Michael Kennedy writes on the criticism page:
Popper’s falsifibility criterion is only a necessary condition for scientific status. If by demarcation criterion we mean a frontier with scientific statements on one side of the line and non-science on the other then falsifibility does not work. For any old prophesy such as the world will end tomorrow or I will win a gold medal at Rio will be scientific. What falsifiability does is distinguish the empirical from the non-empirical. And that is well worth doing. Popper did use term “demarcation” but he was not as clear as he might have been, and I am not sure of quite where he stood. Did he perhaps confuse himself, or was he thinking of less strict meaning of the term “demarcation”?
(1) You say that ‘falsifiability does not work’ and provide two examples meant to demonstrate this, but you don’t explain why it’s unacceptable consider either statement scientific. When Popper wrote The Logic of Scientific Discovery, the problem of demarcation was about the classification of statements. The logical positivist criterion stipulated that statements must be ‘conclusively decidable’, verifiable and falsifiable, by empirical observation. Popper argued that such a criterion was too narrow, and that statements need only be ‘unilaterally decidable’ or falsifiable. I’ll readily admit that Popper’s criterion admits some rather silly and false claims into the class of scientific statements, but so what? It could hardly be otherwise.
(2) Nobody has ever claimed, least of all Popper, that his criterion of falsifiability is exhaustive of good scientific practice. Indeed, Popper, and other critical rationalists, have suggested numerous methodological rules and concepts to guide scientific investigation, e.g. Bartley’s ‘check of the problem’ and Deutsch’s concept of a ‘good explanation’. These help explain why most ‘scientific statements’, like your two examples, are not seriously investigated by scientists. In other words, just because someone is investigating a ‘scientific statement’, it doesn’t follow they are practicing good science.
(3) As Popper regularly affirmed, nothing depends on words. Falsifiability was originally proposed to help clarify the difference between scientific from non- or pseudo-scientific statements–to set apart statements that are open to empirical criticism. Whether it is sufficient or necessary (or neither) to classify scientific statements depends on what you want to communicate. In any case, the distinction remains an important one to anyone who pretends to be interested in empirical investigation.
Personally, following our occasional commenter, ‘d’, I think the criterion of falsifiability is best thought of in two ways: ‘(1) as a criticism of theories that claim to give empirical predictions, and yet are compatible with any state of affairs, or (2) as a criticism of actions taken to surreptitiously protect theories from testing through increasingly ad hoc adjustments.’
Falsificationism never existed.
I should have said that my somewhat spontaneous contribution to this site was prompted by a reading of
Miller, David (2007), ‘The Objectives of Science’, Philosophia Scientiae, 11 (1) 2007, 21-43.
which is partly taken from his Out of Error. Miller asks the question: ‘What do we know from experience?
On comments (1) and (2) I readily admit that I just took it for granted that oracular statements like ‘The world will end tomorrow; are not scientific. The problem is what else besides falsifiability should be included in our concept of science? What are the necessary and sufficient conditions of scientific state,ents? This surely is a question worth asking. I take the point that Popper was originally after an alternative to verifiability, but he came back to falsifiability again and again – notably in his 1953 lecture, and in Realism and the aim of Science. But I do seem to recall that somewhere he said that falsifiability ws ”only a proposal”. I think he should have made it clear just waht he meant by ‘demarcation criterion,, and that scientific statements are a subset of empirical statements.
On (3) Nothing turns on the meaning of words? Popper may have attacked essentialism and linguistic analysis, but he also expected critical rationalists to understand each other. So I don’t think he would object to being challenged on the meaning, or ratjer his meaning, of demarcation.
I am not sure that “science” is so important in itself. Science is reified in our culture because the focus is on the body of findings and consequent technology rather than the critical method – which tends to leave the poor motor mechanic out of the institution of science, even though he does great science when he diagnoses a faulty fuel injector. Science is the metaphysics that has appended the critical method: that has evolved in terms of truth testing. Some problems are trivial, some are complex. Popper’s demarcation liberated metaphysics and the creative imagination and “science” from the the false barrier of meaning and meaninglessness. It is an injunction to unpack our knowledge claims by turning them into objects that can be stress-tested, potentially by anyone.
“The world will end tomorrow” is a scientific claim only if you use it the right way. Actually, you can use this statement at least two ways.
You can assert it and wait for tomorrow…
A1. The world really ends tomorrow and you were right. Unfortunately you forget to tell WHY it should happen so your statement had obviously no explanatory power for anybody to listen to it. Maybe if you explained it a bit more in detail then somebody might have listened to it and that somebody might had an idea and the power to stop the world ending. But no, you were secretive enough and now everyone including you became extinct even if this could have been avoided. 7 million people should blame you. Happy afterlife in Hell, my friend.
B. The world does not end tomorrow and you were wrong. Now if you admit that you had a bad theory and come up with an explanation WHY you had to have this bad theory then someone might tell you WHY you were wrong. It’s hard to face it when you’re wrong but at least you’re open to criticism and with it to some knowledge.
C. The world does not end tomorrow but you still think you were right. Now if you DON’T admit that you had a bad theory and DON’T come up with an explanation WHY you had to have this bad theory then unfortunately I have to tell the doctor. But hey, don’t worry you might get a nice new straitjacket.
I believe B is the scientific usage and C is the non-scientific usage.
If by “sufficient condition” you mean something like a final assurance that your next arbitrary statement is good enough, or true enough even before said it than there is no way to get this kind of assurance from anywhere. There is no magic condition that can guarantee the truth of your arbitrary statement. This was officially proved in 1931 by Gödel and independently a bit later by Tarski.
Z rauses some interesting points.
My first response is to affirm that I described my proposition ”The world will end to morrow’ as a prophecy. By this I meant a completely arbitrary statement, one which could have been the utterance of a madman or a religious zealot. In no way was it attached to a theory or a proposed explanatory law.
The statement would be acceptable into science if, for example, it was implied by some theory about the sun exploding or an asteroid hitting the earth. It would then be a prediction, not a prophecy. It would be a test statement worthy of the scientific label. Also worthy would be predictions from a falsified theory such as Galileo’s theory of falling bodies.
But this leads me to the question of what I would like to call ‘pseudo-science’. Suppose the statement about the end of the world was a prediction from certain and precise configuration of the tea leaves at breakfast. It would now be a test statement? But would ot be scientific? I am quite ready to accept as scientific test statements from primitive theories, such as the theory that plants grow as the sole consequence of absorbing soil into their roots. Tests of these theories, and test statements from them, are in my view entirely scientific.
And, if I may digress, this theory was tested by placing a plant in a bucket, weighing it and its soil at the outset, and again after 5 years growth. The plant had put on weight whereas the weight of the soil was unchanged. So who said falsificationism does not exist? This was falsificationism in the 17th century.
But to get back to the point. If the soil theory of plant growth can be accepted as scientific, why not the tea leaves theory of the end of the world? If the statement about the world ending was seen as an implication of the pattern of tea leaves, then it might be seen as scientific. But as a pure prophecy it is not.
My original problem was with a demarcation line or frontier between science and non-science, but I am beginning to wonder if such a line can be drawn at all. Perhaps there is no set of sufficiency conditions for scientific status. Perhaps scientific status is a matter of degree. We would probably accept that Galileo’s theory of gravity is more scientific than the soil theory of plant growth, and that the latter was more scientific than the tea leaves theory. But our agreement on degrees would be a matter of convention rather than of logic.
Michael Kennedy,
Let’s suppose you enumerated a set of properties which all scientific theories must have–multiple criteria of demarcation. Perhaps some of these properties are not binary, but incremental. In that case, you might say that some scientific theories are more scientific than others–some criteria of demarcation are graded. For a theory to be have scientific status, it is not enough to just satisfy these criteria, but it must be relatively more scientific than competitors. Why? Because we want to some theories to lose their scientific status when surpassed by new theories. For example, we want to acknowledge that phlogiston theory was once scientific but is no longer, because today only a pseudo-scientist would continue to advocate it.
What might these criteria of demarcation look like. Here are a few candidates from a critical rationalist perspective:
(1) must be logically consistent.
(2) must be falsifiable (see Popper).
(3) must be a good explanation (see Deutsch).
(4) must be relevant to a problem-situation (see Bartley).
(5) preferable if more falsifiable.
(6) preferable if a better explanation.
(7) preferable if relevant to multiple problem-situations.
(8) preferable if highly corroborated.
(9) preferable if more consistent with background knowledge.
Perhaps we might add to this some more social criteria, such as:
(10) must be taken seriously by at least a significant minority of the scientific community.
(11) must be part of an ongoing research program.
(12) must require years of specialised study and knowledge to understand.
This is a significant list, and it captures a lot of qualities that people associate with science. However, besides the conditions which concern falsifiability, everything here is also true of mathematics and philosophy. All this list of demarcation criteria really do is demarcate ‘scientific’ in such a way that a scientific theory is, by definition, a good or preferable theory–there simply cannot be a bad scientific theory, because to be scientific it must satisfy all these desirable qualities.
If all you want to do is use to ‘scientific’ to refer to good theories that happen to be falsifiable, then I have no objection, in principle, because words can mean whatever we want them to mean. However, personally, I would like to be able to talk about bad theories which are nonetheless scientific without contradicting myself, because I see little purpose in using ‘scientific’ as a synonym for ‘good’ when talking about theories.
To Lee Kelly
I have not the slightest wish to equate ‘scientific’ with ‘good’. My problem is that falsifiability is not a sufficient condition of scientific status. The prophesy that the world will end tomorrow is not science but is highly falsifiable. It is an empirical statement which is not scientific. Popper’s example of astrology was similar, but his case was that its predictions were so vague that they always escaped falsification. So what if astrologers made precise predictions? This would not, I think, make astrology a science.
So there is something to be said for thinking up a list of conditions which if met would be sufficient for scientific status. This is my purpose, and I have no interest in relegating old or disused scientific theories (e.g.Galileo’s law of falling bodies) to non-science. That is a bit like censorship.
I take it that your list is suggestive only. If we are looking for sufficiency then the ‘preferably this or that’ cannot possibly be included. I have doubts about ‘good explanation’ since any explanation in terms of a universal law is satisfactory. I have even graver doubts about your (10) and (11) – taken seriously by other scientists or part of a research programme. This again strikes of censorship and peer review. Einstein would have got nowhere.
I would add to your list of candidates:
1. A theory must be claimed to be a universal law.
2. The theory or proposition must state conditions for its predictions, and be of the form ‘if X then Y, subject to named conditions a,b,c, …’. This could rule out propositions like ;the world will end tomorrow’.
I do not pretend to be sure about these, but feel they are worth discussing.
I am afraid I have to disagree with you when you claim that philosophy and mathematics subscribe to your list. Take falsifiability. I can see that arithmetic might be entirely falsifiable with empirical examples just as logic is with the method of counter-example. Popper is on record as saying that all philosophy is metaphysical (but still criticisable), and I can hardly see how determinism or realism can be falsiiable or ethics either.
But it would make astrology amenable to scientific investigation. Personally, I think that is all ‘scientific’ need describe. I would prefer not to bog it down with additional complications. The scientific method is, however, far more than a criterion of demarcating statements.
In any case, I think Deutsch’s thesis of good explanations is extremely relevant. His argument is that good explanations have the property of being hard to vary. That is, each component of the explanation is dependent on others, so that one cannot easily vary the explanation without breaking it. If we change one component of the explanation, then a cascading sequence of changes is necessary to restore consistency.
Deutsch’s notion of good explanation has similarities with Popper’s idea of degrees of falsifiability. That is, hard to vary explanations tend to be more falsifiable, but I believe the two ideas are logically distinct. Perhaps it’s closer to Popper’s ideas about metaphysical research programs. That is, astrology is sponsored by a degenerative metaphysical research program because its explanations are easy to vary.
Scientists, then, seek highly falsifiable theories that produce explanations which are hard to vary. That seems to be very close to saying that they select theories couched in fruitful metaphysical research programs.
Michael,
By the way, I wrote: ‘besides the conditions which concern falsifiability, everything here is also true of mathematics and philosophy.’
Apologies. I did not read your penultimate paragraph on maths and philosophy carefully enough. It would be best if my mistaken disagreement with you on this Was erased – since it completely misrepresents you.
I also wish to comment on censorship.
Why would any of what I listed be ‘a bit like censorship’. Why can’t we say that Einstein’s theory of relativity was not a scientific theory when he was just scribbling notes in a Swiss patent office, but it later became a scientific theory when the scientific community began to take it seriously? Only if the label ‘scientific’ is a requisite for being taken seriously would it be like censorship, because then Einstein’s theories would never have got a fair hearing.
It’s precisely this kind of thinking that I’m pushing back against. The idea that ‘scientific’ is like an label of authenticity for good theories. I know you just denied that you wish to use the term ‘scientific’ in this way, but now you’re suggesting that to call something ‘unscientific’ would ‘strike of censorship’. This only makes sense to me if ‘unscientific’ generally denotes a bad theory, because only then would its use amount to anything like censorship.
I am not saying that to label a theory as unscientific is censorship. I am saying that to strip a theory of its scientific medal – to demote it from scientific to non-scientific is like censorship. And theories that are not taken seriously by the scientific community often turn out to widely accepted. But my problem is with the interpretation of Popper. Is falsifiability a divide or a necessary condition. Thecword ‘demarcation’ suggestsv the former, but I think he intended the latter.
He does say in Unended Quest that “logically speaking” falsifiability “cannot be regarded as a very sharp criterion”.
Michael,
I believe Popper states in several publications that his criterion of falsifiability is not altogether as clean-cut as he initially thought. Perhaps we should just avoid the word ‘scientific’ and say that falsifiability is about whether theories can, in principle, be refuted by empirical testing.
Yes, perhaps it is not only the statement that is scientific but rather the practice. Science is fermented in an open culture. It is the more the manner with which we handle our beliefs and suppositions that is rational rather than the belief-objects that we are manipulating.
A statement ceases to be scientific even if it starts that way if it becomes vague and ceases to be open to risky tests.
Joe Barnhart’s essay in the “About” section of the Critical Rationalism blog is a nice essay to revisit occasionally to be reminded of the poetic coherence and relevance of Popper’s world view. We certainly don’t want to inadvertently drift into a form of censorship by defining an “essence” of knowledge – scientific or otherwise. Popper answered the question “How do we know?”, “We do not know. We have no guarantees”.
Falsifiability is about truth. Meaning is something else. The necessary condition for science is that it is concerned with truth testing. Many things are meaningful and important but not truth testable. I suppose this is what demarcation is concerned with.
My problem is that still I can’t get my head around that “sufficient condition” for a scientific theory. To me, it seriously sounds like the strict criterion of demarcation the verificationalists used back in the 1920’s.
In my reading, no such goal like a “sufficient condition” can be found in Poppers theories. On the contrary. We can only say that we have better or worse scientific theories. In my understanding of Poppers view every theory is only more or less scientific. Every theory has some unscientific, metaphysical content, assumption which later turn out to be simply false.
Some are utterly/necessarily bad (those which can’t be falsified), some are just bad compared to others (like your “tea leaves” theory). But I can’t see how anyone could state a theory that’s sufficiently scientific. For me this would be like “Yeah, we’ve found the Holy Grail theory at last. It’s scientific enough.” If something is scientific enough, why would you want to do a scientific research anymore?
Falsifiability entails that there is something to falsify i.e. that the scientist is investigating a real world problem. The aim of science is not just to prove statements wrong but also to explain i.e. to produce universal laws. It is necessary that these are testable but surviving tests is hardly sufficient for satisfactory explanations (as suggested in the lists above). We never know if our explanations are right (sufficiency like truth is a regulative idea) but they can be less wrong than alternatives.
Popper’s section 5 “Aim of Science” section in “Objective Knowledge” 1972 hones in on this.
Replying to Z and Bruce Caithness
Falsifiability is crucial because we need to know whether the statements we make are true, false or just potentially true. Singular statements can to some extent be verified, but universals and large-scale generalisations refer to too many instances for verification to be possible. But if they can be falsified, then we can, at least, find out, after severe testing, that they have truth potential, or that they are false and need modification..
In Conjectures and Refutations Popper says his problem is one of
”drawing a line (as well as this can be done) between the statements, or systems of statements, of the empirical sciences, and all other statements–whether they are of a religious or of a metaphysical character, or simply pseudo-scientific. . . . I called this first problem of mine the ‘problem of demarcation’.”
This says that he is looking for a dividing line between empirical science, on the one hand, and non-science, on the other.
This seems to imply that falsifiability is a sufficient condition for admission to the scientific side of the line; whilst non-falsifiability is sufficient for the other side. But he goes on to say:
”The criterion of falsifiability is a solution to this problem of demarcation, for it says that statements or systems of statements, in order to be ranked as scientific, must be capable of conflicting with possible or conceivable observations.” (p.39, my italics).
Here the word ”must” indicates that falsifiability is a necessary condition for scientific status. I think that Popper’s earlier mention of ”drawing a line” overstated his position, and that this passage is confusing.
So I agree with Z that Popper was not searching for a sufficient condition. I would like to agree that some theories are ”more scientific” than others – except that this raises the question of how we judge the degrees of scientific. When we say that astronomy is more scientific than astrology or tea leaves we mean more than that it is more testable. We have in mind a number of characteristics, which together might constitute a grading system or even a sufficient condition for scientific status.
This, however, is just an offshoot of my main concern, which was with the logical status of the falsifiability criterion. I do not think that to look for a sufficient condition of scientific status amounts to a search for absolute scientific truth, or the Holy Grail.
Popper never said that the aim of science is to falsify theories – although some of his critics like to think he said it. The aim is to explain the world, and the method is to produce theories, to test them, and if they fail, to produce new theories.
Some further musings, open to dissection:
Albert Einstein stated: As far as the laws of mathematics refer to reality. they are not certain; and as far as they are certain, they do not refer to reality.
The success of Eddington’s experiment corroborated Einstein’s theory, Einstein would be the last person to say it was conclusive or irrefutable proof.
Generalisation from instances to universal statements is a conjecture; if anyone can produce a valid logical relation between the instances and the generalisations it would be of great interest. It is not enough to say things are so for them to be so.
Popper’s falsifiability is a logical term; falsification in practice is abused in all sorts of ways.
Falsifiability is valid according to the rules of deductive logic.
“Falsifiable” does not entail being necessarily wrong but rather open to counter evidence and not shielded by ad hoc evasion.
Postsript: in his 1982 Introduction to Realism and the Aim of Science, Popper wrote;
My proposal was that a statement (a theory, a conjecture) has the status of belonging to the empirical sciences if and only if it is falsifiable.
David Miller gas it right when he states: ‘The criterion of demarcation . . . is a negative one, telling us what is not scientific’. ( ‘The Objectives of Science’, Philosophia Scientiae, 11 (1) 2007, 21-43.) Available on
http://go.warwick.ac.uk/dwmiller/poincare.pdf
“Popper never said that the aim of science is to falsify theories – although some of his critics like to think he said it. The aim is to explain the world, and the method is to produce theories, to test them, and if they fail, to produce new theories.”
Yes so many critics they lose sight of the main game, which is about producing deeper and deeper theories, chasing answers to deeper and deeper questions.
I completely agree. It is a pernicious travesty of Popper to pretend that he said the objective of science was to falsify its theories – as if the ultimate aim was wholesale destruction. As you say the aim was explanation.
I don’t think the demarcation criterion has any bearing on the aim of science. I am uneasy about Popper’s ‘if and only if’ in the quotation because it confers the stamp of scientific status on any testable statement; such as Nostradamus’ prediction about the end of the world, my claim that I shall be President tomorrow, and any
astrological prediction which happens to be testable.
Falsifiability, in my opinion, is best seen as a necessary condition, not as a sufficient condition.
I always thought that Popper made too much out of the demarcation criterion and I could not see how it was the most fundamental problem in the philosophy of science. There is a partial explanation in the original book that was cut in half to become Logik der Forschung, I think it was fundamental at the time when Wittgenstein was such a big influence.
My take is that you no more boast that your scientific theory is testable than you boast that your car has a steering wheel.
Attempted falsification is just the logically strong was to use evidence, and also coincidentally the most cost-effective way as well, given the cost of collecting data.
The Austrian writer, Robert Musil (1880-1942) worked on his unfinished novel “The Man without Qualities” for more than twenty years and it was unfinished when he died in exile in Switzerland in 1942. The English edition I own has 1774 pages. He seemed to be attempting a synthesis between science and the mystical. It mirrors Viennese life through at least twenty often quite strange characters.
The fellow Austrian, Sigmund Freud’s (1856-1939) approach to the mystical can be summed up in his railing “against the black tide of mud of occultism”.
Another Austrian, Ludwig Wittgenstein (1889-1951) in at least the first phase of his career mirrored Schopenhauer’s world of will and representation but tried forlornly to differentiate science from metaphysics or mysticism in terms of meaning or sense.
Freud’s estranged Swiss friend, Carl Jung (1875-1961) had a less positivistic and more realistic answer, “We constantly use symbolic terms to represent concepts that we cannot define or fully comprehend.”
I look on Karl Popper’s (1902-1994) emphasis on the problem of demarcation as addressing, in part at least, this problem, successfully I believe. Popper in some ways was the man without qualities who saw knowledge as not a problem of personal meaning but rather as an objective problem. Its objectivity lay in its being open to intersubjective testing and always impossible to tie down as permanently verified. Falsifiability liberates meaning, mysticism and metaphysics from the artificial constraints of verificationism. The logical positivists got beyond themselves as do the current flock of “true belief science” advocates in their ongoing war against “the black tide of occultism”.
I too have been puzzled by Popper’s statement that the problem of demarcation was more fundamental than the problem of induction, but I think the answer can be found in the following quotation:
I have discussed the problem of demarcation
in some detail because I believe that its
solution is the key to most of the fundamental
problems of the philosophy of science.
(Popper in ‘Science: conjectures and
refutations’, Section IV of. Ch.1 of
Conjectures and Refutations (Routledge,
1963).
The solution is the key, and solution is, of course, falsifiability The ability to falsify makes scientific investigation possible. The weaknesses of a theory can be exposed empirically, and new theories proposed in response, which in turn can lead to growth in knowledge. The demarcation problem is not trivial or verbal because it decides between what can be worked on empirically and what can only be argued about. Nor is it a problem of status – which, I confess, is the way I have sometimes seen it.
Falsiiability is the answer to fallibilism, non-jjustification and the absence of verification. Much as we would like to be able to establish final and certain truth we cannot do so. But we need not despair or just sit on our hands because we can test our theories and work on a trial and error basis.
Yes, as Popper said in “The Two Fundamental Problems of the Theory of Knowledge” Routledge, 2009 in English:
Problem of demarcation (Kant’s problem) : “When is a science not science?” (page 467)
Problem of induction (Hume’s problem) : “Can we know more than we know?”(page 466)