|
Author
|
Topic: Is intelligence compositional?
|
Bruce Schuman
Member
Member # 1048
|
posted 02. January 2004 17:56
Does it make sense to understand intelligence as "compositional" -- in the sense that it might be defined as a "composite" that is constructed from chains or networks or recursively nested hierarchies of some lowest-level unit/element?
Poll Information
This poll contains 1 question(s). 5 user(s) have voted.
You may not view the results of this poll without voting.
| |
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 02. January 2004 18:36
Are you asking whether more sophisticated intelligence can be built out of simpler pieces, or whether there exists a unique minimal unit of intelligence, of which all intelligence is (can be) constructed?
I think I would give different answers to those two questions.
IP: Logged
|
|
Bruce Schuman
Member
Member # 1048
|
posted 02. January 2004 21:06
Hi, thanks for the comment.
I just ran into this site today, and thought it was pretty interesting. My "poll" was just a stab to figure out how this place works -- the explanation I offered doesn't seem to show. Or if it does, I don't know where.
Rex Kerr: Are you asking whether more sophisticated intelligence can be built out of simpler pieces, or whether there exists a unique minimal unit of intelligence, of which all intelligence is (can be) constructed?
I think I would give different answers to those two questions.
Bruce Schuman: Yes, this is a very abstract subject, and those options were very minimal. I was prompted to create this question because of a thread I saw posted here in the Brainstorms -- regarding the "smallest possible unit of intelligence". The author of that thread was arguing for a "mega-intelligence" approach, that involves combining (if I understand it right)collective inteligence from a wide array of "life forms".
My idea -- and a theme I have been exploring -- is that intelligence (and perhaps meaning and understanding and "all conceptual structure") can be constructed from a primitive logical element -- which, for the moment, I am calling a "distinction". Generally, the idea is -- any idea can be built up as a composite of nested distinctions.
So -- my thought about intelligence was -- maybe intelligence could be defined in a "recursively compositional" format -- perhaps as the capacity to make a distinction. So, if we have some irreducible "monad" of intelligence, capable of making some single lowest-level fine-grained distinction -- then maybe all other functions of intelligence are built up from these lower-level constructive units.
Adding "life" to this process -- suppose we said that "living intelligence is the capacity to make (and perhaps act on?) a distinction". Increasing levels of intelligence might then, in this scheme, involve greater and greater numbers of nested distinctions.
This idea somewhat reminds me of the "Society of Mind" idea proposed by Marvin Minsky in the 1980's -- and feels to me like a workable notion. We could make a mechanistic analogy to a "one dimensional sensor" -- like a thermostat -- that reads one dimension of sensory experience, and is capable of making distinctions along that dimension. If this makes sense, then we might say that increasing intelligence involves increasing the number of simultaneous dimensions.
All of this might map quite neatly into computer logic -- and indeed, "the entire internet" (a theme of interest to the creator of the thread I was responding to) -- because it is solidly algebraic and digital, and directly correlates to the nature of information.
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 02. January 2004 21:28
I assume from your comment that you meant the "unique minimal element" version of the question, so now I can answer the poll.
It does seem to me that there is a difference between intelligence and knowledge, and that knowledge is what is more readily built up from distinctions, not necessarily intelligence.
IP: Logged
|
|
Bruce Schuman
Member
Member # 1048
|
posted 02. January 2004 22:00
RK: I assume from your comment that you meant the "unique minimal element" version of the question, so now I can answer the poll.
It does seem to me that there is a difference between intelligence and knowledge, and that knowledge is what is more readily built up from distinctions, not necessarily intelligence.
Bruce: I have been bouncing around this site, looking at a few other threads -- thanks for this response. I think this theme of "motive" might be instructive....
Your distinction between knowledge and intelligence makes sense; we might see knowledge as static, and intelligence seems dynamic and active. For me, in my explorations, it has always seemed clear that any act of cognitive discrimination or judgment is "motivated" -- there is some reason, some "purpose" -- indeed, some "energy" -- behind that act of judgement.
So, perhaps what I am suggesting/exploring is the idea that intelligence involves motivation --"something that cares" -- something that is motivated to make a distinction (and thus, in this sense, create "knowledge").
This might be a model of hypothesis formation: something catches our eye, "for some reason" -- and we attempt to characterize it, describe it, measure it, etc. As that process unfolds, we create a model of this thing we think we are experiencing -- and then we act on the basis of this model, testing it, to see whether it "really works". This seems to be a primitive and natural form of "the scientific method" -- how people create models of reality, test them to see if they really work -- and then, if they pass the test, go on to build the new idea into their personal worldview or pantheon....
So, maybe to expand on the initial concept -- intelligence might involve the active and motivated energy involved in making a distinction (??) And maybe a kind of "self-consciousness" emerges as an artifact of the feedback and reality-testing phase....
"I test, therefore I am"
Or better yet, "I test, therefore I converge".
*
But is the motivation in the lowest-level "module" ("monad") of distinction -- like a one-cell organism that reacts to light or heat -- or does it reside in the entire compositional (cellular) structure of some larger whole? I like the idea of "isolating a motivational vector" -- and maybe something like that could be defined in the very smallest unit of analysis. But in a holistic/composite ("gestalt") sense -- that motivation would also seem to exist within the larger whole -- perhaps additively assembled from all the lower-level units of energy that make up the whole.
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 03. January 2004 01:27
Bruce Shuman wrote quote:
Or better yet, "I test, therefore I converge".
In one of the seminal books of the 'cognitive revolution,' Plans and the Structure of Behavior, Miller, Galanter, and Pribram proposed (in 1960) a primitive called "TOTE" - Test-Operate-Test-Exit. Their conception was that TOTE units were organized in a nested hierarchical fashion. They were looking at cognitive systems in general, not just "intelligence" (whatever that is). The notion bears an obvious relationship to the kind of conjectures Norbert Weiner made in several works after WWII.
RBH
Hm. Since this is still the last posting, I'll edit it very late to add that in the mid-60s Eric Lenneberg proposed categorization as the fundamental cognitive operation.
My object in noting these predecessors is to suggest that there have been attempts in the literature to provide an "atom" of cognition, and that discussions of them are in that literature. One doesn't have to reinvent some things de novo.
[ 03. January 2004, 12:49: Message edited by: RBH ]
IP: Logged
|
|
Bruce Schuman
Member
Member # 1048
|
posted 03. January 2004 15:10
quote:
In one of the seminal books of the 'cognitive revolution,' Plans and the Structure of Behavior, Miller, Galanter, and Pribram proposed (in 1960) a primitive called "TOTE" - Test-Operate-Test-Exit. Their conception was that TOTE units were organized in a nested hierarchical fashion. They were looking at cognitive systems in general, not just "intelligence" (whatever that is). The notion bears an obvious relationship to the kind of conjectures Norbert Weiner made in several works after WWII.
(I did show up here early this morning, and was in the midst of a response when I somehow clicked the wrong combination of keys, and vaporized my message -- maybe control-w, that closes a browesr??)
It's been interesting to wander around this site, and find the kind of references that people tend to offer here. It seems that I come from a very similar background, and have read a lot of this material. I was working on "cybernetic models of cognitive structure" back in the late 1960's at UC Santa Cruz, and Plans and the Structure of Behavior was a book that influenced me.
And I did read Norbert Weiner, of course, as well. I was fascinated by that approach to cybernetic control, based on continuously variable mathematics. Very tricky stuff.
I also saw references to "Parallel Distributed Processing"; I had both of those books in the mid-1980's, and conscientiously beat my head against them -- though I tend to take a different approach....
As to whether I am trying to "reinvent" something -- well, maybe that is possible. But, just very casually -- I would say that the suggestion that "categorization" is a primary element of cognition -- for me, is not a very satisfactory approach. Why? Because categorization is a complex process, that can be analyzed in detail. Categories are "built up from more primary elements" -- and my concern is to understand how that happens. So -- I am not disagreeing with that proposal -- just saying that for me, it doesn't go far enough. If I am missing a point here, perhaps you can expand on the suggestion.
And the idea that "intelligence" can be built up from composite units of some type -- maybe that has been discussed and developed -- but in my fairly wide-ranging research, I never saw an approach that I found satisfactory. I suppose we all have different ideas about what constitutes a "primitive" element. But for me -- it's got to be an irreducible thing. My own instincts hover somewhere around the "Dedekind cut" that is used to define a point on the real number line. I think we can build up all "categorization" from there.
I'll be back. This is an interesting site, with a lot of valuable and suggestive stuff. Thanks to all.
Further links to some of my work in this area:
http://originresearch.com/sd/home.cfm
[ 03. January 2004, 16:19: Message edited by: Bruce Schuman ]
IP: Logged
|
|
Noel Rude
Member
Member # 516
|
posted 06. January 2004 11:29
The results of such a poll taken here at ISCID would be interesting. But it would be hard to answer if the choices seem to be the reductionism of philosophical materialism versus the dualism of Greco-Roman Christendom. The prototype intelligence implies a complex of features – consciousness, perception, reason, volition, information – intelligence is already a complex notion. Maybe it would be better to ask whether or not we think intelligence has components not reducible to chance and/or necessity. If what you mean is logics – “a primitive logical element” – then maybe you should ask whether we think this would exist in a disembodied state (are we mathematical Platonists?) and whether we would link consciousness to this or whether we would consider consciousness another matter? Actually I think it would be helpful were some bright ISCIDite with some philosophical sophistication to put together an easy to understand questionnaire that asks all “the right questions” thereby helping each of us to see where we stand.
IP: Logged
|
|
g125
Member
Member # 1129
|
posted 23. February 2004 21:44
Having just wandered through the thoughts set out above it would seem that yes intelligence is compositional... in the sense that yes consciousness exists in a disembodied state but is linked to an organic 'home'... and intelligence is activated (fused consciousness) when it engages in a reflex action with another complimentary 'home'.
IP: Logged
|
|
|
Printer-friendly view of this topic