Miriam Solomon on Medical Epistemology

Miriam Solomon has recently published a book on Medical Epistemology (Making Medical Knowledge, Oxford University Press, pp. 261 + xvi).  This gives a good critical exposition of trends in the last 40 or so years.  There are three chapters on Consensus Conferences, two on Evidence-Based Medicine, one on Translational Medicine, and one on Narrative Medicine.  Solomon devotes the last chapter of her book to her own position which she characterises as a ‘Developing, Untidy, Methodological Pluralism’, though, as she points out on p. 208, this has the unfortunate acronym: DUMP.

Miriam Solomon’s book is certainly to be recommended to anyone involved in the philosophy of medicine.  The purpose of this blog, however, is to point out to its readers that she gives an inaccurate account of the EBM+ position.  Now of course members of the EBM+ consortium have different opinions on many issues.  However, there is a broad area of agreement, and this is given in two multi-authored papers, namely:

Solomon refers to the first of these papers, which is very short, but not to the second longer paper.  Regarding the first paper she says (p. 120):

Although there are differences between Howick (2011b), Andersen (2012), and Clarke et al. (2013), they all share the assumption that mechanistic reasoning should be regarded as mechanistic evidence, moreover, evidence that has a place in the hierarchy.  This is an assumption that I challenge.

As a matter of fact we (i.e. Clarke et al.) do not share the assumption here stated.  This is an assumption, which is held by mainstream EBM, and also by Howick (2011b) who follows the mainstream EBM in this respect.  However, it is an assumption, which we criticize.

First of all we do not regard mechanistic reasoning as form of evidence, which it patently is not.  For us, evidence is empirical evidence, i.e. the result of observation and experiment.  It is the EBM movement, which places mechanism-based reasoning in their evidence hierarchy and so presumably regards mechanism-based reasoning as a form of evidence.  We never use the phrase ‘mechanism-based reasoning’ except when quoting from EBM supporters, and when we do so, we put the phrase in inverted commas to indicate that it seems to us an inappropriate one in the context.  Instead, we systematically use either evidence of mechanisms or mechanistic evidence.  In our longer (2014) paper, when quoting from the Oxford Centre for Evidence-Based Medicine, we add in footnote 2, p. 341:

‘Mechanism-based reasoning’ seems to refer roughly to what we call evidence of mechanisms.

The implication here is that ‘mechanism-based reasoning’ is, in this context, an inappropriate expression, which we think should be replaced by evidence of mechanisms.

We also do not agree with the second part of the assumption attributed to us by Solomon, namely that “mechanistic evidence … has a place in the hierarchy.”  This part of the assumption is indeed held by some followers of EBM, but our view is that (2013, p. 747):

Evidence of mechanisms and evidence of correlation are complementary: each addresses the primary weakness of the other.

In our longer 2014 paper, we illustrate this by the analogy to reinforced concrete (p. 351); and in our diagrammatic representation in Figure 1, p. 355, evidence of mechanisms is treated alongside evidence of correlation rather than as part of the same hierarchy.

While Miriam Solomon does not give a correct account of our position, some of her remarks about mechanistic reasoning seem to me to advance the subject.  As explained, we regarded ‘mechanism-based reasoning’ as just an inappropriate way of referring to what is more correctly described as ‘evidence of mechanism’.  We do not discuss ‘mechanism-based reasoning’ any further than that.  However, Solomon points out that, if ‘mechanism-based reasoning’ or ‘mechanistic reasoning’ is regarded not as a form of evidence, but as a form of reasoning, it may be a useful concept.  She analyses this concept on pp. 122-4 of her book, and makes a number of interesting points, saying, for example (pp. 123-4): “ … mechanistic reasoning … has important non-evidential roles.  In particular, I will argue, mechanistic reasoning is a tool of discovery.”  This seems to me correct, and, more generally, it seems worthwhile trying to develop a theory of mechanistic reasoning, provided it is not confused with evidence of mechanism.

Donald Gillies, University College London (donald.gillies@ucl.ac.uk)


Leave a Reply

Your email address will not be published. Required fields are marked *