A Critical Second Look: Steven Shapin and Simon Schaffer, Leviathan and the Air Pump (1985)

Overinflated? The second (2011) edition of LAP

There are many books that polarise their readership, causing some readers to swoon and others to spew (so to speak). There are not many books that polarise individual readers, causing them to swoon and spew simultaneously. I place Leviathan and the Air Pump: Hobbes, Boyle and the Experimental Life (LAP), the classic 1985 work on the history of science by Steven Shapin and Simon Schaffer, in the latter category. When I read this book for the first time, as a Masters student in History and Philosophy of Science, I was doubly impressed—firstly by the scope and subtlety of the argument, secondly by the conviction that said argument was seriously flawed. Time has not dulled these impressions, as I discovered when I reread the book a few months ago. The latest number of Isis carries ten essays on the book, none of which mention the deep problems that account for the negative part of my reaction—a reaction I have not just to this book but to many of the works that have been, and continue to be, influenced by it.

The declared aim of the ten essays in Isis, which are collectively titled ‘A Second Look,’ is to ‘reflect on [the value of the text] not only in the past but also in the present.’ Now, I am all in favour of reflecting on old books—the older the better. But when the book has entered the disciplinary unconscious as thoroughly as LAP has done, the interesting question is not where its value lies (we all know it is valuable) but where its defects lie. We owe it to Shapin and Schaffer (SS for short) to treat LAP the way they treated Robert Boyle’s New Experiments on the Spring of the Air—as an exemplary text that we should read with a gimlet eye precisely because its methods and assumptions have become part of our intellectual culture.

The most common criticism in the Isis essays is that SS generalise too confidently from their case study. They ignore earlier disputes between Thomas Hobbes and the Royal Society of London (Mordechai Feingold, Paul Wood), they eschew detailed comparisons to France and Italy (I. Bernard Cohen, John Heilbron), they ignore natural history (Paul Wood), they confuse the origins of the experimental method with its nature (Ian Hacking), and they state without argument that ‘solutions to the problem of knowledge’ are always ‘solutions to the problem of social order’ (Heilbron again). There is a grab-bag of other criticisms: too much jargon (various), too little philosophy of science (Sargent), unnecessary use of Hobbes to cast light on Boyle (Sargent again), and underestimation of Hobbes’ backwardness (Wood).

These criticisms have their merits, but they do not go far enough. Before explaining why, let me summarise the virtues of the book, in case my aforementioned admiration seems insincere and in case any readers of this essay are unfamiliar with the book’s argument.

The first virtue is the bold and original thesis, namely that the experimental method was favourably received in England in the 1660s because it was seen as a viable solution to the problem of managing disagreement in a disputatious age. Experiment was politically attractive because it implied ‘limited toleration’ (299), ie. the practice of tolerating dissent in a community insofar as it does not tear the community apart. The alternative, endorsed by Thomas Hobbes, was to eliminate all dissent by coercion. More abstractly, SS’s thesis is that scientific methods are political entities. Different methods cause scientists to organise themselves in different ways, and these ‘ways’ correspond to different ways of organising people in society at large. In a phrase, there are such things as ‘intellectual polities.’

The second virtue is that this thesis is defended in a coherent manner across all eight chapters. The third virtue is that each of these chapters is memorable for the novelty of its concepts and the thoroughness of its empirical research. I remember chapter one for its witty denunciation of past studies of Hobbes’ natural philosophy; chapter two for the phrase ‘virtual witnessing’ and for its microscopic study of Boyle’s moralistic language; chapters three and four for excavating Hobbes’ natural philosophy and tying it to his political philosophy; chapter five for narrating the tortuous negotiations between Boyle and various critics of experiment; chapter six for showing that even the champions of experiment had trouble replicating them; and chapter seven for explaining why all this was relevant to Restoration disputes about religious toleration, press censorship, and legal expertise.

That’s what SS get right. Here’s where they go wrong.

1. LAP attacks a thesis that no-one has ever held. This is the thesis that scientists can obtain meaningful results without doing anything. SS think that ‘in common speech, as in the philosophy of science, the solidity and permanence of matters of fact reside in the absence of human agency in their coming to be’ (23). This claim is repeated on pages 77 and 150; it is attributed to Boyle (77) and to Hobbes (150, 225, 282); and it is implied by the second-to-last sentence of the book (‘Knowledge, as much as the state, is the product of human actions’).

Surely it is obvious that knowledge is the product of human actions. Certainly this was obvious to James B. Conant, the author of ‘Robert Boyle’s Experiments in Pneumatics,’ the 1948 account of Boyle’s pneumatics against which SS define (4) their own account. I find many human actions, and much human agency, in Conant’s work—indeed I find almost nothing but actions and agency.

2. LAP overstates its findings by equivocating on key terms. Why have critics been so slow to notice that SS were attacking a straw man? Partly, I suspect, because SS equivocate on the phrase ‘matter of fact.’ This equivocation means that SS appear to show, for example, that the pressure of the air is due to human agency (a dramatic claim), when in fact they have only shown that Boyle’s beliefs about the pressure of the air owed a lot to human agency (a claim of crashing banality).

SS also confuse a theory about the nature of truth (ie. the theory that true statements are those that correspond to the world, like a portrait to its subject) with a theory about the acquisition of truths (ie. the theory that these correspondences come about in an unmediated way, as a reflection comes about in a mirror). SS assume, correctly, that many philosophers hold the former theory. They infer, incorrectly, that many hold the latter theory (18, 23, 150).

Finally, SS appear to confuse the idea that beliefs are caused by actions with the idea that they are caused by nothing else. On the last page they write: ‘as we come to recognise the conventional and artefactual status of our forms of knowing, we put ourselves in a position to realise that it is ourselves and not reality that is responsible for what we know’ (344, my italics). In other words: the things we do affect our beliefs, therefore reality does not affect our beliefs. Compare: the things we eat affect our health, therefore the things we drink do not.

3. LAP preaches a method it does not practice. The authors advocate ‘an asymmetrical handling of accepted and rejected knowledge’ (11). They fall short of this ideal, not because they reject Boyle’s views in their last sentence (‘Hobbes was right’) but because the argument of the book hinges on such a rejection.

Why were Boyle’s experiments so hard to replicate? Because, SS argue, all experiments can be turned against their creators by ‘displaying the work’ that goes into them--in other words, because Hobbes was right (chap. 6, esp. 226, 281-2, 186).

Again: why was there ‘nothing inevitable’ (13) about the favourable reception of the experimental method? Because Hobbes’ ideas were ‘believable’ at the time (13). How do SS argue for this? In large part, by playing down the weaknesses in Hobbes’ ideas and playing up the weaknesses in Boyle’s. SS balk at the idea that Hobbes might have ‘misunderstood’ Boyle (that would be asymmetric, they say on p. 12), and they do not point out the inconsistencies in Hobbes’ views on man-made experiments (Hobbes says he rejects man-made experiments, yet he accepts the ones that support his metaphysics, and he says that man can know only what he makes). Yet SS invoke misunderstanding when Hobbes is not involved (‘failed to grasp,’ 266), and they frequently point out the inconsistencies in Boyle’s views (162, 163, 164, 166n., 175, 185, 187, 195n, 196).

I fear that, like Laplace’s demon, SS load the dice by opening and closing their critical faculties at strategic moments. LAP would be more convincing if the authors dropped their pretence of epistemic neutrality—their claim that they are ‘not evaluative’ (12)—and acknowledged that their argument depends on engaging the reader’s evaluations of both Boyle and Hobbes. 

4. LAP creates the false impression that the social and political dimensions of science have some sort of primacy. Yes, SS give blow-by-blow accounts of disputes in which they cover the material, literary and intellectual aspects of science in great detail. Yes, they suggest that the distinction between the socio-political dimension of science, and other dimensions of science, is wrong-headed (see point 5 below). They nevertheless imply that social order, social convention, and Restoration politics were paramount in the resolution of the Hobbes-Boyle dispute.

Recall the book’s main thesis: experimental philosophy flourished in Restoration England because it was seen as an acceptable model for the resolution of disputes in society at large. In SS’s words, the experimenters offered ‘a functioning example of how to organize and sustain a peaceable society between the extremes of tyranny and radical individualism’ (341).

A charitable reader might insist that SS leave the door open to other explanations of the experimenters’ success. Indeed, SS choose their words carefully: they say they are giving ‘an answer’ to the question of Boyle's success; they do not say they are giving ‘the answer’ (341, cf. ‘constituents’ on 283).

And yet, SS seem to rule out the obvious intellectual explanation of Boyle’s success. This explanation is that Boyle's contemporaries valued true theories for their own sake (not just because they were politically expedient), that they valued Boyle’s method because his experiments were careful and numerous (not just because they were part of an open and harmonious community), and that they learned about Boyle’s method through his writings and actions (not just through the propaganda that Thomas Sprat and others wrote on Boyle’s behalf).

SS do not exclude this explanation in so many words, but they drop several hints. They say that Hobbes’ methodological views were just as ‘believable’ as Boyle's, and that there was ‘nothing inherent in [Hobbes’ views] that prevented a different [ie. more favourable] evaluation’ (13). After summarising their thesis—that Hobbes and Boyle disagreed about the relationship between knowledge and social order—they write, emphatically, ‘That is what the Hobbes-Boyle controversies were about’ (15). They give a long list of ‘wants’ that the experimenters catered for, a list that does not include the desire to discover new and striking facts about nature (340-1). They conclude: ‘He who has the most, and the most powerful, allies wins’ (342).

As for the political, so for the social. The common theme of chapters 2, 5 and 6 is that the ‘criteria and rules’ of the epistemologist cannot account for the complexity of scientific disputes. Instead, the ‘ultimate justification’ of the disputants’ statements lies in ‘forms of life’ (read: ‘social order’, as per 14-15) and in ‘social conventions.’ Thus SS ‘add weight to our view of matters of fact as social conventions‘ (52, 226, 281-2).

I say that SS give a ‘false impression’ of the primacy of social and political factors because I find their claims for this primacy to be unclear, and insofar as they are clear, I find them unconvincing. I note the following:

- SS do not show that Hobbes’ ideas were ‘believable’ at the time, only that we can make them appear so by treating Hobbes and Boyle asymmetrically (see point 3 above). 

- SS do show that there is more to scientific debate than the application of a small number of explicit, exceptionless, and unambiguous rules. But they do not show that the ‘something more’ is ultimately social in character, because...

- they ignore plausible alternatives. Why not a large number of rules? Why not a small number of ambiguous rules? Why not epistemic conventions?

- they do not explain how conventions succeed where rules fail. They hint (52) that rules fail to explain Boyle’s statements about causal hypotheses because rules are regular whereas Boyle’s statements are irregular. But conventions are also regular!

- although they drop various hints about the meaning of ‘social,’ ‘convention,’ and ‘social order,’ they do not appear to explicitly define these terms. These terms are placed in italics (52), they appear in definitions of other terms (15), and they appear in statements of the book’s major theses (15, 282, 344). But their meaning is unclear to me.

- the case studies in chapters 5 and 6 do not show that social phenomena are primary. These chapters identify many causes of statements made by Boyle and others. Some of these causes look like ‘social conventions’ to me, but many do not. Are the former causes indeed the ‘ultimate’ ones? I am not convinced.

5. LAP contains a misleading account of the internal/external distinction. Earlier historians had either debated the significance of ‘internal’ versus ‘external’ factors in past science, or dismissed the whole debate as wrongheaded. SS write (342) that they take a new position: we should retain the notions ‘external’ and ‘internal,’ but we should treat them as actors’ categories. We should ask ‘how, as a matter of historical record, scientific actors allocated items with respect to their boundaries (not ours).’ We should ask, for example, why Hobbes maintained that certain topics (such as the existence of the absolute vacuum and the merits of absolute monarchy) were within the bounds of natural philosophy, whereas Boyle believed otherwise.

These claims should be taken with several grains of salt. As SS point out in their introduction to the second edition of LAP (xv, xxvii), the i/e distinction is closely associated with the intellectual/social distinction. And SS do not transcend the latter distinction. They favour one side of that distinction (see point 4 above), and in doing so they use terms (‘social order,’ ‘social convention,’ ‘form of life’) that were alien to both Hobbes and Boyle. In addition, SS’s account of the i/e distinction ignores a rich tradition of boundary studies by historians and philosophers of science such as Pierre Duhem, Alexandre Koyré, and Larry Laudan. These authors wrote extensively about the evolution of scientists’ views on the proper role of metaphysical and theological ideas in the natural sciences.

***

I conclude that the characteristic flaw of LAP is, appropriately, over-inflation. This is an oxygen-rich book that suffers from an admixture of hot air. The challenge is to separate the fresh air from the hot. This is not easy, since the two are thoroughly mixed. Here is my attempt, with the fresh air in roman type and the hot in italics:

The central insight of LAP, that different methods of science sometimes correspond to different sorts of polity, stands firm. In particular, experiment was an attractive model of dispute-resolution in Restoration politics. But LAP does not succeed in discrediting the traditional view that the experimental method flourished in part because it was perceived as a source of persuasive evidence for striking claims about nature (such as the claim that air is springy).

LAP is novel in its emphasis on the role of manual labour in experimental research, and in the sheer detail of its descriptions of experimental work, especially the work of replicating a result with a new-fangled instrument. But there is nothing new in the idea that science is difficult. And SS show only that experiments are difficult, not that they are ‘inherently defeasible’ (111-2) or that experimenters are bound to assume what they are trying to prove (226).

LAP also innovates by taking Hobbes’ ideas seriously and by pointing out some weaknesses in Boyle’s views. But SS have not shown that Hobbes’ views were, on the whole, more believable to his contemporaries than Boyle’s.

LAP reveals new aspects of the social history of early experimentation. Some of these aspects are of wide interest (eg. the strength of testimony depends on the social status of the testifier). Others are specific to their time and place (eg. Boyle hid some inconvenient experiments from his critics). But the concepts ‘form of life’ and ‘social convention’ are unclear, and they are misleading insofar as they suggest that SS have discovered a new mode of historical explanation and/or a new argument for the primacy of social phenomena in science. 

LAP go beyond the internal/external distinction in the sense that they describe debates about the place of political philosophy and religious apologetics in natural philosophy. But the idea of historicising the boundaries of science is at least a century old. And LAP is an externalist book, in the sense that it assigns a special role to social and political phenomena in the resolution of disputes about science.

Expand post.

Malheureusement je ne traduis plus tous les posts que j'écris sur ce blog. Si vous voudriez lire les posts qui ont été traduits, veuillez cliquer ici. Agrandir ce message.