Guest post by Justin Cruickshank on critical realism

Justin Cruickshank is senior lecturer in Sociology in the School of Govt and Society at the University of Birmingham. He researches and teaches in the areas of classical and contemporary social and sociological theory as well as the philosophy of the social sciences. His books include Critical Realism: The Difference That It Makes (edited 2004) and Realism and Sociology: Anti-Foundationalism, Ontology and Social Research (2003), as well as many articles and chapters on critical realism. Here is his profile at UB. This contribution is a response to a lively recent discussion here over the status of transcendental ontology in critical realism, including two posts on Cruickshank's critique of Bhaskar (link, link) and comments and criticisms offered by Dave Elder-Vass, Mervyn Hartwig, and Ruth Groff (link, link). My own contributions to the debate include (link, link). Thanks for contributing, Justin.

Reply to Hartwig and Elder-Vass

By Justin Cruickshank

I would like to thank Daniel Little for inviting me to contribute to this dialogue about the fallibility of critical realism. I’d like to start by quickly considering the philosopher who foregrounded the importance of fallibilism.

Popper counseled against asking ‘what’ questions, in favour of asking ‘how’ questions. For Popper, questions about how phenomena interact were subject to a critical dialogue, whereby fallible categories could be revised through the course of problem-solving. With this approach the recognition of fallibilism led to the claim that justification had to be eschewed and replaced with criticism. As regards ‘what’ questions, which required a definition of reality, Popper objected that such questions return us to the search for justification (contrary to the emphasis on criticism required by the recognition of fallibilism) and did so in a way that ultimately entailed dogmatic metaphysical speculation.

One could argue that Bhaskar’s critical realism avoids being a form of dogmatic metaphysical speculation because, having assumed that science worked, Bhaskar drew out the ontological assumptions from within science. The knowledge produced by science was referred to as a transitive dimension because it, like all knowledge, was held to be fallible, whereas reality itself was referred to as the intransitive dimension. Ontological definitions would therefore not be based on dogmatic metaphysical speculation about the intransitive dimension. Instead the role of philosophy was to render explicit the implicit ontological assumptions within the transitive dimension. In doing this, philosophy would use science to furnish the condition of possibility of science: philosophy would set up a transcendental question and answer that using the ontological assumptions implicit within scientific knowledge. This is the basis of the claims by Elder-Vass and Hartwig that the transcendental argument put forward by Bhaskar is not dogmatic ‘old style’ metaphysics but ‘conditional’, ‘relative’, ‘corrigible’, based on a dialogic approach, etc. In place of dogmatic certainty there would be a recognition of fallibilism. The problem, though, is that the recognition of fallibilism becomes redundant because critical realists are concerned with justification. We can explore this in the three points below.

First, the act of rendering the ontological assumptions explicit would be a fallible interpretive act and so other philosophers influenced by realism may interpret the transitive domain differently. Bhaskar (The Possibility of Naturalism, 3rd edition, p. 170), stated that his ontology ‘at present’ is ‘uniquely consistent’ with the ontological assumptions within science. This raised the question as to how to judge between competing interpretations of the ontological assumptions within science so that one may be in a position to know that a particular philosophy is uniquely consistent with the ontological assumptions of science. For just as different philosophers of science have re-read the history of science to discover that their methodological prescription was implicitly adhered to, so different realist philosophers could read the practice of science in such a way as to read in the ontological assumptions that they took to obtain within science. As those assumptions are metaphysical, specific empirical theories can be read to fit the postulated assumptions.

Underpinning this is the problematic attempt to link a commitment to fallibilism with the attempt to justify a philosophical position. The attempt to justify the position will lead to a monologic exchange because there would be no basis for a critical dialogue. That is, there would be no common framing of the problem or the criteria for its solution. A commitment to justification would lead to different metaphysical schemes being justified by being read into the practice and history of science, with there being no empirical test to decide between them and no logical test (if they were all internally coherent). Consequently the commitment to fallibilism would be rendered redundant, in terms of any critical dialogue over the ontological assumptions taken to obtain in the transitive domain. To be sure, claims to infallibility may clearly be eschewed. Thus Bhaskar holds that ‘at present’ his reading happens to be the superior one. However, there is a difference between not endorsing infallibilism and putting any recognition of fallibilism to work.

Second, if we assume for the sake of argument that the ontological assumptions rendered explicit by Bhaskar were uniquely consistent with those that obtain within the transitive dimension, then the question arises as to what philosophy could do with those assumptions. Scientific theories would produce explanations based on a set of ontological assumptions that were implicit within the practice of science, with those assumptions being the condition of possibility of science. Therefore it is hard to see how the philosophy of science could become more than the history of science, tracking the development of those assumptions, because it would lack any normative force motivated by extra-scientific criteria.

It could be countered that philosophy did have a normative role to play, which was to ensure that science did not err by turning to the wrong ontological assumptions. However, as it is conceded that science is fallible and its ontological assumptions are fallible, then a change in itself does not necessarily mean error. Instead, in order for philosophy to act as an underlabourer, it would need to distinguish correct from incorrect ontological assumptions independently of their manifestation in science. In this case, fallibilism would be eschewed for old style metaphysics in order for the underlabouring claims to be justified. So, if we accept that science furnishes its own conditions of possibility, then philosophy becomes redundant, and if philosophy is to avoid this, the justifications for its prescriptions would avoid any recognition of fallibilism. One could counter and argue that if science was influenced by the outmoded positivist conception of science then philosophy could intervene. However, if a practice was based on incorrect ontological assumptions then it would fail to be science, and science would be self-regulating by eschewing approaches based on ontological assumptions deemed incorrect.

Third, standard transcendental arguments are universalist, and Elder-Vass and Hartwig clearly reject any notion of Bhaskar’s transcendentalist position being universalist, because it would lead, in this case, to old style metaphysical dogmatism. However, the alternative approach to transcendental arguments, which holds that the current and fallible ontological assumptions furnish the current condition of possibility of current science, leads to a Kuhnian conception of science. Here the ontological assumptions would define a period of science in a fashion analogous to Kuhn’s paradigms, with empirical work based on those assumptions being a matter of puzzle-solving. That is, the empirical explanations would be justified by being in conformity with the ontological assumptions, with those assumptions delimiting the range of acceptable explanations. Whereas a problem-solving approach may allow for the revision of implicit ontological assumptions, this puzzle-solving approach would be narrower in scope.

The recognition of fallibilism here would amount to the rejection of infallibilism in the form of dogmatic metaphysical speculation, whereby the assumptions were taken to be definitely correct. However, the recognition of fallibilism would do no more than that. After that the emphasis would be on regarding current empirical explanations as being justified by being in conformity with the current prevailing ontological assumptions. Given this approach to justification, the concept of epistemic progress becomes problematic. Under paradigm A, empirical claims would be justified by being in conformity with the ontological assumptions that furnished the condition of possibility of paradigm A. Under paradigm B, empirical claims would be justified if they were in conformity with the ontological assumptions that furnished the condition of possibility of paradigm B. Fallibilism could be appealed to as a denial of infallibilism with the ontological assumptions not being taken to be definitively correct definitions of reality; but it would do no work after that, with the emphasis being on justification. Dogmatic ‘external’ justification, in the form of an appeal to a definitively correct definition of reality would be replaced by ‘internal’ justification, in the form of an appeal to the current assumptions that justified the current phase of science.

Not only does this make fallibilism redundant, but it also makes any notion of progress problematic, given the emphasis on internal justification. In other words, there are no philosophical or extra-scientific criteria to appeal to, in the attempt to judge one paradigm as better than another. Such an approach also returns us to the problem of dogmatism, because as there are no philosophical / extra-scientific criteria to appeal to, each phase of science would have to rely on conservative justification – conformity to the assumptions would lead to justification and conversely a lack of conformity would negate any justification for an explanation. It may be pointed out that explanations which conform to the ontological assumptions may fail. This is true, but Kuhn recognised this too, and the issue would be that empirical explanations which were taken to be successful would be deemed justified because of their conformity to the current assumptions that defined current science.

In order for critical realist philosophy to do any work in this context it would have to make an appeal to some form of extra-scientific criteria, by turning to universalism, and holding that a set of ontological definitions did correspond to the defining features of the intransitive domain. As has been noted in the posts already, there are places where Bhaskar makes such claims. Like most philosophies there is the place where it is asserted and the place where it is retracted. Bhaskar does engage in old style metaphysics, but his rowing back does not save his philosophy. Accepting the fallibilist reading of his ontology shows that fallibilism becomes redundant because the emphasis swings to justification and, in the process, Bhaskar’s philosophy becomes redundant.

Applying this to the social sciences, there could be a post-Marxist science of structures which was taken to be justified because its explanations were taken to be based on a definitively correct ontology of social structures; or a Kuhnian conservative approach that made social science scientific by supplying some fundamental assumptions to agree upon. The forms of justification would be quite different in both, but in neither case would fallibilism do any work. A problem-solving approach to the sciences may be a better way to go.