From Aaron Sloman Sun Mar  2 11:00:20 GMT 1997
Subject: infallible qualia-detectors (was: Ineffability of qualia)
To: psyche-d@rfmh.org

   Sue Pockett (s.pockett@auckland.ac.nz)
   Sat, 1 Mar 1997 15:41:57 +1200

wrote answering Pat's question:

   On 28 Feb 97 at 13:02, Pat Hayes wrote:

> >Can one be mistaken about one's own qualia?

Sue replied:
> No.

But she didn't say why. I think many people would agree
with her answer, and the argument goes something like
this.

The nature of a quale is defined by the content of your
experience, i.e. how something seems to you. E.g. when
you have a visual quale involving something red and
square, that is defined as something seeming red and
square to you.

(I am not claiming that that is a sufficient condition
for the existence of sensory qualia, just a necessary
condition. Some seemings may come by other routes not
involving sensory qualia.)

Contrast the above with the following case: whether the
table IS red and square or not is quite a separate
matter from whether it SEEMS red and square to you, so
you can be mistaken about the table's colour and shape.

If a quale's being red and square is, by DEFINITION, a
matter of something seeming red and square, then there
can be no divergence between how it seems and how it
is, at least not at the instance of time in question.

So there is a kind of infallibility here.

(You might easily forget how it was or misremember,
e.g. if you are quickly distracted by something else,
or nasty psychologists do experiments, etc. on you. But
then we are talking about your being mistaken *later*.
You may also make some wrong inferences from how things
seem to, e.g. thinking you had an electric shock when
you haven't, as in Pat's case.)

If that's the reason why Sue answered "No", then we can
see that the reason why her answer is correct is a
fairly trivial matter of definition: how things seem is
how things seem. There's no room for error there!

But we can now generalise the above point about
infallible qualia judgements:

If there's anything at all that uses information,
whether it is a person, an animal, a machine, a ghost
in a machine, or whatever, then we can make a
distinction between
    (a) what it takes the content of the information to be
        (at a particular time)
and
    (b) how things actually are at the time.

There could in principle be a discrepancy between the
two, i.e. an error or mistake is possible.

If we invent a name e.g. "infocontent" to refer to the
first item (a), i.e. what the content is taken to be by
the information user, then there cannot be any mistake
about THAT.

How things seem to you is, by definition how things
seem to you. What the machine/animal/person takes the
information to be is by definition how it takes the
information to be. It can't be wrong about THAT, even
if it is wrong about (b).

So, for example, an operating system may be mistaken
about how many processes there are at a certain time
because of a bug in the software that prevents some
entries being removed from the process table when the
process dies (or simply because there is a short delay
before the entries are removed). And it may take a
wrong decision on the basis of the error, e.g. deciding
that it cannot give up the space allocated to any of
those processes for use by other processes.

But although the system can be mistaken about how many
processes there are, it cannot be mistaken about its
infocontents, i.e. what it takes the number of current
processes to be.

We can also say that this information about
infocontents, unlike most other information, is direct
and unmediated (by definition).

(A complication: if the system is complex one, with
many subsystems, then some of the subsystems that use
information may have infocontents that differ from
those of other subsystems and with the infocontents of
the whole system. E.g. if your left and right eyes
process colours differently then they may have
different colour infocontents, without your knowing
anything about that, just as they have different
infocontents regarding relative lateral displacements
of edges in the optic array, a fact that you are
unaware of, but which is used in producing your
experience of depth. Of course, in this case you can
become aware of the difference by alternatively
covering your left and your right eye.)

Much philosophical confusion arises from noticing
things that are true, but not noticing that they are
true by definition, and then thinking there's something
very profound going on, perhaps even a "hard" problem?

Of course, I wouldn't dream of accusing Sue of falling
into such a trap.

Incidentally, I would not say that my demonstration of
the necessary existence of infallible infocontents in
information processesors is a *deduction* from
classical physics. But I don't think it follows that we
need to invoke quantum physics.

There's nothing to stop a classical physical machine
being the underlying implementation for an information
using machine, with infocontents.

And I think the example given above shows why
infocontents will occur in such a system.

Likewise the inevitability that brains will contain
experiences/qualia/sensory contents with a kind of
infallibility.

Cheers.
Aaron