Memory vs. Intelligence [was Re: Doogie Mice]

Robert J. Bradbury (bradbury@www.aeiveos.com)
Sat, 4 Sep 1999 02:11:21 -0700 (PDT)

I objected that the Nature/mice article was stretching things when discussing an enhancement of "intelligence".

On Sat, 4 Sep 1999, John Clark wrote:

> In one test a tone was sounded a few seconds before an electrical
> shock was given, the Doogie mice learned faster than the wild mice
> that the tone meant danger. That's intelligence.

I guess we differ on our definition of intelligence (which was a point I think I made in one message). I would say that remembering a correlation between two events *is* memory, not intelligence. It is the *memory* that two events occur in close time proximity.

> Even more impressive the
> Doogie mice learned much better when their old ideas no longer worked;
> after a while the tone still sounded but the shock no longer came, the wild
> mice were much slower in figuring out that the tone no longer meant danger.

Again, a new "memory" (discorrelation), on top of an old memory. You are implying that understanding that something only works 3 out 10 times is intelligence. I would call it simply majority rules "logic" where the critical improvement is the ability to more accurately store the number of successes and failures.

> To quote from the article " transgenic mice are quicker to learn
> to disassociate the previously paired events". You can't do that
> with improved memory alone.

The comparison & selection part of the thought machine has remained the same, what has improved is the ability to remember the events and recall them for comparison with a higher signal-to-noise ratio.

>
> Speaking of memory, this is the clearest indication yet that Long
> Term Potentiation is memory, or at least a good part of it is.

Won't argue with that.

> It's been shown before that slowing down LTP over what's found in
> nature hurt memory, but this is the first time it's been shown that
> speeding it up helps memory. That alone is worth a Nobel Prize.

Maybe, I'd say a careful review of the literature would be warrented. If these individuals were to get a prize, and the person who thought up LTP as a memory mechanism didn't, then I would argue that it would be fairly unjust.

> The article explains what changes were made to the mice and how
> they were made, then they report on how the modified mice did
> when they took several standard tests, what is your objection to that?

The major part I objected to was the suggested *extension* of the findings of *memory* improvement to *intelligence* improvement in humans. One of the purposes of peer review is to not allow papers to make claims that are not supported by the actual data/experiments in the paper. But as I said, I understand why the let it slip through.

> By the way, the article is in Nature not Science.
Yep, I corrected that mis-statement.

It would be interesting to list various types of thought or intelligence and see which of them are really memory based and which involve significant other factors (like the ability to map patterns from one field onto another field). To me intelligence is much less the ability to recall stuff and much more the ability to do mental transforms.

Robert