Again a quick thought - but "definition" is (it follows from what I've been 
saying) a v. unproductive way  to think about these things.

"Define the map of Italy".

The best "definition" of the map is the map itself.

How do you go about one of the simplest episodes - let's say reaching for a cup.

Is that controlled by a definition, or a logical set of properties?

I would hypothesize that it's directed by a fluid *outline*/map of  grasping 
for a cup - a fluid outline which is directly matched to (and convertible into) 
the outlines of the relevant body actions. Schematic/iconic/diagrammatic 
thinking is key here.


From: Piaget Modeler 
Sent: Thursday, July 19, 2012 2:48 PM
To: AGI 
Subject: RE: [agi] Episodes


What I was looking for was a definition, which you were providing. 


If you're saying that things can't be defined, modeled or characterized then 
that is an unreasonable position.


I think we were doing pretty well just now. Your feedback was helpful for my 
model of episodes.





--------------------------------------------------------------------------------
From: [email protected]
To: [email protected]
Subject: Re: [agi] Episodes
Date: Wed, 18 Jul 2012 22:45:09 +0100


Dunno what you mean. I suspect you're still trying to do this logically - in 
some form of logical sequence.

The "terrifying" thing is the brain does it imaginatively - by storing and 
editing image sequences - terrifying because it requires still technically 
awesome powers.

What any logical approach to a sequence can't do is "fill in the gaps" in 
examining an episode.


From: Piaget Modeler 
Sent: Wednesday, July 18, 2012 10:16 PM
To: AGI 
Subject: RE: [agi] Episodes



I'll skip the sex imagery, but your point is taken. 


It appears that from moment to moment we match memory to stimuli to predict 
what happens next.


So at any moment we receive stimuli.  That's the trigger. Let's rewrite our 
concept then:




          Episode(stimuli, series)




Do we need anything else, of course there is a process which  maps the stimuli 
to the episodes. 
And perhaps another process that is constantly forming episodes every moment...


What next...



--------------------------------------------------------------------------------
From: [email protected]
To: [email protected]
Subject: Re: [agi] Episodes
Date: Wed, 18 Jul 2012 22:05:26 +0100


PM:Now we just need to know, given a multi-modal stimulus stream, when the 
triggers occur

Again a quick thought, but the remarkable thing about the brain is that it can 
reconstruct a sequence - in principle - from almost any point.

Think of an image of two people's clothes on the floor by a bed, or two people 
lying in bed after sex ... no problems to construct what comes after, or what 
went before.

This works spatially too - classically via metonymy. Shoes under a curtain 
immediately denote someone standing behind it.

The brain remembers sequences temporal and spatial v. well. Play any tune at 
random on my ipod and at the end I can't help remembering what comes next.

Ditto an urge - a rumbling tummy - can immediately prompt the subsequent 
solution - say you at the fridge scoffing.

So virtually anything can be a temporal or spatial trigger.,I guess


From: Piaget Modeler 
Sent: Tuesday, July 17, 2012 3:03 AM
To: AGI 
Subject: RE: [agi] Episodes


Read the cliff notes:   


So, thus far we have an episode as follows:


      Episode(trigger, series)


where trigger is a core memory that caused the episode, and series is a 
subsequent action sequence.




Now we just need to know, given a multi-modal stimulus stream, when the 
triggers occur.




Interesting....



      

      AGI | Archives  | Modify Your Subscription   

      AGI | Archives  | Modify Your Subscription  

      AGI | Archives  | Modify Your Subscription   



-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968
Powered by Listbox: http://www.listbox.com

Reply via email to