Intelligent Behavior

(Wikipedia) Definitions:

Intelligence comes from the Latin verb "intellegere", which means "to understand". By this rationale, intelligence (as understanding) is arguably different from being "smart" (able to adapt to one's environment), or being "clever" (able to creatively adapt).

At least two major "consensus" definitions of intelligence have been proposed. First, from Intelligence: Knowns and Unknowns, a report of a task force convened by the American Psychological Association in 1995:

Individuals differ from one another in their ability to understand complex ideas, to adapt effectively to the environment, to learn from experience, to engage in various forms of reasoning, to overcome obstacles by taking thought. Although these individual differences can be substantial, they are never entirely consistent: a given person’s intellectual performance will vary on different occasions, in different domains, as judged by different criteria. Concepts of "intelligence" are attempts to clarify and organize this complex set of phenomena. Although considerable clarity has been achieved in some areas, no such conceptualization has yet answered all the important questions and none commands universal assent. Indeed, when two dozen prominent theorists were recently asked to define intelligence, they gave two dozen somewhat different definitions.^[1]

A second definition of intelligence comes from "Mainstream Science on Intelligence", which was signed by 52 intelligence researchers in 1994:

A very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings—"catching on", "making sense" of things, or "figuring out" what to do.^[2]

Can we define intelligent behavior, give an analysis, norms, criteria? This is important if we want to get epistemology beyond the definition or conceptual analysis stage. Maybe epistemology – “The Theory of Knowledge”- is really, or should be regarded as a non- anthropomorphic theory of intelligent behavior, or more generally as theory of “intelligence” per se. But all these terms are prejudiced epistemologically , what do we do to get around this?

The Turing test is a test, an interactive communicative test.(It was for computers after all.) Can we not define intelligence without interactive communications? Can we have a purely observational test for intelligence, involving no interaction? Deciding on the basis of actions is hard. Suppose we have a really good robot that makes car parts and assembles cars, does this make it intelligent? Suppose it starts with raw materials, iron ore, silica, raw rubber and produces a Ferrari, does this make it smart or just well programmed and capable? Could it not pass all the Wikipedia tests above and still not be intelligent. It could be adaptive, it could learn, modify its behavior, even be said to grasp abstract ideas.. These are performance criteria which could in principle be met by machines not much more advanced than those that exist today .Didn’t Big Blue, the chess playing program grasp, in effect, the abstractions in chess?

A system can evidence something we might, primitively, want to call intelligence, via its external behavior ,i.e., behavior we can observe without reference to the internal states of the system. Why would we call this intelligent behavior?. Probably because we observe that it is reactive, the systems performance depends on the external situation. (We actually generally do this on inference maybe, if it acts like us its smart.) If we don’t know its reactive we have a problem. Imagine an mechanical arm coming through a wall making marks on paper, a simpler Turing test.. Whether or not we think it intelligent depends what we make of the marks on the paper whether we think they are wrote learning or creative, assuming we can make any sense out of them at all.. It is impossible to determine externally whether the system is intelligent or not unless it is reactive. And this reactivity must be information based.

Phenomenal Properties (IMPD): Revised Summary -3/9/08

Explaining a difference requires a difference, but how much difference is enough, and when do you know if you have the right kind of difference?

The most fundamental fact about sensory experiences, aside from the fact that we have them, is that they are all qualitatively different. Hearing the violin is nothing at all like seeing it which is nothing at all like touching it or smelling the varnish on it. Sensory modalities are fundamentally different experientially, yet this is hard to explain on the neurological level. Let us call these differences: Inter- Modal Phenomenal Differences, or IMPD. It is easy to explain IMPD if you don’t know much- e.g., seeing is different from hearing because eyes are different from ears; easy and obvious.

This “explanation” becomes a little less obvious if we learn some physiology however and even more difficult to hold if we adopt a theory of the brain as an information system.. (Scientific developments produce new philosophical problems? Not unheard of- quantum mechanics.) Neurological activity is all basically ionic flow through semi- permeable membranes. The patterns may change, but so what? Is this suppose to account for IMPD? Why, if IMPD is basically a cortical phenomenon, is a pattern of activity in the auditory cortex so different at the experiential level from a pattern of activity in the visual cortex? The sensory organs are different off course, but we are not aware of the states of our sensory organs without higher level neurological activity. Further, from brain stimulation experiments, it looks like we can have sensory experience without sensory organ simulation. (Is there retrograde activity out this far –to the sense organs- with direct electrical simulation of the brain?) So there is a suggestion here that experience is a cortical phenomenon.

“ Cortical only” theories are a problem since they fail the “difference tests”. “Transducer/sense organ only” theories fail also perhaps because, at least, I can’t know what’s going on without cortical operations. But it is not at all obvious that this counts. Having a sensation and being aware of it or being able to respond to it are two different things. But other problems with transducer theory abound. First of all, moving the experiential level out to the sensors does no good. Its still the same sort of neuro- biochemistry out there; this too fails the difference test. Combining the two, holding that sensation involves both transducer and cortical neurological activity doesn’t help if their both basically the same thing.

Am I directly aware of the state of my sensory organs? This would seem hard to maintain-who is this “I”?. Our operating hypothesis is this: I am the sensation in that having the sensation is a state that I’m in, I am the sensing thing and the sensation. The total state of the organism is that of having or being aware of the sensation. If –sense organ and central neurological activity apart, each separately, or together are not enough for experience, which seems likely. So the sensation is a totals state of the system which involves the transducers which are different from one another, e.g. organ of corti, cones in retina, and the required difference now lies at the sense organ or transduction level.. (This is a great philosophical position since it is confusing, and explains nothing and with luck is irrefutable!)

. Including the interface level would seem to be required. Right now I’d say that the interface is were its happening, where the sensation is, or what it is but I find this close to incomprehensible. One problem is that sensation involves more than commutable aspects, more than information type operations and I seem to be aware of non-informational aspects of the process in sensing. At the sense organs, at the interface with the external world information is converted and phenomenological aspects of sensation are eliminated. (An important question here is : Is the cortex , or the nervous system in total, an information (only) system or not.)

((Note that we think that are visual world perceptions or experiences mirror or reflect in some sense optical reality or that they reflect actual aspects of it, e.g., spatial relations, colors, depth of field. Are their analogous views for the other modality? ))

Can brain stimulated people know that their current experience is different from that which occurs due to external stimulation? They can figure it out by consideration of their total situation –“I’m in an operating room now, not out in the woods listening to a bubbling brook”, but can they sense the difference?.

This is the best argument yet for the ghost in the machine, but the ghost has to have a really great home entertainment system for presenting the external world show. Introducing the ghost here really only ends the explanation; metaphysics as the terminator of enquiry rather than as an initiator- a god of the gaps introduced at some electro-physiological level.

The problem again is that it is not clear what sort of account of the variations in neurological activity, and especially in neural network activity could be seen as sufficient explanation for IMPD. IMPDs are as different as different gets, comparisons are impossible, yet the underlying physiology does not seem to admit to this order of qualitative difference.

Sense Data

A casual inspection of sense data theory suggests that the IMPD problem is not central, nor perhaps even considered. Does any one ask why sense datum have the secondary properties they do?

Dualism will not survive on quantum-mechanical tricks (Popper –Eccles), little miracles going on inside our heads against an informative theory of network function. Network operation is the key problem, especially the problem of syntactic (timed, structured, sequenced, controlled) behavior. The problem with the single photon response example is that in order to get this sensitivity, a whole lot of other things have to be going on, everything else has to be shut down or inhibited. Banging on a single receptor on a single neuron won’t do it, this selectivity, concentration, focus, whatever has to be generated by a whole lot of other activity. Focus or sensitivity has to be dialed in, that’s the problem, the system is too sensitive and has to be globally inhibited, that’s why we have GABA all over the place.

( The theory ((a theory?)) of IMPD is important because this will be the last refuge of serious mentalism (dualism). But the theory –I am the sensation; not I have the sensation- must accommodate other aspects of our mental life.)

Information and Qualia

Information has no qualia, no phenomenal properties, it is pure causation. More specifically, in the case of the aptly named action potential, it is a causation vector, a potential cause of a particular origin. This is another aspect of the problem of IMPD, since the information stream in the brain has no phenomenal characteristics, it cannot explain how variations in them can arise. It can have no phenomenal characteristics as a matter of empirical fact because action potentials of different sorts, say visual and auditory can impinge on single cell and we don’t have a see/hear sensation, we see and hear at the same time, but we don’t see/hear. Seeing is because of visual tract activity, but it is not that activity, this because seeing is more than having the information associated with seeing, it is an experience.

Not only can’t information theory explain IMPD, it can’t explain why qualia very within the modality. It can’t explain why orange is different from red, much less why red is different from loud or sweet. The problem is not that we know their different, the problem is that they look or are experienced as different. The problem isn’t that we can know, say or think they’re different; the problem is the difference in the qualia. Saying, knowing, realizing, acting as though, being aware of…these are cognitive, information based, cortical. Orange and red are different wavelengths, this can be picked up by the nervous system, but the fact that they look different is not a CNS phenomenon. Knowing that they’re different is CNS, seeing them differently is not..

We are Process

We are process although we tend not to realize it. This is one of the problems we have in understanding sensory experience. We tend to view ourselves as discrete entities with essentially episodic existences. This because memory and reactive (conscious?) experiences involve discrete and non-continuous incidents Understanding ourselves as ongoing processes is physiology, modern and not inherent to a natural language. Saying we are process, if metaphysics be damned- no mentalism- is trite. The question is: What are the philosophical implications of this scientific insight.

So to say we are the experience (of) as opposed to saying we have the experience (of) sounds strange. So what? It’s a philosophical insight. But does it explain anything? We have all the information, at least in terms of kind, that we are ever going to get concerning IMPD. It is hard to see how greater understanding of transducer operation , network operations, information flow or details of the biomechanical processes would be of any explanatory value. Somebody might say that some new facts explain it, but why should we believe them? We would have to adopt so new sort of criteria, but based on what? Do we have a theory of possible criteria? It seems silly to say we do, but again, epistemology is primary. It tells us what can be said by things like us. ( It’s always about the driving functions.)

Information and Qualia

Information has no qualia, no phenomenal properties, it is pure causation. More specifically, in the case of the aptly named action potential, it is a causation vector, a potential cause of a particular origin. This is another aspect of the problem of IMPD, since the information stream in the brain has no phenomenal characteristics, it cannot explain how variations in them can arise. It can have no phenomenal characteristics as a matter of empirical fact because action potentials of different sorts, say visual and auditory can impinge on single cell and we don’t have a see/hear sensation, we see and hear at the same time, but we don’t see/hear. Seeing is because of visual tract activity, but it is not that activity, this also because seeing is more than having the qualia of seeing..

Not only can’t information theory explain IMPD, it can’t explain why qualia very within the modality. It can’t explain why orange is different from red, much less why red is different from loud or sweet. The problem is not that we know their different, the problem is that they appear different. The problem isn’t that we can know, say or think they’re different; the problem is the difference in the qualia. Saying, knowing, realizing, acting as though, being aware of…these are cognitive, information based, cortical. Now, noting the difference assumes there’s a difference to be noted. But this difference is not the phenomenal difference, this difference associated with the qualia it is the difference in the physical phenomenon. Orange and red are different wavelengths, this can be picked up by the nervous system, the fact that they look different is not a CNS phenomenon. Knowing that they’re different is CNS, seeing them differently is not.

“Ought” is Subordinate to “Is”

Epistemology is subordinate to ontology, because “ought” is subordinate to “is” . The question of primacy in epistemology is settled easily –ought is again subordinate to is (possible) , or to what can or could be. Any word must have a driving function whose fundamental nature can be delineated. For words not immediately related to experience of the external world, the menu is limited to neurological operations, to combinations and variations of parameters related to various neurological and neurological system states. There are only so many things I can put behind words such as true, explanation, knowledge etc., the question then becomes what are these things now, and what should they be.

A given speakers epistemology then becomes a question involving an individual lexicon at a given moment.

How does language work? As it is internalized either visually or audibly we convert from the sensible to the neurological. What is the next level- syntactic? Notice that “Colorless green ideas sleep furiously. “ is at some sense comprehensible just as it is simultaneously meaningless. What is the evidence that semantics follows syntax? That the processing isn’t simultaneous? Syntax and semantics maybe logical as opposed to neurological distinctions. There maybe some neuro- research on this. Fmri’s on the difference.

Grant that I comprehend a proposition and that it is meaningful, e.g. it is syntactically and semantically acceptable. Does it convey information? Can it do so outside of a wider context? “The cat is on the mat.” Is this informative unless I know what cat and what mat is being talked about? I need to know something about the referenced entities. “Lincoln was a great president.” Maybe, maybe not, perhaps I have a theory to the effect that he could have prevented the civil war, or that he put the union above the abolition of slavery and thus was morally deficient.

The Lincoln proposition is a value judgment, I understand it as such if I know English and importantly, history. I know that somebody who asserts it, if I understand that they are asserting it; they could be doing something else; is making a value judgment and is probably in their view representing a fact.

.

How Does the Nervous System Work?

I know what the brain does, I just don’t know how it does it. Why is it not enough to know what the brain does- namely generates behavior and determines experience? Why do we have to know how it does it? I don’t have to know how my car or computer works to know what they can or will do based on my controlling inputs; this because they are not agents. I know or can determine all behaviors, capacities, performance limits, modes of operation… based on experience, and a few fundamental realizations. I don’t really have to know how they work. I know that my car can’t fly, and probably won’t float well, and I don’t have to be an expert to know this. The ”fly in the bottle” problem for philosophy is that we haven’t applied this simple observation to human beings.

The fact that people are agents reduces predictability and allows for novelty, but it doesn’t introduce transcendent possibility. You will never fly by flapping your arms, swim like the dolphins naked or jump into outer space. We simple haven’t realized that same biologically based limits apply to cognitive activities also. Nor do we understand what these limits are.

Information Based Science

Can an information based science do analysis, prediction, or cosmology? How far can we get without invoking rules, principles or laws or mathematical analysis.. Consider the rollbar problem –does the rollbar fail on ground contact?. Contemporary multiphysics- FEA computes stresses on an itinerative basis using geometric constraints mechanical principles and kinematics. Can I get the same results without mathematical analysis?

Engineering issues can always ((?)–what about space craft?) be resolved by testing, indeed the point of modern simulation techniques is ultimately to eliminate testing. But I could test until I got close enough from some point of view. Can I adequately determine performance without mathematical analysis however? How much testing would I have to do to establish performance on the basis of comparative results?

An information based epistemic technology comparable to our science and engineering would have to know something about the properties of materials. This could be gotten experimentally, but how do you normalize it ? E.g., how do you get something like our nominal tensile strength numbers in pounds per square inch. You could adopt a particular sample as a norm or standard unit and work from that- scale up or down. (Note this systems information is not limited to the visually apparent- it has chemical knowledge/information extraction techniques also or it can’t do much steel metallurgy.)

What’s the information based alternative to itinerative mathematical analysis? It probably depends on the required level of accuracy or certainty. If were smart we can do a worse case scenario. Proof it doesn’t fail (e.g., exhibit plastic hinge failure ) in the most difficult case and you have a successful design. But I still want to optimize- so I work backwards, reducing weight and cost. Most engineering can be done this way. Analysis is more an economic consideration than an epistemic one in applied science. The “understanding” aspect is psychological. Why can’t my understanding be based on countless experiments and observations?

Still, analysis is quicker and more cost efficient, but this doesn’t mean epistemic exclusivity.