Notes from a 3rd year module in the philosophy of science. I've also contributed a fair bit to the Wikipedia entries that cover this area of philosophy very well.
Falsificationism starts with the rejection of induction as a basis for scientific knowledge, which it regards as irrational despite being a part of scientific method. Since science cannot be irrational, there must be another way for scientific method to proceed.
Karl Popper suggested that scientific knowledge is a category that is falsifiable, rejecting metaphysics, much of psychology, existential statements (e.g. "there are such things as electrons") and so on.
Popper developed the theory of falsification in response to the problem of induction. He supposed that induction cannot be part of rational thought, but that we cannot be irrationalist about science, and so we must be able to find another rational (deductive) foundation for scientific knowledge and method. This was falsificationism, according to which a theory cannot be proved true but it can be falsified, allowing scientists to progress by removing false hypotheses. Because of the dogmatic limits of this demarcation criterion, Popper and his followers also developed a methodology and set of criteria for evaluating falsifiable but unfalsified hypotheses. Ultimately, Popper's criterion of demarcation is too dogmatic to work, and the tertiary methodology and criteria leave him open to the very criticism that he leveled on induction.
The basis of Popper's theory, as I have said, is that we cannot infer new truths, but we can infer that a theory is false. If we observe a set of phenomena, it would be irrational to induce that a hypothesis that predicted the phenomena is true. If, however, the set of observed phenomena contradict the hypothesis, because we can be sure of our observations – Popper was an empiricist – we can be sure that the hypothesis must be false. Popper then took this statement and suggested that if a hypothesis is falsifiable, and it has not yet been falsified, then scientists can in some way accept or at least preserve it in their system of theories. Moreover, he claimed that an unfalsified falsifiable hypothesis is in some way close to the truth, and that the more stringent the testing of the hypothesis, the closer to the truth it must be. The body of science therefore consists in all the theories that are falsifiable and not false (Lakatos, p. 12). And within this body of science, proximity to the truth is a methodology available to discerning scientists. This approach borders on making truth semantically meaningless in science, which can either be a serious problem or an acceptable consequence.
The falsifiability criterion immediately raises the question: what is it for a hypothesis to be falsifiable? For an empiricist like Popper, the obvious answer is that it can, in principle, be tested experimentally such that a set of observations could conceivably contradict it. But how can we know if it is possible for us to experimentally test a hypothesis? We may think it possible but fail to construct an appropriate experiment, even after a thousand years of attempts. But this doesn't mean that in principle it isn't falsifiable... the simple possibility that it is false makes it so. The falsifiability criterion was supposed to mark scientific theories as distinct from metaphysical theories, since the latter are not in principle falsifiable. But it seems problematic.
A solution offered by Lakatos was to say that it must be falsifiable with available scientific methods (Lakatos, pp. 12-13). But even here, the methodology raises problems.
To begin with, it isn't clear that we can distinguish so clearly between observation statements and theories. Most scientific “observation†is done with the help of instruments, and that data that is collected is then interpreted with the help of theories. So if scientists must accept observation statements, they must also be able to accept the theories upon which they are based. But according to Popper's dogmatic falsificationism we cannot accept theories as being true. Furthermore, both in theorising and observing the scientist will have expectations and hypotheses that are based on both past theory and past observation. So we cannot use pure observation as a basis for scientific methodology (Lakatos, pp. 14-15).
To make things even more problematic for Popper, elementary logic – which he used as the basis for his rejection of induction – dictates that propositions can only be derived from other propositions. Observations cannot, therefore, have any logical bearing on propositions and so cannot form the basis for falsification (Lakatos, pp. 15-16).
A further problem with falsificationism is, again, with the use of observation statements. If a scientist formulates a hypothesis and then finds conflicting evidence, E1, he can always posit the existence of a second unobserved phenomenon, E2, that will save his hypothesis. For example, a planet whose orbit appears distorted – from its expected ellipse – could be under the gravitational influence of another large mass like another planet. If we charitably assume that there are no further factors that could be taken into account, and that the scientist then takes E2 into account he cannot be sure about E2 until it is observedand so cannot use it to support the hypothesis. If he fails to observe E2 and thereby falsifies his hypothesis, we must conclude that the conjunction E1 of E2 falsified the hypothesis, not E2 alone, which requires us to be certain about the truth of both E1 and E2.
In reality, the set of evidence will include E1 and E2 and E3...n where n can be any number. Now the scientist must find a way to test each of the pieces of evidence in isolation, or rather in an experiment in which none of the other pieces of evidence are factors, to establish which of the pieces of evidence is false. This can only be satisfactory by Popper's pragmatic standards if n is small and if E1...n can all be separately falsified, i.e. if you have falsified a sufficiently small set of evidence that the result yields scientific insight.
Popper's solution to these problems, which make dogmatic falsificationism unviable, was to adopt a conventionalist stance, creating methodological falsificationism. According to conventionalism, we can accept a statement as being “true†if there is general agreement amongst the community of intellectuals studying it. So if scientists can reach a general agreement about a particular observation statement they can claim it is “trueâ€, therefore allowing them to “falsify†hypotheses. Note the qualification on the terminology of falsificationism – Lakatos emphasises that by Popper's strict logical standards we cannot claim that “true†observation statements in this sense are true, so we must qualify the entire body of terminology (Lakatos, p. 23).
But if a general agreement amongst scientists can establish a convention, which in turn can, though revisable, “falsify†a hypothesis, why can scientists not also “prove†a hypothesis through “induction†based upon convention? (Bird, p. 242) In other words, if we can weaken falsificationism to this extent, can we not also accept a similarly weakened form of induction? Popper could still reject this move on the basis that a weakened falsificationism is still stronger than a weakened inductivism because it is closely related to deduction.
An important consequence of this methodological falsificationism is that all observation statements must be considered revisable. We cannot, therefore, “disprove†a hypothesis, we can only reject it. A critic who wanted to preserve a strict criterion of epistemological justification would object that Popper has therefore failed. But Popper would reply that methodological falsificationism is pragmatic, in accordance with scientific method. If we are to reject induction and preserve the empiricist basis of scientific method, Popper suggested, this is the only route available.
If we accept, for the moment, Popper's framework for the problem of demarcation – providing a sound, pragmatic basis for scientific method – we still find methodological falsificationism wanting. Take the probabilistic methodology, for example. If we have a hypothesis that in a certain experiment a result will occur at a rate of 75%, and the experiment is accepted as a convention, then we can falsify statement of the form: “the rate of occurance is greater than 80%†and “the rate of occurance is less than 70%†and thereby deduce that the rate is 75±5. But according to Bird, this is simply a strong form of induction, which means that Popper's criterion of demarcation relies on induction “when the going gets tough†(Bird, p. 180)
Popper might reply that the observations corroborate the hypothesis and indicate a proximity to the truth, whereas induction must always confirm a truth. Popper is therefore giving a pragmatic basis for preference on top of his criterion for demarcation. One could say then that falsificationism provides a strong rational basis for science, from which methodological falsificationism can provide a basis for scientific method. Bird, however, isn't convinced.
Popper formulated this proximity to the truth, which he called verisimilitude, in the following way: We have two propositions, A and B, which are close to the truth, proposition T. Assume that A is closer to T than B, that A is greater than B, and that T is greater than A:
1.(T – A) > (T – B)
2.A > B
3.T > A
In all formulations of A, B and T, A will have more true consequences than B, and so can therefore be said to have a greater verisimilitude. In other words, “the true consequences of B are included among the trust consequences of A; and the false consequences of A are included among the false consequences of B†(Bird, p. 244). This provides a rational answer to the problem of probabilistic statements, but only where the consequences of the statements in question are comparable. Where hypotheses predict different kinds of consequences, this formulation cannot provide any indication of relative verisimilitude. This means that the criterion of demarcation, upon which this notion is based, is again reduced to an extremely strict, dogmatic falsificationism that cannot provide a rational basis for scientific method.
There is one further worry for falsificationism in general. The method depends upon scientists starting with a set of hypotheses and subjecting each to scrutiny, whittling the body of hypotheses down and thereby making progress. But scientists evidently create more hypotheses as they work, expanding the body rather than reducing it. And if we accept the linguistic assertion that language is creative, then there are, theoretically, an infinite number of hypotheses that scientists can try to falsify. So in terms of methodology, Popper can offer little in the way of insight; he claims that “coming up with a hypothesis is an exercise of the imagination†(Bird, p. 239), And as a criterion of demarcation, it can only say what is and what isn't scientific by reference to the possibility of experimental testing with available methods; in other words it cannot mark a limited body of hypotheses as constituting the body of scientific knowledge, nor even of “scientific†knowledge.
So to conclude, Popper's falsifiability criterion of demarcation seems attractive insofar as it rejects inductivism and accords to an extent with scientific method. But both as a criterion of demarcation, and as a basis for a scientific methodology – something it ought to be able to provide – it seems to either fall back on inductivism or to undermine the place of empiricism in science. The most it can offer is a new category of “science†that is pragmatic but philosophically tangential to science.
A.Bird, Philosophy of Science, (London: Routledge, 2003)
I.Lakatos, The Methodology of Scientific Research Programmes: Philosophical Papers, Volume 1 (Cambridge: Cambridge University Press, 1978)
W.H.Newton-Smith, A Companion to the Philosophy of Science, (Oxford: Blackwell, 2001)
Scientific realism begins with the assumption that induction alone is suspect, and that the ability of a hypothesis to explain empirical facts strengthens it. This is called abductive inference. Realists can go further to suggest that if a particular hypothesis, H, explains a set of facts, F, better than any competing hypothesis then this could warrant belief in the resulting theory.
Objections to this idea are numerous and varied. The most obvious and pressing objection, proposed by Mill when he was arguing against wave theories of light in the nineteenth century, is as follows: abductive inference is used where the subject matter isn't experientially accessible, e.g. to predict that a particular phenomenon is causing various effects even though that phenomenon cannot be observed. As such, no matter how important and useful the theory, and no matter how bad the rival theories, it cannot be accepted as a basis for belief. Hypotheses can only be of instrumental use for scientists in trying to predict outcomes and construct models.
Instrumentalism therefore states that all theoretical language used in abductive inference is itself instrumental, bearing no descriptive or referential properties (to observations). Where observation statements are insufficient to support a theory, instrumental propositions can be used, and judged according to their efficacy, however that is to be decided.
1. Quine - Any given body of evidence can have more than one instrumental explanation (Thesis of Empirical Equivalence). Therefore no body of evidence supports one theory to the exclusion of all others, and there are no epistemic grounds on which one can decide between equivalent theories. The only move open to realists is to choose between theories on a purely pragmatic basis, which becomes the domain of scientists not philosophers. In other words, abductive inference ceases to have any epistemic import, and becomes a mere methodology based on pragma.
2. van Fraasen - If we rely on pragma as a basis for analysis of explanation then we confront the problem of context. Our hypothesis, H, may have explanatory power in one context and none in another, e.g. because of factors unaccounted for in the hypothesis. We have no objective perspective from which to judge when a hypothesis is successful in this respect, and so H can only be considered instrumentally useful given a specific context, including our interests and the direction of its arguments. Worse still, a change in context may reverse the status of explanation and outcome, such that we cannot say, without context, what has explanatory power. The result of this is that, decontextualised, explanation has no epistemic import, making explanatory status itself pragmatic.
3. Laudan - Any explanatory virtues may have failed in the past, or may support theories that are incompatible with current science. Thus, though a theory with greater explanatory scope and power than any other may be instrumentally useful, this success cannot, for historical reasons, warrant belief. Only observations can be believed. Again, this extreme form of instrumentalism is fine for scientists, but still gives us no reason to believe the outcome of science except where theories are based wholly on observations.
4. Fine - There are two levels of reasoning involved here. On the first level we have scientific reasoning, on which theories are proposed and evaluated in light of experimental evidence and, according to realism, abductive inference. On the second level we have philosophical reasoning, where general theories about the epistemic status of the conclusions reached on the first level are evaluated. If philosophical theories purport to judge the epistemic import of modes of inference used in science then it cannot use those modes itself, and the methods of philosophy must be deemed stronger. However, abductive inference violates this rule by legitimising itself according to it's own criteria -- it states that it has epistemic import because it is successful. It is therefore a circular argument.
One might reply that this shouldn't concern us. After all, any epistemic theory will itself need to be justified, and we could adopt a stance similar to coherentism or reliabilism and say that the success of any given theory has epistemic import.
An essay I wrote on relativism in Kuhn's theory of scientific revolutions.
Thomas Kuhn developed a theory of science in his book The Structure of Scientific Revolutions. It has a hostile reception amongst philosophers, especially those who advocated an objective standard or framework for science. Philosophers like Popper and Lakatos attacked Kuhn for developing an unacceptable relativism the undermined their preferred conception of the discipline. Ultimately they are right insofar as they claim it is relativistic, but I don't think it constitutes an unacceptable relativism. Rather I sympathise with those who claim that his relativism did not go far enough.
Kuhn's central thesis is that science operates in a cycle of revolutions with three phases. In the first pre-scientific phase, scientists work with several incompatible and incomplete theories. Eventually one theory becomes accepted by enough scientists that it becomes dominant and so scientists begin to use it as their basic methodology. Kuhn called this dominant framework a paradigm, borrowing from the linguistic term meaning systematic arrangement of all the inflections of a term. Scientists then practice in the second phase, which Kuhn called normal science. In this phase, scientists engage in puzzle solving, trying to expand the paradigm. Inevitably, anomalies will arise that cannot be explained with any theory that would fit the dominant paradigm. They will generally be dismissed or put to one side, until they gain sufficient weight for a crisis of revolutionary science. During this phase the paradigm and its axioms can be re-examined, leading to a new paradigm that has all of the explanatory power of its predecessor in addition to the ability to explain some or all of the anomalies. The change to a new paradigm was called a paradigm shift.
The obvious consequence of this theory is that science has no permanent, objective methodological or semantic standards. It rejects standards like falsification and confirmation, and posits social context – the struggle for dominance in the scientific culture – as the ultimate arbiter of scientific knowledge and progress.
The first problem that this raises relates to a definition of science. How one can distinguish science from other intellectual enterprises, since there is nothing inherent in the methodology or meaning of the discipline? Kuhn might reply that science is simply that which studies the natural world, or even the discipline that can be defined by the methodology of revolutions, i.e. by reference to the phases of pre-theory, normal science and revolutionary science. Positivists like Popper would object to this, since it allows for fields like psychology and sociology to be counted as scientific.
In fact, many other enterprises could conceivably involve these phases, including philosophy itself. Kuhn later revised his theory to admit the possibility of normal and revolutionary science occurring in parallel, which in turn allows for a narrow interpretation of almost any intellectual enterprise as being scientific. If I settle upon one theoretical framework and try to solve philosophical problems within that perspective, am I not practicing normal science? And if I then reject that framework and develop a new one that exceeds its predecessor's explanatory power, am I not undergoing a paradigm shift?
It would seem that Kuhn can only resort to defining science by reference to its subject matter – the natural world – but that itself is problematic due to the incommensurability of terms, a consequence of Kuhn's theory.
According to Kuhn, we judge a theory by reference to a paradigmatic theory, rather than to some permanent, theory-independent rules (such as whether or not it is falsifiable). This has several important consequences. First, scientific theories and methodology are relative to their paradigm, there is nothing permanent about the discipline, as I have explored briefly already. Second, theories from different paradigms are incommensurable since they have no shared paradigmatic theory against which they can be judged. I cannot easily compare Newton's theory of gravity with Einstein's since the two theories can only be explained by reference to their paradigm. I am unable, in other words, to escape any given paradigm and compare two paradigmatically different theories. This is methodological incommensurability (Bird, 4.1).
Furthermore, scientists must suffer observational incommensurability. Observation, Kuhn suggested, is theory-dependent as it is influenced by prior beliefs based in the current paradigm; scientists in different paradigms will perceive the same event differently because of the theoretical framework through which they view the world (Bird, 4.2). So when a scientist in a pre-Newtonian paradigm sees an apply fall to the ground, he may observe an apple moving towards the ground. A scientist observing the same phenomena in a post-Newtonian paradigm will see an apple under the force of gravity changing its inertia. The two observations, Kuhn says, cannot be compared. So one cannot even rely on the evidence of previous paradigms, it must be reinterpreted. It may even have to be re observed, if one takes this form of incommensurability seriously, since we have no reason to believe the observation statements of scientists from different paradigms.
Finally, scientists must contend with semantic incommensurability. Fields of science, Kuhn said, are governed by taxonomies that divide its subject matter into kinds, with an associated lexical network (of terminology). Scientific revolutions involve a change in meaning of terminology, and since Kuhn thought that meaning was holistic this would involve a change in the whole lexical network and therefore also the taxonomy. So terminology and taxonomies cannot be translated between paradigms (Bird, 4.4).
A realist might then object that if scientific theories are potential descriptions of the world “involving reference to worldly entities, kind, and properties†(Bird, 6.2) then how can we take them to be incommensurable? The sense (the meaning) may change, but the reference cannot. They still refer to same concepts, even where the understanding is different (e.g. Newton and Einstein and their understanding of “massâ€). To a realist, one understanding is simply closer to the truth and so the two different understandings are still commensurable, even if only in reference and not in meaning.
Kuhn might reply that with different conceptualisations of the world that describe the same phenomena, the references may be the same but the scientist cannot interpret or understand these independent of his subjective theoretical framework – i.e. depending upon the sense of the terminology. One cannot say that mass refers to the same thing for Newton and Einstein since one can only understand the term “mass†from one of their theoretical frameworks.
The realist would content that natural kind terms are sense-free, particularly causal theories of reference. For example, two theories that explain how gravity causes that apple to fall towards the earth are commensurable because they refer to kinds that are wholly objective as far as scientists are concerned (we have no scientists who can dispute terms like “apple†and “towards the earthâ€). Though the observations may be different insofar as they are embedded in a theoretical framework, the natural kind terms are objective. Similarly natural kind terms like “mass†and “acceleration†change in sense but not in reference (Bird, 6.2).
Kuhn might finally reply that this only shows that reference can be retained, not that it must be, so one can still defend an extremely relativistic conceptualisation of scientific revolutions as being possible if not necessary. In other words, there is nothing necessarily objective about science, and so if its objectivity is contingent then Kuhn can claim that his study of the history of science supports his theory of the structure of scientific revolutions.
Furthermore, for a philosopher or scientist to be aware of any such continuity one would need to posit some external guarantor of truth so that he one could observe the continuity within the current paradigm (i.e. independent of its standards of evaluation). This could either be what Kuhn termed an “Archimedean platformâ€, upon which the philosopher can compare two theories without any theoretical framework that might bias his analysis. Or it could be an epistemic standard such as reliability, according to which the philosopher could judge how science has improved or degenerated in between paradigms.
Kuhn's arguments seem convincing. For while it may be useful to think of science as being a cumulative discipline that adheres to strict, objective standards, the failure of Kuhn's adversaries to posit any convincing account of such a set of standards suggests that science is indeed more humanised than they want to think.
Paul Feyerabend took this point further to suggest that it is good that scientists aren't restricted by any particular methodological, observational or semantic framework, since such a framework would limit the activities of the scientist and hence restrict scientific progress. Although he took issue with Kuhn's strict delineation of normal and revolutionary science, he thought that scientists engaged in puzzle solving could not adhere to the truth. Rather, scientists ought to operate in a kind of methodological anarchism, so that they are free to pursue any line of thought they choose (Preston).
Strictly speaking Feyerabend rejects Kuhn's model of scientific revolutions outright, dismissing it as simplistic. In fact, he contends that philosophers have nothing useful to say about scientific methodology, and advised scientists to ignore philosophers. But as a defense against those who attack Kuhn's relativism, Feyerabend provides compelling arguments. He idealised Kuhn's descriptions of the pre-scientific and revolutionary phases of science, and claimed that all good science occurred during these phases. He therefore rejected outright the phases in which there is any continuity, and so denied not only incommensurability between paradigms but also between theories.
Even if one rejects Feyerabend's extreme incommensurability thesis, and even if one rejects Kuhn's particular structure of scientific revolutions, one can still preserve his theory of incommensurability between paradigms. His detractors have failed to develop a convincing account of how theories from different paradigms can be compared, and so I must conclude that Kuhn's portrayal of science doesn't involve an unacceptable relativism.
Alexander Bird, 2004, Thomas Kuhn, entry in the Standard Encyclopedia of Philosophy: http://www.seop.leeds.ac.uk/entries/kuhn, on file with author
W.H.Newton-Smith, 2001, A Companion to the Philosophy of Science, Orxord: Blackwel
John Preston, 2002, Paul Feyerabend, entry in the Standard Encyclopedia of Philosophy: http://www.seop.leeds.ac.uk/entries/feyerabend, on file with author