Social Studies of Science http://sss.sagepub.


Language and practice
Harry Collins Social Studies of Science 2011 41: 271 originally published online 22 February 2011 DOI: 10.1177/0306312711399665 The online version of this article can be found at:

Published by:

Additional services and information for Social Studies of Science can be found at: Email Alerts: Subscriptions: Reprints: Permissions: Citations:

>> Version of Record - Mar 25, 2011 OnlineFirst Version of Record - Feb 22, 2011 What is This?

Downloaded from by guest on March 18, 2013

Discussion paper

Language and practice
Harry Collins

Social Studies of Science 41(2) 271–300 © The Author(s) 2011 Reprints and permission: sagepub. DOI: 10.1177/0306312711399665

School of Social Sciences, University of Cardiff, Cardiff, UK

What are the relative contributions of language and physical practice to practical understanding? The resolution of a series of puzzles depends upon the answer. I argue that language is, and must be, more central than physical practice in individual acquisition of practical understanding. Only this makes it possible for there to be a sociology of scientific knowledge, for there to be scientific specialities, for there to be a division of labour in society and for there to be a society that is more than a set of narrow and isolated worlds. Physical practice remains central to human culture but its influence is at the collective level at which languages are formed, rather than the individual level at which practical abilities are acquired. Domain languages ‘contain’ practices, and it is from these that individuals draw much, usually most, of their practical understanding. Because the individual level and the domain level have not previously been distinguished, certain philosophical problems have been wrongly cast and mistakes have been made. Domains of practice/language are embedded within one another in fractal-like relationships, and this is how we can make sense of higher levels of coordinated action. The ideas of ‘special interactional expert’, ‘practice language’ and ‘methodological interactionalism’ are introduced.

interactional expertise, language, linguistic socialization, methodological interactionalism, practice, ‘practice language’

To what extent does one have to practice in order to understand a physical practice? The prevailing view seems to have changed over the last half-century. In what, for argument’s sake, can be called ‘the 1950s’, when it seemed that computers would soon be capable of displacing human thought, understanding things through practice and experience was mostly thought of as a deficient or partially formed version of formal, scientific understanding.1 Where there was no properly developed formula or theory, rules-of-thumb

Corresponding author: Harry Collins, School of Social Sciences, University of Cardiff, Glamorgan Building, Cardiff CF10 3WT, Wales, UK Email:

Downloaded from by guest on March 18, 2013


Social Studies of Science 41(2)

or the fruits of experience could serve as second-best until such time as scientists and technologists worked things out properly. In the latter part of the 20th century the role of practice came to be seen as more important. Polanyi argued that even in science there were ‘tacit’ elements that could not be represented formally, and the sociology of scientific knowledge produced detailed case studies to show more clearly why this was bound to be so. Formal reasoning and experimental procedures came to be seen as meaningful only in social settings. The Fleckian Denkkollectiv, the Kuhnian ‘paradigm’ and the Wittgensteinian ‘form-of-life’ (understood in the way it has generally been understood within SSK), each implies that what is formalized and what counts as an observation or an experimental result is made meaningful only when embedded in a taken-for-granted social reality (Collins, 2010; Fleck 1979 [1935]; Kuhn, 1962; Polanyi, 1958; Wittgenstein, 1953). What is sometimes forgotten is that taken-for-granted realities are as much a product of shared languages as of shared practice. For example, Peter Winch’s (1958: 121) brilliantly perceptive, Wittgensteinian analysis of the antics of the surgeons and nurses in the anteroom of an operating theatre, with their exaggerated scrubbing and choreographed donning of gloves, can only make sense in terms of the germ theory – ‘the language of germs’. The germs themselves are not forcing the surgeons to scrub and glove!2 The importance of language is already, as it were, in the ‘mother’s milk’ of anyone brought up in the academic traditions of science studies though, nowadays, it is practice that is most often the main focus of discussion. In some recent approaches, language has been entirely ignored and practice alone has been taken to be what makes it possible to understand practice. Philosophers such as Merleau-Ponty and Heidegger have emphasized the role of the body in understanding, while others, such as Dreyfus, used such insights to criticize attempts to build symbolbased machines to reproduce the full range of human capacities. As a consequence, language, seen as belonging to the domain of symbols, has been pushed to the margins (Merleau-Ponty, 1962: 143).3 Here I argue for a transformation in the way we think about these relationships. I argue that lived language is not just to be balanced with practice but is more central to individuals’ practical understanding than physical practice itself. I argue that were this not the case we could make no analytic sense of the world as we know it – in fact, there would be no world as we know it. Before getting on to the main argument, however, the thesis needs to be clarified along with the terms in which it is cast. In addition, the way in which the question relates to immediately previous work needs to be explained.4

Here I contrast ‘practical understanding’ with ‘practice’. It has been claimed in earlier publications that practical understanding can be acquired through ‘linguistic socialization’ alone without the need to engage in the physical practices themselves. This has been called the acquisition of ‘interactional expertise’.5 The idea of interactional expertise implies that it would, for example, be possible to come to understand, say, tennis – to have a practical understanding of tennis – without ever having played tennis, held

Downloaded from by guest on March 18, 2013

Therefore the contrast is not really between language and practice but between ‘linguistic practice’ and ‘physical practice’. ‘What does it feel like when you hit a hard serve really sweetly?’ If the judge cannot distinguish the wheelchair-bound person from the tennis player we say the wheelchair-bound person has exhibited practical understanding even though he or she could never actually make a line call or execute a serve. 2007).com by guest on March 18. At the same time. via the possession of interactional expertise. however. We can give meaning to this kind of understanding with a thought experiment that is also. without watching tennis or stirring from their wheelchair. which is a practice. and something similar must hold for certain levels of peer review when the job is being done properly. We imagine that a person who has played tennis all their lives asks questions about tennis of the person in the wheelchair and another person who has played tennis all their lives. Furthermore.sagepub. Upon it depends the very ability for non-scientists to accomplish deep and authentic analysis of sciences they do not actually practice. and ‘language’ as located in dictionaries and grammar books. that is.8 The ability to make practical judgements as a result of linguistic socialization alone. We say that the wheelchair-bound person is as good at making practical judgements in discursive settings as the tennis player. Another important contrast is between ‘lived language’. in principle. it will be suggested that language is a practice. the intended meanings of terms should be clear from the context. Imagine a person who has been blind and confined to a wheelchair from birth. roughly what sort of distance from point of bounce to line makes it difficult to decide on whether the serve was “in” and “out”?’ or. so the Imitation Game described above will probably remain a thought experiment. Downloaded from sss. experts in practical domains spend much of their time in purely conversational settings making important judgments about what and how to practice. managers of scientific projects can do their work only because they acquire interactional expertise in the specialities in respect of which they must make decisions (Collins and Sanders. this is the sense in which ‘language’ and ‘practice’ are intended throughout. to some extent.Collins 273 a tennis racket or bounced a tennis ball. The claim is that such a person could acquire a practical understanding of tennis solely from extended and intensive discussion of tennis in the company of tennis players. In these experiments judges could only rarely distinguish the blind from the sighted – as per the hypothesis. 2013 . In addition. The ‘judge’ has to work out who is who from the answers to question such as ‘In the case of a fast serve.6 The ‘Imitation Game’ is similar to the Turing Test. Some questions concerned line-calling in tennis and the like. understand tennis as well as someone who had played it all their lives.7 It is hard to find congenitally blind and wheelchair-bound persons who have spent many years in intense discourse with lifetime tennis players. the latter are better thought of as strings of symbols rather than linguistic practices. such a person could. But experiments very like it have been done with persons who were registered blind in childhood. has been argued to be of great importance in science studies. Sighted ‘judges’ tried to tell the difference between them and sighted persons. In every case. a real experiment. What follows is an analysis of what can come to be known through language and what can come to be known through practice.

‘special interactional expert’. In the long term. 2007: 79ff). language dominates practice (nearly) everywhere. What is argued here. is that interactional expertise is: (1) the main component in the acquisition of most practical abilities. Contributory experts are. for the individual. the class of contributory experts is entirely included in the only very slightly larger class of interactional experts (see Fig. Since. as will be argued. There would be no interactional experts (I should say ‘special interactional experts’ – see below) without contributory experts (those who practice). The opening question concerned the relative contributions of language and practice to practical understanding. There can be tennis language only if there is the practice of tennis.274 Social Studies of Science 41(2) What is new What is new here in respect of the overall programme known as Studies of Expertise and Experience (SEE) is the central importance given to language. Thus. Tennis language is an example of what I am going to call a ‘practice language’ – which is a language related to a practice or set of practices. 1). however. we are all interactional experts. In early discussions of interactional expertise it was taken that ‘interactional experts’ – those who gained their practical understanding from linguistic discourse alone – were rare and exotic. Interactional expertise is found everywhere. language dominates practice. but also what bridges disparate worlds of practical activity. if a community of interactional experts were isolated for any significant time from the contributory experts who give rise to the corresponding practice-language. then. at the collective level.9 Another innovation is that here it is argued that language is not only central to practical understanding in any one domain. Another iconic example of an interactional expert was taken to be someone like Collins. the language would begin to degrade and die. and (3) the basis of human societies. 2013 . who had to spend decades acquiring the interactional expertise of gravitational wave (GW) physics. if we were all blind and congenitally bound to wheelchairs there would be no talk about tennis to learn from – there would be no ‘tennis language’. The answer offered here is that for the individual. This makes humans quite different from animals and other non-humans. While. language dominates practice for the individual. Downloaded from sss. it would become a kind of cargo-cult by guest on March 18. This is what has been called the social embodiment thesis (Collins and Evans. A point made in earlier publications is that the relationship between language and practice found at the individual level is not the same as the relationship found at the collective level. like the imagined blind person in the wheelchair. where language is formed and maintained.sagepub. without such bridges our lives would be bounded by our practical experiences and we would each live in a condition not far from social isolation. practice is a vitally important driver of the language. Furthermore. (2) the foundation of any complex division of labour. even those classed in earlier treatments as contributory experts. This means that it is necessary to invent a new term for the special group of interactional experts who are not contributory experts. the obvious term is. Another innovation is the changed relationship between contributory experts and interactional experts. rather. interactional experts. too – the two classes do not contrast. This is what is meant by saying that interactional expertise is parasitic.

Practice languages.Collins 275 Special interactional experts Contributory/ Interactional experts Figure 1. an example of a practice language is ‘tennis language’.11 (2) There is an indefinite number of practice languages. (5) Practice languages are embedded within one another in fractal-like relationships (see below). someone with practical understanding of a domain. As intended here.sagepub. The important feature of a practice language is its substantive (often tacit) content. another is GW physics language.  Special interactional experts Definitions: ‘Experts’ and ‘practice languages’ Before embarking on the main argument.10 Second. and so forth. This contrasts with the following common usages not deployed here: (1) someone with more true and justified beliefs than someone who is less expert. There are indefinitely more practice languages than there are natural languages. amateur cricket. 2002. 2007). as understood here. But the term ‘expert’ has a variety of uses and the one intended here needs to be distinguished from others. Other examples are the languages of professional cricket. an expert is someone who possesses the tacit knowledge pertaining to a domain of expertise (Collins and Evans. (3) someone who gives expert testimony to government enquiries and the like. two further clarifications are necessary. with new ones coming into existence all the time. differ from languages as understood by other academic disciplines in some or all of the following ways. Downloaded from sss. (4) There is taken to be no analytically significant difference between children’s learning of practice languages and adults’ learning of practice languages. or (4) experts at innovating and inventing. that is the social embodiment thesis. (2) someone whom other people believe to be an expert. As has been seen. (1) The practice languages of groups are intimately related to their practices in the world. the term ‘expert’ is associated with SEE: thus. will be described as an ‘expert’. whether they can also physically practice or not. ‘practice languages’ must be distinguished from ‘language’ as it is understood by other academic by guest on March 18. (3) There is taken to be no analytically significant difference between learning a first language and learning subsequent practice languages. 2013 . Many practice-based languages can be found within a single natural language speaking community. First.

Groups of scientists’ actual physical engagements with the world are represented by stick figures using hammers and anvils.12 The central argument: Language is central to practical understanding The central argument of this paper follows from observations of scientific practice in GW physics. and so on. still another might be the analysis of GW waveforms. A formal description does not capture a practice – only a practice language can capture it (see below). the recovery would involve the reinvention of the entire practice language.276 Social Studies of Science 41(2) (6) A natural language is. 2007). a high-level practice language formed from the joint practices of the natural language speaking community. different practical specialities are indicated by the numbers 1– by guest on March 18.sagepub. In Fig. reason why language is central to practice. not just resuscitating some practice based on a ‘recipe’. 2013 . 2004b.14 In a field such as GW physics there will be many such specialities – ‘n’ of them! (Also in the diagram is a stick figure with no hammer and anvil but engaged in the same kind of talk as all the others: 1 3 5 2 4 … n Figure 2. GW physics is a big science that brings together many practical specialists. (7) The death of practice languages is particularly interesting because it means the death of a domain of physical practice without easy hope of recovery. The practice of GW physics Downloaded from sss. Collins and Sanders. another might be laser-development. the domain language is represented by the little bundles of waves. but strangely overlooked. One of these specialities might be mirror-suspension design. among other things.13 Here. Such observations reveal that there is a simple and obvious-once-stated. 2011. 2 the domain bounded by the irregular line is meant to be GW physics. Consider the practice of GW physics (see Collins.

3. who has spent many years embedded in the discourse of GW physics. at one time. This is a special interactional expert. in this case.Collins 277 this figure. share the same physical space in a university. then. but has no practical expertise. though it can and must be accomplished through this intense mutual linguistic immersion. understanding the work of those who belong to the other 1–n specialities within a practice such as GW physics. To sum up. That this process leads to the sharing of a language has been made clear in a simple Imitation Game experiment. 2007: 108). 2013 . It has been shown that even groups. 1–n. ‘contains’ the practice of tennis). The form of the GW practice language is. In contrast. the very notion of specialization would make no sense. particularly in relativistic physics. in a sense. often with video links. An individual’s grasp of GW physics language has many sources. This is reinforced by endless email lists. the majority of the whole international community gets together several times a year at conferences or workshops. and then even more particularly in GW by guest on March 18. even when not everyone is together in the same place at the same time. The language of GW physics. so the only way they can gain such understanding is via a shared practice language. the language of a domain of practice. did. who are disciplinarily close to GW physics and. ‘contains’ the practices of all those specialists (just as tennis language. such as astrophysicists. who might be a sociologist or a manager. in so far as it is a recognizable practice Downloaded from sss. GW physics language is a language heavily based in practice and that practice covers all the individual specialities that are part of GW physics. the author. Crucially. backed up by networked materials and emails. so the language continually filters around the network as a whole.sagepub. some feedback). as the social embodiment thesis makes clear. knows the language and has practical understanding. and found he could act as a judge in the astrophysics Imitation Games and identify the participants very easily (Collins and Evans. Furthermore. were unable to pass as GW physicists in Imitation Games. were this not the case. To repeat. They do not do each other’s work. the individual’s grasp will have developed in a good part through the discursive contributions of those who are inventing and developing the specialist practices. the posting of materials on the net and teleconferences.15 Now this central argument will be repeated a little more fully. is not the same as being able to do the work of the others. The way this comes about is through apprenticeship in groups distributed in universities across the world. in a sense. A hard-working GW scientist might participate in two or three teleconferences a week. with members of those universities meeting other members on a regular basis to discuss their work on shared or interacting specialities (the 1–n). based in the n practices that make up the field and it is only this that makes us able to distinguish and identify it as a discrete practice as a whole. Turning to GW physics language as a whole. It is developed during the specialists’ training in science. 2006). and in physics as whole. pass such a test (Giles. We can imagine arrows coming out of each of the specialists and representing the ways each of their practices has contributed and continues to contribute to the ‘sea’ of the practice language (with. of course. The separate contributions of each of these n practices has contributed to the language as a whole and this must have been through the interactions of the individual practitioners and their specialist subgroups with the collectivity of GW physics language speakers. as in Fig.) Each of the ‘n’ specialists must understand the work of the others if they are to cooperate so as to form the big science –‘the practice’ of GW physics.

The ratio of language to practice will vary from domain Downloaded from sss. If there are n specialities. once more.sagepub. Comparing the two bolded stick-figures. But there are also arrows. Also shown in Fig. the special interactional expert. not coming out of the specialists but going into them. The leftmost bolded stick figure in Fig. each specialist. one can see that the only difference between them is that the contributory expert on the left engages in an nth of the physical practices pertaining to the field whereas the special interactional expert does not practice that nth and contributes nothing at all. These downward arrows represent the way language gives meaning to and shapes practice as individuals are inducted into the field. 3 is the special interactional expert from Fig. certain things follow. Methodological interactionalism Once the importance of language to practice is grasped. each specialist has learned the language of ‘the practice’ – the language that enables him or her to work within the practice and to declare ‘I am a practising gravitational wave physicist’ – in the course of practising only about 1/nth of the range of physical practices pertaining to the practice. such as the bolded hammerer to the right in Fig. 2. 4. that person also contributes only a tiny part to the formation of the practice language.  Formation of the GW physics practice language language. But first the ideas need a little refinement. 2013 . As an individual. is formed largely from the typical physical and theoretical embodied practices of the individuals within the specialist groups belonging to that domain. These arrows go from the language to the ‘hammerers’ (Fig. but he or she makes no contribution to the formation of the practice language. 4 is. 4).com by guest on March 18. learns the language of GW physics while practising only a small part of the physical activities that comprise the entire practice’s physical engagement with the world.278 Social Studies of Science 41(2) 1 2 3 4 5 … n Figure 3. That is why the possibility of the existence of special interactional experts should come as no surprise – there is almost no difference between interactional experts and contributory experts as far as their relationship to the practice language is concerned. To repeat.

2003) need careful reconsideration. the point remains that even in relatively individualistic domains the meaning of the physical activity. position.16 Methodological interactionalism embodies what Popper would call a ‘bold conjecture’ – namely that a practice can never be learned from someone else in the absence of shared language. act as though language is always the learning mechanism. or default. as the sociology of knowledge points out.  Language dominates practice for the individual to domain. as central.Collins 279 Figure 4. In some domains – and this might well include tennis – there is little in the way of division of labour and so the relative importance of language might be less than in. and therefore the way we conceive of it and practice it. physical immersion in practice should be thought of only as the condition for immersion in the practice language. then. In the first instance. rather than the ‘human-as-animal’.sagepub. these need to looked at again. individual-encounter-with-the-physical. cases where practical understanding really is acquired largely from practice Downloaded from sss. say. as cases of the quintessential collective way of human learning. Nevertheless. In other words. If there are said to be cases where no language is necessary. If they exist. methodological interactionalism is intended to derail the commonsense. This is to say that when we human individuals engage with the world for the purpose of doing anything more complicated than is done by animals. Like ‘methodological relativism’. this must become the new default position. way. In this case what has to be derailed is the idea that practice is a sufficient explanation for understanding practice and for the acquisition of practical skills. existing descriptions of apprenticeship regimes that appear not to depend on language (for example. 2013 . the way we cut it up differs from language community to language community. as far as possible. Imagine a group that appears to learn entirely through deep immersion in physical practices. in the first instance. That this is the case gives rise to a methodological principle: ‘methodological interactionalism’. we know how to dissect it into discrete entities through our language by guest on March 18. all cases of human acquisition of expertise should be treated. When investigating any practice-learning environment one should. even in such a case the role of language should be treated. is formed by language. GW physics. Jordan and Weston.

Having acquired verb placement. even though they are generally not explicated during early learning of a language. like all rules. can never be translated (nor even used). and the place of physical practice in the absence of language becomes a topic for research. we send a series of different strings of symbols each time so that the receiver can look at different ways of expressing the intended meaning. remedy or measure loss in strings of symbols we transmit the same string of symbols over and over so as to create redundancy in the information. at least not until much later. that there is no equivalent meaning theory has to do with the fact that languages are not meaningless. such as balancing on a bicycle. even though they usually cannot say what they know. Languages. without the risk of irremediable and immeasurable loss – there is no ‘meaning theory’ equivalent to information by guest on March 18. the child then ‘knows’ how to do something. The child learns where to place the verb just as the child learns to ride a bicycle – by doing it. To that extent and more. English letters can be transformed into binary code and back. Elsewhere the difference between languages proper and strings of symbols has been explored under the heading of the ‘transformation–translation distinction’ (Collins. nor do they even ‘know’ that they know it. being the property of living societies. on the other hand. among the things that can be made roughly explicit. Where there are losses. Traditionally. however. In learning language. 2010: 25): strings of symbols.280 Social Studies of Science 41(2) alone become interesting exceptions. Furthermore.sagepub. to try to reduce loss of meaning in language. That is.18 For example. can be transformed backward and forward into other symbols either without loss or with losses that can be measured and sometimes remedied. But acquiring native fluency in a language – as opposed to learning the shell of a language from its explicit grammar and the dictionary – always involves the acquisition of tacit knowledge. That there is a worthwhile information theory has to do with the fact that strings of symbols are by themselves meaningless. chess pieces can be made of carved wood or bottle tops and matchsticks. The tacit is often exemplified by a practical ability. 2013 .19 Children learning their first language acquire the knowledge of where to put the verb only tacitly. can be broken in the right context (‘In the right Downloaded from sss. the child learns things that are never said. Thus. The child learns verb placement by learning to perform the language. and so on. to say that it is possible is not to reintroduce symbolic representation as the pre-eminent form of practical understanding. are certain grammatical rules. information theory deals with them. the second law of thermodynamics aside. at least in its relationship to practical and technical matters rather than artistic or expressive matters. this is because the fluent lived language discussed here is itself laden with tacit knowledge. Can we imagine how language might ‘contain’ practice? How is it even conceivable that practical understanding could be contained in language?17 First. since rules such as those about verb placement. it involves the acquisition of far more in the way of understanding than what is said or even could be said. language speaking is a practice. language. Note that to reduce. Fluent lived language is full of meaning that is continually in flux. has been thought of as the domain of the explicit as opposed to the tacit. we know that there are overall rules for the placement of verbs in sentences in natural languages such as ‘put the verb in the middle’ in English.

can transmute calculations into real stuff.’ – the editor will not correct that sentence even though Word has warned me about a grammar problem with a jagged green line). and the delegates learned that the right response was to quietly move on to the next paper. In retrospect.20 He should have been in Pisa to see what could be done with black holes – everyone was spraying them all over the place. In the first passage. this process might have been illustrated in a couple of passages of Gravity’s Shadow (Collins. during which I listened to every paper. Theorizing and computer modeling. and ‘if you can spray them they’re real’. 2004b: 452) Downloaded from sss. … Yet at the Pisa conference. in conference after conference.Collins 281 context the rule broken can be. it learns what words and usages are properly uttered in polite company. The modalities surrounding the term black hole were those having to do with certainty. It might also be contained in wider patterns and frequencies of usage in the community as a whole. when it is conducted by consenting adults in public.sagepub. it would be impossible to learn virtuoso verb placement from rules alone. We do not know how practical understanding is contained in language. Becoming an interactional expert in a practice involves coming to know aspects of the practice through acquiring fluency in the language. For example. They were using spit. a whole new slant on the notion of ‘social construction of reality’. explaining that he had found gravity waves long ago. Thus. the frequency with which a word or phrase is uttered by the entire body of speakers may indicate something of practical importance. It might be contained in the arrangements of words. ( by guest on March 18. black holes were as comfortable and familiar as cups and saucers. and nowhere was it more apparent than here. The less literal side to truth making is still more interesting. In my first day at the [1996] Pisa [GW] conference. (Collins. in passing. Conferences are the places where the community learns the etiquette of today’s truth. 2004b). Weber’s name was mentioned just once. What is argued here is that learning a practice language – such as the language of GW physics – might carry with it an understanding of the practice of GW physics in something like the way that learning any language carries understanding of where to put the verb. And later. phrases and sentences. it is suggested that the silences of speakers can be seen to indicate their understanding of the (low) flux of gravitational radiation that is to be found in the universe. one learns how to make practical judgments. 2004b: 451–452) In the second passage (referring to the same conference) we see that talk is used to establish existence: And what else was being established? No black hole has ever been seen. Joe Weber [a pioneering GW scientist whose claims to have seen high fluxes of GW were discredited by 1975] would stand up and present his papers. Theorists have hijacked the discourse of discovery. The theory of black holes was a matter of fact. this or that feature of black holes has not been postulated but ‘discovered’. So the language learner has to learn things that cannot be said as well as things that are simply happen not to be said. conferences would happen without the physical presence of Joe Weber or even his virtual presence in the vibrations of the airwaves that constitute words. Just as one learns where to put the verb. Philosophers have tried to define the real: Ian Hacking says that he thinks electrons are ‘real’ because experimentalists talk of spraying electrons. 2013 . and some scientists refuse to believe in them at all.

as illustrated by Fig. one is learning things of practical importance. as the tacit knowledge of the understanding of a practice is acquired through the development of fluency in the domain language.24 No amount of explanation will enable the novice to get on a bike and ride it at the first time of trying. nerve pathways and synapses. It is just a difference in social roles rather than in the grounds of knowledge – it is a sociological difference rather than an epistemological difference. in recognition of what they can contribute. in the subtle changes to muscles. 2013 . by guest on March 18. of course. comes with the acquisition of the collective language.22 All that can be said is that. metaphorical. Mirror neurons aside (see notes 23 and 24). Contributory experts are those who gain their interactional expertise in the normal way. the difference between practicebased domains and non-practice-based domains remains as sharp as ever at the collective level. One is learning what and who is to be taken seriously. the kind of understanding that enables sound technical judgments to be made and physical activities to be coordinated. then – in terms of epistemology – interactional and contributory experts are almost identical. In terms of the way knowledge is made and acquired. where nearly all understanding. and in this case most would be unable to provide a scientific explanation of why this could be said. but almost disappear at the individual level. practical or otherwise.282 Social Studies of Science 41(2) In learning to use words as the community around one uses words.sagepub. there are subtle changes to the metaphorical muscles. such as bicycle-balancing. The skill of bicycle-balancing (as opposed to riding in traffic – see note 22) is individually embodied rather than collectively embodied.21 What is written above is no more than an invitation to develop a full theory of how language contains practical understanding.23 Practical understanding developed though linguistic discourse alone does not. Physical skills of this kind require changes in the material form of body and brain. Yet the pattern of usages they acquire and promulgate contains things that affect practical judgements and physical practices. carry with it the ability to execute embodied practices. The sense of difference between individuals’ experience in a practical domain can be recaptured. Unfortunately it is likely to be even more difficult than developing a theory of the location of embodied pieces of tacit knowledge. 4. and metaphorical nerve pathways of the spoken language and in the. The same is true of what we might call ‘embrained’ abilities such as mathematically expressed theorizing – this requires ‘mental muscles’ to be trained and exercised as the tacit abilities are acquired. where practices feed into the language. and such things are some of the crucial components of practical judgments – they teach what does and does not exist and what can and cannot be done. the understanding that comes with language is not of this kind. they need make no special effort to become Downloaded from sss. it is. however. Speakers are mostly unaware of the frequency and nuance of words and names in their speech. they are granted access to roles that facilitate their socialization into the specialist domain. ‘synapses of society’. They bring some narrow practical expertise that makes them valued members of that domain and this automatically immerses them deeper and deeper in the practice language. nor how they were creating scientific understanding in the course of their speech. What is special about ‘special interactional experts’? In terms of the grounds of their knowledge. and that could not be said.

2013 . on the other hand. emphasis added)26 In this respect. 2006. anthropologists. Indeed. it is hard to see why mutual incomprehension would not go right down to the level of individual personal experience.25 Some wider implications of the importance of language The notorious ‘science wars’ turned in part on the claim that only active scientists could properly understand science. if not fully. ‘If that’s your goal you need a knowledge of the field that is virtually. The weasel word ‘virtually’ gives the game away because without it no one would understand anyone else’s world: we would all be isolates in our little specialities.Collins 283 fluent interactional experts – it comes naturally with their practical contribution and consequent immersion in the community and the discourse of the community.sagepub. and the only way to understand something practical was to practice it. Sokal says that to probe how cultural and scientific factors shape science you need a knowledge of the field that is virtually. insisting that such a subject cannot be properly understood without full immersion in the practices: Sokal says he is struck by Collins’s skills in physics. who are also special interactional experts. So is the sportscaster who can’t tell a strike from a ball until the umpire has announced it.’ says Sokal. high-level journalists. if not fully. Alan Sokal expressed surprise that a sociologist could learn enough to pass as a GW physicist without practising. and so forth. they get paid to fill it. Dreyfus insists that the only through practical immersion in a domain can its practices be fully understood: You may have mastered the way surgeons talk to each other but you don’t understand surgery unless you can tell thousands of different cuts from each other and judge which is appropriate. but the social role is easier to attain – indeed. Downloaded from sss. Managers. in Kuhn’s (1962) terms. Drawing on Heidegger. language is more important to practical understanding than practice. if Sokal and Dreyfus were right. incommensurable with it. As late as 2006. however. If they are sociologists. also have to work hard and self-consciously to acquire the interactional by guest on March 18. (Dreyfus. Every narrow domain of physical practice would be incomprehensible to every other or even. at the level of researchers in the field. they have to work hard to attain a role that gives them sufficient immersion in the day-to-day life of the language community to enable them to become fluent.27 Furthermore. 2007: 737) If. are those who gain their interactional expertise without the advantage of having a valued embodied skill that automatically maintains and enhances their immersion in the language community. Special interactional experts. philosopher Hubert Dreyfus is Sokal’s intellectual bedfellow. then each of us would be isolated in narrow domains of understanding bounded by the specific physical practices in which we had engaged. but notes that such understanding would not be enough for more ambitious sociology research that attempts to probe how cultural and scientific factors shape science. Sokal and Dreyfus must be wrong.’ (Giles. at the level of researchers in the field. ‘Unless you understand the science you can’t get into the theories. in Selinger et al. In the domain of surgery no matter how well we can pass the word along we are just dumb..

2013 . And so on: one may become a special interactional expert in any domain of practice so long as the possibility of immersion in its discourse is available. but it is not impossible to understand. This. It is possible to come to understand it with enough immersion in the discourse – though this is not a trivial task. Many mistakes can be avoided if sociological and logistical barriers are no longer taken to be epistemological barriers.284 Social Studies of Science 41(2) The same applies to Dreyfus’s model. foxhounds. which would be impenetrably different from the world of the liver by guest on March 18. ‘orthopaedic surgery’.29 Interactional expertise is not something possessed by odd characters such as social scientists and managers – the special interactional experts – it is everywhere: it is the ‘glue’ of human social life. there would be only ‘this person who does things with a knife’ and ‘that person who does things with a knife’. is an idea that social studies of science can feed back into politics writ large. each of which would be as incomprehensible to practitioners of the others as the Azande poison-oracle is to Westerners.30 The difference between the old and new views can also be seen by considering the contrast between humans and non-humans. Various different embodied specialities are also distributed among non-humans.28 That nearly the whole of even a contributory expert’s expertise in a practice is interactional expertise applies not only in science and technology but also to complex ­ division of labour wherever it is encountered in society. and so on. In so far as there are obstacles to the spread of mutual understanding across disparate groups. At worst.sagepub. To take an example from a domain apparently close to human ones. and so on. It may be true that each of these specialists would be reluctant to take on each others’ jobs ‘at the drop of a hat’ but if their worlds were impenetrably closed to each other in terms of understanding how would the domain of surgery work? There would be no such thing as ‘surgery’. they are logistical or sociological. only ‘heart surgery’. then the world of the heart surgeon becomes impenetrably different from the world of the orthopaedic surgeon. nothing that can link the differing experiences of these dog specialists into a joint domain of dog practice. That is why we are not bound to be social isolates. there would be. The idea of interactional expertise is vital if the complex division of labour. not epistemological. perhaps. the foxhound Downloaded from sss. is to be understood. It is because of this that we have a certain choices about how we live our political lives and our academic lives: accidents of birth no longer bestow quite the unquestionable cultural authority they have sometimes been said to bestow (for an example drawn from science studies see the discussion of Maori science in Rip (2003)). ‘liver surgery’. and so forth. balancing STS’s tendency always to draw its metaphors and models from the world of politics. The pointer is unable to come to know anything of the foxhound’s world per foxhound. however. specialities are distributed among dogs: there are pointers. The experience of being a woman is hard to understand without being a woman. such as depends on coordination of ­ practical activities that cannot be formally described. the stomach surgeon. If it is necessary to have made the cut in order to understand the cut. That we are not so bounded has political implications. There is. The experience of the descendant of a black slave is hard to understand without being the descendant of a black slave. chihuahuas. our comprehension narrowly restricted to our particular set of physical practices. but it is not impossible if the circumstances for deep sharing in the discourse are available even though the practical experiences can never be shared. at best.

For example. The mirror-suspension subgroup is related to the GW group as a whole as in a ‘fractal’ – the scale is reduced but the form is the same. and so forth. vibration analysis and attenuation. There are varied techniques for drawing fibres. has its own set of sub-specialities. and so on. Common understanding. then. One of the many steps required to build a state of the art mirror suspension is the drawing of quartz fibres into suspension ‘wires’. astrophysics. one would not expect any member of the wider GW physics community to be able to take part in a discussion of drawing of fibres in the same by guest on March 18.sagepub. feedback systems. not just science. and they will come to understand ‘well’ and ‘reliably’ from the normal GW practice language discourse. creates human social collectivities. of which one has to be chosen – a technical judgement. expect every member of the group to be able to draw fibres. 2 – let us say the one involved with mirror-suspension design. One can see that the mirror-suspension speciality is a scale model of the wider GW physics practice. This is why sociology should be principally the study of the human realm. they need only understand that fibres must be drawn well and reliably. Mirror-suspension design. GW physics is a reduced-scale model of physics as a whole. One would expect every member of the mirror-suspension speciality to be able to take part in the discussion of the best method for drawing fibres and take part in the subsequent decision.Collins 285 is unable to come to know the world of the chihuahua. Physics as a whole includes the sub-specialities – GW physics. electrostatic control and driving. Downloaded from sss. 2013 . is the ‘glue’ that holds together practices such as GW physics. This fractal relationship is true of all forms-of-life. There is also a weaker form of cross-practice communication. ‘1–n’.31 The fractal model Language. high-energy physics. and where there is no language there is no drawing together of disparate experiences into a common understanding. On the other hand. Forms-of-life contain other forms-of-life within them and these in turn contain others. including adhesion of quartz materials. and direct and coordinate their actions with those of the sub-group accordingly. members of some other GW speciality could learn to take part in such discussions by immersing themselves in the sub-practice language of mirror-suspension design. Consider one of the n groups in Fig. damping. the human realm is crucially different to that of non-humans. as we have seen. however. They also learn from that discourse the extent to which these goals have been achieved within the subgroup. fusing and dislocations in materials. while the form-of-life structure in terms of specialities is preserved at every scale. one would not. but the logistics of the matter implies that no one can get this degree of detailed understanding for all n specialities. The world of the dog cannot combine the embodied specialities of all dogs because there are no doggy practice languages – practical activities among dogs cannot be glued together by language. If they wanted. Non-humans have no language (where the term language is properly understood as being distinct from information exchange). analysis of magnetic and electrical properties of materials. created and captured by language.32 The same relationship applies if one goes up in scale. To work as a GW contributory expert they do not need this degree of understanding.

2 could equally represent any of these levels. This nomenclature is meant to express the idea that every GW physicist possesses it but that it is special compared to the ubiquitous expertise found in ordinary social life. the Figure is nested within itself again and again. highenergy physics. But science cannot be done unless it is possible to talk in terms of classes and ignore boundary problems for the sake of analysis. And within the form-of-life of cricket there are the forms-of-life of professional cricket. and so forth. and so on. adhesion science. and it is possible to move up to the whole of Western society in which science is just one speciality. 2. And each figure within the reduced scale representation contains a smaller representation of the Figure as a whole and so on until we reach the bottom or the top. How do we define the boundaries of specialities. as has been explained. (This is an experiment that has actually been carried out. the bottom level might be all those practical sub-specialities that.33 If we move up a fractal level or two to physics as a whole. in which physics is one of the specialities 1–n. The middle level might be.286 Social Studies of Science 41(2) among native English-speakers. To ask whether there is a boundary between astrophysics and GW physics we need only imagine that a GW-accomplished judge is asked to distinguish in an Imitation Game between an astrophysicist and a GW physicist (and vice versa). sub-specialities. The cascade of levels is shown in Fig. In Fig. Something like Fig. Downloaded from sss. vibration analysis. 5 each group of stick figures who together practice a speciality x contains within it another sub-speciality that can be diagrammatically rendered in the same way as the group in which speciality x is itself represented. And so on – all the way down to where the groups are so small that they can no longer be said to support a distinctive discourse or practice language. GW physics. Progress has been made in STS by focussing on boundary transgressions. Invoking the Imitation Game as a thought experiment helps. we can say that there is a ‘ubiquitous expertise(physics)’. amateur cricket and school-kids’ cricket.sagepub. which is itself inside the form-of-life of physics. the top level might be physics. with stick figures also representing astrophysics.) We can always ask whether it makes sense to think of a boundary between collectivities by imagining the outcome of an Imitation Game. At the next level up there is ‘ubiquitous expertise(science)’. say. it is found inside the form-of-life of astrophysics. making the problem more intractable since it now seems politically suspect to talk of sharp boundaries anywhere. then there is a boundary between their domains of practice or forms-of-life. If the participants are distinguishable. and so forth? This is an old question. astronomy. in which GW physics falls back to being merely one of the specialities equivalent to the original 1–n in Fig. contribute to mirror-suspension design – glass-drawing. The interactional expertise – the practice language – that holds GW physics together as a practice could be called ‘ubiquitous expertise(GW physics)’. 5. each associated with its own by guest on March 18. there is the form-of-life of sport and there is the formof-life of war. Within the form-of-life of sport there are the forms-of-life of football and of cricket. then there is a boundary that can be made visible between their universes of discourse. which is in turn inside the form-of-life of science and so on. It is middle-level forms-of-life which have been referred to here as ‘practices’. If we believe the participants would be distinguishable. 2013 . missing out an intermediate level. or practice language. GW physics is a kind of middle-level form-of-life.

2013 .sagepub.Collins 287 … n 1 2 3 5 4 … n 1 2 3 4 5 … n Figure 5. The solution is to delegate particular Downloaded from sss. The fractal model Two kinds of linguistic bridge between practices The argument so far has shown how interactional expertise makes sense of middle-level practices such as GW physics. enabling each sub-specialist within the practice to act as something other than an isolate. so a bridge between GW physics and astronomy is by guest on March 18. This is not a trivial matter. as Imitation Game experiments have shown – such groups of physicists do not speak each other’s practice languages. GW detection involves a search for correlations with electromagnetic signals such as might be seen by astronomers watching the explosions of stars. For example. Sometimes interactional expertise can also be used to bridge middle-level practices.

We could make visible the unity of the higher-level physics practice language by asking Imitation Game judges to distinguish between physicists and non-physicists.36 Intermingling practices and the collective versus the individual Using the Imitation Game to think about boundaries reveals another important feature of practices that goes to the very heart of the philosophical argument of this paper – that human life is essentially collective. bridges can be built in the same way between the different groups found in social life as a whole. in so far as they succeed.sagepub.35 Then we could step up a level and show the difference between science and religion in the same way again. x-ray astronomers. biology in general. one for visible light emissions.288 Social Studies of Science 41(2) individuals belonging to the GW physics-practice to learn some astronomy practice language. and the language of GW physics in particular. and they will have a good sense of what the scientific pay-off might be from a certain expenditure. the practices that make them up. it is inconceivable that the distinction between. however. From what we already know. It is still the case that practice languages that pertain at the different levels are built from. 2013 . It is just that the practices are not ‘mirror suspension design’ or ‘waveform calculation’. which still requires specific bridging mechanisms. say. but more general things such as ‘large-scale experimentation’ or ‘theoretical model-building’. can then answer technical questions and queries from GW physicists on behalf of. physics and biology could not be revealed in this way and that the ‘native members’ asked to act as judges would not understand the difference between the language of physics in general. Once more. because they can understand them as a result of their fluency in the higher level practice by guest on March 18. and so on. peer review of big projects will often involve an understanding at this level – all appropriate reviewers will know what doing a physics project is like. The fractal model allows.34 As has been argued. to gain interactional expertise. one for neutrino bursts. For example. To repeat. To know these things they do not need to be able to engage in each others’ practices. and how teams are brought together and maintained. not every physicist has to engage in every one of these practices. Downloaded from sss. The delegated individuals. and to form bridges with different kinds of astronomer: one bridge for those investigating x-ray emissions. and ‘contain’. and adjust their questions accordingly. Each delegate has to become a special interactional expert with respect to the community to which he or she is to build a bridge. Not only are practices embedded within one another. without always referring back to those astronomers – this is how one detail of the technical cooperation between these middle-level practices is made possible. they need speak only the practice language of physics which links all middle-level practices into the practice of physics as a whole. for a weaker kind of communication between middle-level-physics practices mediated by the higher level practice language – the practice language of physics. mutual understanding is possible at this higher level – without it we cannot make sense of the world – but it does not solve the problem of more detailed technical cooperation. how hard and fallible (or reliable) experiment is. say. The practice language of physics is not enough to facilitate detailed technical collaboration but is enough to enable one physicist to understand another at a more general level. nor even speak each others’ middle-level practice languages.

not the individual. be thought about in terms of the Imitation Game. such as high-energy physics versus fluid dynamics. One might subdivide physics into specializations. one treats the collectivities as the basic units – the units that cannot be further subdivided – and the individual as composed of these units. One cannot divide the world up into practices in an indefinite number of ways. If. however. often associated with radical differences in geographical location and material conditions. The appropriate unit of analysis is the collectivity – the collectivity is the stuff of social analysis. but the individuals involved are divided into different but intermingled sets. It is an empirical question but we can be sure that many such groups would intermingle and overlap. The number of different possible combinations of collectivities at intersections nicely accounts for the individuality of individuals. however. Here the Kuhnian position is generalized to all practice languages – how do those whose practice differs understand each other? The problem is that Fig. Compare this with the atomic theory of the elements. each one is different.37 The problem is completely general. not the individual. but if it is so divided then the Imitation Game will show it. where there is a limited number Downloaded from sss. for example. been the pretext for a television comedy in the UK. One only encounters the problem of intermingling if one considers that there are atomic units – individuals – in common between the collectivities. the implicit model is often of two distinct groups.sagepub. Thus there is no intermingling of the dimensions of different physics specialities on the one hand. Indeed. Each individual is the intersection of a set of collectivities. we have had to work with the idea that there can be radical conceptual difference alongside geographical. say. The problem of bridging across practical specialities applies in the same way. How many ways one can do it can. The way to make sense of this intermingling is to take the basic unit of analysis as the collectivity. Thus. that the UK is ever so slightly divided between ginger-haired people and the rest – such a view has. then the Imitation Game would not reveal a boundary. instead. experimental scientists and theoretical scientists. Radical distinctiveness of worldview is often discussed as though it coincided with radical sociometric by guest on March 18. material and sociometric co-location. such as theorists versus experimentalists. Thus. 2013 . It might well be. In discussions of the incommensurability of cultures and the like. The way to resolve the problem. If the world were not divided up by practice language into. played in the UK. again. the Catherine Tate show. the problem goes away. the Imitation Game would not reveal any boundary between those with light brown hair and those with dark brown hair – these groups are not socially organized in such a way as to have specialist languages pertaining to their different ways of being in the world. 5 could be drawn in different ways. is to take a unit of analysis that does not lead to the notion of intermingling of practices. or via styles of work.Collins 289 but they also intermingle. I suggest. At least since Kuhn. 5. The alternative choices would yield alternative partitions of the individuals involved and very different versions of Fig. such as the British and the Azande. and theory versus practice on the other – they are simply different collectivities. one might divide any population into tennis players versus hockey players or choose chess players versus bridge players or manual workers versus white-collar workers. this way of looking at things fits much better with the standard notion of an atomic theory because if we take individuals as the atomic units. whatever the division.

Summary and conclusions The central point of the argument presented here is that language is central to practice because the quintessential human mode of practice is collective. This makes it obvious that there is no philosophical problem presented by the existence of special interactional experts. Everything else follows from this point. There are two ways in which human social life can function without breaking down into islands of mutual incomprehension bounded by their own physical practices. that which is found Downloaded from sss. (5) Animal-like. practices should not be confused with quintessentially human practice.290 Social Studies of Science 41(2) of types of atoms. political and logistic obstacles that prevent this from happening are often strong and may be insurmountable in by guest on March 18. individualistic. This may happen in the normal process of socialization within a practice. Nevertheless. This model is homologous with the treatment as collectivities as ‘atoms’ and individuals being ‘molecules’ made up of varying proportions of some of them. enables only weaker and less ramified coordination. we can recapture familiar features of life. (4) Practicing cannot be said to be essential to understanding practice without qualifying the claim by reference to the fractal model. This is what sociology should be about. though the social.sagepub. The level of collectivity being referred to must be made clear. there is no hope of bridging practice because there are no practice languages. 2013 . interactional expertise and the fractal model in hand. the principle of ‘methodological interactionalism’ follows from what has been argued. With these ideas of practice language. and there is every reason to think that they also work in life as a whole. described by the fractal model. (2) Special interactional experts are exotic because they come by their expertise in an unusual way – but they are sociologically special rather than philosophically special. with each member belonging to a class being identical with every other. It has been shown how these processes work in science. (1) There is little distinction between contributory experts and interactional experts. The first is deep technical understanding across islands of practice made possible by technical practice languages.38 It may be worthwhile to list some of the claims made in this paper. (3) Given this. The second way in which disparate groups can communicate is through their common knowledge of a higher-level practice language. In the case of non-humans. There is incommensurability and/or mutual incomprehension between domains of practice. or by the building of bridges using special interactional experts. it becomes obvious that there is no epistemological barrier to mutual understanding across groups with disparate forms of life. because contributory experts typically contribute in only narrow domains of practice. (6) As explained. but it resolves some of the problems of incommensurability. This. of course.

sagepub. but they apply to every sphere of collective human life. 2013 . 6). These results have been developed from close examination of the workings of a science. Fig. not just draw upon it. It is the difference between the impact on language of the body that typifies the species – the social embodiment thesis that operates at the collective level – and the relationship of the individual to language. in this way. The first of these has already been by guest on March 18. 6 shows two models of the relationship between immersion in practices (made possible by the body) and potential linguistic fluency (for example.39 Appendix Reconsideration of the relationships between body. contribute to political understanding. I believe that existing debates are often confounded in two ways. The quadrant bounded by the axes contains all language-speaking entities. as shown by the heavy dotted diagonal line. This view is illustrated by Dreyfus’s statement about Potential fluency Potential fluency Practice Practice Dreyfusian Model Interactional Expertise Model Figure 6. Science studies can. and it takes for granted that potential linguistic fluency increases only with increased practice. Furthermore. The easiest way to explain the second is to begin with a diagram (Fig. practice and language I now want to suggest that the theory of interactional expertise means that the relationship between language and the body needs to be more carefully thought out. even at the lower level the incomprehension can be resolved by a determined enough effort to form bridges using special interactional experts. What I am calling the ‘Dreyfusian Model’ is shown on the left. The second confusion concerns two different arguments about individuals. Two models of the relationship between language and practice Downloaded from sss.Collins 291 at the lower levels of the fractal does not prevent regular communication at the higher levels. as measured by Imitation Games).

For ‘to be intelligent’.. Dreyfus’s first shot in his justly famous battle against the ‘artificial intelligentsia’ was entitled ‘Why computers must have bodies to be intelligent’. with the hypotenuse (the Dreyfusian route) being the most frequently encountered. insists that so long as any language can be acquired then. 1972. let us substitute ‘acquire fluency in language’ – which can.292 Social Studies of Science 41(2) the surgeons (above). The ‘Interactional Expertise Model’. This covers the various different routes to linguistic fluency. then this supports the idea that a body is necessary for the acquisition of any language at all. 2008. Under the Interactional Expertise Model. Why cannot all elements of language be acquired without a body? Under the Dreyfusian model the ability to acquire a practice language is confounded with the ability to acquire basic language. involving more or less practice. It is not easy to acquire practice languages in the way that the Interactional Expertise model says is possible.41 The Dreyfusian Model implies that to acquire a practice language it is not only necessary to have a body. 2013 . Were the topic the relationship between the typical bodies and practices and the languages of whole communities.sagepub. a brain-surgeon’s body to acquire brain-surgery language. fluency can be attained in any practice language without practice. Let us call this ‘the embodiment thesis’. and therefore the question arises. Crucially. The world of language acquisition as we actually encounter it is represented by the black triangle in the Interactional Expertise Model. in addition it must be one that can engage in the corresponding practice – for example. of course. then we know that some elements of language can be acquired without the corresponding body. however. be done without doing too much violence to the original intention. the use of a hammer or a blind person’s stick. under the Dreyfusian Model. just enough in the way of a body to be able to engage in the discourse that makes linguistic socialization possible. nor do the right circumstances occur very often. merely to restate the social embodiment by guest on March 18. Selinger et al. But under the Interactional Expertise model the claim that to have language at all one must have a body is no longer supported by the examples of tennis players. not philosophical or logical constraints. given the right circumstances.40 Note that the two models apply to language acquisition by individuals. we can take away all the bits of individuals’ bodies needed to do these things. 1992). a tennis player’s body is needed to acquire tennis-language. but these are sociological and logistic constraints. 2007). The question of whether computers must have bodies to be intelligent is not about collectivities of computers – we can all agree that any collectivity of computers that developed its own human-like language would have to have fully competent human-like Downloaded from sss. and the corresponding practice languages can still be acquired. brain surgeons. shown on the right. hence the dotted line rises vertically then reaches a plateau. this is the ‘minimal embodiment thesis’ (Collins. the diagram associated with the Dreyfusian model would be correct in all cases – this is. and this theme was to continue through his subsequent writings (Dreyfus 1967. every example of the relationship between practice and fluency supports the embodiment thesis. The Interactional Expertise Model breaks this link: if practice languages can be acquired without practice. there is no need for an individual to have a practice-capable body to acquire a practice language. All that is needed under this model is a ‘minimal body’ – that is. perhaps. If it is true that to understand any practice it is necessary to have a practising body.

body of work that describes itself as dealing with ‘practice theory’ (for a review see Rouse. but to some extent that has had to be reversed in the light of referees’ comments. The Einstein. analysis of expertise and tacit knowledge. suggesting it should relate back less to earlier work. 2006.. see Collins and Evans.. well-marshalled by the editors. For example. To make good judgments within an unfolding practical situation may require presence on the ground and embodied skill at observing such situations. Harvey. and the collection of papers in Collins (ed. became a real experiment that revealed that non-locality was measurable – it is nowadays known as quantum-entanglement (see for example. see Collins and Sanders (2007). When I say previous work. meant to show the impossibility of quantum theory because of the paradox of non-locality. For the special role of managers. I thank all those who commented subsequently and the three anonymous referees who provided very conscientious feedback. Downloaded from sss. impossible to capture the intellectual ‘spirit’ of a time without inviting attention to a multitude of exceptions. Collins. It has become clear that some of the vital background understandings have not been readily available to an STS readership.   1. more philosophically driven.   2. 2013 . There is nothing special about this. (Mike Gorman provided these examples taking them from Gary Klein – see Ross et al. Collins et al. 1992. see also Giles (2006). Podolsky. empirically driven. Dreyfus’s first publication on the matter (1967) was entitled ‘Why Computers must have bodies to be intelligent.   3. 1981). 2009).) 2007. In the light of the idea of interactional expertise. 6 – now needs new and careful argument. Language can be sufficient only in so far as such decisions are made in discursive settings rather than in practical courses of events. we now know that such a computer does not have to have much of a body. Collins and Pinch. There is another. of course. Thus to make the right judgement on when to pull a fire-fighting team out of a fire or to pull a platoon out of combat may require the presence of an expert with experience with observing such situations.42 Notes The initial impetus for preparing this paper came from the discussions at the Third International Workshop on Studies of Expertise and Experience (SEESHOP3) held in Cardiff on 15 and 16 November 2009. Evan Selinger triggered a major re-write of the initial draft. 1982. What he meant by that is that what you see is determined by what you have learned prior to seeing and that can only have been learned via discourse. 2002.Collins 293 bodies – but is about whether an individual computer can acquire language. Results of many of the arguments and demonstrations are gathered together in Collins and Evans (2007). but that is a price that has to be paid if there is to be any attempt to characterize broad intellectual change.’ by guest on March 18. which has led to significant changes. Rosen thought experiment. I am referring to analyses in the specific tradition of the sociological. Dreyfus and Dreyfus (1986).sagepub. so exactly how much it needs to reach the threshold that allows the vertical dotted line to begin – the horizontal position of the intersection of the axes of Fig. My understanding of Heidegger is taken from the writings of Dreyfus referred to throughout this text: Dreyfus (1972.)   6. The kind of sound practical judgment that interactional expertise makes possible can be applied only in discursive settings. so some ideas that have been published outside the regular STS sphere have been revisited in the introductory section. just as embodied skill and the right language are needed to observe through a microscope. 2004a. It is. 2007). Hanson (1958) also insisted that ‘observation is theory laden’.   5. 2006. the question becomes far more demanding and interesting.

For the purposes of defining a contributory expert. H. Wells (1934 [1904]: 474) illustrates the idea of a practice language in his story ‘The Country of the Blind’: ‘For fourteen generations these people had been blind and cut off from all the seeing world. I would like to see their argument re-worked to take into account the role of the practice language of weapons testing.sagepub. In the quasi-control condition. For the derailing of the default position that truth is its own explanation see. as well as Collins’s passing as a GW physicist. in its current instantiation anyway. the stick figures with hammers are sometimes taken to be individuals and sometimes specialist groups of individuals. Does language contain ‘practice’ or ‘practical understanding’? In so far as practical understanding is necessary for practice.294 Social Studies of Science 41(2) Nevertheless. such as calculating waveforms. 2007: 38. 12. Of course. 2010). who appears to have become fully fluent in the native language. 2006). the names for all the things of sight had faded and changed. Thus the calculators of the waveforms of inspiralling binary neutron stars are contributory experts within GW by guest on March 18. are described in Collins et al. 17. Note that the argument presented above should not be confused with the argument about whether any individual can acquire language at all unless they have a body. I believe the context will disambiguate in each case. In this text. for example. 16. (2006). theoretical practices are included. 13. the Imitation Game will often have to remain a thought experiment since. I could do none of the work pertaining to any speciality. the term ‘discursive settings’ means discussion distant from or unaffected by the material and practical context of the domain. Because blind persons spend their lives immersed in the discourse of the sighted. is germane.   8. 14.’. Mackenzie and Spinardi (1995) argue that nuclear weapons will become ‘uninvented’ if they can no longer be tested. implying a high degree of practical understanding. 11. It should be noted that a contributory expert’s interactional expertise may be ‘latent’ in the sense that they may not be able to express it (Collins and Evans. it requires that the participants are fairly literate and fluent – have ‘interactive ability’ – though the notion of expertise extends to those who are not. they should be able to make the same judgements as the sighted and pass as sighted.. This is not an exhaustive list of definitions of expertise.   9.G. 2011). Oliver Sacks’s (1985) example of the congenitally blind. we could say this – indeed we have been saying it – but it seems odd. ‘Madeleine’. and almost totally disabled. Related experiments on the colour-blind and on those with perfect pitch. See Appendix (above) for an attempt to sort out the four distinct theses about the relationship between language and practice that are usually confounded. wheelchair-bound.. in containing practical understanding it also contains   7. If the ‘special’ were not added we would be in the awkward position of having to say that a contributory expert has interactional expertise but is not an interactional expert. but some clarification is worthwhile since confusion between the SEE definition and other definitions has given rise to confusion in earlier projects (see Selinger et al. Bloor (1973) and Collins (1981). blind persons could easily identify a sighted person trying to pretend to be blind. The stick-figure convention is taken from Collins and Evans (2007). 2013 . 10.. the story of the outer world was faded and changed to a child’s story . In physics. though their conclusions would remain the same. because the sighted person had not spent a lifetime immersed in the discourse of the blind (see Collins and Evans. 15. Downloaded from sss. practical specialities include those where the practice is theoretical. Note that in the text. Selinger and Mix.

make a small contribution to the specialist domain itself. 2010). The existence of an innate generative grammar as proposed by Chomsky is not relevant to this example. Language-speaking is ‘polimorphic’ and polimorphic actions cannot be described formally (Collins and Kusch. One might ask why the manager should not be said to be a contributory expert rather than a special interactional expert. Hardly any of them would have been involved in the non-observation of Weberstyle GWs with all its subtleties. see Collins (2010). 25. Of course. and on the other. 20. this similarity is one of the more nicely revealing and initially counter-intuitive aspects of the interactional expertise idea. This way of looking at things also resolves the difficulty that even the sociologist or other kind of outsider may. Contributory experts. The implication of what Sokal says is that you have to be a scientist (contributory expert) to understand science. but rather that they. it does not contain the ‘somatic tacit knowledge’ aspects of practice (Collins. still has to acquire interactional expertise in an unusual way. not an observational force. 22.) Most literally represented by the occasional activity of mirror-neurons in the individual (Schilhab. 24. 2013 . Downloaded from sss. the idea of mirror-neurons might lead us to believe that the execution of an already acquired physical ability could be improved with enough talk but not learned in the first place.. Contributory experts are those who have gained their expertise by being granted a role in the heart of the practice language community because of their ability to engage in one of the physical (including theoretical) practices pertaining to the domain. on the one hand. 19. what cannot be said. practice. 1996. are not defined as ‘contributors’ to a domain of practice – that would be far too broad a category. 1998). given that the manager is paid precisely because he or she makes such significant contributions to the practice. since ‘bicycle-riding’ involves understanding the social conventions of traffic. 26. This is from the report in Nature of Collins’s successful attempt to pass as a GW physicist in an Imitation Game (Giles. 2010). The quote is from Hacking (1983: 24). on rare occasions. For a discussion of tacit knowledge as meaning. I refer to ‘bicycle-balancing’. by guest on March 18. Imagine finding a bike for the first time on a desert island! How would one come to understand that this spindly thing could be balanced and ridden? (See Pinch et al. On the other hand. for a development of this point. whose only claim to domain expertise is sharing the language. 8) for a description of the first use of a discourse of ‘discovery’ applied to computer modelling of atomic weapons. what is not said. had learned that rationale from the discourse.sagepub. which is a far more complicated thing. even learning to ride a bike involves being inducted into the language. at least one who enters at the top from another scientific specialization. 21. 2007) but only a metaphor in the case of collective tacit knowledge. learning practical understanding from a language depends on the equivalent of the acquisition of the ‘collective tacit knowledge’ of bicycle riding rather than the ‘somatic tacit knowledge’ of balancing (Collins.Collins 295 18. See also Galison (1997: ch. Here we are concerned with the differences between languages that have to be learned during the course of linguistic socialization. The label ‘special interactional expert’ places managers in the same class as the sociologists and others like them. The force that prevents one from talking of Weber is a discursive force. 2006). The position is not stated clearly. This is not to say that the speakers could not provide a scientific rationale for why Joe Weber was no longer mentioned. Again. The answer is that ‘special interactional expert’ in the way the term is used here. in turn. then. connotes the fact that the manager.

Thus a new student GW physicist might more easily catch Collins out in a GW Imitation Game than could a full-blown GW physicist (thanks to Luis Galindo for this point). 2013 . (I do not think Goodall’s position does insist on this.) This subsection arises out of a personal communication from Will Thomas in response to an earlier draft of this paper. To point out the obvious. Latour. 34. 1986. For this general use of the term ‘ubiquitous expertise’ see Collins and Evans (2007). In short. (2007). 30. For an analysis of Galison’s notion of trading zones see Collins et al. as explained above in the section of methodological interactionalism. Darrin Durant (2010) indicates that the ‘Third Wave’ has wider political significance than is discussed here. the argument is incompatible with ‘Actor Network Theory’ (for example. Collins and Pinch (1982) and Galison (1997). Though the ratio of language to practice in different practices might be different in different places. Collins (2010) makes a related point under the heading of ‘Social Cartesianism’. 2007) or any other theory that does not accept that there is a deep and fundamental difference between humans and non-humans. Oddly enough. It might appear to be in opposition to the interesting study by Goodall (2009). compared with these other physicists Collins probably does not possess much in the way of ubiquitous expertise(physics).296 Social Studies of Science 41(2) perhaps because it is incoherent given that under a strong version of this rule no one scientific specialist could ever understand another kind of specialist. Goodall shows that universities (and basketball teams) generally do better when led (coached) by those with high level experience in the relevant practice. 33. Wilson (1970). though Dreyfus’s does. 37. 31. An anonymous referee then suggested that as soon as one admitted that the division between sub-specialities could be drawn in different ways the whole model ceased to make sense. I set out his comment in a footnote to the paper initially submitted for publication. Callon. For the first use of the fractal metaphor for forms-of-life see Collins and Kusch (1998: 16–17). Downloaded from sss. 32. from a philosophical point of view the crucial thing is that practical experience is not always a necessity. There may be other ways of managing this bridging – these possibilities are discussed in Collins et al. 29. and judges could possibly have trapped Collins in the GW Imitation Games by asking him more general physics questions rather than specialized GW physics questions. this now becomes a topic for investigation. 35. higher levels could be subdivided into lower levels in different ways. From the management point of view it makes sense to place ones ‘bet’ on someone with practical experience. See the Appendix for an analysis of some of the confusions that may have contributed to this incorrect view. What has been argued here goes against Dreyfus’s view that the only good sports coaches and commentators are those who have actually played the sport. but wrong in so far as their philosophies insist that they cannot exist. But this statistical relationship is exactly what we would expect given the sparseness of the roles that allow one to acquire interactional expertise in the absence of practical experience. This subsection is intended to resolve what I believe 27. he says its political philosophy is by guest on March 18. 36. 28. The early debate can be found in such places as Kuhn (1962). He said that under the fractal model. The question goes back to the very beginning of the sociology of scientific knowledge where Kuhnian ‘incommensurability’ was related to the ‘problem of rationality’ in anthropology. See Ribeiro (2007) for an example from the steel industry.sagepub. (2007). Dreyfus and Goodall are right in so far as special interactional experts are rare.

most clearly expressed in Collins (1998). the study of embodied action by philosophers still tends to take the lead from animal behaviour. my critiques of artificial intelligence and my analyses of tacit knowledge. in particular in one case via conscious understanding and in another case with no understanding at all. Ribeiro is exploring the shaded triangle as it is found in practice in his work on ‘levels of immersion’. see Dreyfus (2009) and the response in Collins (2009). 2010) will always fail if the test is properly designed. buzzing confusion that is ordinary speech. Loebner Prize. See also Collins et al. perhaps. Verbot Jabberwacky. Mark V Shaney. 42. Collins (1990: Chs 13. (2007) and Ribeiro (2007). to be a very profound and revealing problem. in Selinger et al. Having a language means having the ability to immerse oneself in living discourse. Artificial Linguistic Internet Computer Entity. Downloaded from sss. I have used Will Thomas’s example to begin the analysis. 2013 . 39. References Bloor D (1973) Wittgenstein and Mannheim on the sociology of mathematics. On the other hand I cannot imagine how a computer with no experience at all could ever make sense of the blooming. I now think that the minimal body could be still more minimal than I suggested in that paper. In so far as the idea of interactional expertise is not thoroughly established.Collins 297 38. It should be clear that the Sokal/Dreyfus model is far too crude. possessing a language is also not equivalent to the information exchange of bees and other insects and animals (Crist. As explained above. Unknown: Books Llc. and that is why machines built to perform in such tests (Books Llc.. Studies in History and Philosophy of Science 4: 173–191. It seems to me impossible to believe any longer that it has to follow the diagonal represented by the hypotenuse. To see how. Much of this is argued. But to prove this he had to describe a mechanism that would provide identical performance to a human – the Chinese Room – hence his argument depends on the Chinese Room performing identically to a human. Being able to play at language in a 5-minute test has been taken to be an indicator of the intelligence of symbol-manipulating machines (Turing. be mimicked by symbol by guest on March 18. I should add that the position argued here is not something dreamed up solely for the purpose of resolving an awkward problem but has been a constant theme of my work in terms of my earlier analyses of the nature of science. Books Llc (2010) Chatterbots: Eliza. contrary to what Searle’s (1980) famous ‘Chinese Room’ thought experiment seems to imply. We do know that having a body is not sufficient to accomplish the trick of language. 1950). nevertheless. the philosophical objection has to be about whether the dotted line can ever be truly vertical or whether it must stray across to the right a bit. this also implies that to look at bodies is to look in the wrong place.sagepub. however. (Note that the main point of Searle’s argument was to point out that identical language performance could mask quite different mechanisms by which it was achieved. so long as it was equipped with suitable prostheses. for example. but by no means fully resolved. It is. based. on common experimental procedures. 2007. The performance of a lived language cannot. 40. 14) has shown that this mechanism is impossible. Chatterbot. 2004). as the splendid bodies of dogs and other animals reveal. 41. The approach also explains how it is even possible for there to be Galison-type trading zones. so I can see the problem posed by Selinger and Dreyfus – but I don’t think they have a satisfactory solution. continually acquiring its evolving tacit meanings.

Science. animals. Collins HM (1981) What is TRASP: The radical programme as a methodological imperative. Collins HM. Dordrecht: Springer. Collins HM (1998) Socialness and the undersocialised conception of society. Dreyfus HL (1972) What Computers Can’t Do. Collins HM and Evans R (2010) Quantifying the tacit: The Imitation Game and social by guest on March 18. MA: MIT Press. In: Collins HM (ed. Cambridge. Collins HM and Kusch M (1998) The Shape of Actions: What Humans and Machines Can Do. Social Studies of Science 34: 7–43. Unpublished paper. In: Collins HM (ed.298 Social Studies of Science 41(2) Callon M (1986) Some elements of a sociology of translation: Domestication of the scallops and the fishermen of St Brieuc Bay. Collins HM (2010) Tacit and Explicit Knowledge. cardiff. Review of Metaphysics XXI: 13–32. Dreyfus HL (1967) Why computers must have bodies in order to be intelligent. Downloaded from sss. Dreyfus HL (1992) What Computers Still Can’t Do. Studies in History and Philosophy of Science 37: 656–674. Collins HM (1990) Artificial Experts: Social Knowledge and Intelligent Machines. Phenomenology and the Cognitive Sciences 7: 309–311.sagepub. Collins HM and Sanders G (2007) They give you the keys and say ‘Drive it’: Managers’ referred expertise and other expertises. Ribeiro R and Hall M (2006) Experiments with interactional expertise. Cambridge MA: MIT Press.) Case studies of expertise and experience: Special issue of Studies in History and Philosophy of Science 38: 621–641. Collins HM and Evans R (2002) The third wave of science studies: Studies of expertise and experience.) (2007) Case studies of expertise and experience: Special issue of Studies in History and Philosophy of Science 38(4) (December). Collins HM (2009) The new orthodoxy: Humans. Collins HM (2008) Response to Selinger on Dreyfus. Collins HM (2004a) Interactional expertise as a third kind of knowledge. New York: Harper and Row. Chicago: University of Chicago Press. 2013 .html. Heidegger and Dreyfus. University of & Human Values 23: 494–516. Evans R and Gorman M (2007) Trading zones and interactional expertise. Chicago: University of Chicago Press (in press). 196–233. School of Social Sciences. Collins HM (2011) Gravity’s Ghost: The Equinox Event and Science in the 21st Century. Phenomenology and the Cognitive Sciences 3: 125–143. 75–85. In: Leidlmair K (ed) After Cognitivism: A Reassessment of Cognitive Science and Philosophy. Chicago: University of Chicago Press.) Case studies of expertise and experience: Special issue of Studies in History and Philosophy of Science 38: 657–666. Available at: www. Chicago: University of Chicago Press. London: Routledge and Kegan Paul. MA: MIT Press. Philosophy of the Social Sciences 11: 215–224. Cambridge. Social Studies of Science 32: 235–296. Collins HM (ed. Technology. Crist E (2004) Can an insect speak? The case of the honeybee dance language. Collins HM and Pinch TJ (1982) Frames of Meaning: The Social Construction of Extraordinary Evans R. Collins HM (2004b) Gravity’s Shadow: The Search for Gravitational Waves. Collins HM. Collins HM and Evans R (2007) Rethinking Expertise. Accessed 29 January 2011. In: Law J (ed) Power Action and Belief: A New Sociology of Knowledge? London: Routledge & Kegan Paul.

In: Leidlmair K (ed. Ross KGJ. Schilhab T (2007) Interactional expertise through the looking glass: A peek at mirror neurons. Shafer L and Klein G (2006) Decision making expertise. Social Studies of Science 11(1): 95–130. Ribeiro R (2007) The language barrier as an aid to communication. American Journal of Sociology 101: 44–99. Cambridge: Cambridge University Press. Merleau-Ponty M (1962) Phenomenology of Perception. 403–420.) After Cognitivism: A Reassessment of Cognitive Science and Philosophy. Durant. MacKenzie D and Spinardi G (1995) Tacit knowledge. Sacks O (1985) The Man Who Mistook his Wife for a Hat. The Behavioural and Brain Sciences 3: 417–424. In: Ericsson KA. Kuhn TS (1962) The Structure of Scientific Revolutions. Charness N. Nature 442: 8. Chicago: University of Chicago Press. Cambridge: Cambridge University Press. Galison P (1997) Image and Logic: A Material Culture of Microphysics. Rouse J (2007) Practice theory. Hanson NR (1958) Patterns of Discovery. D ‘Models of Democracy in Social Studies of Science’ Social Studies of Science (forthcoming) Durant D (2010) Public participation in the making of science policy. Fleck L (1979 [1935]) Genesis and Development of a Scientific Fact. Chicago: University of Chicago Press (first published in German in 1935). brains and programs. Polanyi M (1958) Personal Knowledge. Selinger E. Dreyfus HL and Collins HM (2007) Embodiment and interactional expertise. Feltovich PJ and Hoffman RR (eds)  The Cambridge Handbook of Expertise and Expert Performance. Honolulu: University of Hawaii by guest on March 18. Dreyfus HL and Dreyfus SE (1986) Mind Over Machine: The Power of Human Intuition and Expertise in the Era of the Computer. 2013 . Downloaded from sss. London: Duckworth. Sociological Review 44: 163–186. Searle J (1980) Minds. Oxford: Oxford University Press. Cambridge: Cambridge University Press. Rip A (2003) Constructing expertise in a Third Wave of Science Studies? Social Studies of Science 33(3): 419–434. Perspectives on Science 18: 189–225. Jordan BG and Weston VL (eds) (2003) Copying the Master and Stealing His Secrets: Talent and Training in Japanese Painting. Giles J (2006) Sociologist fools physics judges. 39–73.Collins 299 Dreyfus HL (2009) How representational cognitivism failed and is being replaced by body/world coupling. Dordrecht: Springer. 499–540. weapons design and the uninvention of nuclear weapons.) Case studies of expertise and experience: Special issue of Studies in History and Philosophy of Science 38: 722–740. Hacking I (1983) Representing and Intervening: Introductory Topics in the Philosophy of Natural Science. Harvey B (1981) Plausibility and the evaluation of knowledge: A case study in experimental quantum mechanics.sagepub. Social Studies of Science 37: 561–584. Latour B (2007) Reassembling the Social: An Introduction to Actor-Network-Theory. In: Turner SP and Risjord MW (eds) Handbook of the Philosophy of Science: Philosophy of Anthropology and Sociology. New York: Free Press. Pinch T. Chicago: University of Chicago Press. Princeton NJ: Princeton University Press. Collins HM and Carbone L (1996) Inside knowledge: Second order measures of skill. London: Elsevier. Goodall AH (2009) Socrates in the Boardroom: Why Research Universities Should be Led by Top Scholars. Studies in History and Philosophy of Science 38: 741–747. London: Routledge. London: Routledge and Kegan Paul. In: Collins HM (ed.

53–66). Harmondworth: Penguin. Oxford: Blackwell. He also analyses and researches the nature of expertise.) (1970) Rationality. New York: Columbia University Press. Turing AM (1950) Computing machinery and intelligence. Selinger E. plus another 27. Downloaded from sss. Mind LIX: 433–460 (reprinted in: Hofstadter D and Dennet D (eds) (1982) The Minds I. Ltd. 1904). Wells HG (1934 [1904]) The country of the blind. be further developed with the aid of a grant from the ERC. Biographical note Harry Collins is Distinguished Research Professor and Director of the Centre for the Study of Knowledge. The ‘Imitation Game’. The first of his approximately150 papers. Thompson P and Collins HM (2011) Catastrophe ethics and activist speech: Reflections on moral norms. Metaphilosophy 42(1/2): 118–144. a technique pioneered for gauging by guest on March 18. London: Odhams Press. are published in this journal.300 Social Studies of Science 41(2) Selinger E and Mix J (2006) On interactional expertise: Pragmatic and ontological considerations. Oxford: Blackwell. will. London: Routledge and Kegan Paul. 2013 . Winch PG (1958) The Idea of a Social Science and Its Relation to Philosophy. His continuing research on the sociology of GW detection is now funded by the NSF. from June 2011. He has written 11 books and edited five. Wittgenstein L (1953) Philosophical Investigations. In: Wells HG Joan and Peter & The Country of the Blind. Wilson B (ed. In: Selinger E and Crease RP (eds) The Philosophy of Expertise. 467–486 (originally published. He was President of the Society for Social Studies of Science from 1991 to 1993 and was awarded its Bernal prize in 1997. advocacy and technical judgment. Expertise and Science (KES) at Cardiff University.sagepub. 302–321.

Sign up to vote on this title
UsefulNot useful