Re: COMP & Self-awareness

From: Colin Geoffrey Hales <>
Date: Wed, 26 Jul 2006 16:39:39 +1000 (EST)

<mega snip>
(a) Phenomenal awareness (experience inclusive of a self model)
(b) Psychological awareness (knowledge inclusive of a self model)
<more snip>

Brent Meeker wrote:
> Maybe...with some more explication. You're saying that phenomenal
awareness (a) is perception that
> includes a model of oneself as the percipient.

(a) _may_ include phenomenal depiction of self. 'Self' is also implicit in
the location of the experiencer within the phenomenal depiction - eg your
vision feels like it's generated by your eyes - in reality you're 'being'
an occipital lobe persiscope. The periscope effect centres the visual
field apparent locus at your eyeballs. Imagine your own visual field
depicting you from your dog's perspective. Imagine your own visual field
with everything in it except your own body (look in a mirror - you are not
there are you a vampire? :-) ).

These 'self' aspects are optional implicit/explicit constructs in your
visual phenomenal life. They are, in effect, collections of intrinsic
knowledge... knowledge that is 'known' through the act of experiencing it.
The presence or absence of 'self' in this knowledge is moot.

>But I don't see what (b) is?...knowing you're six foot tall and live in

Yes... but in the way that you 'know' an old phone number... no experience
until recall, and an ability to use the number on recall - to behave in
response to the holding of that knowledge.

> Have you read any of John McCarthy's
> essays (see his website) on making a conscious robot?

My main goal in life is Artifical General Intelligence. Making a concious
robot is entirely what I am about. I know John McCarthy's stuff. He's one
of my AI heroes. BUT....He has the traditional
computationalist/functionalist/eliminativist views "if it look like an X,
acts like an X then it must be experiencing what X's experience". WRONG.

Like all computer scientists J McC forgets that the physics of phenomenal
consciousness (a) does not exist as a body of knowledge. Nobody can make a
scientifically justified statement as to the causal basis or predict the
actual experiential life of a rock, computer, scientist, the internet or
anything else. Computer science is particularly deluded that real
artifical general intelligence (AGI) can arise without the role of the
AGI's phenomenal life being sorted definitively.

> If a robot knows where it is (say via GPS) and
> senses its surroundigs
> (say by IR cameras) then it's got consciousness (a).

No. There is no phenomenal (visual) field here. There are photons hitting
a semiconductor array. Outer electron shells being perturbed, causing a
cascade of electronic causality within which a deductive electronic
analysis is enacted based on a-priori rules. Similarly, in a real retina
the photons cause a protein isomerisation and a different electrical
causality cascade (an action potential pulse train)...this too has nothing
directly to do with the generation of the visual experience - which
happens elsewhere in cortex.

Whatever the physics of biology has that generates a visual experience in
brain material - that physics is justifyably proven absent in the
discussed robot electronics. Not even close! There may be experiential
content, but whether this is even remotely like a visual experience
science cannot say. It could be the experience of a hot silicon rock. If
not, why not? If so, why so?

> If it also knows it weighs 5000lbs and has
> enough fuel to go 200miles it's got consciousness (b)

Side note: I'd say (b) was not consciousness at all. Nevertheless....
Experiencing something proportional to weight is (a)
Behaving because you have data corresponding to weight is (b)

> (I'm not just making these up - they're things
> a vehicle in the DARPA challenge would have).
> Now suppose that it also has a memory of what obstacles it crossed in
the past and which ones it
> failed to cross; and when it detects a new obstacle it uses this memory
to decide whether to go around or not.
> What kind of consciousness is that?

(b) = it's just a bunch of learning rules installed based on the a-priori
knowledge of beings who used phenomenal experience (a) to create the
rules, abstract them and make a type (b) zombie-critter live by them.


The existence and use of phenomenal fields is no natural and omnipresent
in humans it's hard to see them even though it's all there is to see. They
are seeing! Close your eyes. Your visual phenomenal field is radically
altered (assuming you have eyelids!). Imagine you had 'eyelids' on your
entire body surface touch-field (this field is actually generated in your
anterior parietal lobe or was it posterior matter).... You
would go numb all over when you 'closed' them.

The subtlety with (a) consciousness is, I think, it's actual role in
biological critters that combine (a) and (b) or more specifically use (a)
to generate (b). (a) was clearly found necessary by evolution or we
wouldn't have it. So what role has it got? My guess is that in providing
the source of all derived knowledge (b), it actually serves the purpose of
handling novelty. Generate all the rules you want... unless you have rules
for handling arbitrary levels of novelty your critter is going to
eventually crash and burn. Symbolic grounding... that's why I think (a)
was found so compelling by evolution.

No amount of abstracted (as-if) computation can replicate the 'natural
world' computation of (a) consciousness. IMO (a) consciousness is raw
physics-as-computation and it's application in the construction of 'self'
is just an option commensurate to its need. You only need a model of self
to behave intelligently in relation to self - a survival imperative. I can
imagine a servile AGI that had a radically complex phenomenal life (say
one that experiences neutrinos directly) but almost no explicit self
model. I'm not sure how useful it would be, but it seems very possible.

colin hales

You received this message because you are subscribed to the Google Groups "Everything List" group.
To post to this group, send email to
To unsubscribe from this group, send email to
For more options, visit this group at
Received on Wed Jul 26 2006 - 02:41:55 PDT

This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:11 PST