On 12/20/2014 6:19 PM, John Clark wrote:
On Sat, Dec 20, 2014 at 3:07 PM, <[email protected]
<mailto:[email protected]>> wrote:
> Part of how you sum up your core insight: that consciousness has no
detectable
objective reality
No, I'm saying that consciousness DOES have a detectable objective reality if and only
if it's a brute fact that consciousness is the way data feels like when it is processed
intelligently. And I'm saying that human beings can detect intelligent behavior and so
can the process that produced them, Evolution.
> evolution cannot detect consciousness.
If I accept that Darwin was correct and if I also accept that John K Clark is conscious
then I am forced by logic to conclude that consciousness is indeed the way data feels
like when it is processed intelligently.
As a corollary I MUST also conclude that to whatever degree the Turing Test is
successful at detecting intelligence then it must be equally successful at detecting
consciousness.
John K Clark
In one of his recent papers, http://arxiv.org/pdf/1405.0493v1.pdf, Max Tegmark comments on
this and reaches a somewhat different conclusion. He starts from Tononi's "Integrated
Information Theory" (ITT):
/ IIT has generated significant interest in the neuro-//
//science community, because it offers answers to many//
//intriguing questions. For example, why do some infor-//
//mation processing systems in our brains appear to be//
//unconscious? Based on extensive research correlating//
//brain measurements with subjectively reported experi-//
//ence, neuroscientist Christof Koch and others have con-//
//cluded that the cerebellum a brain area whose roles in-//
//clude motor control is not conscious, but is an uncon-//
//scious information processor that helps other parts of the//
//brain with certain computational tasks.//
// The IIT explanation for this is that the cerebellum//
//is mainly a collection of “feed-forward” neural networks//
//in which information flows like water down a river, and//
//each neuron affects mostly those downstream. If there//
//is no feedback, there is no integration and hence no con-//
//sciousness. The same would apply to Googles recent feed-//
//forward artificial neural network that processed millions//
//of YouTube video frames to determine whether they con-//
//tained cats. In contrast, the brain systems linked to con-//
//sciousness are strongly integrated, with all parts able to//
//affect one another.//
//*IIT thus offers an answer to the question of whether*/*/
/**/a superintelligent computer would be conscious: it de-/**/
/**/pends. A part of its information processing system that/**/
/**/is highly integrated will indeed be conscious. However,/**/
/**/IIT research has shown that for many integrated systems,/**/
/**/one can design a functionally equivalent feed-forward sys-/**/
/*/*tem that will be unconscious.* *This means that so-called*/*/
/**/“p-zombies” can, in principle, exist: systems that behave/**/
/**/like a human and pass the Turing test for machine intel-/**/
/**/ligence, yet lack any conscious experience whatsoever./*/
//Many current “deep learning” AI systems are of this p-//
//zombie type. Fortunately, integrated systems such as//
//those in our brains typically require much fewer computa-//
//tional resources than their feed-forward “zombie” equiv-//
//alents, which may explain why evolution has favoured//
//them and made us conscious.//
/So contrary to most on this list, he thinks a philosophical zombie is possible. But
maybe it's materially inefficient such that consciousness is favored. Personally, I think
"integrated" is a vague concept and amounts to "and then a miracle happens" without some
further elucidation.
Brent
--
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.