Echoing the intellectual concerns of other philosophers, Sir Karl Popper was initially motivated to draw a line of demarcation between science and pseudo-science (Popper 2002, 344). Popper is not convinced by the scientific status quo, which argued that science was based on induction (Popper 2002b, 3-7). He agrees with Hume that to use a limited set of experiences to conclude for an unobserved experience would not give rise to certainty (Hume 2002, 298-310). Popper mirrors Hume by stating that,
“it is far from obvious, from a logical point of view, that we are justified in inferring universal statements from singular ones, no matter how numerous; for any conclusion drawn in this way may always turn out to be false: no matter how many instances of white swans we may have observed, it does not justify the conclusion that all swans are white.” (Popper 2002b, 4)
Hume does not restrict his argument to the uncertainty of induction; he claims that induction is not justified in any way. Inductive arguments are based on an assumption that nature is uniform, or as Hume asserts, “the future will resemble the past” (Hume 2002, 305). However, the only way to justify this assumption would be to use induction. Popper reflects Hume’s position by asserting that to logically justify induction “we should have to employ inductive inferences” (Popper 2002b, 5). Hume argues that this reasoning is circular because the assumption is based on the thing that we are seeking to justify. To justify induction with this assumption would be tantamount to justifying induction with induction. After all, it could be that nature is not uniform at all (Hume 2002, 304-305).
Popper’s View of the Scientific Method
Popper’s method may be considered as an “anti-inductivist version of the hypothetico-deductive method” (Sankey 2010, 253). His model is based on two stages. The first is based on falsifiability, which is the formulation of a testable hypothesis that explains the empirical data. The second is falsification, which entails that the hypothesis is tested by trying to find empirical data it cannot explain. The essence of this model is to not justify the hypothesis with instances of empirical data that support it; rather it is to find empirical data that proves the hypothesis wrong. If we do find data that the hypothesis cannot explain, we therefore falsify our hypothesis, and we ought to give up the hypothesis for another one that comprehensively explains that new data. Popper argues that this model represents the practicalities of the scientific method more accurately, and that it provides us with a deductive argument, which gives us with certain knowledge. Popper’s falsification is sometimes called “deductivism”, a term he also used to describe his approach (Sankey 2010, 253-254; Popper 2002b, 7).
Popper understands that in rejecting induction, his critics would argue that he removed “the barriers which separate science from metaphysical speculation” (Popper 2002b, 11). Popper retorts to this criticism by arguing that induction does not “provide a suitable ‘criterion of demarcation’” (Popper 2002b, 11). He maintains that falsifiability “should be taken as the criterion of demarcation” (Popper 2002, 345; Popper 2002b, 18). Popper asserts that science should not focus on piling up observational evidence to support its hypotheses, because no amount of observational evidence would prove a theory to be true; rather science should look for evidence against its hypothetical claims (Rosenberg 2012, 207). Popper’s idea of demarcation was premised on the logical asymmetry between verification or confirmation and falsification. Due to the logical problems inherent in the inductive method, it is impossible to confirm a hypothesis or a theory using limited experience. Conversely, a single experience or observation that refutes the theory’s claim would conclusively falsify it.
Popper’s view was that a scientific claim was only valid if it can be falsified; he maintains that “the possibility of refuting theories by observations is the basis of all empirical tests.” (Popper 2002, 192). Falsifiability entails that in order for scientific statements or hypotheses to “convey information about the empirical world” they have to be “capable of clashing with experience” (Popper 2002b, 313-314). So according to the concept of falsifiability a hypothesis must be testable.
Popper maintains that it was the progress and growth of science that gave it its rational and empirical character (Popper 2002, 291). Popper’s idea of progress and growth was that “scientists discriminate between available theories and choose the better one” (Popper 2002, 291) and he argues that his idea of progress should not be mistaken for the “historical law of progress” (Popper 2002, 293), rather, it relates to the fact that we can learn from our scientific errors and mistakes (Popper 2002, 293).
According to Popper scientists should “propose bold, speculative theories” (Sankey 2010, 253). He asserts that for scientific theories to have “potential progressiveness” (Popper 2002, 294) they have to be informative, which means that the theories contain “greater amount of empirical information or content” (Popper 2002, 294). Theories must also have “greater explanatory and predictive power” (Popper 2002, 294) which allows them to be “more severely tested” (Popper 2002, 294).
Popper distinguishes between falsifiability and falsification (Popper 2002b, 66). Falsifiability asserts that scientific statements must be falsifiable, in other words testable. Falsification states that if a particular hypothesis predicts that under certain conditions X will happen, and if under those conditions occur X does not happen, then the hypothesis is falsified.
Popper viewed modus tollens [MT] as “a good representation of the logic of testing a hypothesis (H) by seeing if its observational implications (O) hold true” (Sober 2008, 129):
- If H then O
- Not O
- No H
This doesn’t mean that if a prediction were to come true that the theory is true. Such an argument would commit the fallacy of affirming the consequent [FAC], which would be deductively invalid (Sober 2008, 129):
- If H then O
Since MT is deductively valid and the FAC is not, Popper argued that you can prove theories false, but you could not prove them to be true (Sober 2008, 129).
Consider the following example,
- All sheep are white.
- There are black sheep in Europe.
- Therefore, all sheep are not white.
The above conclusion necessarily follows from the premises, therefore it is deductively valid. It starts with a hypothesis (all sheep are white) and it is proven wrong (there are black sheep in Europe). The conclusion (all sheep are not white) necessarily follows. Hypotheses cannot be proven to be true, they can only be falsified. Under this model we attempt to find a fault in the generalisations we have made from these limited observations. This only leads to certainty in what is false, not what is true.
Popper’s rejection of induction does not mean that there is no way of accepting a theory. Popper maintains that theories are not “‘true’ statements” (Popper 2002b, 264) and that they are “provisional conjectures” (Popper 2002b, 264). Popper argues that a theory is strengthened by the fact that its passes falsification tests. This non-inductive warrant for a theory is described as corroboration. Corroboration is not as simple as enumerating the number of tests a theory has passed (Popper 2002b, 265), rather, it includes the testability of a theory. In other words, the more testable a theory is, described as “the degree of falsifiability” (Popper 2002b, 267), the greater the corroboration. A theory that has passed its tests can only be replaced by another if it is better testable (Popper 2002b, 268). The higher the degree of falsifiability, and the greater the number of tests it has passed, increases its degree of corroboration. Popper defines his idea of corroboration as “quasi-inductive” (Popper 2002b, 276), however, he wants to make sure that he is not misconstrued as arguing inductively, and maintains that corroboration is an evaluation of a theories past performance (Keuth 2005, 114).
Is Popper’s View Satisfactory?
The Problem of Demarcation
Elliot Sober argues that Popper’s falsifiability as a criterion for demarcation would render some unscientific theories as scientifically valid. Sober argues that falsifiability “entails that some creationist claims are falsifiable and, hence, are scientific” (Sober 2008, 129). For example, the claim that a Divine being created every living thing with three legs, would be falsified by the fact that living things have less or more than three legs. Since this claim is falsifiable, according to Popper it is a scientific theory. However, creationism is considered unscientific by the majority of scientists (Hansson 2015). Hansson Sven Ove argues that a strict interpretation of Popper’s ideas “excludes the possibility that there can be a pseudoscientific claim that is refutable” (Hansson 2015). Hansson postulates that according to falsifiability unscientific hypotheses are not rejected on the grounds that they are unscientific, but rather they are accepted as scientific then tested and subsequently refuted (Hansson 2015).
Herbert Keuth argues that Popper’s falsifiability is contingent on time. Falsifiability entails that something is testable, but how do we distinguish between testable and non-testable hypothesis or theories? The idea of falsifiability or testability is contingent on the technology available to us at any particular period of time (Keuth 2005, 43). Take into consideration a historical hypothesis that there are other galaxies in the universe. This theory could not be tested until in 1923 when Edwin Hubble used telescope technology, and he observed empirical support for the existence of other galaxies. This implies that Popper’s corroboration, which is premised on falsifiability is “relative to a particular time” (Keuth 2005, 43).
Can Confirmation Be Dismissed?
Scientists use confirmation to make decisions as whether one theory is to be taken over another. The focus of confirmation “is to understand what it means to say that datum E confirms or supports a hypothesis H” (Hajek and Joyce, 115). The main concepts of confirmation involve the following:
Absolute: A hypothesis, H, is highly supported given evidence E.
Incremental: E increases the evidential support for H.
(Adapted from Hajek and Joyce 2010, 115)
The above concepts involve E which is acquired via induction. However, Popper rejected induction, he argued that no matter how many instances of E we find that seem to confirm H, it doesn’t prove H to be true. The conclusion for Popper is that confirmation is to be dismissed as a way to accept one hypothesis or theory over another.
Poppers denial of confirmation seems to be untenable in light of mainstream scientific practice, as some argue that “rational decision and science would be impossible without it” (Hajek and Joyce 2010, 115). Popper’s rejection of confirmation does not reflect the scientific enterprise because confirmation is used by scientists, and to understand science one would have to understand the “logic of confirmation” (Crupi 2015). Practically speaking, if we were to dive into a swimming pool, would we sink into the water or bounce back like a tennis ball? Regardless of the philosophical gymnastics being played by philosophers of science, there is little doubt concerning well-confirmed hypothesis from a pragmatic point of view. Well-confirmed scientific theories are used by other fields such as engineering; consider the construction of buildings that use the well-confirmed laws of physics. They are true in a practical sense, because they work.
Popper’s falsification does not only make some unscientific theories scientific, it renders some scientific theories unscientific. This applies to “all of the probabilistic theories that have been developed in different scientific disciplines” (Sober 2008, 130). Probabilistic theories in science do not attempt to prove a hypothesis of theory as true, because observational evidence does not logically entail the theory. There are a range of interpretations for probability; the classic, the frequency, the propensity, the logical and the subjective interpretations (Galavotti 2010, 417-423). Galavotti postulates that the frequency interpretation has “long been considered the natural candidate” due to its “empirical and objective character” (Galavotti 2010, 423).
Probabilistic theories entail that an increase in empirical support of a theory, increases the probability that the theory is true. For example in medicine, explanations for the causes of diseases “employ epidemiological relations which are reported in statistical form” (Rosenberg 2012, 90). This is why there is a consensus that smoking causes lung cancer, even though it is common knowledge that statistical correlations do not represent causal connections.
The Problem of Corroboration
Popper’s idea of corroboration is based on the number of tests a theory has passed and its level of falsifiability. This proves to be problematic for competing theories with different levels of empirical confirmation. Generally, the standard practice is that scientists take a well-confirmed theory over another that attempt to explain the same phenomenon. However, Popper’s corroboration does not consider confirmation as a basis to differentiate one theory over another. No matter how well-confirmed a theory is, it will only be taken over another if it passes falsification tests and if a theory that has a higher level of falsifiability.
Corroboration is at odds with scientific practice. During the nineteenth century there were competing theories on the cause of Cholera; the Miasma theory and Germ theory. Miasma theory postulated that diseases were caused by a poisonous vapour in the air. Germ theory maintained that diseases were caused by microorganisms. A Popperian would not take the empirical confirmation of Germ theory as evidence for its truth and would still accept Miasma as an epistemically equal theory because it (at the time) has not been falsified.
Inference to the best explanation
Many of the conclusions scientists make are based on inference to the best explanation (IBE). IBE share the same problems as induction, as new data can alter which of the competing inferences best explain the overall data. IBE attempts to comprehensively and sufficiently explain observational data or background knowledge that we hold (Lipton 2004, 56).
Lipton postulates that scientific data can have competing explanations and we attempt to differentiate between them by trying “to find data that discriminate between them…An inference may be defeated when someone suggests a better alternative explanation, even though the evidence does not change.” (Lipton 2004, 64-65)
Other conditions for IBE includes that the best explanation is one that is the simplest. There must also be a balance between simplicity and comprehensiveness. Comprehensiveness means that an explanation considers all of the data, including unique observations. Another criterion in evaluating the best explanation is that, given our background beliefs, it is the likeliest to be true (Harman 1965, 88-95).
IBE is an obvious and indispensable way of forming conclusions about the data that we have observed. However, since our observations and data are limited, it follows that there can always be new observations and data that can change which of the competing explanations are the best. According to the Popperian, IBE is rejected because it is premised on induction. A Popperian also objects to inference to the best explanation as it is deemed as too subjective. Theories, hypotheses or explanations can never be described as likely or good. According to Popper, hypotheses are either falsified or have a degree of corroboration. This highlights how Popper’s philosophy is at odds with main scientific practices.
The Assumption of Induction
An interesting argument against corroboration is that it assumes induction. Popper denied induction but may have inadvertently required its use in his criteria for accepting one theory over another. Wesley Salmon provides a cogent argument to highlight this point. Salmon argues that since the degree of corroboration is what informs a scientist’s decision to accept one theory over another, it implies induction. Because corroboration informs the scientist to take a future action; the acceptance of theory over another.
Popper seems to be aware of this objection and he postulates that we should not rely on any theory, and since we have to choose between competing theories then it is rational to pick the one with a higher degree of corroboration (Salmon 1981, 121-122). Popper responds to further criticism by arguing that degrees of corroboration do not have predictive import (Salmon 1981, 123). However, Salmon address this response by arguing a theory’s degree of corroboration does inform the future action of a scientist, which is to choose the best theory. Therefore, corroboration has inductive power (Salmon 1981, 123-124). Howard Sankey summarises Salmon’s argument well, he says that since corroboration is used to decide the future action of accepting one theory over another “this amounts to an inductive inference from past success in surviving tests to the likely continuation of such success into the future. It, therefore, appears that Popper’s falsificationist philosophy of science rests at base on an assumption that is inductive in nature.” (Sankey, 253)
Are Falsified Theories False?
From a logical point of view a falsified theory can be revived by referring to their auxiliary assumptions. According to Pierre Duhem,
“Physical science is a system that must be taken as a whole; it is an organism in which one part cannot be made to function except when the parts that are most remote from it are called into play, some are more so than others, but all to some degree.” (Duhem 1991, 187-188)
Take the discovery of Neptune as an example. When previously established Newtonian laws failed to form predictions concerning Uranus’ orbit, scientists did not reject the laws because of the mere existence of an anomaly. The perturbations of Uranus could have been explained with reference to another celestial body whose mass would affect Uranus’ orbit. In 1945 Le Verrier and Adams applied their calculations and a new planet was observed in line with their predictions. According to Popper’s falsification, Newtonian physics would have to be discarded, however, this would mean that Le Verrier’s and Adams’ predictions would have to be described as unscientific. Instead of rejecting their theory, they adhered to it and sought to explain the conflicting data. This example is not uncommon. Scientists do not reject theories just because they encounter some conflict with new observations. They seek ways to reconcile the data without dismissing their theories all together.
Popper’s view of science ignores one of its hidden pillars. The increasing size of the scientific body of knowledge necessitates that scientists refer to other scientists’ experiments and results to aid their understanding of a particular phenomenon. Scientific knowledge, as it stands today, would not be possible if there was an requirement for a scientist to attempt to falsify every theory upon which their theory was based. Simply, we cannot test everything. Even if the argument were to be put forward that since the theory has already been tested, it still assumes testimony. Because a scientist would have to believe in the testimony (which includes written) of another scientist or group of scientists, without doing the actual testing themselves. It seems, therefore, that testimonial evidence is fundamental to scientific theories. Theoretical explanations do not exist in a philosophical vacuum; they are premised on other scientific theories.
Testimony is a branch of epistemology “concerned with how we acquire knowledge and justified belief from the say-so of other people” (McMyler 2011, 3). Scientific knowledge is impossible without it. C. A. Coady argues for the justification for testimony, and addresses Hume’s reductionist account of testimony. (Coady 1992, 79 – 81) The reductionist asserts that testimony is justified via other sources of knowledge such as observation. Testimony on its own has no justification and must be supported by experience. However, Hume highlights the importance of testimony, “We may observe, that there is no species of reasoning more common, more useful, and even necessary to human life, than that which is derived from the testimony of men…” (Hume 2008, 56-57) Hume argues that our trust in testimony is based on a conformity between testimonial knowledge and experience (Hume 2008, 57). Coady views testimony as fundamental; he argues that testimony is justified without appealing to other sources of knowledge. This account of testimony is known as the anti-reductionist thesis. Coady seeks to dismantle the basis of the anti-reductionist approach.
Coady argues that Hume’s argument is circular. Hume asserts that testimony can only be justified if the testimonial transmission, agrees with observation. However, for Hume, observation is not individual personal observation, rather it is collective experience. Coady dismantles this argument because we can only capture collective observations based on other people’s testimonial transmission. Referring to individual observations would be insufficient, and it would be not be provide much scientific knowledge. Therefore, the reductionist thesis is mistaken. Testimonial transmission cannot be justified via other sources of knowledge, such as observation, because it assumes that testimony is fundamental. In order to capture collective observations, you must refer to other people’s testimony, as we have not observed them ourselves. (Coady 1992, 80 – 81)
Science requires the interweaving of one theory into another. However, a single scientist cannot practically test all theories or sub-theories that their research programme is based on. They have to rely on some form of testimonial evidence, even if it appears in the form of a peer review book or journal. This highlights that testimony, as a fundamental source of knowledge, is necessary for science to progress. However, a Popperian account of science would not be able to account for testimonial evidence. To highlight this further, consider the experts in other fields of science. Scientists usually have to refer to expertise to ensure their theories or experiments – that may inevitably overlap into other fields – are robust. This dependency on other experts is a form of testimonial knowledge. Elizabeth Fricker explains,
“But given my cognitive and physical limitations as parametric, there is no room for rational regret about my extended but canny trust in the word of others, and enormous epistemic and consequent other riches to be gained from it.” (Fricker 2006, 244)
Popper’s account of the scientific method is based on the falsifiability and falsification of theories. One theory is preferred over another based on how much it has been corroborated. Popper dismisses confirmation due to the problem of induction, yet he requires inductive inference when laying down the principle of corroboration for accepting one theory over another. Popper’s philosophy is useful if taken as a guide, however, a hard Popperian account of the scientific method dismisses well established scientific practices and assumptions.
Coady, C. A. J. Testimony: A Philosophical Study.New York: Oxford University Press, 1992.
Corvi, Roberta. An Introduction to the Thought of Karl Popper. London and New York: Routledge, 1997.
Crupi, Vincenzo, “Confirmation”, The Stanford Encyclopedia of Philosophy (Fall 2015 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/fall2015/entries/confirmation/>.
Duhem, Pierre. The Aim and Structure of Physical Theory, trans. from 2nd ed. by P. W. Wiener. Princeton, NJ: Princeton University Press, 1991.
Fricker, Elizabeth. Testimony and Epistemic Autonomy in The Epistemology of Testimony. Edited by Jennifer Lackey and Ernest Sosa. Clarendon Press: Oxford. 2006.
Galavotti, Maria, “Probability” in The Routledge Companion to Philosophy of Science, ed. Stathis Psillos and Martin Curd. (Oxford: Routledge, 2010).
Hajek, Alan and Joyce, M. James, “Confirmation” in The Routledge Companion to Philosophy of Science, ed. Stathis Psillos and Martin Curd. (Oxford: Routledge, 2010).
Hansson, Sven Ove, “Science and Pseudo-Science”, The Stanford Encyclopedia of Philosophy (Spring 2015 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/spr2015/entries/pseudo-science/>.
Hilary Putnam. (1979). Mathematics, Matter and Method. In: Mathematics, Matter and Method. [Online]. 2nd ed. Cambridge: Cambridge University Press. Available from: Cambridge Books Online <http://ebooks.cambridge.org//chapter.jsf?bid=CBO9780511625268&cid=CBO9780511625268> [Accessed 14 April 2016].
Hume, David. 2002. “Of scepticism with regard to reason” In Epistemology: Contemporary Readings, edited by Michael Huemer, 298-310. Routledge. Originally published in David Hume, “Sceptical doubts concerning the operations of understanding” In An enquiry concerning human understanding, in Enquiries concerning human understanding and concerning the principles of morals, 2nd edn, ed. L.A. Selby-Bigge (Oxford: Clarendon Press, 1902).
Hume, David. 2008. An enquiry concerning human understanding. http://www.earlymoderntexts.com/assets/pdfs/hume1748.pdf.
Keuth, Herbert. The Philosophy of Karl Popper. Cambridge: Cambridge University Press, 2005.
Kuhn, Thomas. The Structure of Scientific Revolutions. Fourth Edition. Chicago: University of Chicago Press, 2012.
Lipton, Peter. Inference to the Best Explanation. Second Edition. London: Routledge, 2004.
McMyler, Benjamin. Testimony, Truth and Authority. Oxford University Press, 2011.
Popper, Karl. Conjectures and Refutations. Oxford: Routledge, 2002.
Popper, Karl. The Logic of Scientific Discovery. Oxford: Routledge, 2002b.
Reichenbach, Hans. 2009. “The pragmatic justification of induction” In Knowledge: Readings in Contemporary Epistemology, edited by Sven Bernecker and Fred Dretske, 552-55. Oxford University Press. Originally published in Hans Reichenbach, “The pragmatic justification of induction” In Experience and Prediction: An analysis of the foundation and the structure of knowledge. (Chicago: University of Chicago Press, 1938) 341-2, 346-51, 355-7.
Rosenberg, Alex. Philosophy of Science: A Contemporary Introduction. New York: Routledge 2012.
Salmon, Wesley C. “Rational Prediction.” The British Journal for the Philosophy of Science, 32 (1981): 115-125.
Sankey, Howard, “Scientific Method” in The Routledge Companion to Philosophy of Science, ed. Stathis Psillos and Martin Curd. (Oxford: Routledge, 2010).
Sober, Elliot. Evidence and Evolution: The Logic Behind the Science. Cambridge: Cambridge University Press, 2008.
Thornton, Stephen, “Karl Popper”, The Stanford Encyclopedia of Philosophy (Winter 2015 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/win2015/entries/popper/>.