Monday, August 01, 2005

The God of Reason   posted by Razib @ 8/01/2005 02:26:00 PM
Share/Bookmark

The Robot's Rebellion: Finding Meaning in the Age of Darwin is an ambitious book that fails in its aim ("finding meaning"), but of course that's not a particular concern, I think if 3,000 years of philosophy and philosophical religion tell us anything it is that the journey is the true destination. Written by a Keith E. Stanovoich, a cognitive psychologist, in less than 300 pages (though with detailed notes!) the book shoots from basic cognitive science all the way to philosphical ethics, with long jaunts to philosophy of the mind in between. Readers of this weblog will guess that I found the cognitive science more interesting than the other more philosophical fields (I prefer fiction to most of philosophy when I read it for philosophy's sake as opposed to getting a bearing on the intellectual environment of a historical period).

If you have read introductory cognitive science texts or any of Daniel Dennett's books there won't be any surprises, the book hammers home the point that TASS (The Autonomous Set of Systems) is as ubiquitous in our life as the analytic mind (or more so). He asserts that the conscious mind is in many ways an illusion, that we live in a "Cartesian Theater" of self-delusion as to the importance of the central processing unit in mental function. "Self" is just an emergent propery of subsystems and their synergy.

One of the great things about cognitive science is that you can illustrate the experiments, and have your readers test them, in book form. Of course, you aren't getting many repetitions (no statistical analysis), but you get a feel for what the author is getting at by doing the experiments which led to the concepts being exposited. About three years ago I reviewed the book The Number Sense, which had a neat little experiment, and I'll reproduce it here.

Answer the following questions as fast as you can:

-2+2
-4+4
-8+8
-16+16
Now quick! Pick a number between 12 and 5

Click here to find out what you should have picked.

If you didn't choose the number that pops up, you are in a minority. My point above is that people often think of TASS systems as motor reflexes, innate and hardwired, like pulling your hand away from the fire, but they are more than that. Arithmetic is learned, but by the time they are adults for many people (but not all) it is almost reflexive a skill.

Fodor's definition of a modular process (roughtly analogous to TASS) list the following traits:
  1. fast
  2. mandatory
  3. domain specific
  4. informationally encapsulated
  5. cognitive impenetrable
  6. subserved by specific neural architecture
  7. subject to idiosyncratic pathological breakdown
  8. ontogenetically deterministic
Obviously this list is too restrictive for something like basic mathematical skills, as there is a great deal of evidence that the last five criteria do not apply to mathematical functioning, and the extent of the first three are subject to specifics (what type of math, how "fast" or "mandatory"). But the point is that "higher cognitive functions" can become part of TASS over time, and the mandatory part really undermines our conceit that we are by and large rational and conscious beings, always in the cockpit controlling everything manually. This insight is "obvious" if you are talking about motor skills (shooting a basketball), but we tend to not acknowledge its reality in functions which demand some g (figuring out the best route to integrate a nasty function).


Though many elements of TASS are learned, some are innate and genetically programmed. Capacity for language seems to be one of these, as is facial recognition. This is important, because these "competencies" can disappear if there is a genetic or developmental problem, and it highlights how changes in TASS can diminish or distort "higher" aspects of our intellect. For me, the most interesting example the author gave to illustrate the importance of TASS is Capgras Syndrome, where individuals believe that those around them are "imposters." The explanation is simple: the facial recognition system still works, so you recognize people, but the "wiring" which connects this system to that which results in emotional response cued to familiar faces is broken. So, you see your father, you recognize your father, but you feel no concomitant response of filial affection. A problem emerges when the analytic system of our brain "explains" this disjointed perception: everyone must be an imposter, that is the only solution to the cognitive disconnect. They look like people you know, but they can't be the people you know. This causes bizarre paranioas, sometimes resulting in the death of the "imposters" at the hands of those suffering from this illness.

This is "confabulation" writ large, the creation of a story by the analytic mind to explain choices which are rooted in TASS. When people guess or make random choices usually there are plenty of inputs which are shifting them toward a particular selection (that's why advertising works, people don't want to acknowledge that choices are contingent upon outside inputs rather than spur of the moment). Conversely, many choices which we think are made freely and rationally are also dependent on inputs, if not TASS, at least environmental and contexual variables. For example, Stanovoich reports that over half of individuals surveyed in the USA assert they would adhere to the religion they profess no matter what the environment or social inputs were (that is, they were born to a different family which practiced a different religion). Bouchard's twin studies indicate that this isn't so, people who are genetically the same tend to track for "zeal" (half of the variation seems genetically heritable) but not specific denomination.

What does any of this have to do with evolution and Darwin? Well, it's all about TASS. Some evolutionary psychologists take TASS and to some extent almost argue away any other aspects of our mind as having any relevance, even domain-general intelligence (the analytic mind). Massive modularists are especially guilty of this, in particular when they push the idea of "psychic unity" of mankind via the monomorphism of salient psychological traits (human universals which show no heritable variation). Stanovoich's contention is that in many ways the human mind is subdivided in two, between the various aspects of TASS, and the analytic mind, which is serial, slow, kludgy and a later evolutionary development. Unlike TASS the analytic mind is a general purpose tool which is under conscious control and can be brought to bear on any problem (in theory). He goes on to argue that while TASS serves our genetic interests, the analytic mind can serve the interests of the vehicle (though it originally evolved to give more flexibility to the phenotype to advance the replication of the genotype). The author also believes that evolutionary psychologists tend to conflate the interests of vehicle (the "robot") and the genes which it carries because of their fixation on the natural rationality of TASS. Related to this, over the last generation evolutionary psychologists have attacked the heuristics and biases research of cognitive psychologists that suggest that humans are not pure "rational actors" and have a faulty conception of probability. Evolutionary psychologists who reject the heuristics and biases research offer that the mind is adapted to particular tasks, that its skillset is bounded and biased toward evolutionarily relevant concerns, and that one shouldn't presume that it would be a standard utility maximizing device (ie; that transitivity (def. 2) would hold, etc.). From this they formulate policies and tools to mitigate the confusions of the human mind in the modern world, which to some extent does assume a utility maximizing consciousness. But many evolutionary psychologists move further and rhetorically assert that the choices that most people make on various tests that suggest irrationality are the correct answers because they are the evolutionarily adapted answers. This mindset is also common once the ideas of evolutionary psychology make it to the general public, the naturalistic fallacy seems like a default cognitive template that is easily triggered. But there is a problem with this. First, modern society is not an Enviornment of Evolutionary Adaptedness (if there was one), so for example our craving for sweets is now a serious maladaptation. Second, there is a strong correlation between those with higher general intelligence behaving as if they were instrumental utility maximizers. In other words, those with a high I.Q. are much more likely to allow their analytic mind to overrule their "gut instinct." This shouldn't be surprising, as it seems that the extreme right end of the bell curve is also the least likely to submit to TASS instincts which result in fecundity (or at least, the analytic mind always manages to take appropriate precautions). Additionally, the book argues that our analytic mind characterizes our human uniqueness more than TASS (some TASS systems like language are pretty unique too of course). There is no reason to privilege TASS, and no reason to privilege the genes which it serves. Like science fiction accounts of humans endowing their android servants with sentience and regretting the decision later when the robots rebel against their masters, so it is that Stanovoich argues that the genes have lost control of their servants.

The last chapter of the book makes the case for a post-TASS analytic system of ethics. I'm not going to repeat in detail what Stanovoich argues, because it is familiar to many, basically a system based on reason and a humane utilitarianism. He comes close to railing against capitalism as being driven by first order rationality that appeals only to TASS hedonic interests (I am not totally deaf to this complaint, but the author offers no productive alternative). He argues for the socially embedded nature of utilitarianism (ie; hooking yourself up to a pleasure machine for your whole life is not cool for various philosphical reasons, most famously articulated by the late Robert Nozick). The process of reason looms large, and the rational analytic mind takes center stage.

But I think there are problems with this thesis. First, there is nothing new in rising above our baser natures. Since the emergence of Buddhism axiomatic philosophical-religions have periodically made waves. The Buddha offered the Four Noble Truths and the Eightfold Path in a non-theistic system of self-cultivation which allowed one to transcend our baser instincts. Christ was the Prince of Peace, and the early Christian communities were filled with idealists, culminating in the reunciates who fled to the deserts of Egypt. The original Ummah was predicated on justice for all. Marxism was about allowing government to wither as one reaches the final anarchic utopia. Here the issue: systems concocted by those who privilege their analytic minds tend to morph once they become popularized. 1,500 years after Buddha Mahayana Buddhism had imported gods back into the equation via Boddhisattvas. 1,100 years after the Prince of Peace Jersalem was bathed in blood in his name (or his father, "God Wills It!"). Somehow, Muhammed's particular message morphed into a worldwide jihad to bring the "Peace of Islam" to every corner of the earth (with fits). The anarchist utopia somehow led to totalitarian Marxist-Leninism.

The reformation of the mind that Stanovoich proposes is contingent upon faith in the analytic mind. Those who are cognitively nimble, who have had success in school solving equations or writing essays, they will trust that mind. Those who have failed at these tasks, or gotten by on rote memory and "tricks," will not. The majority of the human race is in the latter category. I do believe that with the coming biological engineering revolution and cybernetic extensions, this could change, but Stanovoich doesn't go into this issue (though he alludes to it, and not always positively from what I can gather).

Another point he makes is that he assumes that the cognitive revolution has debunked traditional religion. He agrees with Daniel Dennett that the "universal acid" of Darwinism will eat away all the ancient pre-scientific paradigms. I am skeptical because I do not think that religionists are anywhere close to abandoning their "code base" because of the unwieldiness of the extensions that are being added by modernity. In fact, I don't think that many people actually have within them many high level densely coded applications that need lots of spare processing power and RAM which novel extensions can be added on to. Most people just run the basic house-keeping applications which are a 'good enough' solution for the replicators. The scientific revolutions alluded to above might change that, but I suspect that the first result won't be a streamlined rationalization or a homogenization of worldviews converging upon the analytic mind, rather, I suspect there will be a profusion of mad and somewhat insane mentalities which divergent values and sense of purpose as humans boot-strap their minds into wildly different directions. Let a thousand flowers bloom....

Related: I think some readers might question the use of the term "robot" to define humans. I have a tendency to visualize people as nodes with a landscape of gene flow, for example. But note that Jennifer Aniston recently asserted that Brad Pitt has a sensitivity chip missing. I think on a very basic level humans do perceive themselves as a sort of chimera, though verbally and in our idealizations we tend to project the reality of a unitary self.