[Robotgroup] the allure of walking robots

[Bryan Bishop]

On Tuesday 08 July 2008, Def Egge wrote:
> I have a student this Summer who is a very difficult subject to
> interact with ... "it" (FERPA) asks questions but, upon receiving
> answers, exhibits neither facial gestures nor body language that
> might indicate comprehension, confusion, or any other human
> condition.  I am left with neither cues nor clues about what else is
> needed, expected or desired.  I understand what is meant by "talking
> to a wall" now.

Funny how this resembles the behavior of individuals with autism. 
http://en.wikipedia.org/wiki/Autism#Communication

"About a third to a half of individuals with autism do not develop 
enough natural speech to meet their daily communication needs.[35] 
Differences in communication may be present from the first year of 
life, and may include delayed onset of babbling, unusual gestures, 
diminished responsiveness, and the desynchronization of vocal patterns 
with the caregiver. In the second and third years, autistic children 
have less frequent and less diverse babbling, consonants, words, and 
word combinations; their gestures are less often integrated with words. 
Autistic children are less likely to make requests or share 
experiences, and are more likely to simply repeat others' words 
(echolalia)[23][36] or reverse pronouns.[37] Joint attention seems to 
be necessary for functional speech, and deficits in joint attention 
seem to distinguish infants with ASD:[1] for example, they may look at 
a pointing hand instead of the pointed-at object,[24][36] and they 
consistently fail to initiate a pointing gesture to "comment" about 
or "share" an experience at age-appropriate times.[1] Autistic children 
may have difficulty with imaginative play and with developing symbols 
into language.[23][36]

In a pair of studies, high-functioning autistic children aged 8–15 
performed equally well, and adults better than individually matched 
controls at basic language tasks involving vocabulary and spelling. 
Both autistic groups performed worse than controls at complex language 
tasks such as figurative language, comprehension and inference. As 
people are often sized up initially from their basic language skills, 
these studies suggest that people speaking to autistic individuals are 
more likely to overestimate what their audience comprehends.[38]"

That's actually an interesting story behind all that, when it comes to 
robotics and artificial intelligence it turns out that a particular 
research group, the one simulating large regions of the brain on an IBM 
cluster somewhere in Switzerland, also does significant research in 
autism and the microcircuit basis of the frontal cortices. 
	http://heybryan.org/intense_world_syndrome.html
	http://heybryan.org/mediawiki/index.php/Henry_Markram

There's an awesome hour-long video presentation from Henry:
http://video.google.com/videoplay?docid=-2874207418572601262&q=almaden+cognitive+computing

It's mostly computational neuroscience, and not necessarily robotics, 
but it might explain why the EEG spiking models aren't going anywhere, 
to say nothing of AGI models with formal logic and other silly 
hierarchical goal systems (UIAX and such). 

Maybe it could be on the agenda for the afterwards-sitting-around 
section?

- Bryan
________________________________________
http://heybryan.org/