harvard, 1980); Chalmers, Does Conceivability Entail Possibility? in T.
Gendler and J. Hawthorne, eds., Conceivability and Possibility (Oxford: Oxford
University Press, 2002): 145-200; and other essays in Gendler and Hawthorne, eds.
2 Indirect objections: Gilbert Ryle, The Concept of Mind (London: Hutchinson, 1949,
and Ludwig Wittgenstein, Philosophical Investigations, tr. G. E. M. Anscombe (Oxford:
Blackwell, 1953); direct ones: e.g. Daniel C. Dennett, Consciousness Explained
(Boston: Little, Brown, 1991) and The Unimagined Preposterousness of Zombies,
Journal of Consciousness Studies, 2 (1995): 322-6; R. Kirk, Zombies and
Consciousness: 37-57; Sydney Shoemaker, Absent Qualia are Impossible, Philosophical
Review, 90 (1981): 581-599, and On David Chalmers’s The Conscious Mind,
Philosophy and Phenomenological Research, 59 (1999): 439-444; Michael Tye, Absent
Qualia and the Mind-Body Problem, Philosophical Review, 115 (2006): 139-68.

3 This outline mirrors that of Kirk, Zombies and Consciousness: 39-55. However, the
arguments here are significantly different (and I think clearer and more cogent) and take
account of objections not considered in the book.

4 G. F. Stout, Mind and Matter (Cambridge: Cambridge University Press, 1931): 138f.

5 Some authors use “zombie” for merely behavioral duplicates, or for systems
resembling us only in input-output functions; but those senses too are irrelevant here.
Physicalists can consistently concede that behavioral and dispositional similarity is
insufficient for mental similarity (pace Dennett 1991: 438-440): Kirk, Zombies and
Consciousness: 97-118. The definition of zombie twin worlds is refined in section X
below.

6 See for example John Perry, Knowledge, Possibility, and Consciousness, (Cambridge
Mass: MIT Press, 2001). Interactionist zombists must deny that causal closure holds in
our world, hence cannot define zombies as physically like us. Chalmers says the
conclusion of his anti-materialist argument is not epiphenomenalism, but “the
disjunction of panprotopsychism, epiphenomenalism, and interactionism” (Materialism
and the Metaphysics of Modality, Philosophy and Phenomenological Research, 59

22
(1999): 475-496, p. 493. See also his The Conscious Mind: 150-160), and sections X
and XI below.

7 Even those epiphenomenalists who maintain that God intervenes in the world can
accept it is conceivable that the physical world should have been causally closed.

8 The Conscious Mind, p. 359, n.2.

9 Chalmers uses ‘epistemic contact’ for what he calls ‘acquaintance’ with qualia (The
Conscious Mind, p. 197). I will consider this notion later (section IX, Objection 3); but I
am using ‘epistemic contact’ in the broader sense indicated.

10 It might be suggested that causation is not necessary: mere counterfactual dependence


would be enough. But the need for unconsciously persisting traces and structures would
still prevent e-qualia from being capable of the necessary cognitive processing. See also
section VII.

11 E-qualists cannot deny that such a situation is possible. Since e-qualia are non-
physical, neither their existence nor their non-existence can entail or be entailed by
anything physical (but see n. 18).

12 Epiphenomenalists typically assume isomorphism. See e.g. Chalmers, The Conscious


Mind, p. 243.

13 It would not help e-qualists to say e-qualia are subjectless: that would only support
my claim that no one could be in epistemic contact with them. On the other hand, I see
no objection to the notion of integrated, nonrelational processes of having-qualia (see
Zombies and Consciousness: 154-8) – but that is obviously inconsistent with the e-
qualia story.

14 Thanks to a reviewer for this and the preceding objection, and for the phrasing of the
suggested perspective on consciousness in an E-world.

15 See n. 8 above.

23
16 ‘Logically’ independent in the sense that its existence was not entailed (or a priori
necessitated) by the physical component.

17 Conscification need not be thought of as zombies becoming conscious, but only as the
coming into existence of conscious subjects whose physical components had been
zombies.

18 Cartesian zombists might demur. If thinking – a kind of activity – is essential to the


soul’s existence, then ψ* cannot cease to be a cause without ceasing to exist. Also, some
interactionist zombists might maintain it is a priori necessary that consciousness
involves non-physical causes. See, however, the discussion of premiss (2) below. (Note
that zombists cannot resist condition (i) by invoking causal essentialism, according to
which a thing’s causal dispositions are essential to it. For if that doctrine is taken to
entail that what physicalists count as physical items cannot conceivably exist without
causing or being caused by conscious states, then zombies are inconceivable for that
reason; while if it lacks that entailment, then it lets in (i).)

19 It would not be enough to claim that conceivably the changes might prevent ψ* from
continuing to make our successors conscious; (2) says only that (ZI) entails it is
conceivable that it should do so.

20 Here I take into account the worries mentioned in note 18.

21 Thanks to an anonymous reviewer for this suggestion.

22 The Conscious Mind, p. 197.

23 Zombist interactionists might consider maintaining that ψ*’s loss of causal efficacy
would cut off not only epistemic contact but consciousness too. But by the argument of
the last paragraph that would prevent them from being zombists.

24
24 (ZI) does not appear to entail that ψ* is caused by physical items in i, only that it is
affected by them. However, given causation is contingent (n. 18), (ZI) does entail it is
conceivable that ψ* should be caused physically.

25 Special thanks to Bill Fish for much detailed discussion and correspondence, and to
David Chalmers for comments, suggestions, and encouragement through several
revisions.

25

You might also like