Monday 9 June 2008

Rethinking Expertise

Whilst away, first on holiday (Corsica) and at a conference (Finland), I’ve been reading Harry Collins’ new co-authored book (Collins, H. and Evans, R. (2007) Rethinking Experience, Chicago: University of Chicago Press).

Harry Collins has been one of the sociologists of science whose work I have most admired and most often cited. His Changing Order (1985 London: Sage) is a brilliant balance of small scale qualitative empirical work with a really strong philosophical narrative both substantiating and being substantiated by the empirical picture. That earlier book makes a very strong case for the central importance of tacit knowledge at the centre of knowledge claims even in hard sciences where publication is a mark of objectivity.

The new book is less dramatic, less directly empirical and probably, for me at least, less interesting. But it sets out to begin a debate about expertise in the assessment of scientific claims by articulating a taxonomy of knowledge which includes varieties of degrees of ground level knowledge or expertise (from ‘beer-mat knowledge’ to ‘contributory expertise’) and ‘meta-expertises’ (from the ‘ubiquitous discrimination’ that allows most of us to conclude that the moon landings could not have been faked to ‘technical connoisseurship’ and then ‘referred expertise’ that underpins science management).

The taxonomy is called a ‘periodic table of expertises’. But unlike the Periodic Table proper, this is not underpinned by a very thorough theory of expertise. It is rather a rough and ready tool for thinking about how technical expertise can be judged by those who do not directly share it. The proof of the pudding will be whether others find the distinctions helpful ways to frame further empirically informed work or public debate.

The main new idea is that short of contributory expertise there is a form of tacit knowledge which Collins calls ‘interactional expertise’ and which is the expertise that is constituted by a thorough going grasp of the language of a specialism or subject area. Interactional experts cannot make a direct contribution to the discipline (they cannot carry out experiments, for example) but they have a genuine expertise and according to a further claim advanced (the ‘strong interactional hypothesis’), their linguistic abilities need not be discriminable from those who also have contributory expertise.

This claim motivates two chapters which sit rather uneasily with the tone of the rest of the book. One attempts to engage with Dreyfus’ arguments that embodiment is a necessary precondition for genuine linguistic ability. This is, in effect, a philosophical chapter, although informed by a case study described by Oliver Sacks. Sadly, the treatment is much too short to make a convincing case against the variety of philosophical claims made for embodiment but long enough to disrupt the previous brisk flow of the book.

The other motivated chapter is an empirical test of the ‘strong interactional hypothesis’ by examining colour blind and non-colour blind subjects playing the imitation game (which underpins the Turing Test) and contrasting this case with those with and those without perfect pitch. The hope is that, because they have been surrounded by the non-colour-blind community, colour blind subjects will have fully mastered its language (though they do not have contributory expertise) whilst those without perfect pitch will not have mastered the language of the perfect pitch minority.

I had a concern that a different explanation for the findings was not considered. In both cases, one group of subjects has an ability the other does not have and it might always be easier for those with the extra ability to fool those without. But the examples given suggest instead that a ‘deficit model’ of colour blindness does not help. It is more as though colour blindness has its own phenomenology (as does perfect pitch) which outsiders cannot access. (Though, of course, this is not the point of the example which is that the colour blind, having been surrounded by the language of normal colour perception, have fully mastered it.) But although this chapter supports the thesis considered with direct (if relatively small scale) experimental work, I am not sure that it really adds to the argument of the whole book which turns less on whether the strong interactional hypothesis can be true either ever or in these cases and more on the role of tacitly informed interactional expertise in both assessing and managing scientific knowledge.

One further aspect of the book that pleased me was that it anticipated a number of worries I had. Thus I began to worry that the relationship between art critics and artists is not that between interactional expertise and contributory expertise as rather between two distinct kinds of contributory expertise. (And the claim that the latter kind is sufficient for the former would seem particularly implausible given the poor quality of some artists’ discussions of their own work.) But whilst not fully worked out, this idea was at least addressed.

Especially as it is presented as intended to start a debate rather than close it off, this is a valuable short book suggesting some ways to think about expertise and the broader issues of the accountability of science to public opinion. Whether it has a lasting significance rather depends on what use other social scientists, scientists etc make of its terms.

PS: see report on a workshop organised by Collins on expertise here.

For a change to the official picture, see this update.