Colin Geoffrey Hales wrote:
> This discussion is a hybrid of a number of very famous thought
> experiments. Unlike those thought experiments, however, this experiment is
> aimed purely and only at scientists. The intent is to demonstrate clearly
> and definitively the nature of subjective experience (phenomenal
> consciousness) and its primal role in a scientist's ability to function as
> a scientist.
> Imagine a room. Its walls and celing and floor are matt black. There are
> no doors or windows. All over the walls are digital displays which
> announce numbers in warm, friendly colours. Up against all four walls and
> up to international standard control desk height (roughly 750mm) is a
> sloping console. Covering the console around all four walls are
> pushbuttons. The number of displays is equal to the number of sensory
> nerves entering a typical human brain from the peripheral nervous systems
> including all sensory input. The number of pushbuttons is equal to the
> entire set of effector nerves emanating from a typical brain. This total
> number of displays and pushbuttons is in the millions.
> There is a comfortable chair upon which is seated the room's sole
> occupant, Marvin the human. Marvin is normal except for never having been
> outside the room and having never otherwise acquired any knowledge of
> anything other than that of the room and its contents. He knows absolutely
> nothing of any sort of external world or any other people. He has no clue
> about the external world except for what he can surmise from the displays
> and buttons.
> Marvin, like all healthy humans, has an experiential life which is the
> collection of private phenomenal scenes delivered by his brain material.
> These are operating normally except that Marvin has been deprived of all
> the phenomenal content (experiences) that he ever could have received had
> he been allowed outside the room. He has had a lifetime of experiences,
> but all of them have been confined to depicting the room.
> Marvin is a scientist. He studies the science of not-room. His life
> consists of experiments which involve the pressing of buttons and the
> recording of patterns in the displays. Over time an enormous volume of
> data begins to show patterns for which Marvin constructs models. The
> models are generalisations of the behaviour of the displays after buttons
> are pushed in a certain way. He tests and refines the models and has
> developed a form of symbolic representation of the behaviour of the
> displays. Certain features on the display occur regularly enough that
> names have been given to notional not-room phenomena. One is called the
> tronelec. The mathematics of not-room includes a lot of rules about the
> behaviour of tronelecs. In time Marvin realises that his exploration of
> not-room can be done according to routine rules and he sets about a
> systematic, exhaustive assay of the entire range of possible
> button/display relationships.
> Everything is going nicely but then one day a well known pattern does not
> occur the way it used to. After a while the old pattern resumes. The whole
> mathematics of not-room is undermined. It takes a long time for Marvin to
> construct a new model that accounts for the novel behaviour. A new entity
> called gytravi is needed. Then things settle down and routine systematic
> exploration resumes.
> In time a massive collection of not-room entities and behavioural rules is
> constructed and begins to repeat itself. So much so that Marvin, after
> checking and rechecking, finds that the model seems to have stabilised.
> At this point we stop to survey the situation.
> The first thing to note is that the room is an empirically verified
> physiologically accurate representation of the sensory circumstances of a
> human brain. Nervous activity effect/affect is replicated by
> buttons/displays and all signals are standardised. There are no
> experiential qualities (perceptual sensation qualities) associated with
> this nervous activity. This is a physiologically verified, well
> established empirical fact. Indeed the room is a little too kind - If you
> discarded all the buttons and displays, then that is a more
> physiologically accurate circumstance .
> The main point is that this collection of totally sensationless signals is
> exactly what is available to a human brain and through which all sense
> measurement arrives. Consider the room without a Marvin in it. Based on
> what evidence is there any reason to even conceptualise the existence of
> an external world?
There is nothing in the room to conceptualise an
external world because Marvin does all the conceptualising.
Is this supposed to show that an you need phenomenality is
needed to conceptualise an external world? It doesn't.
We could imagine the control panel delivering non-phenomenal
information to Marvin (such as he slips of paper used in the Chinese
Room).
Or we could split Marvin into two people, A Watcher who has the
phenomena,
and a Theorist that the watcher reports to non-phenomenally. IOW,
The philosophical zombie is an exercise in conceivability.
We can conceive of a version of your Room in which the interpretation
(i.e. the basis for supposing that there is an external world)
is separated from the phenomenality. We can still conceive of a version
in which all the functionality is separate from all the phenomenality.
It might be the case that the conceivability is not your concern,
your concern is real possibility. But then you are no
contradicting anything the philosophers are saying. And real
possibility is not established by armchair thought experiments.
> Nothing in that room gives any indication of it.
That depends on what you mean by "indication". The Room contains
all the information a real human has to go on, by hypothesis.
There isn't a single light that goes on to say "Hey! There is an
external world!". But Marvin seems to be able to process his data
on the assumption that they are caused by some stable formation
like a gytravi or a tronelec. Perhaps Marvin needs a apriori
belief ore assumption that there is an external world, in order
to proceed with an investigation of the signals "as if" they had
external sources. There is no reason Marvin should
lack such an assumption, there is no reason a real person should,
there is no reason a zombie should lack it, and no reason
a suitably programmed AI should.
> There
> is no a-priori knowledge of not-room.
So you say. But is that scientifically accurate? The tabula
rasa theory is rather out of date. And there is no reason
a zombie or anything else should lack such knowledge, since
it is not a phenomenon. And why would Marvin need *knowledge*
anyway? Isn't an assumption or hypothesis enough?
> No possible way to interpret any of
> the signals. All there is is correlated behaviour - the behaviour of
> displays correlated with other displays and buttons, none of which have
> any other discernable impact.
If you say that Marvin is the Theorist, the Interpreter,
and you unplug him, there is no interpretation. Does that mean
that interpretation and phenomena are all done by the same module
in humans? I don't see any evidence for that, it all
depends on your say-so. In any case, it doesn't
affect conceivability arguments. It is still
conceivable that phenomenology and functionality could be prized apart.
> The room is quintessentially a zombie in the
> sense it has no experiential life at all despite being wired as
> extravagently as a complete human.
> Marvin thinks of himself as a scientist. Exactly what is he a scientist
> of? His science is, he thinks, the science of 'not-room'.
Why would that be? You said the Room+Marvin system is supposed
to be an accurate representation of a person. Well, unless the person
is a solipsist, the Room+Marvin sytem would believe in an external
world.
> But is it? All
> he has is a very sophisticated model of display/button correlations.
So you say. According to you he has a "flat", or instrumental model
that doesn't posit anything behind the phenomenal instrument-readings.
But it is far from clear why tha must be the case. Since real-world
scientists don't *have* to be instrumentalists, why should inner
homunculi like
Marvin? (More below [*])
> His
> science is not about 'not-room' at all! His science is the science of room
> display/buttons. He has absolutely no justification to any claim about
> anything going on anywhere else, although that is how Marvin thinks of it.
A predictive model is justification for whatever entities it posits.
Your claim of "absolutely no justification" is not itself justified.
> His science is very very predictive of button/display behaviour. As to the
> reality of the model he has developed? He has no way of contextualising
> any of the abstractions he has created with the actual state of affairs in
> 'not-room'.
What does *that* mean? You mean can't step outside the room-world
system,
hover above them, and compare them directly. No he doesn't know in that
sense. But if that is a problem
(and I don't think it is, not as much as is claimed) it is a problem
that affects everything. You cannot claim there is a magical solution,
if that is
what you are claiming. And it is an exaggeration that Marvin does not
know
at all because he lacks "transcendent realism". He still has
"empirical;
realism". He can construct robust, predictive models.
> Absolutely anything could be out there driving the displays and Marvin
> will never know.
Since Marvin can reject theories as over complex (Occam's razor) or
falsify the predictively, "absolutely anything" is unwarranted.
But he still has limitations. If he has observed that phenomenon X is
followed
by phenomenon Y, he can suppose that there is an external
process characterised by the tendency to cause those phenomena.
But he doesn't know what that "thing" might be in itself.
He is no in a state of absolute ignorance, of total
knowledge. He has good-enough knowledge about certain things.
> Indeed worse than that, there may be an infinite number
> of different ways that not-room could present the same display values and
> Marvin would never know. The display data is fundamentally, intrinsically
> ambiguous.
If the external world has in-itself properties then it is undetermined
by Marvin's knowledge. If it is completely characterised by
the tendency to cause phenomena in onlookers, it isn't. You have
to make an assumption about the metaphysical nature of the world
to come to a conclusion about how ignorant we are (or Marvin is) about
it.
> Even if the displays and buttons were nicely grouped and
> labeled all that would happen is that there would be expectations of
> related behaviours in a group, which Marvin does not have to name himself.
> It could even be named 'sight', 'sound' and so on and Marvin may
> understand what that might mean, but it delivers no claim to any
> definitive or unambiguous knowledge of not-room.
> And that, fundamentally, is the real nub of the matter.
> What is outside the room? It doesn't really matter. It could be an
> elaborate multistory building with a ratsnest of electrical
> interconnections done with a computer. It could be a control room in the
> head of a giant robot. It could be literally wired up to a human body with
> no brain. It could be a spaceship. None of that matters except in the
> details. Marvin can never know because he has no experience available to
> even imagine it.
Marvin can exclude a lot of possibilities with Occam's' razor and
falsification.
> To finish off the room scenario we now take Marvin out of the room. In his
> place we leave a machine that is based on his systematic scientific
> behaviour. It runs with clockwork and pneumatics from punched cards. None
> of the displays are needed, none of the buttons are needed - they are all
> directly connected to the new machine. The massively complex model Marvin
> has constructed, which has nothing whatever to do with 'not-room' except
> in the most indirect of abstracted ways,
i.e. accept inasmuch as it *does* have something to do with the
not-room...
> goes on being verified and
> occasionally amended using more rules for amending the model, also devised
> by Marvin.
> The model can never say anything about any changes in
> 'not-room'. All it can sense is novel (unexpected according to the model)
> behaviour in the sensing displays, which is an entirely different thing.
Not at all. If the model start with the assumption that the
phenomenal displays are driven by some sort of external world, and
there is no
reason why it shouldn't, then the most
obvious explanation for a change in the behaviour of the phenomena
is a change in the world. A rule-driven, self-modifying model is just
as capable of formulating hypotheses about what external-world
entity could be causing its phenomena as a scientist.
(or course if it isn't self-modifying, it is a bit stuck. By why
make that assumption? It has nothing to do with
phemomena. Self-modification, is functionality, not phenomenality)
> The main message to take from this is that this bizarre concoction is the
> necessary circumstance that would exist in a human if it weren't for what
> phenomenal consciousness provides a human scientist.
You haven't demonstrated anything of the kind. The conclusion that
"Marvin" plays a vital role is just produced ready-made
at the end. It is not justified.
-------------------------------------------------------------------------
[*] See separate posting "Phenomenology"
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Everything List" group.
To post to this group, send email to everything-list.domain.name.hidden
To unsubscribe from this group, send email to everything-list-unsubscribe.domain.name.hidden
For more options, visit this group at
http://groups.google.com/group/everything-list?hl=en
-~----------~----~----~----~------~----~------~--~---
Received on Tue Nov 28 2006 - 13:31:58 PST