Wednesday, September 06, 2006

Forget About Falsification

Dear Reader, I am going to be painfully abstruse here just for once. Please bear with me.

Science works well as a way of finding out about the world, and philosophers have thought long and hard about how it works. As a research scholar in a discipline where there is a lot of controversy over this issue, I have also had reason to read and think quite a bit about it.

One of the most widely read philosophers of science is Karl Popper. He's associated with an attitude called falsificationism. It means two things.
  1. A hypothesis about the world is only scientific if it could in principle be proven wrong ("falsified") if untrue. This could also be phrased, "A lot of things are unknowable. Don't waste time and ink on speculations that can't be tested".
  2. The way to do good science is to formulate hypotheses and then try to falsify them. If a hypothesis survives repeated attempts at falsification with different data and angles of approach, then it may be taken to be true.
The other day I read a new paper by Sven Ove Hansson where he argues that the second sense of Popper's falsificationism is incorrect.

It's an unusual philosophical paper in that it has an empirical base. Hansson has looked at how the 70 articles in the journal Nature for the year 2000 are built from the perspective of the philosophy of science. Good science is after all produced continually, and, argues Hansson, to get into Nature studies must be very good indeed. So if Popper was right about how good science is done, then most of the studies in Nature should be structured around hypotheses and attempts to falsify them. As it turns out, they're not. Quite the contrary.

70% of the studies don't start out from any hypothesis at all. Hansson calls them explorative: they begin with simple questions such as "What is the molecular structure of this protein?" or "What is the base-pair sequence of that gene?". Formulating hypotheses would just be a waste of time here. What the scientists did was to make tricky and time-consuming observations and then report what they had seen and inferred from it.

Only 24% of the studies start from a Popperian favoured hypothesis that the scientists involved could try to falsify if they wanted to. But half of these hypotheses are framed in such a way that they would give equally conclusive knowledge regardless of whether they are confirmed or falsified. Only two studies (3% of the total 70) start from hypotheses that would give more conclusive knowledge if falsified than if confirmed.

So, Dear Reader, if you're doing research, never mind trying to falsify your own hypotheses: the people who publish in Nature don't. The thing to do is apparently to either a) go exploring, find out some useful/cool data, report it and suggest a well-argued interpretation of it, or b) make up a hypothesis and collect the best experimental evidence and arguments you can to support it.

But Hansson's paper leaves the first sense of falsification mentioned above untouched. A lot of things are in fact unknowable. Don't waste time and ink on speculations that can't be tested. And that pretty much kills off a lot of archaeological interpretation, particularly when it comes to attempts at reconstructing what prehistoric people thought and believed. Such a cautious, anti-speculative stance, by the way, is called positivism. It's extremely distasteful to some scholars, particularly the ones who profess to seek understanding rather than truth. But I'd like to see much more of it in archaeology.


Hansson, S.O. 2006 (antedated to 2004). Ealsificationism falsified. Foundations of Science 152. Kluwer.
[More blog entries about , , ; , , .]

14 Comments:

Blogger Tor said...

Good post! Let me just add that Falsification 1, as formulated by Popper, is pretty hopeless as well. Take the sentence, "There are planets around other stars than Sol". No reasonable person would condemn this hypothesis as "unscientific". Yet Popper's criterion would seem to rule it out. For how could it ever be proven wrong? By checking, in turn, every single star in the universe for planet-ownership and finally emerging empty-handed? Even if _in fact_ you managed to pull off such a feat, how could you ever be a hundred per cent _sure_ that you hadn't missed something---some star coyly stowing away with its planets in some yet-undiscovered part of the Universe?

Proving the planets-around-other-stars hypothesis to be _true_ is of course a different matter---as far as I understand, it has already been done to astronomers' satisfaction.

In order to attain any plausibility, Popper's criterion would have to be weakened at least to something like: "A hypothesis about the world is only scientific if it is possible in principle to obtain reasonably strong evidence for or against it"---but I wouldn't be surprised to learn that even this criterion is too strong. Indeed, I find it prima facie unlikely that any simply formulated criterion could ever succeed in demarcating the scientific from the unscientific---which is not to deny that there are obvious examples of each!

Popper seems, for some reason, to be held in high regard among natural scientists regardless of the extent to which their practice accords with his theory. Philosophers of science mostly write him off.

07 September, 2006 10:57  
Blogger Martin said...

Plz tell me your views on, shall we say, uninhibited interpretations in the historical humanities.

07 September, 2006 11:14  
Blogger Tor said...

I suspect it would be difficult to find a piece of reasoning that I would be prepared to call both "uninhibited" and "scientific". Hard to say in advance, though. What kind of inhibitions did you have in mind?

07 September, 2006 11:35  
Blogger Martin said...

In my opinion, Scandinavian archaeology in the past few decades has cultivated a climate where interpretations about the past may be published though their support in source material is flimsy or non-existent. In the annual Current Swedish Archaeology, for instance, you'll find many papers that have clearly made it into publication not thanks to the empirical strength behind the suggested interpretations, but due to the prestigious theoretical models employed.

And then I haven't even mentioned the amateur Continental philosophy produced by archaeologists that makes no attempt at all to speak concretely about the past.

07 September, 2006 12:03  
Blogger Tor said...

Sounds like rubbish to me. Is it science? Possibly; judging from your description, bad science.

The word 'scientific' is ambiguous. Sometimes it's used as a term of praise, sometimes it serves, in a non-evaluative way, to delineate a certain field or method of inquiry---it isn't always clear what the writer has in mind.

Personally, I don't find much use for the term. For evaluation, I use 'good', 'reasonable', 'idiotic', etc.; for specification of fields and methods of inquiry, 'astronomy', 'philosophy', 'shirt-sniffing', etc.

07 September, 2006 12:31  
Anonymous Victor said...

Tor, your argument on the falsification of the hypothesis "There are planets around other stars than Sol". falls on the single premiss that "you don't know till you know." I suppose you have misinterpeted what falsification really is about. Falsification is not wheter it's really POSSIBLE to falsificate. But wheter if it's theoretically possible. Theoretically we CAN observe every star in the universe. If we during these observations don't find any star having planets orbiting round em, well then I guess this does not exists... til we'll find one.

Well luckily we don't have to

It doesn't matter really what we can't percieve or observe, we don't know till we know. How can't you be 100% sure that there's some invicible little gnome living in your TV that casts spells of bad luck on you everyday? when you can't percieve that in any way.
There's a very big different what's theoretically percievable and what isn't.

07 September, 2006 14:28  
Anonymous Victor said...

Well luckily we don't have to falsify this hypotesis very long because just as we observe one planet orbiting around another star the hypothesis is defalsified. What matter is that it CAN be falsified.

07 September, 2006 14:30  
Blogger Tor said...

Victor, I'll agree for the sake of the argument that it is "theoretically" possible to check all the stars in the universe (though I guess it depends on how you spell out "theoretically"). The problem is that in order to _prove_ that no planet except Sol has planets, it is not enough to _actually_ check all the stars---you also have to _prove_ that the stars you have checked are in fact all the stars there are. And I don't see how that could be done, even "theoretically".

Of course, the fact that I don't see how it could be done doesn't mean that it can't be done. Maybe there is some way which I haven't thought of; if you have a suggestion, I'd be interested in seeing it.

07 September, 2006 15:19  
Anonymous Victor said...

TOR wrote:
"The problem is that in order to _prove_ that no planet except Sol has planets, it is not enough to _actually_ check all the stars---you also have to _prove_ that the stars you have checked are in fact all the stars there are. And I don't see how that could be done, even "theoretically"."

Well it's not a problem at all, if you run out of stars to check, I'll suppose there's no more stars, unless you find a way to percieve more stars. You CAN'T and don't have to prove that there's any more stars, that's just leaving it up to speculation.
If we can't see or "sense" more stars then there's all the stars there is.
I can't prove that there's more milk in my fridge than I can percieve.

Even though this. These arguments are pretty irrelevant. Have we proved that just one other star than sol has planets (which we have) around them, further falsification of the hypothesis would be a waste of time.
Either we see stars or we don't.

This is a typical criticism against falsification from many unexperienced science philosophers that is based on the (in my meaning pretty stupid) presumption that we CAN and SHOULD know EVERYTHING. Which in fact we can't really KNOW EVERYTHING. No one can't.
We can percieve the world wrongly and we might not see everyhting that's there.. All we have to do is to come up with new ways to percieve the world e.g we need good technology and methodology to make as good science as possible. In this matter falsification is exceptionally a good method and realistic.

07 September, 2006 20:24  
Anonymous David Hedenberg said...

Its seem to me that the statement in question is unfalsfiable. It is an existential statement (stating that x exist), and such are from Poppers view metaphysical (or 'unscientific, but _not_ meaningless as the possitivists would have it.)

To see this we need to be aware of the asymmetri between existential statements and universal (general) statements (that is, have the form 'all x are (or have property) y')

The general statement, which is the form scientific hypothesis should be in, are falsifiable but not verifiable. Existential statements are verifiable but not falsifiable.

Observations are necessarily only about particular facts, and cannot bear on general statements directly. So observation verifies existential statements of various meanings, and it is these existential statement that the general statements must cohere.

So the statement "there exist one white raven" contradicts the statement "all ravens are black", and once the existential statement is verified, the second is falsified.

The statement up for discussion now is clearly existential :"there exist a planet other than Sol such that it has plantes around it"

It is not falsifiable, because to falsify it would demand a verified general statement of the form "all stars other than Sol are such that none of them has any planets around them". But general statements are not verifiable. No possible observation can give support to such a statement.

And if one tries to modify Popper on this point, beware that a lot depends on this sharp distinction between general/existential and falsifiable/verifiable. For instance a lot of the strenght behind Popper's critique of the logical possitivists, verificationists, and of induction in general.

By the way, I wrote (in Swedish) just a couple of days ago about problems concerning 'degrees of falsifiability'. http://www.lagerkransen.se/the-open-society/files/2cec02346d2c5cb8d2b48bcba94d78bb-30.html

08 September, 2006 09:23  
Anonymous David Hedenberg said...

mistake. Here is the link

08 September, 2006 09:26  
Anonymous dr. dave said...

I think all this talk about the "planets around other stars" comment is misguided. As I understand Popper, he is talking about falsifying scientific THEORIES. A theory is a broad overarching explanation that encompasses a great number of facts and observations. The statement "there are stars around other planets" is not a theory... it is a simple factual statement that is either true or false. One observation is all that is needed to make it true.

Popper is talking about much bigger ideas than this - think instead of the nebular theory of planetary formation, which attempts to EXPLAIN all the observed facts about the planets in our solar system (and would include the prediction that there SHOULD BE planets around other stars.) It are these explanatory frameworks that are open to falsification by collecting more data.

14 September, 2006 02:33  
Anonymous David Hedenberg said...

You are most certainly right. But personally I just felt a need to clarify the logical properties of general and existential statements.

If I remember correctly it was just the kind situation you describe that insprired Popper to develop his theory of falsifiabilty as demarkating property between science and metaphysics. I'm thinking of one of Einstein's experiments.

The key thing for a theory is to rule out some of the possible ways the world could be like, and that is basically what we have in your example. Namely that with the theory in question, every possible world without a star that has planets around it are ruled out by the theory.

Still I feel abit unsure. For what would falsify the big theory is the very brave claim that one have actually succeeded in looking through the whole universe withou finding such a star.

/david hedenberg

14 September, 2006 14:37  
Anonymous naturalist.atheist said...

I applaud the approach of Sven Ove Hansson. By actually testing the claims of Popper against actual scientific activity he is not doing "philosophy of science", but such as it is, he is doing the science of science. Given the persistence and long standing history of presumtions of reality that philosophy is so prone to I welcome the change.

23 June, 2007 00:43  

Post a Comment

Links to this post:

Create a Link

<< Home

eXTReMe Tracker