anthropic arguments

Addendum on Anthropocism

No one would say that the cause existed because of the effect.

F.P. Ramsey

Both the advocates and the opponents of some sort of anthropic principle have twisted themselves into pretzels over a pseudo issue whose untangling is in fact pretty simple. There are three straightforward problems with any anthropic hypothesis:

(1) The notion of coincidence is not a scientifically respectable concept and cannot be used as a premise for drawing any scientifically valid conclusions. The musings of Weyl and Eddington were probably no more than that and hardly intended to spur any kind of theorizing. The only way to lend scientific validity to the concept of coincidence is to interpret it in terms of probability theory which is scientifically rigorous. But probability in the sense of mathematical probability does not apply to true facts and past events.

(2) Anthropocists who imagine some aspect of the initial state of the universe being other than it was always assume that all the other aspects remain unchanged. If one aspect were to have been different there is no reason some other aspect could also have been different in such a way that observers would have evolved.

(3) Anthropocists confuse necessary and sufficient conditions. Just because some initial state of the universe was a sufficient condition for observers to have evolved, it does not follow that it was a necessary condition. If that initial condition had not been the case, then observers might still have evolved because some other fact about the initial universe might also have been sufficient for observers to evolve (Think of the silica based life forms of Star Trek).

In The New Encyclopedia of Unbelief Victor Stenger (pp. 65-70) summarizes versions of the anthropic principle that do not involve propositional attitudes. Roughly: Certain physical conditions had to obtain in order for there to be life (alternatively observers). There is life. Therefore those physical conditions necessarily obtained. One way of formulating this starts with the Weak Anthropic Principle of Barrow and Tipler. Stenger more or less correctly says that the Weak Anthropic Principle is a tautology (Of course, by "tautology" he means "analytically true"): Since there is life the conditions necessary to produce life obtain(ed). However, in case the conditions cited by the Anthropic Principle did not obtain but different life producing conditions did obtain, the Principle is simply false. As stated by Barrow and Tipler, the Weak Anthropic Principle also asks us to believe that it makes sense to inquire after the probability of a past fact and to even assign a random variable in a situation where the sample space is arithmetically dense and so contains an infinite number of possible values. Setting aside probability, anthropic reasoning turns into the Tipler and Barrow Strong Anthropic Principle. But once we get rid of the confusion surrounding the invalid use of mathematical probability the simple or strong anthropic principle wears on its face a logical error that mathematicians designate by the following technical term – “howler” (Some say “schoolboy mistake”). It starts from a conditional assertion that is true but only contingently true, viz. if certain conditions obtain then life comes about. Formally (using my standard notation), If p then q. But anthropocists conclude that since there is life, the initial conditions necessarily obtained. That is q, therefore necessarily p. (“Necessarily” is a redundant addition to the truth functional statement. More on this below.) This conclusion is based on two mistakes. The first is to assume that “If p then q” is true, then its logical converse “If q then p” is true. It is not. If q is false and p is true, then the first material conditional is false while the second is true. Of course, by modus tollens, if “If p then q” is true, then its contrapositive “If ¬q then ¬p” is true (barring - valid - qualms about some substitutions for the propositional variables). But “If ¬q then ¬p” is an entirely different proposition from “If q then p” which is entailed by neither of the other two propositions. The second anthropocist mistake is to take a contingent implication as necessary. That accounts for the conclusion that it is necessarily true that certain initial physical conditions obtained. A more correct conclusion would be that certain physical conditions contingently obtained, i.e. just happened to obtain. And so, assuming the truth of the conditional, life followed. A minor ambiguity attaching to the second mistake is that anthropocists don’t specify whether they mean logical or physical (if there is such a thing) necessity. Of course, the logical errors remain no matter what kind of necessity they have in mind.

Let me reprise in slightly different language. We can state a causal relation in the form of an if-then proposition (material implication) in this way (setting aside temporal indices in the subordinate clauses which, for these purposes, are superfluous):

(a) If conditions C occurred in the universe then life appeared in the universe.

(Stenger, pp. 65-66, gives a few examples of physical conditions that qualify as C or constituents of C.)

In pure propositional logic (a) is equivalent by modus tollens to:

(b) If life did not appear in the universe, then conditions C did not occur in the universe.

A third version of the same conditional is:

(c) Only if life appeared in the universe did conditions C occur.

(a) and (c) are truth functionally equivalent. However, the conclusion that, if there is life in the universe, then conditions C had to occur is not the same as (c). It states something quite different, namely:

(d) If life occurred in the universe, then conditions C occurred in the universe.

(d) is not equivalent to (a), (b) or (c) as I explained above because if life did not occur in the universe but conditions C did occur, then (a), (b) and (c) could all be false and (d) true. On the other hand, if conditions C did not occur but life appeared, then (d) could be false and (a), (b) and (c) all true. The anthropic conclusion that, since life did occur in the universe, the conditions causing its appearance necessarily occurred confuses (d) with (a)-(c). As I stated above, C may be a sufficient but not necessary condition for the appearance of life. Other conditions may have led to the same result. In case C lists necessary and sufficient conditions for life, then (a)-(d) are all truth functionally equivalent. (a)-(c) simply state that C is a sufficient condition for life. (d) states that C is a necessary condition for life.  The anthropocist has to prove necessity if he is to validly conclude that certain initial conditions had to obtain if their purported consequences obtain. The empirical data only give us sufficiency.

A much larger issue is lurking in the background about the nature of scientific explanations. There is a zone of murkiness surrounding the scientific imperative to search for an explanation for observed phenomena (such as the size of the pure numbers which occur in the description of certain physical properties) and what a legitimate explanation would look like. Standard scientific practice is to look for explanations in conditions obtaining in the physical universe. Metaphysical, theological and indeed anthropic explanations typically look outside the physical universe. Theological explanations constitute the most clear cut case of non scientific explanations (which are ultimately non-explanatory for reasons Dawkins among others have offered). Other cases are not so clear. The much maligned élan vital might actually qualify as a physical explanation from some perspectives. Anthropocism further clouds the distinction by proposing what appears to be a physical explanation: The existence of life is a physical fact. Certain conditions had to obtain for life to exist. Therefore the existence of life explains why those conditions obtained. Suspicion arises from the fact that this explanation relies on subsequent conditions whereas standard scientific explanations typically look for prior conditions (which may also be concurrent) or from a theory constructed to explain other conditions. In principle there is nothing wrong with inferring a cause from its effect (This is what Fred Hoyle did). But, as we saw, error arises when you confuse sufficient with necessary conditions, causes which will result in the effect vs. causes without which there is no effect. This is the error of anthropocism.

Incidentally, Hoyle’s success in discovering initial conditions based on an “anthropic” assumption was not obviously an error. He was simply working on the legitimate methodological assumption that those initial conditions could have occurred, not that they necessarily obtained (Actually he didn’t even have to assume that they were sufficient; just that they were contributory.) And the fact that they had obtained was subsequently verified experimentally. If the connection in this case was necessary as is supposedly the case for anthropic principles, experimental verification would not be needed. The fact that the effect exists would be proof enough. And exactly what, may I ask, would constitute an experimental verification of an anthropic principle?

By way of another aside, your average goddist rarely rises to the level of anthropic reasoning. However, one could imagine what he would say. “Only by ascending beyond mere logical understanding does the sage reach true oneness with God.” Or else, “Jayzuz don’t do no logic! Dat is de work ov Satans!” I daresay. If Jayzuz don’t do no logic, why do you try to come up with these proofs in the first place?