In 1972 Jeremy Shearmur reminded me to be alert to the difference between falsifiability and falsification. He was Popper’s research assistant at the time so he was well placed to offer this kind of advice. Unfortunately I did not take it on board, being obsessed with the bigger picture of Popper’s ideas and what they could do for problem-solving of all kinds, from the theory of literature to the integration of the human sciences and humanitarian social reforms.It is only in the last year or so that the import of that reminder has dawned upon me, as I contemplate with fresh eyes the mess that philosophers have made of Popper’s ideas by defining him as a “falsificationist” without attention to the bigger picture and especially the difference between the LOGIC of testing and his proposals regarding the SCIENTIFIC METHOD . The latter is something which he insisted did not exist, despite the fact, as he reminded his students, that he occupied a chair of Logic and Scientific Method!
The dual responsibilities of that Chair can be used as a hint to make a demarcation, or at least a clear distinction between two aspects of Popper’s thoughts – on the one side the logical critique of induction and on the other his work on conventions, traditions and institutions, the “rules of the game” of science and his start on the constitution for the republic of science (this side of his contribution is beautifully described by Ian Jarvie).
My first extended paper on Popper was published in Honi Soit, the weekly student newspaper at the University of Sydney, in 1971. It was the first of three papers in a series “Towards a Liberal Education”. In that 1971 version I wrote “If the facts do not corroborate the theory we know we have missed the truth, so we must think again. We may even be pleased when our theory is refuted because it gives us the chance to build another”.
Most of the time we cringe when we re-visit out juvenile efforts but in this case I am gratified to find that apart from that particular error the paper can stand with very minor improvements. This is the version on my website at present.
The offending section now reads “If the tests yield negative results (which themselves withstand counter-arguments) or if the theory has internal inconsistencies, or clashes with other well-tested theories, then we know that something is wrong somewhere (even if it can be difficult to locate the source of error). In this situation new ideas are required, or a reinterpretation of old ideas.”
The old version contained a “saving clause”.
“Falsified hypotheses do not need to be discarded, they are a part of the history of ideas, also they may have instrumental value, they may persist as a part of a larger structure”. The current version adds that “they may even stage a revival if they are revised or reformulated.”
Actually that can be improved because a falsified theory is merely rendered problematic, and it may still be the main game in town. Newton’s theory was known to have problems long before Einstein came up with a rival theory of comparable scope.
The reason for putting up this post is to make a kind of excuse for philosophers like Dan Hausman who have not done justice to the two components of Popper’s contribution. Surrounded by other people who all make the same mistakes in interpreting Popper’s work, how could they do better?
To clarify the distinction it helps to read the introductory pages that Popper write in Realism and the Aim of Science (1983), the first of three volumes of The Postscript to the Logic of Scientific Discovery. He noted the way that Lakatos and Kuhn were widely regarded as effective critics of his views, but they had failed to make the critical distinction.
In brief, falsifiability is a logical property of a proposition that is vulnerable to refutation by a true existential statement. The propositions of concern to Popper were universal laws in the form “all swans are white”. This is falsified by the statement (if true) “here is a black swan”.
Falsification is the process of demonstrating that a proposition has been falsified. Unlike the decisive logic of the process, the real-world process of falsification can never be decisive due to the Duhem problem, the uncertainty of observations and the many more or less disreputable ways that people can protect their views.
Even though falsification cannot be decisive it is still an essential mode of criticism (with various other forms of criticism:does the theory solve the problem? is it internally consistent? is it consistent with other well-tested theories? etc).
The point of Popper’s approach is to maximise the effectiveness of testing by working out conventions and rules of the game to detect and minimise the use of disreputable avoidance strategies. And when you get to the problem of working out when probabilistic theories are refuted, this is no easy task. That was the subtext of The Logic of Scientific Investigation (1935), getting to grips with the challenge of quantum phyiscs to conventional views on testing.
Popper’s social turn.
This is a scan on Ian Jarvie’s book to show how Popper took a “social turn” along with the “conjectural turn”, long before people like Kuhn started to take an interest in the social context of science. You can get the message from the closing sections of The Poverty of Historicism where Popper very briefly discussed the role of institutional factors that impact on the growth of knowledge. See also Chapter 23 of The Open Society which launches a preemptive strike on the sociology of knowledge, with a critique of central economic planning, which he saw as a part of the same mindset.
Right.
The logic and method of science are not one and the same. The method of science introduces extra-logical rules–norms and conventions. For example, Popper offered no logical solution to the Duhem-Quine problem, because there is no such thing. Popper instead proposed a methdological solution–we should voluntary abstain from ad hoc dismissals of recalcitrant evidence and “conventionalist stratagems.”
Popper’s methodological solution is no solution at all to mainstream philosophers, because they perceive a different problem. A rule that does not carry the authority of logic, but must be voluntarily acceeded to, lacks force–such a convention cannot assure us that we are embracing truth and rejecting falsity. In this view, the method of science must be reduced to pure logic, because the introduction of extra-logical rules contaminates scientific knowledge with doubt.
The pursuit of an inductive logic is a consequence of this perceived problem. Every convention, norm, decision, thought process, must be reduced to unassailable logical implication. A scientist following this method is no creative decision-maker, but an automaton connecting dots. The directive to “think logically” transforms into an imperative that each thought should be a logical consequence of prior experiences and thoughts.
By offering social solutions to logical problems, Popper was rejecting the implicit assumption that the scientific method be reducible to pure logic. Thus, he was at cross-purposes with more traditional philosophers–now if only Popper’s critics would understand and acknowledge this!
(By the way, I think it is okay to talk about a “scientific method,” even though it may only be a component of a larger method of rational investigation.)
To sum up the point of my comment above:
Philosophers usually do not appreciate Popper’s distinction between the logic of falsifiability and the social conventions of falsification. By making such a distinction, Popper was confronting a whole different set of questions. To the average philosopher, presuming they recognise this departure from tradition, Popper is merely missing the “real” problem, i.e. how can the conclusions of science be imbued with the same legitimacy and certitude as the conclusions of pure logic and mathematics? But since problems are only problems relative to goals and obstacles, there is nothing especially “real” about this problem. The problem is the consequence of a decision (often unconscious), and it can be undone by simply deciding upon a new goal.
BTW on the Duhem problem, I wrote a whole thesis on the topic under the supervision of Alan Chalmers at Sydney Uni. Like my other wonderful supervisor, Keith Barley, he got the balance right between allowing individual initiative and keeping on track to get the job done on time.
The thesis was written for a Master of History and Philosophy of Science. This was a time when I gave up on promoting Popper’s ideas (as an independent scholar), having hit brick walls for years I decided to take a step back and approach the field in a formal course, with an open mind, to find what people learn when they enter through a course in the normal and approved manner.
Alan was revising his book “What is this thing called science?” for the second time, so the third edition appeared soon after. I realised later that some of the essays that he gave me to write were on topics that he was re-visiting for the book, like the real problem with Kuhn.
The thesis is on line.
Introduction and Popperians http://www.the-rathouse.com/Theses/Duhem-QuineIntroPopperians.html
Bayes and the New Experimentalism http://www.the-rathouse.com/Theses/Duhem-QuineInBayesNewExperimentalism.html
Duhem response, conclusions, references http://www.the-rathouse.com/Theses/Duhem-QuineResponseConclusions.html
I did not find the topic exciting because I really wanted to work on the implications of non-justificationism or the theory of metaphysical research programs applied to Austrian economics, or the fertility of the three world theory, but nobody was available to supervise those topics.
Still the thesis turns out to be more interesting and useful than I realised at the time. The point is that there is no logical solution, as Lee said, scientists just have to do more work, using ingenuity with experiments or tinkering with the details of the theory. And sometimes the results of experiments are so dramatic that the field is convinced very rapidly, despite all the logical possibilities that exist to quibble.
It was also useful to explore the implications of Bayesian subjectivism which has become the great hope of the justificationists and inductivists. One striking result was that the Bayesian formula does not work when you have to choose between two rival theories of comparable scope. That is apart from a heap of unresolved technical problems. But those problems just point people in the wrong direction, towards the formal machinery of the trade and away from the things that really matter “outside the window” in Pete Boettke’s fertile phrase, the problems that theories solve and the problems that they do not solve, etc.
My major problem with Bayesianism (as opposed to Bayesian probability) is that it fails to overcome the problem of induction. It seems to me that Bayesianists are too liberal with their interpretation of “h.” The symbol “h” is chosen because it stands for “hypothesis,” but I can’t make sense of Bayesian equations unless the hypothesis takes a particular form. Whenever “h” is interpreted as a universal, the Bayesian equation doesn’t mean what it appears to.
Where “p(e|h) = 1,” it is always possible to define “h1” as the logical content of “h” minus the logical content of “e.” The evidence is neutral toward the new hypothesis, but the new hypothesis contains all unverified predictions of “h.” In other words, although “h” increases in probability, the unverified predictions of “h” do not. The problem of induction cannot be overcome by an appeal to probability.
It is commonplace to claim that evidence can change the probability of a theory being true. Predicted evidence increases probability, and contradictory evidence decreases probability. An apparent corollary is that when evidence changes the probability of a theory, the change of probability is transmitted to each of the theory’s other predictions. This assumption seems false. It could be described as a fallacy of division.
This objection is probably terribly misinformed–I am no mathematician!
Don’t waste too much time getting better informed until they can explain what difference it makes to some actual living, breathing research program to put probability figures on the rival hypotheses. Just another academic industry, thriving on elaboration of technical details.