Wednesday, November 12, 2008
A Tale to Pinto a Donkey
I suppose I should preface this with the statement that I really admired Stephen Jay Gould's take on science and his writing. Also, The Panda's Thumb is one of my favorite blogs. But the fact is that Gould's use of the panda's thumb as an argument in favor of science and against creationism was deeply flawed. This is not anything previously left unsaid, but Elliot Sober, in his new book, Evidence and Evolution (in case you haven't noticed, I really like it) has a good discussion of the difficulties with what he calls the no-designer-worth-his-salt argument.Panda's have what might be called a "thumb" but, in fact, it is not a digit:
Rather, they have a spur of bone that sticks out from their wrists. The thumb and the paw together form a V through which the panda repeatedly runs branches of bamboo, laboriously stripping the stalks to get them ready to eat. Pandas spend a large portion of their waking lives at this task. The thumb is extremely inefficient. Gould's point is that no designer worth his salt (the phrase is due to Raddick 2005) would have given the panda this device for preparing its food. A truly intelligent designer would have done better. On the other hand, Darwin's theory of evolution by natural selection says that inefficient devices of this kind are not at all surprising. Darwin thought of natural selection as a gradual process that improves adaptedness; natural selection does not necessarily lead to perfect adaptation, whatever that might mean. Selection modifies the traits found in ancestors by small changes; the result is not that the best of all conceivable adaptations evolves; rather, natural selection causes traits to evolve that do a better job than the alternatives that are actually present in the evolving lineage.
... [Gould] claims that the hypothesis of intelligent design makes the panda's thumb very improbable, whereas the hypothesis of evolution by natural selection makes the result much more probable. Creationists have a serious objection to Gould's argument. It can be expressed by a rhetorical question: How does Gould know what God (or some unspecified designer) would have wanted to achieve in building the panda? Gould is assuming that an intelligent designer would have wanted to supply pandas with a superefficient device (like a stainless-steel can opener) for preparing bamboo and would have had the ability to achieve this objective. But why is it so clear that God would have wanted to do this? Perhaps God realized that if pandas had better tools, they would eat all the bamboo, which would cause the extinction of the bamboo forest and of pandas as well. And maybe these two extinctions would have triggered a cascade of others. Perhaps God realized that these bad consequences would follow if pandas had better tools, and so he decided to slow them down. Creationists don't need to assert that they know what God would have had in mind if he had built the panda. All they need to say is that Gould does not know this. Gould adopts assumptions about the designer's goals and abilities that help him reach the conclusion he wants - that intelligent design is implausible and Darwinian evolution is plausible as an explanation of the panda's thumb. But it is no good simply inventing assumptions that help one defend one's pet theory. Rather, what is needed is independent evidence concerning what God (or some other intelligent designer) would have wanted to achieve if he had built the panda. And this is something that Gould does not have.
But does that mean the no-designer-worth-his-salt argument is useless? Not necessarily. Consider what Francisco Ayala makes of it:
The implications of the theory of intelligent design are not even compatible with the idea of a personal God, a creator, "which is all powerful and all knowing and benevolent, that is omnipotent, omniscient," said the Spanish-American Ayala.
"It would have to be imperfect design or incompetent design or inept design, but not intelligent design," he said, noting for example that human beings have more teeth than can fit in the jaw, so dentists have to remove the wisdom teeth.
In the end, the panda's thumb is a theological argument against a particular version of God. And the attempt to avoid the argument leaves that God, not as the infinitely competent creator of a magnificent universe and all that is in it but, instead, as a bumbler on a par with, as famously argued by the Discovery Institute's Casey Luskin, the designers of the Ford Pinto.
If that is the sort of God you want to worship, you might want to make sure your insurance is paid up.
.
Labels: Sober - Evidence and Evolution
Sunday, September 28, 2008
Purpose and Science
This is just one of my musings about philosophy that is, doubtless, more interesting to me than to others ... but it's my blog! Feel free to spare your glazed-overed eyes.There is a common argument for atheism that runs like this:
If purpose was part of the universe then there's no reason at all why science couldn't have detected it. It follows that, if there was purpose to the universe, then we would have scientific evidence of purpose. The facile nonsense of Intelligent Design Creationism notwithstanding, we don't have scientific evidence for purpose, therefore [it is reasonable to conclude that] there is no purpose to the universe [and, therefore, that there is no God].
Now, I'm an Apathetic Agnostic ("We don't know and we don't care") . Among the "tenets" of apathetic agnosticism (if we cared enough to actually hold to something as tiring as a tenet) is:
To all appearances, any purported Supreme Being is indifferent to our Universe and to its inhabitants. If there is a God, and that God does not appear to care, then there is no reason to concern ourselves with whether or not a Supreme Being exists.
The atheist argument above actually takes the form of what Sober calls a "probabilistic modus tollens." The standard modus tollens is a rule of deductive reasoning as follows:
If H, then O
not O
————————————
not H
... which reads: if the hypothesis H entails the observation O, and O turns out to be false, then it follows that H is false. The sticking point is that "entails." In order for the above syllogism to be valid, it must be certain that, if H is true, you will make the observation O.
But science does not work like that. Science is fallibilistic and incomplete. That is the basis of the common saying "the absence of evidence is not evidence of absence." Otherwise, it would follow, for example, that the universe was not expanding until Edwin Hubble made his observations of stellar redshifts, since, until then, there was an absence of evidence for expansion.
This aspect of science means that the very best that the atheistic argument can state is:
Pr(O \ H) is very high
not O
————————————————————
not H
... which reads: if the hypothesis H says that O will very probably be true, and the scientific observation, O, is not made, then H is false [or, at least, very improbable and, therefore, should be rejected]. Sober explains the problem:
Friends of [probabilistic modus tollens] need to say where the probability cutoff for rejection is located. How low must Pr(O H) be for O to justify rejecting H? Richard Dawkins [The Blind Watchmaker, p. 144-46] addresses this question in the context of discussing how theories of the origin of life should be evaluated. He says that an acceptable theory can say that the origin of life on Earth was somewhat improbable, but it cannot go too far. If there are n planets in the universe that are "suitable" locales for life to originate, then an acceptable theory of the origin of life on Earth must say that that event had a probability of at least 1/n. Theories that say that terrestrial life was less probable than this should be rejected. Creationists also have set cutoffs. For example, Henry Morris [King of Creation] says that theories that assign to an event a probability less than 1/10111 should be rejected, and William Dembski [The Design Revolution] says that a theory that assigns to a "specified event" (a technical term in Dembski's framework) a probability less than 1/10150 should be rejected. Morris and Dembski obtain these numbers by attempting to calculate how many times elementary particles could have changed state since the universe began.
Dawkins, Dembski, and Morris have all made the same mistake. It isn't that they have glommed on to the wrong cutoff. The problem is deeper: There is no such cutoff. Probabilistic modus tollens is an incorrect form of inference. Lots of perfectly reasonable hypotheses say that the observations are very improbable.
Pr(O \ H) is very high
not O
———————————————————————————
not O is evidence against H
... which reads: if the hypothesis H says that O will very probably be true, and the scientific observation, O, is not made, then that failure is evidence against H. But consider the counterexample Sober gives:
Suppose I send my valet to bring me one of my urns. I want to test the hypothesis (H) that the urn he returns with contains 0.2 percent white balls. I draw a ball from the urn and find that it is white. Is this evidence against H? It may not be. Suppose I have only two urns; one of them contains 0.2 percent white balls, while the other contains 0.01 percent white balls. In this instance, drawing a white ball is evidence in favor of H, not evidence against it.
[J]udgments about evidential meaning are essentially contrastive. To decide whether an observation is evidence against H, you need to know what the alternative hypotheses are; to test a hypothesis requires testing it against alternatives. In the story about the valet, observing a white ball is very improbable according to H, but in fact that outcome is evidence in favor of H, not evidence against it. This is because O is even more improbable according to the alternative hypothesis.
In fact, the atheist argument runs up against the very problem that the IDers face when they claim to demonstrate disembodied "design." Without knowledge of the motives and abilities of the purported "designer" or "purposer," there is no way to differentiate the action (or non-action) of such a being from our state of ignorance.
There is much more that can be said on the subject and Sober exhaustively exposes the philosophical weaknesses of ID, much of which applies as well to this argument for atheism. But it is only fair to ask why, if I think that the argument fails logically and scientifically, do I then accept the lack of apparent purpose in the universe as any basis for agnosticism? For one thing, the question that the lack of evidence bears on is (perhaps subtlety) different. Instead of considering whether there is, in fact, purpose behind the universe, the issue is whether whatever purpose-giver there may be is interested in humans knowing and reacting with that purpose. Given that any being capable of imparting purpose to an entire universe is presumably capable of making its existence and intentions known in unambiguous ways, a tentative conclusion that any such being is uninterested in human participation is, if not logically or scientifically rigorous, at least not unreasonable.
And that leads to my final point. Science is not an appropriate tool to determine matters that are, ultimately, nothing more than personal preferences as to how to approach life. Certainly, science's methods and results can be taken into account by anyone considering those preferences but, in the end, our weltanschauung is made up of a bundle of emotions, attitudes learned in childhood and unconscious value judgments with, at most, only partial imput by reason. Furthermore, science delivers information on what is, not what ought to be. Any assertion that we should determine our values and preferences by scientific rationalism is not, itself, a result of the scientific method, making the claim circular and self-defeating.
When it comes to values, we're on our own.
.
Labels: Sober - Evidence and Evolution
