view entire entry:
2001_05_13:01: Falsifiability Take Two
to see pictures, view
I posted again to the psychology list. There seems to me to be a bit of guilt by association and over-generalization going on: some people who use the falsifiability criterion to evaluate theories also subscribe to the analytic/synthetic dichotomy or some of its manifestations, and therefore, the falsifiability criterion is just a manifestation of this fetid dichotomy, and anyone who thinks falsifiability is a useful concept needs to Go ReRead Rand, or Peikoff, on the analytic/synthetic dichotomy. There will be excommunications at noon.
One particularly distressing aspect of this is that, despite my preceeding post and Tom Radcliffe's regarding faith, people have latched onto the idea of imaginability and won't be shaken loose. I guess it is just unfalsifiable, for them, that falsifiability and random imaginative excursions are inextricably bound together.
So on I forged:
Abstract: I attempt to thwart this unfortunate detour into imagination and the analytic/synthetic dichotomy, and answer Adam Reed and Andrew Schwartz on another round of objections to this useful criterion.
Adam Reed wrote:
> I would like to thank James Lett for his clarification on > falsifiability. His response pretty much confirms my initial > impression: the principle of falsifiability is derived, like all of > positivism, from the analytic-synthetic dichotomy, and specifically > depends on one's subjective ability to conceive of contingent > observations outside the context of reality.
After my initial post on this subject, I immediately pasted the text of that post in my online journal, and prefaced it with this thought:
"One thing I didn't say in my post to the list but thought of afterwards, is that the aversion to this criterion might be due to a common bad practice in the use of thought experiments. I talk about this bad practice in my dissertation. The problem is that some people try to use fantasy thought experiments to discount empirically-supported theories: if one can conceive, without self-contradiction, disconfirming evidence, then the theory might be false. This same language, I gather, is now being used to show that a theory is good because it is falsifiable: if I can't conceive of disconfirming evidence, then the theory holds up under the
falsifiability criterion. It's very twisted language and hard for
laypeople to follow, and hard for objectivists to swallow..."
The reason I'm quoting, Randlike, from my own writing, is that I had to laugh when Lett's post, to which Reed refers above, landed in my box shortly after I wrote the above. I got another note privately, in which falsifiability was condemned on the very grounds that Lett was using to defend it.
Well, at least we are all predicting the same thing, but I for a very
different reason than others, apparently.
We should keep firmly in mind that, while there are people who accept the analytic/synthetic dichotomy, logical possibility, "conceivability," definition by biconditional (instead of by genus and differentia--I talk about all these things in my tirade against science fiction thought experiments in identity theory), etc, and will of course put everything in these terms, there are people like me. I don't fit in that box. Please see Chapter Two:
"The Things That Are Not: Against Science Fiction Counterexamples" of my dissertation, for the discussion of these inter-related issues; definition by biconditional, and necessary and sufficient conditions as criteria of identity, are discussed throughout. It's on the Enlightenment web site.
There are people who would defend the virtue of honesty as being what God wants for us. There are people who condemn murder because it is selfish. Merely using these kinds of terms does not mean that the virtue of honesty is invalid because it is rooted in religion, or that classical liberal law is invalid because it is rooted in altruism. I think this is similar to Reed's earlier contention that behaviorism is no longer the current trend in psychology, and that blasting the entire discipline on the basis of behavioristic premises is beating a dead horse. I feel I'm in the same position here; while I admit that there is a lot of conceivability/logical possibility stuff still going on in philosophy, I really must insist that the mere fact that someone uses logical possibility to defend falsifiability doesn't mean that there is no other defense of it.
Despite the way Lett has defended the application of the falsifiability criterion, I defended it quite differently in my post. My defense stands unaddressed.
As I pointed out in that post, the point is not to sit around trying to dream up nutty counterexamples. The point is your attitude toward reality: if you were to be presented with disconfirming evidence, would you consider it, or would you ask not to be bothered with facts because you had made up your mind? A coherent mind is one that is always open to evidence, always open to facts; an incoherent one holds onto a theory for dear life, because it doesn't really understand causality, doesn't understand how the theory was arrived at in the first place, and is afraid that the article of faith will be yanked out from under it leaving it with nothing to sit upon.
That's my theory, I know it to be certain and absolute, and I refuse to consider any counterarguments or disconfirming evidence. :-|
Andrew Schwartz followed up with this question:
> Here is my question: instead of requiring a claim to be
> "falsifiable," why not simply require it to be evidential? For
> purposes of acquiring knowledge, in what way is the concept
> "falsifiable" superior to the concept "evidential"?
Karl Popper coined the term, I believe. The term he chose was 'falsifiable,' and the reason that he chose that term was the direction from which he was coming: he was attempting to provide a criterion for distinguishing a scientific theory from an unscientific one, and he thought that one distinguishing mark of an unscientific idea was that its proponent would be unwilling to even _state_ (or acknowledge a statement of) conditions under which she might reject the theory. The concept that the word represents is the one I've been trying to explain: that when there is valid counter evidence, it is not evaded, ignored, suppressed, or worked-around--it is taken into account, and if the theory needs to be adjusted, then so be it. If this is what one means by 'evidential', then these two _words_ refer, for that person, to the same _concept_ and the same methods. But I use them to refer to slightly different things. It's not that one concept is superior to the other; it's that each is useful in its context.
One may say that the word 'falsifiable' is unfortunate because it is so easily misunderstood by lay-people, and because it is now being used by people who are committed to logical possibilism and the analytic/synthetic dichotomy. But then again, one may say similar things about the word 'selfishness'.
view entire entry:
to see pictures, view
in thread: whatever2001_05_09:00:22
The Falsifiability CriterionI responded today to some
discourse on the Psychology list
regarding the falsifiability criterion for judging theories. I think it might be common amongst objectivists to worry about this criterion. The worry goes, "I have good evidence that this theory is true--lots of it. So it can't be false. It's silly to say that my theory is only a good one if it could be false. So of course it doesn't fulfill that criterion; in fact, the criterion is suspect." I don't know how widespread this interpretation of the falsifiability criterion is. Since it is relevant to my work on propositions, I plunged in. One thing I didn't say in my post to the list but thought of afterwards, is that the aversion to this criterion might be due to a common bad practice in the use of thought experiments. I talk about this bad practice in my
dissertation. The problem is that some people try to use fantasy thought experiments to discount empirically-supported theories: if one can conceive, without self-contradiction, disconfirming evidence, then the theory might be false. This same language, I gather, is now being used to show that a theory is good because it is falsifiable: if I can't conceive of disconfirming evidence, then the theory holds up under the falsifiability criterion. It's very twisted language and hard for laypeople to follow, and hard for objectivists to swallow, so I worked through the tangles like this:
My understanding of the falsifiability criterion is this. Consider some theory you've just proposed. Could I say _anything_ that would make you question the theory, in principle? People who subscribe to a genuinely unfalsifiable idea will always answer "no". People who subscribe to a genuinely falsifiable idea will always say "yes." That doesn't mean that they think their theory is _false_; if they did, it wouldn't be their theory.
The most widely-used example of such an idea is God's infinite benevolence and love for humanity. The good job with a fat salary that you just acquired is attributable to omnibenevolence (OB) in the this theist's view. But what about the job you lost 4 months prior and the financial hardship that ensued? Oh, well, that was God's setup to get you this greater job. OK, how about the fact that my child was born with spina bifida just after the layoff? You see, that was your incentive to strive as hard as you did to get x, y, and z benefits. What about the millions of children who die slow, agonizing deaths from disease and starvation? They provide a good contrast object for you--if there weren't pain and suffering, you wouldn't know how happy you are. But what about them? How does OB benefit _them_? I'm tempted to say that I'm the protege of a Really Benevolent God, while they are the victims of a Pretty Powerful Sadist. Oh, no, you see, there are things that God does that we don't understand. This is one of them. God _is_ OB; you just don't understand all that that entails.
So, I don't see it as a question of being able to _imagine_ evidence that would falsify the theory; in the above examples, we actually _have_ those facts at hand. Falsifiability comes into play when the defender of the theory would be unwilling to give up the theory no matter _what_ evidence, real or imagined, is suggested, and is willing to change even the meanings of the terms in order to make the evidence fit the theory. In other words, the theory fluidly changes whenever disconfirming evidence is suggested, all the while continuing to call itself the same theory. Marxism is similarly accused of being unfalsifiable. One red warning flag is the claim that a theory explains
_everything_, including two contradictory pieces of evidence, which OB
does. The warning flag is reason to take a closer look at the theorist's
methods and claims.
Evolutionary theory is falsifiable in this sense: Suppose we got evidence that those "bones" that we have in museums were just very interesting objects d'art, and we've all the while thought that they were from real animals? If evolutionary theory is falsifiable, then we'd have to say, yes, that would count as evidence that the theory wasn't true; and if we got enough evidence like that, we'd say the theory wasn't true. If someone's "evolutionary" theory is unfalsifiable, then he would say, "No, that evidence MUST be mistaken, because my theory is correct; I don't care how much of it you get, I won't admit it as disconfirming the theory." This is what it means to say that disconfirming evidence cannot be conceived. Conceived by whom, is the question.
Like any proposition, a theory exists only in virtue of the mind that proposes it. In part, we have to judge the way the theory is held, not just the plain linguistic statement of it. I personally proposed that evolutionary theory is true and falsifiable: if you can give me evidence that aliens visited the planet and left the bones of their deceased pets here, then I'll happily rethink my theory of evolution. Or if you can show that they are really ancient pottery shards, or that they aren't really that that old. Any of these new facts, were they facts, I would take as disconfirming.
But if I believe that God's love is simply beyond my understanding, I don't care what kind of evidence, real or imagined, you present; if I can't tell you how it fits the picture of an OB God, then I will say, "I don't know how this demonstrates OB, but it does, and if you were omniscient like God you'd see how." A person with a falsifiable theory does not say this. Note that this is different from what Tom Radcliffe said in his last post: Our theist is not saying, "I don't know why my theory of OB doesn't account for that data;" he's saying, "My theory DOES account for that, I just don't know how; or else it isn't real data."
In conclusion, I don't think theories can be judged without reference to the minds that hold them. I can't just say "The theory of evolution is that human beings evolved from apes." I also have to say, "And I WOULD consider disconfirming evidence," or "I WOULD NOT consider disconfirming evidence."