Tuesday, June 02, 2009

What's So "Intelligent" About "Intelligent Design"?

Many of the debates about "intelligent design" (ID) that I have read online have focused on the defintion of "intelligent". This is not necessarily because we all agree what "design" means, but rather because we know even less about that quality we refer to with the term "intelligent". If one cannot define what one means by "intelligent", then any attempt to define or investigate "intelligent design" would seem to me to be a futile exercise.

Some ID supporters have suggested substituting the term "purposeful design" for the term "intelligent design". To me, this sounds almost redundant; after all, design is all about "purpose", isn't it? And if that's the case, then "purposeful design" reduces to "purposeful purpose" or "designed design". Furthermore, it's not clear to me that the terms "intelligent" and "purposeful" are necessarily interchangeable, or mean even similar things.

Many ID supporters seem most upset about the implication that evolutionary theory is "random". That is, the processes by which new characteristics of living organisms come into being are not necessarily the result of intentional design. To many of them, this would eliminate a supernatural force or deity as the causal factor in biological evolution. Ergo, if one is committed to the intervention in nature of a supernatural force or deity, one must deny a priori the possibility that new characteristics of living organisms can come into being without "intention".

However, it is not necessarily the case that "purposeful" (i.e. teleological) objects and processes are necessarily non-random. First of all, it seems to me that "purposeful" is not an antonym for "random". For example, consider a falling rock: its movement as it falls is most definitely not random. Neither its trajectory nor its acceleration are "random" at all. On the contrary, they are predictable to such a degree that we call the mathematical description by which we can predict the movement of falling objects a "law" - the "law of gravity".

Ergo, it seems to me that the best antonym for "random" is "predictable", in the sense of being able to predict successive states in a dynamically changing system.

Given the foregoing, what is the best antonym for "purposeful"? Forgive me, but I think the only reasonable answer is "non-purposeful". This then forces one to define what one means by "purposeful". To me, the best definition of a "purposeful" (or "teleological", if you prefer the more technical term) object or process is "a dynamical process (or component of a dynamical process) in which the dynamical entity's actions are actively and homeotelically regulated by a cybernetic process that functions according to a pre-existing program, the outcome of which is a specified end state.
A homeotelic process is one in which a dynamical entity reacts to external perturbations from its original trajectory in such a way as to regain its original goal orientation. For example, an arrow fired from a bow is not homeotelic, whereas a heat-seeking missile is. By the same logic, a snowflake growing in a supercooled cloud is not homeotelic, whereas a virus replicating in a host cell is.

In my opinion, most of the arguments about "intelligent design" founder, not on the definition of "intelligent" but rather on the definition of "design". If one focuses not on "design" but rather on "purpose" (i.e. teleology), much of the disagreement (like a boojum) vanishes softly and silently away.

Indeed, I think the qualifier "intelligent" is unnecessary, and quite possibly redundant. Why argue over something – that is, "intelligence" – that is indefinable without self-reference?

That is to say, "purpose" is very clearly and unambiguously defined in cybernetics, as Gregory Bateson and Norbert Weiner pointed out a half a century ago. "Purpose" (aka "teleology") are what this argument is really about, and so it would help immensely if all of the participants on both sides of the debate would define it in such a way as to render its presence or absence empirically verifiable.

The same could also be wished about "intelligence", but I see no real hope for this, given that virtually every definition of "intelligence" given in this thread (and all previous threads) is neither empirically verifiable nor applicable to simple systems such as those found in viruses or very simple cells. How "intelligent" is the lambda bacteriophage? Compared to a human, not much; compared to a crystal of sodium chloride, tremendously so. Indeed, what separates crystallized viruses from crystallized salts is precisely the "quality" that separates life from non-life and "purposeful" from "non-purposeful" things.

Termites build termite mounds using a surprisingly simple set of "decision rules". For example, one decision rule (which is clearly "wired in" to the nervous system of worker termites) is the rule to stack particles of sand on top of each other and glue them together using a material like saliva in such a way as to produce an arch (this is beautifully illustrated in E. O. Wilson's masterpiece, The Insect Societies). In Höldobler and Wilson's new book, Superorganism, they explain in detail how insect societies produce astonishingly complex, adaptive, functional dwelling places, "highways" (army and driver ants), "farms" and "pharmacies" (leaf-cutter ants), etc. without anything that remotely resembles what we would call "intelligence" or "consciousness" (remember, their brains are smaller than a poppy seed and their life spans are measured in days).

Furthermore, none of the instructions for doing all of this "design" is encoded directly into the DNA of any given social insect. Rather, the instructions are "compiled" from the individual activities of thousands of individual insects performing very simple, stereotyped actions (mostly coordinated by chemical pheromones). In other words, the "intelligence" that produces the marvelous structures and functions of insect societies is a collective "intelligence" consisting of a small set of "decision rules" hard-wired into the nervous systems of individual insects.

Might it not be the case that this same process is the paradigm for all biological complexity? This would not only explain where the "designer" is (it's all around / inside us) and who the "designer" is (it's everyone, interacting collectively in producing the "superorganism"), it would also present what ID has so far completely lacked: an empirical research program. That is, one could search for the "decision rules" that produce biological complexity, in viruses, cells, insect societies, primate societies, and human societies, and figure out how the interaction of such rules produces biological complexity. And when you did that, you would have recreated the already-existing field of biology known as sociobiology, which is a branch of evolutionary biology.

Termites do not have "goals and foresight". Rather, they are quite literally programmed (i.e. "hard wired") to perform a surprisingly simple set of simple behaviors. They are born with this capability and do not have to learn it. Furthermore, their behaviors are extremely stereotyped and subject to quite a bit of essentially "random" variation. Despite this, and because there are so many of them (literally millions in some large hives), they collectively produce structures and functions that rival the most complex "artificial" factories and dwelling places designed by humans.

The point here is that "intelligence" is not being defined well at all, if it is restricted to humans and higher vertebrates, but not to insect societies. Each insect is definitely not "intelligent" (any more than each of our individual cells is), but collectively both the insect societies and our multicellular selves are intelligent. "Intelligence" is therefore an emergent property, rather than a pre-existing attribute. And evolution, of course, is all about emergent properties.

One of the points I tried to make earlier is that using human "intelligence" as a yardstick for intelligence in general is like using a Cray XMT as your yardstick for evaluating the "intelligence" of an abacus. In virtually every discussion I have read about "intelligence" at ID blogs, there seems to be an unspoken yet universal assumption that "intelligence" is an either/or phenomenon: either something is at least as intelligent as a human (or the Intelligent Designer aka God) or it isn't intelligent at all.

How "intelligent" a virus like the lambda bacteriophage? If "intelligence" is to be a useful (not to mention empirically measurable) phenomenon, it seems to me that it should fall somewhere along a spectrum, from the "intelligence" manifested by simple viruses up through the "intelligence" manifested by complex animal societies such as ours.

The latter point - that "intelligence" must somehow be massively multiplied as the result of social/collective interactions - is also non-trivial. As I pointed out earlier, an individual termite is extraordinarily "stupid", especially by human standards. Indeed, taken out of their social contexts, the behaviors of most social organisms seem pointless and almost random. However, what appear to be pointless and virtually random behaviors when viewed at the individual level become extraordinarily complex and "hyper-intelligent" when one moves up in organizational levels in animal societies.

How "intelligent" would each of us be, if we were forced to live in complete isolation from all other humans? If we were forced to do so from birth, our "intelligence" would be so limited as to result in almost instant death. Ergo, if one uses "able to live independently" as one's criterion for "intelligence", one would have to conclude that oak trees are immensely more intelligent than humans.

In my opinion, until ID theory comes to grips with the concept of "intelligence" in such a way as to make it both empirically verifiable and quantifiable, ID "theory" will continue to be not much more than unsupported speculation.

As a first approach to an operational definition of intelligence, consider whether learning is a necessary component of intelligence. Several commentators have strongly implied that this is the case. That is, the more an entity is capable of "learning", the more intelligent it is.

However, using the ability to learn as a criterion for intelligence is fraught with difficulties. For example, termites do not learn to build termite mounds, yet virtually everyone in this thread has agreed that mound-building behavior in termites indicates that termites (at least as a group) are indeed intelligent. Ergo, it is quite clear that an entity that is utterly incapable of "learning" can still qualify as being highly "intelligent".

This would also apply to some ID supporter's assertion that the Intelligent Designer is the God of the Abrahamic religions. This entity is universally recognized as being a "4-O deity": that is, He is omnibenevolent, omnipotent, omnipresent, and omniscient. However, this last quality also strongly implies that the ID/God does not learn from His actions, as to do so would be directly contradictory with His being forever omniscient (i.e. from the beginning to the end of time, assuming that time does indeed end). Ergo, the ability to learn is quite clearly not a criterion for determining intelligence, if one assumes that the Intelligent Designer of ID theory is the God of the Abrahamic religions.

If one is familiar with so-called "expert systems" in computing, the same would be the case. Expert systems (ESs) do not "learn" to do anything in the sense that animals with "wet" minds do. On the contrary, an ES performs a complex (sometimes recursive) calculation using data embedded in one or more "truth tables", producing a calculated outcome. This outcome is sometimes hedged with statistical error calculations, but it is a calculated (i.e. not learned) outcome nonetheless. While the final calculation produced by an ES can be modified, this happens only when the values in the "truth tables" are modified. Otherwise, the outcome is simply a calculation. Ergo, expert systems do not actually "learn" anything, at least in the same way that animals (and some other living organisms) do.

So, I believe that it is fair to conclude that the ability to "learn" is quite clearly not a necessary criterion for intelligence. Some highly intelligent entities (such as termite colonies and the God of Abraham) are clearly incapable of true "learning". Conversely, some very unintelligent entities, such as bacteria, are nonetheless capable of changing their behavior over time in response to changes in their environment (the standard operational definition of "learning" in the cognitive sciences).
CONCLUSION: Intelligence is fundamentally unrelated to the ability to learn.

Which brings us back once again to the fundamental question: what is "intelligence", how can it be observed, and can it be quantified in any way? If not, then ID is quite literally a "science" without an empirically definable subject, and therefore a pointless exercise in mental masturbation.

One might also be tempted to define "intelligence" as "adaptability". That is, an "intelligent" entity has the ability to adapt its behavior (and, presumably, its underlying cognitive machinery by means of which its behavior is generated and regulated) in response to changes in its environment. However, this presents two serious problems to an ID supporter:

1) "Adaptability" is what natural selection is all about. Why posit the existence of an "intelligent" entity that is capable of "adapting" to changes in the environment, when this is precisely what natural selection is supposed to be able to do?

2) Since ID is supposed to be a theory that explains adaptation, then saying that the Intelligent Designer (i.e. the entity that moulds adaptations) is adaptable is essentially defining "intelligence" via constructing a tautology:

• "intelligence" = "ability to produce adaptations"

• "intelligent design" = the process by which adaptations are created

Ergo, "intelligent design" reduces to "adaptability producing adaptations".

This is what is sometimes referred to in logic as the "dormative principle" argument, from Moliere's "The Imaginary invalid". When asked how or why opium produces sleep, the learned doctor replies "because it contains a 'dormative principle'"; that is, it causes sleep because it contains a material that causes sleep. In the same way, defining "intelligence" as "the ability to adapt to changes in the environment" (including changes that have not yet happened, i.e. foresight) reduces to "design that is 'adaptable' because it is 'adaptable'".

Where does this leave us in a search for an empirically quantifiable definition of "intelligence"? And if the answer is, "nowhere", then where does this leave "intelligent design"?

In the same line of argument, one clearly cannot define "intelligence" as "that principle/process/quality by which complex specified information is produced". To do so would once more be arguing via tautology:

Question: What produces "complex specified information"?

Answer: Intelligence.

Question: What is "intelligence"?

Answer: That principle/process/quality that produces complex specified information.

Ergo, "the principle/process/quality that produces complex specified information" is what produces "complex specified information".

Again, a pointless exercise in semantic gymnastics.


As always, comments, criticisms, and suggestions are warmly welcomed!


Labels: , , , , , , ,