Browse Search Feedback Other Links Home Home
The Talk.Origins Archive

Can Deduction overturn Induction (repeated observation) in Science to crush Darwinism?

Post of the Month: May 2010


Subject:    | Popper, "prohibited Observations", and Pagano-like Poppycock
Date:       | 28 May 2010
Message-ID: |

There is no doubt that Pagano is a revolutionary genius destined to overturn secular scientific orthodoxy and revitalize the Catholic Church, bringing the wayward cafeteria Catholics back to the fold, and reversing the terrible damage done to the human soul by the so-called "Enlightenment." No doubt. No doubt whatsoever. In the face of his unequalled genius and inevitable victory it is only with the greatest of trepidation that I raise a few questions about his method of deductive falsification, derived, he says, from Popper.

Tony argues that no amount of confirmatory evidence for a theory tells you anything about its......(wait for it)....truthlikeness, but that a single "prohibited observation" suffices to falsify the theory, with all the certainty of rigorous deductive logic. I would prefer that he spoke in terms of predictions, rather than prohibited observations, but he has astutely noticed that clarity of language is not his friend and thus shuns it. Here is his argument in brief.

Consider the chain of reasoning:

  1. Theory A predicts observation X
  2. Observation X is observed
  3. Therefore theory A is true.
Pagano correctly points out that this reasoning will not hold up, since any number of other theories might also predict observation X. And simply accumulating a whole lot of observations correctly predicted by Theory A does nothing for you, since repeating a bad syllogism 10,000 times doesn't make it a good syllogism.

On the other hand, consider the reasoning involved in falsification.

  1. Theory A predicts that X will not be observed
  2. X is observed
  3. Theory A is false.
That reasoning is absolutely airtight. So it seems that while confirmation of a theory is inductive and therefore unreliable and uninformative about the truthlikeness of a theory, falsification is rigorously deductive and reliable.

Well, that's Tony's argument, anyway. With due regard to his undoubted genius, it is the sort of argument that could only be made by someone who'd never done any actual science. In mathematics and abstract logic deductions are ironclad because one can simply make the premises true by definition. In any question regarding the physical world it is rather more difficult to have certain premises. For example, in any specific experimental situation your calculation that observation X is "prohibited" by the theory may be wrong. You may have overlooked sources of background effects that you should have included in figuring out what observations to expect (or not to expect). When you observe X, the allegedly prohibited observation X might actually be X', an observation compatible with Theory A. In trying to be certain of your premises you are stuck using induction; so the conclusion of falsification, which Tony thought was an ironclad, rigorous deduction, rests on premises supported by poor old Humean induction. So falsification is on no better a footing than confirmation.

In fact though, we rarely consider theories in isolation and attempt to falsify them. We always consider competing theories and try to find situations in which they make different predictions. Then we do the experiment and see which theory's predictions are observed. This is true even when it looks like we are simply trying to falsify a single theory. Let's say we find the "prohibited observation" X. We then consider several competing theories

  1. Theory A is false
  2. the observation X was an error
  3. our prediction that X should not be observed if Theory A holds was wrong.
Then we try to find experiments we can do for which each of those possibilities makes a different prediction of the result. And so on. [As a side point, without induction one cannot conclude that observation X was anything but a fluke with no bearing on the falsity of Theory A whatsoever - see, Tony, you cannot get anywhere if you are unwilling to use induction].

Tony thrives on obscurely worded generalities, but it will be easier to understand falsification if we use a specific example. So one case in which it's pretty close to true that physicists falsified a theory in the absence of a better alternative is the case of the classical predictions for the spectrum of electromagnetic radiation from a black body (that's a body that absorbs all EM radiation that shines on it, regardless of frequency). The classical prediction, summarized IIRC in Rayleigh's equation was that the intensity of radiation from a black body at any given wavelength would increase monotonically for shorter and shorter wavelengths. That was wrong. It would have predicted, for example, that if you peeped into the right sort of hot furnace you'd have been instantly killed by high energy gamma radiation. In fact, there was a peak in the intensity vs frequency graph, not a monotonic increase. Ultimately, this helped get quantum mechanics started, but at first it just falsified classical electromagnetics. But even here, the physicists considered alternatives of the sort I mentioned above. Many of them went over and over the derivation of Rayleigh's equation to make sure that it's prediction of a monotonic increase was in fact what classical EM theory predicted. Others looked for errors in the observations - a perfect black body is an idealized thing, and it might have been that the experimental apparatuses differed from an ideal black body in ways that could account for the prohibited observation. It was only when there was a good alternate theory for black body radiation, based on quantum mechanics, that it became clear, in retrospect, that the classical EM theory had been falsified. And quantum mechanics still had to be shown to be compatible with all of the successful predictions made by classical EM theory in other settings. That's where confirmatory evidence has weight. Classical EM explained an awful lot of disparate observations in many different settings - you could not easily dump it without an alternative that explained all of those observations AND ALSO explained black body radiation.

You could summarize it as

  1. Classical EM predicts that there is no peak in the plot of intensity versus frequency in black body radiation
  2. a clear peak in the plot was observed
  3. Classical EM theory is false.

You *could* summarize it that way, and Tony would like to do so, but such a summary would have left out all the competing theories that were actually in play while this was being worked on.

So now, let's put Tony's ID based falsification of the theory of evolution in the same format. The argument would be

  1. NeoDarwinism predicts that the only biological structures that can exist are those which can have evolved as a result of mutation and selection
  2. there is at least one structure, for example the biological flagellum, which cannot possibly have evolved as a result of mutation and natural selection
  3. therefore NeoDarwinism is false.

It's hard to disagree with step (1). But step (2) is the crux of the matter. If this falsification is to work deductively and be ironclad, then we must know with all the certainty of a geometric definition that the evolution of the bacterial flagellum is an impossibility. The burden is on the falsifier to show where the impossible step in any proposed pathway lies. It's actually a really rough road to try to take towards falsifying the ToE. It would be far easier to try to show that morphological and DNA sequencing based phylogenies were contradictory, or that human and trilobite fossils were found in the same strata. Those are potentially objective reproducible observations that would falsify the ToE. Much stronger than "I just don't see how it could have evolved."

So, in brief, contra Tony, confirmatory evidence does matter. The more correct predictions a theory makes in a range of circumstances, the truer the theory is. Falsifying evidence matters, too, of course, but falsification depends on induction and on the consideration of competing hypotheses just as surely as does confirmation. There's nothing deductively airtight about falsification.

In the end it is impressive that in spite of all the philosophical name dropping and the delightfully obscure philosophical posing, Tony's argument just comes down to "I don't see how anything so complicated could have evolved."

Home Page | Browse | Search | Feedback | Links
The FAQ | Must-Read Files | Index | Creationism | Evolution | Age of the Earth | Flood Geology | Catastrophism | Debates