Re: Innoivationm and the Curse of Knowledge, etc

casey wrote:
On Dec 31, 3:38 am, "Wolf K." <wolf...@xxxxxxxxxxxx> wrote:
This may be of intereest:

http://www.nytimes.com/2007/12/30/business/30know.html?th=&emc=th&pag...

Enjoy!


I didn't know you were into pop psychology?

?????

Among other things, the story refers to an experiment:

"Elizabeth Newton, a psychologist, conducted an experiment on the curse of knowledge while working on her doctorate at Stanford in 1990."

Whether the experiment actually proves what is claimed in the NYT story is another issue. Look it up, and judge for yourself. Anecdotally, it's long been known that an outsider often brings just the right kind of questions (sometimes the right kind of "wrong" questions) to a problem that help the experts reframe the problem and solve it differently or better.

An issue raised was the failure for experts
in a field to communicate effectively to
non-experts in that field. An issue I raised
many times with GS and DL without effect.
>
Experts in a field dismiss the potential for
other fields of interest to impact upon an
understanding of the problem at hand such as
GS when he dismisses simple machines as not
being relevant to understanding complex
biological machines.

Well, actually what my reading about simple machines has shown me is biological machines are rather more complex than I had imagined. I first thought that it would be simple to build a conditionable machine when I saws the turtles that the Engineering students had built (this was some 50 years ago). You know the ones I mean: Two or more photocells are coupled to motors through a balancing circuit so that when one cell receives more light than the other the turtle turns until all photocells receive the same amount of light. I was pretty sure that the Engineers were only a few circuits away from building a trainable turtle, one which followed you rather than someone else's flashlight. But that's because I didn't understand that simple machines like phototropic electric turtles aren't complex enough to reproduce even simple conditioning. Not even "association" (quotes because I don't believe any more that I know what AI engineers mean by it.)

Anyhow, it's a two-street: GS notices something that advocates of "simple RL machines" haven't noticed: that "trainable artificial neural networks" (for example) do not reproduce even classical conditioning. Why? Because you can't condition a system (animal) unless it has at least two built in behaviours. Why? Because conditioning means connecting a stimulus to a response that was not connected to that stimulus. But that implies that the s\de4sired response was connected to another stimulus...

All the trainable ANNs I've seen or read of were attempts to reproduce "association" of a stimulus to a response, on the incorrect assumption that conditioning consists of making such associations. What the TANNs have shown is that conditioning is rather more complicated than that. As I've tried to explain.

When I first read about the Skinner box it
seemed to me that the rat was being replaced
by simple measurable variables such as lever
pushing. The rat could be put in a 'black box'
and the relevant input/output be all that was
seen. You could then play the imitation Turing
game with another 'black box' containing
electronic components and play spot the
difference.

The difference would be easy to spot. Just change one or more environmental factors, such the colour of the light, or the timing between lever press and pellet release, or randomly not release a pellet, or whatever. Come to think of it, an electronic box that shows the typical strengthening of the conditioned response with random changes in timing would be an intersting challenge to build, don't you think? Or program...

To me the behaviorists were not studying
animals "in all their complexity" but rather
restricting the environment and selecting
a set variables such as lever pushing with
lights, bells, food, over time, to _simplify_
the system they were studying.


--
JC




You might just as well say the physicists don't study nature in all its complexity, but rather restrict the environment which they study and select set variables such as time of rolling down a slope, or dimming curve of a star or ....

Of course you have to simplify the system you want to study. When observing animals in their habitat, it's damn hard enough figuring out which of X, Y, Z... _might_ be a discriminant (stimulus for a conditioned response), let alone demonstrate that (in this particular environment, with this particular animal) X can be a discriminant. But change just one factor in the environment, and X may no longer function as a discriminant. And that one factor may simply be a troop mate's wandering closer than twenty feet, for example. Or a movement in a thicket some fifty feet off, which you (being merely human) haven't noticed, let alone responded to by marking your notebook. Etc. But it may be that 20 of 30 troop mates change the function of X, and 10 of 30 do not. So what's affecting the response to X? Etc. It's that complexity that makes EAB challenging. Just what does trigger any given response, and how is that effect modified by other factors? It makes it difficult to decide just exactly what you've seen when you've observed interactions in a troop of baboons, for that matter.

See, what you are reading now is a discriminant for you. Or rather, it's a trigger for whole raft of (mostly) internal discriminants, loosely called "concepts, thoughts, memories, sensory images,....." Depending on your mood, what you've read recently, your fatigue or alertness, your general "attitude" to the stance I take, etc etc, you will make some kind of sense of what I've written. And the sense you make will vary more or less drastically from the sense I hope you'll make.

So why do I keep on typing this stuff? Because every sentence I type is a discriminant for my typing behaviours, triggering a whole raft of (mostly) internal discriminants as listed above, which (eventually) result in my typing yet another sentence. And then I look at what I've written, and I see little red underline dashes under some words, and I'm conditioned to invoke the spell checker menu, and either change the spelling or add the word to the dictionary. Since I'm a bad typist, there's a lot of that behaviour... ;-)

I'll give you an example of an animal behaviour "in all its complexity" which reduces to a few simple rules of behaviour: the flight of a flock of birds. How do thousands of birds "know where to go?" How do they "know", for example, that the flock is turning right? To understand it, think of each bird as responding to two cues, namely the distances between its right and left neighbours. To maintain fairly constant spacing, all you need is a rule that be "if the distance between you and left bird becomes much larger than distance between you and right bird, veer left." Add rules for the other two dimensions, an you have enough to steer a large flock (relatively) safely through a maze of tall buildings. Do birds actually respond this way to each other's presence when flying a large flock? I don't know, but it sure looks like they do. And of course "rule" here means "If S then R", which may refer to either conditioned or unconditioned behaviour. To find out which, you'd have to watch a lot of birds from fledging to flying in flocks. Kinda hard to do, which I suppose is a reason no one to to my knowledge has done it.

BTW, the above explanation of the flight of a flock of birds was AFAIK first proposed or taken seriously by computer animators who were trying to simulate the appearance of a flock of birds. Seems to me they "reduced" the complexity of that flight to a very elegant, simple model. A thoroughly behaviourist one, please note.

HTH


PS: I can relate to the remote control with too many buttons. To operate your home entertainment system, five is all you need. I've figured it out. Companies that want my design advice may form a line to the right. ;-)

.

Quantcast