Review
of: Wegner,
Daniel (2003) The Illusion of Conscious
Will. MIT Press.
Multiply
reviewed in: Behavioral
and Brain Sciences
(2004) 27: 649-692
Precis: http://www.bbsonline.org/Preprints/Wegner-05012003/Referees/Wegner.pdf
On the Incommensurability of Feeling and Doing: The Illusion of Free Will
Stevan Harnad
Chaire de recherche du Canada
Institut des sciences
cognitives
Université du
Québec à Montréal
Montréal, Québec
Canada H3C 3P8
and
Department of Electronics
& Computer Science
University of Southampton
Highfield, Southampton, UK
SO17 1BJ
http://www.ecs.soton.ac.uk/~harnad/
The difference between a
conscious entity and a nonconscious entity is
that the conscious entity feels and the nonconscious one does not. What
the
feeler feels is just a detail. That it feels is the critical
property,
and explaining how and why it feels is the so-called 'hard' problem of
explaining consciousness.
Among the kinds of feelings a
feeler might feel are primary sensations
(such as colors, shapes, sounds, smells, touches),
pains, emotions (such as joy, fear), desires
(such as hunger, lust),
kinesthesias (involuntary movements, voluntary movements), and complex
cognitive states (such as knowing that it is raining, believing that it
is
Tuesday, understanding that 2+2=4, and meaning that the cat is on the
mat). These
are all just things a feeler might be able to feel.
There is also a world, to
which those feelings may somehow be related. When a feeler feels that
it is
getting colder, the temperature may or may not be dropping; if it feels
it has
moved its limb, the limb may or may not have moved (if it has a limb at
all).
And if it feels it has moved its
limb voluntarily, it may or may
not have been voluntary: It may have moved because a physician tapped a
reflex
point or even just pushed the limb, and the feeler mistakenly felt it
as a
voluntary movement .
So there are feelers and
feelings and a world in which the feelings take
place, with what is felt sometimes
corresponding to what is actually happening in the world, sometimes
not. Let us
pause to think about this 'correspondence': I said the feeler could
feel it is
getting colder, and the temperature (internal or external) might indeed
be
dropping, or it might not. When the feeler feels it's getting colder,
and it
really is getting colder, we are tempted to say that it has felt the
drop in
temperature: but a drop in temperature is, as we all know, a reduction
in the
average motion of the molecules in the (say) ambient air molecules. But
does
what 'feeling colder' feels like really resemble in any way a reduction
of
average molecular motion? Does what red looks like resemble a property
of
reflected photons of a certain frequency?
Does what palpating a round
shape feels like resemble a property of roundness (equidistance from a
center, or symmetric curvature)?
Does even what it feels like for something to feel more or less intense
resemble what it is for for any physical magnitude to be greater or
less in any
other way than that it is reliably correlated with it?
I didn't really want to embark
on an extended Lockean exercise here,
just to pump some intuitions about the relation between feelings and
their
causes and correlates. Our feelings must correlate with reality closely
enough
to get us by in this Darwinian world, but that is a functional
criterion and not
a geometric one. We know what it means for a drawing of a circle to
resemble a
circle, but what does it mean to say that what it feels like to look at
that
drawing resembles the circle (as opposed to resembling what it feels
like to
look at the circle)?
So in what way might our
feeling of free will be an illusion? We know
there are sensory illusions as well as hallucinations: I can see
something when
there is nothing there. I can see one line as longer when they are both
the
same length. In such cases we would say that the usually reliable
correlation
between the way things really are
in the world and they way the feel has failed us. It does not, of
course, mean
that all perception is an illusion. Just that it can sometimes be
illusory,
i.e., that our feelings can sometimes mislead us.
Wegner's book reviews many
examples of how our sense of agency,
voluntariness and control can also fail us, creating what
one might call kinesiological illusions
(kinesthetic illusions being already an inhabited niche). Under some
conditions
we feel we are voluntarily doing something that we are not, or are not
voluntarily doing something that we are. The effect sometimes even
stretches
from one feeler to another, with one feeler having the illusion that
they are
willing the movements of the other feeler, or vice versa. Here too, the
natural
conclusion is that there can be illusions of will and voluntary
control,
including all the usual perceptual possibilities: false negatives,
false
positives, and misses. It does not, of course, follow that free will is
an
illusion. Just that our feelings can sometimes mislead us.
Yet of course free will is in
another sense -- a sense other than the
usually reliable correlation between the way things feel and they way
things
are -- an illusion. When I make a voluntary movement such as raising my
arm (and
let's say this is not one of the cases when the correlation fails and
creates
an illusion), it feels as if I am causing my arm to raise: I've
raised
it because I felt like it. But could it be that my feelings are no more
the
cause of the motion here than they resemble the motion of molecules,
when I am
feeling the heat or the cold? There is something more fundamental here
than
just the incommensurability between
physical properties and feelings. Psychophysics at least gives
us a
reliable quantitative correlation, if it cannot give us a qualitative
resemblance. But with action we are not talking about mere correlation
or
resemblance, but about causation. What could it mean for feelings to be
causes?
An innocent (and probably
correct) meaning would be that the same neural
event that causes the feelings causes the movement. But then the real
cause is
that neural event, not the feeling (Libet BBS). Could feelings
themselves,
independently, be the causes of a movement? Only if a certain kind of
dualism
were true: Telekinetic ('mind over matter') dualism. But for
telekinetic
dualism to be true, there would have to be reliable empirical evidence
that
feelings can be independent causes, rather like gravitation or
electromagnetic
forces. Yet centuries of efforts by paranormal researchers to
demonstrate that
this was the case have so far failed to generate any reliable evidence
(Alcock
BBS) -- and the evidence would have
to be dramatic indeed, for it would either be in conflict with the
conservation
laws of physics or it would need to add mental forces into the pantheon
of
physical forces (of which there are otherwise currently only four). A
tall
order for a soft science!
In the absence of any evidence
for telekinetic dualism, we are left with
the 'hard' problem: If feelings can't be causes, why are they there at
all, and
how are they implemented, physically? (Not "What are their neural
correlates?", but "How and why does neurophysiological function --
'functed' function -- become felt function -- unlike, say,
cardiac, or
pulmonary, or renal function, which we 'funct' but do not feel or feel
we
cause?") Wegner suggests that 'constructing a virtual agent' somehow
has
some functional utility. If it did have some functional utility, then
free will
would be elevated from the correlational to the causal, and the hard
problem
would be solved. But is it solved?
It is easy to provide a
plausible adaptive function for the capacity to
detect and respond to input, and to learn to interact differentially
with
different kinds of input, via sensorimotor interactions of both the
voluntary
and involuntary kind. (The difference between pushing and being
pushed.) You
could even put a social world into this, with conspecifics, kin,
allies,
competitors, predators, and even natural language communication. A
correlation
between sensory activity and reality would be useful and adaptive, as
would a
correlation between motor activity and reality, including the
all-important
distinction between the causes and effects of when the organism does
something
voluntarily, owing to an internal command, as opposed to when it does
something
involuntarily, owing to some inner automatism or an external force.
Correlations would not be restricted to inert ones: Correlations
between outer
behavior and inner states could evolve and be learned so that such
organisms
could communicate as well as 'mind-read' -- not in the true sense of
telepathy,
but simply through adaptive functional correlations.
There could be mistakes
('illusions') of both the sensory and motor
sort, in which the usually reliable functional correlations fail. But
so far
one element is missing from this rather lifelike and humanlike
scenario:
Feelings. How and why should these organisms be feelers? It will not do
to
propose, as Wegner does, that it is functionally useful to 'construct a
virtual
agent': It is certainly useful to detect information, learn patterns
and
correlations, plan, infer, reason,
compute, simulate, even simulate actions and the actions of
others. But
those are merely functions, and functions we can already implement with
inert
pieces of man-made machinery. Why and how should any of those very
useful and
adaptive functions -- right up to language -- become felt
functions,
rather than just function? For unless we can explain how and why those
functions are felt functions, all the rest of our attempt to explain
how and
why we do things because we feel like it comes to nothing at
all.
FOOTNOTE on "correlation": The
observation that feelings
"correlate quantitatively" with reality even if they do not resemble
it qualitatively" is a cheat. The right construal of Wittgenstein's
argument against the possibility of inventing a "private language" is
that it does not even make sense to say that my sensations correlate
quantitatively with reality: In fact, it only feels (qualitatively)
as if they
correlate
quantitatively with reality! The objective correlation is only between
reality
and what I do, not what I feel. That feeling I have now, that
this sound
is more intense than it seems to have been a second ago, and that this
correlates (objectively and quantitatively) with the increased
amplitude
registered by the audiometer (and is the same correlated feeling of
increased
intensity that I have felt and
have confirmed with meter readings countless times before) is,
objectively
speaking, merely a correlation between the sound and what I (and my
neurons) do.
Whether it is also in some way "correlated" with what I feel
neither I nor anyone else (nor instruments) can determine. It just
feels that
way. Feeling and function are not only causally dissociated, they are
objectively incommensurable.
-------------------------------------
Notes.
The book is not about the
illusion of will -- conscious' is
redundant: 'unconscious will' is
self-contradictory, just as an 'unfelt feeling' would be -- but about illusions
of will: misperceived and misattributed agency and causality
Mostly about illusions or
misattributions of causality: cf hallucination.
This doesn't imply that FW is an illusion any more than sensory
illusions imply
that sensation is illusory. It is logic (and metaphysics) that shows FW
is an
illusions. We don't actually do things because we feel like it; it only
feels
like it!
Sense of agency can be altered
or duped. But that does not make it an
illusion: all perception can be duped. Some times we perceive
correctly,
sometimes incorrectly, whether about the length of two lines or the
voluntariness of a movement. But the fact that no movement is really
voluntary
at all is another matter.
The basic problem is the
causal status of feeling (of which the feeling
of agency, or voluntariness, is just one of an infinity of the same
kind.
Hysteric automatisms can come
from hypnosis, which can suppress pain
too, but is it like neurological automatic writing? (anosognosia)
The only one who believes in
free will is the telekineticist (immaterial
force, mind over matter). Rest of
us are just telling teleological Just-So Stories, post/ad hoc
hermeneutics, not
explanation
Anosognosia for the causes
of our own behavior.
Libet, Alcock, BBS
P 328 P 340: errors
Social Psychology (dissonance)
of rationalization and confabulation
Not about reasons why
but about feeling that
Not FW vs. determinism but the
causal status of feeling: feeling
willing is it phenomenological core: It feels as if I am doing things
because I
feel like it.
Not why we feel will,
but why we feel at all.
Can't separate problem of
consciousness and causation. Sensorimotor feelings
'action projection' is false,
but mind-reading is true.
Telepathy, Telekinesis and
Teleology: Causality and Consciousness:
Feeling and Function: Explanation
Why one me? There are
no unfelt feelings, and we can't really
feel another's feelings (it just feels-as-if feeling another's
feelings, i.e.,
it's an illusion
Cogito: I might feel-as-f I am
another, or many, or nothing, but
clear-headed reflection (w/o anosognosia) shows this is not so: It is
infallible that feeling is felt. And feeling of feeling is part of the
package.
Robot 'morality': feelings
only help because they are correlated with
objective adaptive 'values' -- values are still underdetermined.
'People experience
conscious will when they interpret their own
thoughts as the cause of their action' interpret? Or feel as if they
are doing something
because they feel like it
not like billiard-ball to ball
causation because there objective cause
is inferred, whereas subjective will is felt. Causation may sometimes
be
inferred because we are hard-wired to infer it: 'It looks (and feels)
as if A
causes be, but I might be wrong.'
I might also be wrong that
something that feels voluntary really is
voluntary, but I can't be wrong that it feels voluntary.
'The consistency principle'
How can a thought resemble, or relate to, or 'be consistent with' an
action: It can only feel-as-if it resembles, etc. But
that's the same as feeling as if it causes. Hence it is no explanation.
FW sense is not the same thing
as an illusion of control. FW is just the
illusion of having free will at all, whereas control may be an illusion
in this
case and not in that case. (cf 'illusion of seeming')
That new ideas do not feel
willed goes against the notion that trying in
advance makes the outcome feel willed. Do we will
the answer to 2+2=? Do we will adverse thoughts?
Remembering, or thinking an
old thought, is every bit as sudden and
spontaneous as a 'creative' thoughtÉ (attributed to 'unconscious
mind': does
that have a FW too?)
'Construction of a "virtual
agent'' To be conscious one needs neither to
be able to do, nor to feel one is doing nor feel that one is doing
voluntarily:
one need only feel.
Speculating about functional
(= causal) role of FW is as futile as speculating about the
functional
(causal) role of C (feeling). Why feel anything at all (including, in
particular, feeling as if one is causing what one is doing)?