Re: Cool visual illusion


"feedbackdroids" <feedbackdroids@xxxxxxxxx> wrote in message
news:1133679942.647891.13750@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
>
> Allan C Cybulskie wrote:
>> "feedbackdroids" <feedbackdroids@xxxxxxxxx> wrote in message
>> news:1133638799.972872.252830@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
>> >
>> > Try to understand the list of 8 items, mentioned previously. Try to
>> > understand that the items on the list above lead to the items below.
>> > Try to understand that stopping at items #1 and #2 does not solve the
>> > problem. Try to understand that to design and build a system, you have
>> > to go through the entire process. Try to understand systems-level
>> > problem solving.
>>
>> The problem is that you seem to fail to comprehend -- even though I
>> pointed
>> it out to you point-blank -- what that actually means. Neurology MIGHT
>> come
>> into play at items 4 and 5 on your list, when we get down to doing a
>> specific implementation of intelligence. Since you claim that you have
>> to
>> go through the entire process, do you really want to claim that we've
>> done
>> 1 - 3 satisfactorily to enable us to undertake 4 and 5? Where do you
>> think
>> neurology can help us on items 1 - 3?
>>
>
>
> If you had been following the discussion, you would see that Wolf did
> exactly what hard-core behaviorism would never do, and what philosophy
> is incapable of doing. Which is to attempt to explain outward behavior
> by "going inside" to look at internal mechanisms. This is not just
> what, this relates to how. Where behaviorism ends, physiology begins.
> Even Skinner realized and said as much.
>
> The following comment is irrelevant ... do you really want to claim
> that we've done 1 - 3 satisfactorily to enable us to undertake 4 and 5?
> ... as both philosophy and behaviorism have about given us as much as
> they likely ever will.

Every fact uncovered by the EAB is of importance. Some facts are more
worthwhile for AI than others. For example, there are two process that
modulate the reinforcing efficacy of stimuli as well as making behavior that
has been reinforced by those stimuli more likely. This has a direct bearing
on what animals do in complex environments. These two processes are
"resurgence" and "schedule induction." In resurgence, responses that were
once reinforced, but later extinguished because some other response (B) was
reinforced, are emitted when B is placed on extinction. More important
still, though, is schedule induction - here, the introduction of an
intermittent schedule of reinforcement (often food-presentation) with a
"predictable" period of non-reinforcement following the delivery of a
reinforcer, greatly affects the reinforcing efficacy of other stimuli. No
amount of looking at brain structure will ever yield facts like this, or any
other fact about behavior, and these facts are of overwhelming importance.
Once again, Dan just spews nonsensical garbage.




>
> That's the real truth, however, in practice, the systems-level approach
> continually works top and bottom, revising as it goes along. Anyone
> who's actually designed anything, or written a non-trivial computer
> program, knows top-down and bottom-up are 2 sides to the same coin.
> However, most of the work always comes in at the lower levels in the
> process.
>
>
>
>
>> I have argued that neurology is not going to be that useful in doing
>> steps
>> 1 - 3 as even things like behaviourism are for reasons that should be
>> clear
>> to anyone who actually works with computers.
>
>
> You're just arguing which bits and pieces are more important. Sorry,
> dude, they're all important in their proper place. [see comment above
> about programming]. Systems-level thinking is a total process.
>
>
>
>
> It is extremely difficult to
>> discover algorithms by studying a particular implementation of them.
>> And,
>> usually, it isn't necessary, since studying what the system does -- this
>> would be the behaviourist view -- is usually superior to trying to
>> decipher
>> the implementation. Moreover, if you do so you'll likely copy the bugs
>> as
>> well -- biologically introduced illusions, for example -- which surely
>> isn't
>> what we want. So, while neurology will give us SOME useful
>> implementation
>> and design details, it is not required to create AI ... and so is not as
>> important as you make it out to be.
>


.

Relevant Pages

  • Re: Cool visual illusion
    ... Where behaviorism ends, physiology begins. ... Some facts are more ... > has been reinforced by those stimuli more likely. ... greatly affects the reinforcing efficacy of other stimuli. ...
    (comp.ai.philosophy)
  • Re: A step by step discussion
    ... Both behaviorism and cognitive "science" have their assumptions, ... the facts of Kandel's experiments are what they are and they are ... Behaviorists and cognitivists are not saying the same things in ...
    (comp.ai.philosophy)