On Mar 9, 12:19 am, "Glen M. Sizemore" <gmsizemo... from yahoo.com> wrote:
> JC: When I tried to talk about simple networks you dismissed
> them saying, in essence, you weren't interested as they
> didn't cover conditioning in all its complexity.
>>> GS: The difference is that your description had no complexity.
> You were going on about simulating only the definitional
> properties of conditioning. When I was playing around with
> neural nets I did that in the first 3 minutes. Donahoe's group
> have gotten the network to do many things and now they have
> simulated revaluation.
And have they gone beyond the definitional properties of
In fact I have played with more complex networks to try
and duplicate the behavior of connecting stimuli with
other stimuli or actions with outcomes. Certainly my
first efforts took more than 3 minutes but than maybe
you knew more about ANN's than I did when we first tried
to simulate the definitional properties of conditioning.
> GS: The key term is "phenomena of interest." Who would
> be interested in a network that merely simulated the
> definitional properties of conditioning? Does it show
> faster reacquisition than acquisition? Blocking?
> Overshadowing? Sensory preconditioning? Higher-order
> conditioning? Generalization gradients? Peak shift?
> Fading? Spontaneous recovery?
And are the stimuli complex or just a signal representing
a complex stimuli? Are the responses complex or just a
signal representing a response? I could extend the
modeling to try and get the above behaviors although
I suspect they would emerge from an evolved network if
they were useful and not just quirks of behavior and
not directly relevant to intelligent behavior.
As I have mentioned before, the problem in AI is in getting
the stimuli in the first place from a complex input. Can
I attach an eye and legs to this neural network?
It is also about extracting the relevant principles involved
in intelligent behavior just as it is about extracting
the relevant principles of flight. We don't need to waste
time duplicating bio specific behaviors such as feathers
and flapping wings or even more complex behaviors you will
find in bird flight.
> There are still a lot of things that Donahoe's people
> haven't shown. I would like to see them take on schedule
> effects. But, still, they have accomplished a lot. And,
> of course, they have carried forth the notion that the
> facts uncovered by behavior analysis are what must be
> accounted for.
Did they account for anything or just duplicate a simple
relationship between signals representing complex stimuli?
> JC: Maybe you can explain it, in less technical terms than the
> paper, so that the views and mechanisms of this neural network
> model might find a wider audience?
>>> GS: Seems reasonably straightforward to me.
Yes, but you are not the "wider audience".
You are not on a radical behavioral analysis group.