The latest edition of Synthese is dedicated to “Evolution and its rivals” and includes a contribution by Robert Pennock entitled Can’t philosophers tell the difference between science and religion?: Demarcation revisited (originally published in 2009). This piece is highly critical of Larry Laudan, particularly his famous paper The demise of the demarcation problem, and in this entry I look at some of Pennock’s objections. There are already some good responses available, including Brad Monton’s previously and some recent comments from Brandon Watson, and – since both have mentioned it – in the remarks below I ignore the rhetoric employed by Pennock. I also restrict my attention to his criticisms of Laudan, such as they are.
Pennock’s main protest is that “Laudan’s entire critique of demarcation… expects a precise line that can unambigiously rule any possible theory in or out of science”. Pennock argues that what we actually need is “what might be called a ballpark demarcation that simply identifies a position as violating a basic value, or ground rule… Showing that creationism is not science requires no more complicated notion of demarcation than that – it violates a scientific ground rule and is not even in the ballpark”. For Pennock, this necessary condition is methodological naturalism.
What Laudan actually wrote (in The demise of the demarcation problem) is that
… a philosophical demarcation criterion must be an adequate explication of our ordinary ways of partitioning science from non-science and it must exhibit epistemically significant differences between science and non-science. Additionally… the criterion must have sufficient precision that we can tell whether various activities and beliefs whose status we are investigating do or do not satisfy it; otherwise it is no better than no criterion at all.
Laudan goes on to explain that merely necessary or sufficient criterion are not enough to accomplish demarcation. If we are able to provide a necessary condition then we can use it to establish that particular activities are unscientific, but it does not allow us to say that something is scientific; after all, if a claim has merely satisfied a necessary condition, we are only in a position to say that it might be science. Similarly, if we can find sufficient conditions then these would allow us to say that a claim is scientific, but not that it is unscientific. What we want to do, especially with ideas like Intelligent Design (ID), is say both that something is scientific and that something else is not. Pennock’s insistence that a ground rule of methodological naturalism is enough does not address Laudan’s claim that both necessary and sufficient conditions are needed.
Pennock quotes Laudan (now in the latter’s Science at the bar – Causes for concern) saying that “[claiming] creationism is neither falsifiable nor testable is to assert that creationism makes no empirical assertions whatever. That is surely false”, remarking that this “is certainly a strange statement for someone to make in a discussion of the definition of empirical science who has just rejected both of these criteria for just that purpose.” However, Laudan has not “just rejected both of these criteria”: he has accepted them and used them as the basis of a reductio. His argument against the testability criterion is actually very clear: he allows it and then points out that Creationist claims can and have been tested. The fact that they have failed these tests (and therefore been falsified) is what is actually important, since this speaks in favour of evolutionary theory and against Creationism. That they were testable and thereby falsifiable means that they satisfied these demarcation criteria and hence Creationism was, in this sense, scientific, and so the reductio succeeds.
It is thus not the case that “Laudan’s strategy in criticizing Ruse’s five criteria and presumably any other demarcation criteria was to find counterexamples from the history of science” because his strategy for the first two is a reductio. His non-reductio objection to testability, made later in Science at the bar, is that it is “now widely acknowledged that many scientific claims are not testable in isolation, but only when embedded in a larger system of statements”, which he allows is the case for “some tenets of Creationism” and which Pennock does not address. However, Laudan’s argument is plain: the way to make testability important is to emphasise that Creationism has been tested and has failed the tests, not to insist that the more important point is that claims must be testable. The former requires Creationists to show that the evidence for their claims is compelling and better than that for evolutionary theory, while the latter only requires them to make claims that are testable – regardless of whether the claims fail the test – in order to be granted the status of scientific. Even if Laudan was wrong that Creationism was testable and had failed the tests (that is, that Pennock is right that Laudan had “failed to take into account [its] real supernatural content”), it would not impact upon his objection because it would still make the point – to the Creationists – that offering a testable claim is more important than passing the test.
Laudan re-emphasises this in his reply to Ruse (in More on Creationism), in which he states again that “[o]nce we make [the separation between 'the soundness of creation-science' and 'the [supposed or actual] dogmatism of creationists’] we discover both (a) that creation-science is testable and falsifiable, and (b) that creation-science has been tested and falsified – insofar as any theory can be said to be falsified”. Laudan’s warning in his reply was that if we nevertheless insist on making “extremely weak demands from an epistemic point of view” to demarcate science from non-science then “it would be child’s play for creationists to modify their position slightly – thus making their enterprise [by Overton's lights] ‘scientific’”. It seems that this is exactly what happened with Creationism morphing into ID, assuming we accept the arguments of those opposed to ID that it is related to Creationism in this way.
This is why Laudan’s caution is often pointed to by those who wish to remark that attacking ID on the philosophical grounds of the demarcation problem instead of in terms of confirmation or as a research programme is ill-advised. No doubt it is a source of some annoyance to opponents of ID that Laudan’s comments have been seized upon by its advocates, but this approach only succeeds insofar as the debate remains fixated on the question of whether or not Creationism and/or ID are scientific. Perhaps the correct response to such disagreements is to point out that, if we read the entirety of Laudan’s The demise of the demarcation problem, we find that there is no safe place in it for Creationism or ID: Laudan states repeatedly that ideas should not be disallowed via an initial demarcation but should be assessed in terms of how well-founded they are, on which grounds Creationism and ID fail. It can thus be conceded that the demarcation problem may be intractable without allowing that this makes life any easier for Creationism and ID. If this proves difficult in terms of what should be allowed in a science classroom then perhaps it requires that we frame that debate differently; those who attempted to settle the question via demarcation, or who allowed Creationism and ID advocates to do so, are responsible for this situation, not Laudan.
The irony is that Pennock insists on just this point himself in addressing the “dustbin of history argument”. Defending the view that “creation-science does not qualify as science” even if Laudan’s claims about their testability are granted, he points to the example of geocentrism: Pennock says that although “such a claim was historically scientific or even that it remains scientific in the abstract sense that it is testable, it would nevertheless be fair to conclude, because this claim has been decisively disconfirmed (at least under the assumptions of MN [methodological naturalism]), that it is unscientific to continue to hold and teach it today”. This, of course, is exactly the kind of approach to demarcation that Laudan is advocating: the problem with geocentrism is not that it is unscientific but that it has been disconfirmed and other theories are better confirmed. It is conceivable that this might one day change, just as the fortunes of heliocentrism did, but for the time being we discount geocentrism for reasons other than it failing demarcation criteria.
Another issue Pennock takes Laudan to task for is “the prima facie absurdity of Laudan’s claim that searching for demarcation criteria is a pointless pseudo-problem”, but Laudan does not dispute that we can and do attempt to demarcate. What he does argue in The demise of the demarcation problem is that since “the labeling of a certain activity as ‘scientific’ or ‘unscientific’ has social and political ramifications which go well beyond the taxonomic task of sorting beliefs into two piles”, therefore “it would be wise to insist that the arguments in favor of any demarcation criterion we intend to take seriously” should be as compelling as those for the practical actions we plan to base upon it. Pennock is obviously correct that people distinguish between science and religion, just as they do between science and football; the issue is whether our grounds for doing this are actually compelling or not. The fact that some scientists are happy to demarcate does not speak to the soundness of the criteria they use (if they use any) and this problem of demarcation is a philosophical one whether practised by scientists, philosophers or anyone else; what matters, in terms of the demarcation problem, is that the criteria stand up to philosophical scrutiny. Laudan states plainly that he is “not denying that there are crucial epistemic and methodological questions to be raised about knowledge claims”, nor that “we are never entitled to argue that that a certain piece of science is epistemically warranted and that a certain piece of pseudo-science is not” – these clarifications come immediately after he calls demarcation a pseudo-problem. The important point is that after we have investigated whether a theory is well confirmed and explored it in detail, the question of whether or not it satisfied demarcation criteria initially no longer matters.
Pennock argues that the demarcation used in practise by scientists is the “ground rule” of methodological naturalism that he has advocated. He observes that critics often point to examples from the history of science, such as Newton, to show that science and scientists were not always methodologically restricted by naturalism, but he does not seem to appreciate why this objection is raised. The key issue with methodological naturalism is that it is not an a priori criterion that has any impact upon Laudan’s argument; it is an a posteriori principle adopted because of its success. That most or even all scientists use it, explicitly or otherwise, does not change this: if methodological naturalism is successful then we would expect it to be employed, but this does not mean that everyone – including ID advocates – needs to assume it. Indeed, if people choose not to and then fail to achieve any results, while others stick with methodological naturalism and succeed, then this is more justification for methodological naturalism as an a posteriori principle, notwithstanding that it does not bring us any closer to making it an a priori one.
This is why it is absurd to speak of a claim “violating” methodological naturalism because it swaps and confuses the a posteriori for a priori, as if we can infer that a currently successful principle will always be so and should fix scientific practice. Moreover, only by allowing the investigation of alternatives can we determine if our methodology needs to change. If it turned out that a research programme not employing methodological naturalism achieved results then this would be a problem for methodological naturalism, not the research programme; this would give some justification for working without methodological naturalism in future but we would still have the many other successes continuing to speak in favour of it. Precisely what we mean by success in this context is what Laudan is pointing to by insisting that we should be asking what makes a belief well-founded, rather than whether we can call it scientific or not prior to undertaking the investigation.
Finally, I want to comment briefly on Pennock’s reference to Feyerabend and his values-based demarcation criterion, which I have already written about here and which Pennock approvingly calls “practice-based”. Leaving aside the claim that Feyerabend defended “epistemological anarchism and claim[ed] that in science ‘anything goes’”, a common misconception that I discussed in a short essay here, Pennock points to Feyerabend’s criterion as an example of a philosopher of science (“even someone like Feyerabend”, no less) not agreeing with Laudan that demarcation is a pseudo-problem. However, this values-based approach was never intended to solve the demarcation problem or provide any kind of ground rule. Instead, it characterises Feyerabend’s insistence on the principles of tenacity and proliferation, and it is interesting to note the sections of Feyerabend’s claim that Pennock left out. Immediately before the quote of Feyerabend’s view that cranks do not undertake further investigation, Feyerabend had insisted that “to declare that only plausible theories should be considered is going too far” and that “we never know in advance which theory will be successful and which theory will fail”. For Feyerabend, the distinction between respectable people and cranks therefore “lies in the research done once a certain point of view is adopted”. Even a cursory reading show that this does not permit us to demarcate science from pseudoscience and Feyerabend had no interest in so doing; after all, he had already raised “[t]hree cheers to the fundamentalists in California who succeeded in having a dogmatic formulation of the theory of evolution removed from the textbooks and an account of Genesis included” in his lecture How to defend society against science (another frequently misunderstood remark). What it does do is point to exactly the issue Laudan raises with regard to demarcation: we cannot rule out theories ab initio because of their implausibility relative to our existing ideas, even though we tend to justifiably discount the claims of people who refuse to put them to the test or develop them further.
Ultimately, if someone wishes to investigate ideas like Creationism or ID then we will soon see if they have anything to recommend them if they prove well-founded and become well-confirmed. If the advocates of these ideas will not develop them further then we can ignore them, and if they will then we can make a subsequent assessment – no one needs to demarcate a priori. This is why Pennock has missed Laudan’s argument and why it remains the case that, as Laudan concluded,
[... if we want to] protect ourselves and our fellows from the cardinal sin of believing what we wish were so rather than what there is substantial evidence for (and surely that is what most forms of ‘quackery’ come down to), then our focus should be squarely on the empirical and conceptual credentials for claims about the world. The ‘scientific’ status of those claims is altogether irrelevant.
Feyerabend, P. (1981 ). Realism and Instrumentalism in Realism, Rationalism, and Scientific Method: Philosophical Papers, Volume 1. Cambridge: Cambridge University Press.
Laudan, L. (1982). Science at the bar—Causes for concern in Science, Technology, & Human Values, 7(41), 16–19.
Laudan, L. (1983). More on creationism in Science, Technology, & Human Values, 8(42), 36–38.
Laudan, L. (1983). The demise of the demarcation problem in R. S. Cohen & L. Laudan (Eds.), Physics, Philosophy, and Psychoanalysis (pp. 111–127). Dordrecht: Reidel.
Pennock, R. (2010). Can’t philosophers tell the difference between science and religion?: Demarcation revisited in Synthese, Volume 178, Number 2, 177-206.