Thursday, January 26, 2006

The Intelligence Illusion

Listen to this Essay Instead

Audio Part I

Audio Part II

Audio Part II

A few weeks ago, I listened to George Gilder, the Founder of the Discovery Institute and Richard Dawkins duke it out on a Boston public radio program about Intelligent Design. As with other, now ever increasing, articles and shows on the Intelligent Design/ Evolution debate, I have noticed that for all the points and counter points made, there is one concept that both sides do seem to implicitly accept. While both “design” and “evolution” are questioned, intelligence never is. In fact, if not for the implicit agreement about what this concept means, it would be impossible to even debate whether life is an outcome of an “intelligent” process or a “non-intelligent” process. Which means that this debate actually pivots on our sense of reassurance about the word intelligence.

So what is “intelligence”? If you look up a typical dictionary definition, you will find entries like, “mental capacity”, “capacity to acquire and apply knowledge” etc. Though you will never see it stated explicitly, what these definitions really mean is the “mental capacity” of a human being or the human “capacity to acquire and apply knowledge.” Our baseline understanding of the word intelligence is the human being. We have long used the word intelligence to differentiate ourselves from the rest of nature’s creatures. Even scientists usually speak of intelligence as a human adaptation. The collective consensus of this concept seems to suggest that we believe there exists some ‘singular thingy’, the presence of which in humans makes us intelligent and the absence of which in other creatures or entities, makes them unintelligent.

This idea has certainly driven hoards of psychologist, cognitive scientists and neuroscientists to try and quantify this ‘thingy’ in the human brain. They have studied stroke patients and individuals with selective brain damage, to see what ‘intelligent’ activities they can and cannot perform. They have studied individuals with Downs Syndrome, autism, dementia and savants. They have also studied how healthy individuals perform ‘intelligent tasks’ under varying conditions, like sleep deprivation or high stress. With the advent of a host of neuroimaging techniques, many of these experiments have been repeated with ‘live’ brain scans of subjects as they perform these different tasks.

And the cumulative result of these studies?

First, the human brain is a composite of many parts, each performing different functions. Second, any given observable ‘intelligent’ behavior of an individual involves multiple different parts. So say, taking a geometry test will involve attention functions, retrieval functions, spatial functions, and language functions. Selective damage to some parts may only affect those functions while leaving others intact. Or as the case is with savants, some parts, such as those involved in music work extraordinarily well, allowing them to play and compose, while parts of their brain needed for language are impaired making them unable to read music. It is, in other words, impossible to study human intelligence without parsing it into a series of functions all of which contribute in varying degrees to behaviors we consider to be overtly intelligent.

This fragmented reality of intelligence is also borne out by studies in animal cognition and comparative brain anatomy. Comparative brain anatomy reveals that the architecture of human brain shares much in common with that of other species. This is of course to be expected from common descent. Behavioral studies of animals show that corresponding to these shared structures, animals demonstrate varying degrees of shared cognitive abilities with humans. Behavioral observations of our close cousins the chimpanzees, for example document that they too are capable of basic tool use, a behavior once thought to mark the advent of human intelligence.

Yet another field that poses a challenge to the notion of intelligence as a ‘singular thingy’ is the field of artificial intelligence. While studies of the human brain and animal behavior have taken a tear down approach to dissecting intelligence, the field of artificial intelligence has taken a build up approach. We try to build inorganic systems that can behave like a human being and perform ‘intelligent’ tasks. For example, we can create an inorganic computer system like Deep Blue to perform a very complex human task like playing the game of chess. Other artificial intelligence systems can be built to mimic other human like cognitive tasks.

This means that the concept of intelligence we take for granted, when placed in the MRI machine, does not prove to be the solid word we think it is. In fact, it is more like a sponge that can absorb of a lot of different functions, but does not hold up very well under pressured scrutiny. It is certainly a useful word when used to denote comparative relationships such as, “that man is more intelligent” or “that animal is less intelligent.” But it is not a clear-cut singular concept that can clearly categorize objects into intelligent and unintelligent bins.

If the word intelligence itself cannot be used to clearly separate the intelligent and the unintelligent, then the implicit agreement between the scientists and the ID folks central to the debate on evolution, that the creative process of an “intelligent being” or human being is completely different from the unintelligent creative process, must be reexamined.

How then does the creative process of the “Watchmaker” differ from the creative process of the “Blind Watchmaker”?

To find out, let us take another look at what the “Watchmaker” does when he creates and how evolution creates, this time from an information perspective. The watchmaker generates the first design for a watch using the information he has about components of a watch in his memory. Once the information in his memory has been translated into an actual watch, he then tests it or evaluates whether his design works. If the watch keeps time and keeps ticking he will retain that design in his memory. If the watch does not work he will discard that design. The figure below illustrates this process.

Now let us take a look at what evolution does. Evolution too involves information about the organism. This time the information is stored in the organism’s DNA or genetic memory. During reproduction changes to the DNA’s base pairs, generates variation or new information. Once the new organism with the new information comes to life, the environmental conditions will dictate whether it will survive. In other words environmental conditions will evaluate whether the organism’s genetic information will be retained in the genetic memory of its descendents. If the organism is not able to successfully reproduce and pass on its genes, its genetic information will be lost or discarded from the genetic memory of future life forms. The figure below illustrates this process.

As you can tell by looking at the figures, while we have been caught up with the word intelligence, the creative process of a human being and the evolutionary process are a lot more alike than different. When we look at the two processes in terms of information we find that what really matters is a Memory where information can be stored. The memory is necessary to Generate new information. Information in memory is then brought under some sort of a selection pressure and is Evaluated. Information that ‘passes’ the evaluation is retained in memory and information that does not is discarded.

In effect, as far as information is concerned, it does not matter whether the memory involved is the human brain, DNA or even a computer. The process of information change, save for contextual specifics, remains the same for all systems. To put it another way, the precise mechanism by which new information is generated in the human brain and the DNA might differ, but both are engaged in the generation of new information. Similarly there may be differences in how information in each of these memory systems is evaluated, but nonetheless they are subject to an evaluation process at the end of which some information is retained and others are discarded.

This means that what Darwin articulated nearly a 150 years ago was a lot more than just the process by which new organisms come into being. What he discovered was much greater and far more profound. What he really discovered was the process of information change in nature.


Stuff to Read/Listen:

On Point Program on Intelligent Design

PBS cites about brain geography:

Sleep Deprivation & IQ

Chimpanzee tool use:

Animal Behavior Society

About the AI Deep Blue

BBC In Our Time Show on Artificial Intelligence

More Serious Stuff:

Felicia Gershberg & Arthur Shimamura, “Neuropsychology of Human Learning and Memory” in Neurobiology of Learning and Memory ed. Jose Martinez & Raymond Kesuer (San Diego: Academic Press, 1998), p.333-359;

Christine Sahley & Terry Crow, “Invertebrate Learning: Current Perspectives” in Neurobiology of Learning and Memory, ed. Joe Martinez & Raymond Kesner (San Diego: Academic Press, 1998), p.177-209.

Jane Goodall, The Chimpanzees of Gombe: Patterns of Behavior (Cambridge: Harvard University Press, 1986)

Hirata S, Celli ML “Role of mothers in the acquisition of tool-use behaviours by captive infant chimpanzees” Animal Cognition 6(4) (2003), p.235-44

Sanz C, Morgan D, Gulick S. “New insights into chimpanzees, tools, and termites from the Congo Basin” Animal Nature 64(5) (2004), p.567-81

Darold Treffert, Gregory Wallace “Islands of Genius” Scientific American 14(2004), p.14-23

Claus Hilgetag “Learning from Switched-off Brains” Scientific American 14(2004), p.8-9

Norman A. Krasnegor, G. Reid Lyon, and Patricia S. Goldman-Rakic Development of the prefrontal cortex : evolution, neurobiology, and behavior (Baltimore: P.H. Brookes Pub. Co., 1997.)

John Pearce Animal Learning & Cognition: An Introduction (Hove, East Sussex: Psychology Press, 1997)

Richard Dawkins, The Blind Watchmaker (New York: W.W Norton & Co., 1986)

Very interesting. Thanks!
I'd like to point out one difference between design by evolution and design by intelligence. In evolution, the design and the finished product are the same object, whereas an intelligent being uses abstract models (e.g., drawing boards, CAD programs, human imagination) to represent and refine the design before instantiating it into the finished product. An intelligent designer can consider and discard/refine many candidate designs without even getting up from his chair. Perhaps then intelligence should be defined, at least in part, as the ability to emulate the outside world in an internal model.
Mr, Brameld, that's an interesting analysis, but it's a bit hair-splitting. The distinction need not necessarily be made, and is really just a linguistic one anyway.

In the engineering company I work for, engineers frequently refer to any instance of the physical item as a "design." For example, "I shipped them his newest O-ring design to install in the rig to see if it solved that corrosion problem." Or, to use another context, "The charity ball organizer arrived in one of Christian Dior's designs." (Presumably she did not arrive wrapped in used drafting paper, though that would certainly be original. Heh.)
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?