Misinterpreting Popper
Does modern science philosophy misintetrpret Karl Popper?
I wrote this essay as part of a course on science, ethics and society (MNSES9100). Perhaps it could be useful for a larger audience.
Science has been here all along
An important part of human endeavor is to produce, conserve and share information. It provides continuity and progress across generations and cultures, and it could easily be argued that this activity is to a civilization what learning is to an individual human.
The scope of this information is vast. Some of it is purely descriptive, like which plants are edible or maps of the earth. Other parts are more abstract, like mathematics or music. Others yet are technical or related to method, for example on how maps are drawn in the first place. Fairytales, religion, literature and philosophy provide information on human flourishing (amongst others), and in some cases it is intended purely for entertainment purposes.
Scientific knowledge, or simply science, is obviously part of this activity, no matter which one of it’s many definitions one adhere to. It is not a modern invention - it has been here all along, but it’s definition has been subject to change. From the perspective given this far we may do something different than giving a rigid definition of science; we may instead propose a motivation for such a definition.
While all of this information may be important in it’s own right, not all of it enables us to predict and anticipate nature. This is essential for example in weather forecasting, but is also a prerequisite for constructing equipment with a purpose or understanding how natural processes occur. It would probably not be a good idea to base climate models on the climate data from Middle Earth in Tolkien’s ”Lord of the Rings”, and while the aliens in Star Trek look suspiciously human it may be unwise to expect real aliens to be anything close.
Clearly, if the aim is to make good predictions about our universe on any scale, we should base these predictions on the parts of our knowledge that ensures some accuracy and precision. This motivates a definition of science, or what Karl Popper would refer to as the demarcation criterion.
It should be clear that - at least in this view - science earns it reputation from it’s ability to make accurate and precise predictions. A successful demarcation criterion should be measured on how it provides accuracy and precision in scientific predictions. This will be the framework within which we shall discuss the ideas of Karl Popper and their counterpoints.
Falsification and falsifiability
Demarcation
Karl Popper attributed the demarcation problem to Kant, and found it to be fundamental to all theories concerning knowledge [10, p. 11]. He described it as ”the problem of finding a criterion which would enable us to distinguish between the empirical sciences on the one hand, and mathematics and logic as well as ”metaphysical” systems on the other[…]” [10]. He rejected the idea that truth or falsehoods could be established by means of verification, as proposed by the logical positivists’ [10]. The main proponents of this idea was the Vienna Circle [2], and in their manifesto [6] they state the boundaries of what they consider legitimate science:
- First it is empiricist and positivist: there is knowledge only from experience, which rests on what is immediately given.
- Second, the scientific world-conception is marked by application of a certain method, namely logical analysis.
What they refer to as the scientific world-conception is the set of attitudes, points of view and direction of research associated with science as they see it practiced [6]. As with Popper, the logical positivists’ framework is thus twofold in the way that it first sets the boundaries for what evidence should be considered when doing science, and then describing the best way of using this evidence. More specifically, they proposed that the establishment of truth- or falsehoods within experience-based knowledge was possible by application of logical deduction and -induction.
The problem of induction
As was recognized already by David Hume [7], logical induction comes with an intrinsic problem. The Vienna Circle was aware of this, and in their manifesto [6] they state that ”The method of induction, the inference from yesterday to tomorrow, from here to there, is of course only valid if regularity exists. (…) It may be applied wherever it leads to fruitful results, whether or not it be adequately founded; it never yields certainty.”.
Popper saw this as a fundamental flaw in the framework of the positivists, because it ultimately relied on the possibility of establishing universal truths from experience. Considering the example of black swans, he argues that any conclusion drawn by inferring universal statements from singular ones may turn out to be false. [10]
Popper recognized that by rejecting the ideas of the Vienna Circle, he removed the barriers between science and metaphysical speculation [10]. His solution to this problem is what has earned him a place as an important philosopher of science, namely introducing the concepts of falsifiability and falsification.
Falsifiability
For a theory to be falsifiable there must exists a way of refuting or testing it. According to Popper, any theory that is falsifiable should be considered scientific. Popper’s first proposition is thus to replace the positivist’s criterion for the boundaries of science - which was essentially empiricism - with the criterion of falsifiability. Falsifiability should be seen in contrast with verifiability. As was thoroughly discussed by Popper [10] and others (see for example [7, 9]), we lack a rigid logical framework for establishing truth. Thus, a theory cannot be verifiable. If we still insist on using logic to test the theory, we may however falsify any theory. On this basis, falsifiability provides a logically consistent criterion for demarcation.
In comparison to the positivist’s view, Popper grants scientific status even to theories that has no empirical foundation. The emphasis is put on a theory’s properties instead of it’s origin.
Falsification
Since the criterion of falsifiability by construction permits even false theories for consideration, Popper suggested in analogy to the positivists’ second proposition that the way to distinguish the good theories from the bad ones is by testing and subsequent refutation if the theory does not pass the test. He named this logically consistent framework falsification.
Already in his original work, Popper made a clear distinction between falsifiability and falsification (see for example ref. [10, p. 66]).
Conjectures and refutations
There are details of Popper’s theories that are beyond the scope of this essay, but at this point it is possible to formulate Popper’s philosophy of science in a compact form found in his book ”All Life is Problem Solving” (1994) [11]: \begin{equation} PS_1 \rightarrow TT_1 \rightarrow EE_1 \rightarrow PS_2 \tag{1} \end{equation}
This formula is intended to express the progress of science, where a problem situation (\(PS_1\)) gives rise to a conjecture or tentative theory (\(TT_1\)), which in turn is subject to falsification, refutation or error elimination (\(EE_1\)), finally yielding a new problem situation (\(PS_2\)).
Before we delve any further into the subtle details of Popper’s ideas we shall now explore some of the most common objections to his ideas.
Critiques of Popper
Popper’s theories have been subject to extensive criticism. From his own field, notable philosophers with contradicting views includes Thomas Kuhn [8], Imre Lakatos [2], and Paul Feyerabend [2], but the controversy was not only internal to philosophy of science.
In broad terms, the criticism can be divided into 3 categories:
- The theory is not logically consistent.
- The theory doesn’t work.
- The theory does not describe reality.
An overview of the most common critiques against Popper is given in the book ”What is this think called Science?” by Alan Chalmers [2]. This book is also commonly used as an introduction to philosophy of science [4], and it will form the basis for the following sections.
When first introducing Popper, Chalmers associates him with what he refers to as falsificationism [2, p.55]:
Karl Popper was the most forceful advocate of an alternative to inductivism that I will refer to as ”falsificationism”.
Thus, when reading his chapter 7, ”The limitations of falsificationism”, it is reasonable to assume that the arguments presented are directed at Poppers theories.
Inconsistencies in Popper’s theories
Chalmers argues that falsification does not provide a decisive way of determining whether it is theory or observation that is falsified. [2, p.83]. Popper has, in Chalmer’s view, neglected the possibility of the theory being correct while it is the observation that is false. This argument is further reinforced by the Duhem-Quine thesis [2], which is essentially that predictions are based on a bundle of inextricable hypotheses instead of a single one. As a consequence, falsification in itself cannot decisively refute the tentative theory (\(TT_1\)) of Popper.
Failures of Popper’s theories
Astrology is very often used as an example of a discipline that should be excluded from science by a reasonably formulated demarcation criteria. Chalmers points out that with Poppers criterion, this is not the case: astrology makes falsifiable claims. Chalmers adds in the favour of falsificationists that the practice of falsification may be used to refute claims made by astrology, but he also adds that this does not really resolve anything because of the many other problems associated with falsification [2, p. 102]. (Chalmers here seems to refer to the full body of chapter 7 in his book, although no explicit example is given. Thus, the author wish to point out that in my view, this argument holds only if the rest of the arguments holds.)
Scientific practice
It is reasonable to assume that Popper expects scientific progress in the case where his theories are to be followed. Historical examples however seems to imply that strict adherence to falsificiationism would be counterproductive to scientific progress. The essence of this argument is that at the time of conception of new theories, several observational claims that contradicts the new theory may already be accepted by the scientific community, resulting in a potential refutation by the falsificationists even before any testing is performed.
Chalmers supports this view with historical examples [2, p.85], like how observations regarding the moon or mercury was incompatible with Newton’s gravitational theory. Imre Lakatos [2] also uses Bohr’s original theory of the atom as an example of this, where the electrons were modeled as point-particles. Lakatos’ argument was, analogous to Chalmers, that the observations of stable electron configurations around the nucleus was in contradiction with Bohr’s theory, thus would the falsificationists have to reject Bohr’s theory.
There seems to be an underlying assumption to this argument, where scientific progress is seen as an incremental and continuous process, each iteration dependent on what is previously verified. This is in some contradiction to Thomas Kuhn and his scientific paradigms, who is also often posited as having an opposing view of Popper with regards to how science is actually practiced.
Popper - a falsificationist?
The narrative presented by Chalmers, where Popper is introduced as a falsificationist, and falsificationism is subsequently shown to be problematic, can be found in several books on the subject [5]. This seems to have formed into a consensus view, which may be found for example on the wikipedia page on falsification [12]. But would Popper consider himself a falsificationist, as suggested by Chalmers? When reviewing the criticism of Popper by consulting his original work in ”The Logic of Scientific Discovery” [10], we shall see that this is at best questionable.
Counterpoints to this view are not very common, but one former student of Chalmers has written extensively on the subject. In several blog-posts [5] and books (see for example ref.[3]) writer Rafe Champion has attempted to clarify the at first glance conflicting views held by Popper and his opponents.
Naïve falsificationism
In the first two editions of Chalmers [1] from 1979 and 1982, he define falsificationism by stating that
”Theories can be conclusively falsified in the light of suitable evidence (…) Theory rejection can be decisive. This is the factor that earns falsificationists their title.” (2)
Although changes in subsequent editions of Chalmers refined this definition by including ”sophisticated falsification”, the above definition obviously describes what is often referred to as na ̈ıve falsification. Since chapter 7 on ”The limitations of falsificationism” as well as the pinning of Popper to ”falsificationism” has remained unchanged throughout the editions, it should be clear that this is the theory subject to the criticism in chapter 7.
Popper on the Duhem-Quine thesis
In Popper’s ”The Logic of Scientific Discovery” from 1937, Chalmers could at the time of researching his first edition have read
”In point of fact, no conclusive disproof of a theory can ever be produced; for it is always possible to say that the experimental results are not reliable, or that the discrepancies which are asserted to exist between the experimental results and the theory are only apparent and that they will disappear with the advance of our understanding.” (3)
Thus, it is a direct contradiction between 3 and 2, showing that Popper anticipated the problems of strict application of falsification on empirical sciences. Popper further argues that while falsifiability and falsification provides a logically consistent framework, it is not directly applicable to the empirical sciences. The pragmatic approach he propose for doing science is to set a goal for the empirical method to expose the theory to falsification in every conceivable way [10, 20].
Some of the confusion may thus stem from the fact that while Popper himself is quite clear on the distinction between logic and practical science his opponents are not.
Popper on historical inconsistencies
Revisiting the criticism in section 3.3, we now see that this is clearly a critique of how naïve falsificationism would have impacted scientific progress historically. If we instead consider the logical framework of Popper in this fashion, it can even be shown that the argument is selfcontradictory.
What Popper originally wanted to resolve, was the misconception of verification through induction. If - historically - the body of scientific knowledge was based on an inductivist framework, there would surely come a time were accepted knowledge contradicted new observations - the black swan. Historically this argument does not pass. If we instead base it on falsification such an event is ruled out by construction (…except in the real-world cases were errors surely happen and falsification is indecisive). This latter point is how the argument becomes self-contradictory; for how would the erroneous observation be accepted in the first place?
Paradoxically, this point can be illustrated in Lakatos’ own example of Bohr’s model of the atom. If this model was to be accepted despite the fact that neither the electrons or the nucleus are properly described by point particles, it would give rise to even further inconsistencies with the advent of modern quantum physics. If we instead accept Popper’s approach, Bohr’s theory would have to be rejected as a whole, although constituents like the concepts of electrons and nucleus could by themselves remain unfalsified. This is also what essentially happened.
On the failure of Popper’s theories
Finally we consider the case of how falsifiability, according to Chalmers, does not rule out for example the claims of astrology. As already stated this argument relies on the previously discussed being valid, and could therefore be dismissed right away. There is however some points to be made here.
As astrology bears some rather strong ”unscientific” connotations for many, it is perhaps a somewhat misleading example. In contrast to the logical positivists, Popper puts the emphasis on the properties of the conjecture itself, not it’s origin. While most claims made by astrologers are demonstrably false, they do not actively expose their claims to falsification. Actually, they seem to do quite the opposite, by making vague and ambiguous claims, so falsification becomes really hard. If they were to follow Popper’s pragmatic view on doing science, they would rather make high risk predictions, like what a person born on a certain date would have to do to provoke a highly specific outcome.
Conclusion
Idealisation and reality
Karl Popper was originally a physicist, and as physics relies heavily on mathematical models and constructs it may come natural to many physicists to accept the distinction between model and reality. For example, although the algebra of the Lie group SU(2) may be used to accurately describe the spin of fermions such as electrons, they are not assumed to somehow embody the electrons. Nature just acts, it does not do any algebra to know how to act.
Also, with a background in physics, it may be easier to accept the at times imprecise descriptions of reality that arise from for example quantum mechan- ics. Nature is imprecise but not inaccurate, and this is also what one gets with Popper. When adhering to Popper’s pragmatic view on empiricism and falsification, it is clear that the boundary between science and the rest is fuzzy, and there are degrees of rigor in the falsification process.
Actually, the use of idealised models on real-world problems permeates the whole of science, so it should maybe not come as a surprise that our best model for scientific progress comes in a similar form.
A valid critique of Popper
Although this essay to a large extent argues in favour of Popper, the author does recognize the possibility for a valid critique of Popper’s theories. In the author’s opinion, such a critique should at a minimum distinguish between the logical framework and the pragmatic one. Clearly, it should not associate Popper with na ̈ıve falsificationism.
Further, if a theory is to be seen as an alternative to falsification, it should be possible to somehow hold it up to the same counterarguments that are commonly posed to Popper. Kuhn’s theory of scientific paradigms [8] is for example often contrasted to Popper, but it remains unclear whether Kuhn is even applicable to the intricacies posed by the Bohr-example of Lakatos.
What would be a decisive argument against falsification, was to show that it is somehow internally inconsistent. For example, if it could perhaps be shown that under certain circumstances a tentative theory would have ambiguous status with respect to falsifiability, or that some conditions could render a tentative theory unfalsifiable, making decisive refutation impossible.
Beyond Popper
If we now return to the logically consistent framework of falsifiability and falsification, it should be clear that this is an advancement over the fallacious inductivist approach. However, some may find it unsatisfactory to limit ourselves to establishing falsehoods, but never being able to identify truths. Popper himself had no problem with this and even argued in favour of abandoning the concept of truth [10, p.273], but it may be hard to accept that there could be situations where no unique theory is the correct one.
That Popper was able to come up with these ideas in modern times could be seen as an indication that there are philosophical advancements yet to be made beyond the idea of falsification. We should therefore ask; could we - within the frames of logic - come up with better ways of treating the evidence than Popper? Although the progress could stem from other disciplines like statistical Bayesianism, they should at least within the framework of logic be able to provide more decisive results.
So far, however, falsification seems to be the best game in town.
The bigger picture
Most philosophers of science, except perhaps Feyerabend [2], recognize the need for a well-defined and reasonable criterion of demarcation. But while it is safe to assume that they all want to keep the baby, they have a hard time agreeing upon what is baby and what is bathwater. This could be one of the many reasons for the controversy. Another one could be that philosophers takes delight in disagreement.
In broad terms it is reasonable to rhetorically ask whether there is more to life than science. Popper himself surely did not devaluate contributions even from metaphysics, as he writes[10]
”I am inclined to think that scientific discovery is impossible without faith in ideas which are of a purely speculative kind, and sometimes even quite hazy; a faith which is completely unwarranted from the point of view of science, and which, to that extent, is ”metaphysical””.
The author warmly agrees with this view, since one of the many problems that currently haven’t been dealt thoroughly with in philosophy of science is how tentative theories are discovered or created in the first place. The synergy between the arts, technology and scientific theory is beyond the scope of this essay, but it should be mentioned that human creativity clearly plays an important role in scientific progress.
In the introduction to this essay, we basically demonstrated that the knowledge we refer to as ”scientific” would be largely seen as naturally belonging to a common category whether we enforce some demarcation criterion or not, and that this type of knowledge earns it’s own merit. It should be apparent that the better (and less fuzzy) we are able of identifying theories belonging in this category through a demarcation criterion, the more likely we are to improve scientific progress. It is simply a matter of focusing the effort in the right places. For example, we will know by construction which theories not to spend time testing, and we have some guidelines for which theories to keep and which to throw away.
One could for example hypothesize the ultimate demarcation criterion, which by construction encompasses only the ”true” theories (never mind there may not be such a thing), where no effort beyond the conception of the criterion was needed for knowing all there is to know in science.
A well-chosen demarcation criterion also increases the reliability of, and therefore society’s trust in science. Paradoxically, when proponents of any pseudo-scientific theory purports it to be science, or whenever scientific au- thority is invoked to enforce a certain unscientific view, there is an inflation in the status of science. As it happens, even scientists are known to abuse the scientific status this way, and in these cases they should be reminded that the authority from which they draw their ”power”, so to speak, is reliant on them not acting this way in the first place. In short; science earns its own trust - enforcing it should be superflouous.
References
[1] A. Chalmers. What Is This Thing Called Science? Open University Press, 2nd edition, 1982.
[2] A. Chalmers. What Is This Thing Called Science? Hackett publishing company, 3rd edition, 1999.
[3] R. CHAMPION. Popper: the Champion Guides. CreateSpace Independent Publishing Platform, 2016.
[4] Rafe Champion. A more critical view on Alan Chalmers on Pop- per (blog). http://www.criticalrationalism.net/2010/08/12/a-more-critical-view-of-alan-chalmers-on-popper/, 2017.
[5] Rafe Champion et al. Critical Rationalism (blog). http://www.criticalrationalism.net/, 2017.
[6] Hans Hahn, Otto Neurath, and Rudolf Carnap. The Scientific Conception of the World: The Vienna Circle. 1929.
[7] D. Hume. An Enquiry Concerning Human Understanding. Dover philo- sophical classics. Dover Publications, 2004.
[8] T.S. Kuhn and I. Hacking. The Structure of Scientific Revolutions: 50th Anniversary Edition. University of Chicago Press, 2012.
[9] Christopher Macleod. John stuart mill. In Edward N. Zalta, editor, The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab, Stanford University, spring 2017 edition, 2017.
[10] K. Popper. The Logic of Scientific Discovery. Routledge Classics. Taylor & Francis, 2005.
[11] K.R. Popper. All Life is Problem Solving. All Life is Problem Solving. Routledge, 1999.
[12] Wikipedia. Article on Falsifiability. https://en.wikipedia.org/wiki/Falsifiability, 2017.
