[FOM] The Lucas-Penrose Thesis

[John McCarthy]

I don't think consciousness presents any logical problems provided the
thinker doesn't purport to be able to guarantee answers to "will I
ever" and "are my thoughts consistent".  Humans can observe some of
our mental state, but computer programs can be much more conscious
than that.  My 1995 Machine Intelligence 15 article "Making robots
conscious of their mental states" is available as
http://www-formal.stanford.edu/jmc/consciousness.html.
It goes into much detail about consciousness of the past, intentions,
hopes, knowledge and non-knowledge.

Provided one has an adequate language to express what humans are
conscious of and what robots should be made conscious of, the actual
programming should be straightforward.  In particular, no exotic
physics is relevant - not even ordinary quantum mechanics.  However,
we still haven't enough of the right predicates for human-level
consciousness.

> On Oct 2, 2006, at 10:20 AM, John McCarthy wrote:
> 
> > Don't forget Penrose's speculation in ENM that quantum gravity was
> > involved in consciousnesss.
> 
> Penrose recognizes a problem that Turing also recognized in  
> consciousness. Turing wrote:
> "I do not wish to give the impression that I think there is no  
> mystery about consciousness. There is, for instance, something of a  
> paradox connected with any attempt to localize it. But I do not think  
> these mysteries necessarily need to be solved before we can answer  
> the question with which we are concerned in this paper." Computing  
> Machinery and Intelligence, Alan Turing (1950)
> 
> I think Turing was wrong in his dismissal of the "paradox" here.  
> However, it seems plausible for a physicist such as Penrose to  
> explore the connection between the question of locality in  
> consciousness and the observed non-locality in quantum observations.
> 
> The locality issue remains a fundamental question in solutions that  
> use either constructive integration or constructive differentiation.  
> Carnap preferred the latter approach, as do I. There is some question  
> about how well Peirce and Kant defined this problem - they can be  
> interpreted both ways - but appear to have preferred the integrative  
> view [see Kant's "Pure Reason" and Peirce's "New List of Categories"].
> 
> Professor McCarthy will recognize constructive integration - it is  
> the common methodology of symbolic systems -  but one can argue that  
> it leads inevitably to dualism on this question since it relies on  
> epiphenomena.
> 
> One might take a third approach and dismiss the locality issue. This  
> would then require that one accounts for the physical latencies (that  
> appear to be contradicted by the evidence [Libet et al.]) and the  
> nature of the integrated illusion.

I do dismiss the locality issue and expect that the people arguing
against the "poverty of the stimulus" argument will succeed.  Here's
an abstract of a relevant forthcoming seminar.  Perhaps the lesson is
that Chomsky was unimaginative in dismissing the possibility of new
discoveries in grammar learning.

                              *******************************
                                      Cog Sci Lunch
                              *******************************
                         noon, Thurs., Oct. 5, Cordura 100


                                   Amy Perfors, MIT
                            http://www.mit.edu/~perfors/

                 Poverty of the Stimulus: A Rational Approach

The Poverty of the Stimulus (PoS) argument holds that children do not
receive enough evidence to infer the existence of core aspects of
language, such as the dependence of linguistic rules on hierarchical
phrase structure.  We reevaluate one version of this argument using a
Bayesian model of grammar induction, and show that a rational learner
faced with typical child-directed input and without any initial
language-specific biases could learn this dependency. This enables the
learner to master aspects of syntax, such as the auxiliary fronting
rule in interrogative formation, even without having heard directly
relevant data ( e.g., interrogatives containing an auxiliary in a
relative clause in the subject NP). The hierarchical Bayesian
framework is also applicable to other innateness questions involving
core aspects of linguistic structure.