Page 2 of 2 <12
Topic Options
#26450 - 06/30/09 07:11 AM Re: Analyzing Methods of Measuring Intelligence [Re: coelentrate]
god.over.djinn Offline
pledge


Registered: 06/23/09
Posts: 75
Loc: Melbourne
 Originally Posted By: coelentrate
The ultimate way to see if the synaptic path and the electrical flux are sufficient to encode memory, is to build a system with only that.


Artificial neural nets learn. Data that is used to train them will continue to have an effect long after the training data has been discarded.

It is hard to say that this is memory per se, but then, what is memory? And in any case, ANNs can't hold a candle to the complexity of their biological counterparts.

Whether or not memory has a physical substrate beyond what synaptic connections encode for requires a null hypothesis along the following lines: that in fact, the neural/synaptic model is sufficient. The onus is then on anyone who wishes to test (or challenge) this hypothesis to seek conclusive experimental evidence favouring some alternative.

But maybe this evidence has already been found and presented? I don't know, I am not a neuroscientist. It is possible, but I doubt that any such findings would be uncontroversial.

If we can assume that coelentrate, as a biologist, is familiar with biology in general, then it seems reasonable to suppose that if the synaptic model of memory faced severe opposition, then he would probably have heard whispers of this at some point.

On the other hand, to take the angle that "crystallisation of memory within the brain" actually refers to some abstract or distributed process that follows the principle of crystallisation, I don't think that sounds far-fetched. It would, however, require some creative experimentation to actually demonstrate it. Perhaps something to do with measuring the change in entropy of a brain as it approaches some phase transition associated with the storage of a fresh memory?
_________________________
SATAN, a recursive acronym invented by GOD: "SATAN: Advocating The Adversarial Nihilist"

Top
#26453 - 06/30/09 07:40 AM Re: Analyzing Methods of Measuring Intelligence [Re: Ethophobia]
Dimitri Offline
stalker


Registered: 07/13/08
Posts: 3151
Maybe a little off topic but I'd like to mention this:
Intelligence can hardly be measured by ANY test. There are too many factors which have their influence. (Mood, sick/healthy, heat/cold, stress,...)
IQ-test such as mensa uses to see if you fit in, I call hardly intelligent due to the above parameters. Some seem to forget intelligence is also experience of life. Of the actions one undertakes, of the motoric skills you attained thanks to intensive practice and so on. Intelligence is not limited to "brainpower".

If you really want to measure intelligence you'll have to follow a person his whole life and see the actions he takes in life together with the excuses why he/she does so.


Edited by Dimitri (06/30/09 07:43 AM)
_________________________
Ut vivat, crescat et floreat

Top
#26461 - 06/30/09 09:53 AM Re: Analyzing Methods of Measuring Intelligence [Re: god.over.djinn]
coelentrate Offline
member


Registered: 07/07/08
Posts: 164
Loc: Dundee, Scotland
 Originally Posted By: god.over.djinn

Whether or not memory has a physical substrate beyond what synaptic connections encode for requires a null hypothesis along the following lines: that in fact, the neural/synaptic model is sufficient. The onus is then on anyone who wishes to test (or challenge) this hypothesis to seek conclusive experimental evidence favouring some alternative.


There's no evidence yet that synapses and electrical currents are sufficient for human memory. It's just a model at this point, definitely not some standard to beat.



Morgan, that is a really good story. I can't access the original work right now to find what the function of that protein is. I'll definitely check it out later. I wonder if that protein synthesis is a cause or an effect of that memory, and how important it is. I'm reasonably sure that formation of short term memories is faster than protein synthesis.


Edited by coelentrate (06/30/09 09:54 AM)

Top
#26506 - 07/01/09 04:46 AM Re: Analyzing Methods of Measuring Intelligence [Re: coelentrate]
god.over.djinn Offline
pledge


Registered: 06/23/09
Posts: 75
Loc: Melbourne
 Originally Posted By: coelentrate

There's no evidence yet that synapses and electrical currents are sufficient for human memory. It's just a model at this point, definitely not some standard to beat.


Null hypotheses are not about standards to beat but about retaining the simplest models that have sufficient explanatory power to match our knowledge to date. Neural nets are simple models with good explanatory power.

The principle of parsimony is an essential part of forming robust scientific models.

G.O.D.
_________________________
SATAN, a recursive acronym invented by GOD: "SATAN: Advocating The Adversarial Nihilist"

Top
Page 2 of 2 <12


Moderator:  Woland, TV is God, fakepropht, SkaffenAmtiskaw, Asmedious, Fist 
Hop to:

Generated in 0.014 seconds of which 0.001 seconds were spent on 17 queries. Zlib compression disabled.