[Functions associated with social spaces in which observers negotiate interpretations of each others’ actions]
language as moderator o f the private/public distinction: experimental uncertainty or disagreement with other experimenters allows for the reconstruction and reinterpretation of one’s own experience
language as link between experience and competence: the evidential status of results depends on negotiations about the quality of the setup and competence of experimentalists
language as measure o f understanding: a scientist must elicit responses from other people so as to tell whether they have grasped what he or she intends them
to — only then can outcomes (such as images and devices) be disseminated
language as pathway to veracity and fidelity: the potential of language to bring about trust in scientists’ honesty and accuracy in reporting their experience (and therefore in the experience itself)
language as currency o f an economy that depends on the division o f labour: the fixing of an image obviates the need for other observers to acquire the skills embodied in it
In this five-fold classification we find descriptions of functions of language that take us all the way from world A at the interface with world B, to world A as it appears in published reports. In the course of this journey language becomes involved in ‘producing phenomenal possibilities; inventing ways of making them into intelligible, shared experiences; constructing interpretations which make them relevant to theories; and verifying and demonstrating this relevance by reconstructing the activity as the brainchild of hypotheses or theories’. T a b l e 4, below, provides an illustration of the earlier stages of this sequence.
Table 4. Adaptation of Gooding’s abstraction of a sequence of material/conceptual operations from the first page of Faraday’s laboratory notes for 3 September 1821. Functions of language involved with each move appear in brackets as references to the classification of Table 3,‘^
set up apparatus to explore wire/needle interaction [1]
query battery configuration [3]
set up reference frame to record positions of wire and needle [1,2]
recall/invent representation of wire [2, 3]
record position of wire [1,4]
plan exploration of region of wire [2, 3]
ascertain needle behaviour [1,4]
recall/invent representation of needle behaviour [2, 3]
record needle behaviour [1,4]
repeat/re-examine previous sequence [1,3]
record needle behaviour (side view and top view) [1,4]
transform — represent effects as wire positions instead of needle positions [ 2, 3 , 4]
record effects as wire positions [3, 4]
construct new possibilities — ‘indicated’ by needle movements [3]
imagine continuous motion of wire perpendicular to fixed needle [3]
(...etc.)
Even though exploratory and incomplete, tables 3 and 4 demonstrate above all else that the ‘referential’ function of language is completely insignificant in science (it is a rhetorical, derivative function), i^i This is important because if world B is to be under-
•'♦0Adapted from ibid.. Table 6.3, p. 140.
•^'In Goodman’s Ways ofworldmaking, we find a sketchy and unelaborated classification of functions
of language that is in the spirit of Table 3. It is four-fold: (i) synthesis and analysis of
entities: ‘Much but by no means all worldmaking consists of taking apart and putting together, often conjointly: on the one hand, of dividing wholes into parts and partitioning kinds into subspecies, analysing complexes into component features, drawing distinctions; on the other hand, of composing wholes and kinds out of parts and members and subclasses, combining features into complexes, and
stood as a source of ‘resistances’ (as I think it should), my position invites the follow ing (dualist) rebuke: even if we accept that language contaminates resistances in the process of referring to them, who is to say that there are no things-in-themselves beyond those inputs that we are simply unable to refer to directly?
The first clause of this question is based on a factual error: laboratory practice reveals that language is never used to refer to inputs, nor is it employed in the vain hope of establishing reference to ontologies beyond world A. Table 3 underpins the conclusion that ‘the results scientists present as bold, self-evident facts about nature have been painstakingly elicited, shaped, and made visible through the invention of methods of observation and representation’. Language does not, as a matter of fact, ‘refer’ to world A or in any way reach out to world A — language is firmly part o f that very
world. In other words, dualism, which forges a disjunction of language and world and
reimposes a connection that is referential (see the drawings of Table 1), is refuted. 2.2.5
Phenomena, natural facts, and their effects have ‘no fixed, essential nature that can be assessed independently of the manipulations that [the construals of scientists] construe’. A t the same time they are not mere fictions. Gooding takes his analysis of the experimental making of meaning to support a special kind of realism — as he puts it ‘a modestly realistic interpretation of what representations are about’. I shall try to explain what this non-traditional realism amounts to, and demonstrate its irreducibility to the dualist conception of language.
Gooding’s discussion of realism proceeds from his aforementioned notion of the ‘convergence’ of the main features of the experimenter’s space. After Faraday
kinds of the one world, rather than being absent from the other, are present as irrelevant kinds; some differences among worlds are not so much in entities comprised as in emphasis or accent, and these differences are no less consequential’ (p. 11); (iii) deleting and supplem enting: T h e scientist [goes about] rejecting or purifying most of the entities and events of the world of ordinary things while generating quantities o f filling for curves suggested by sparse data, and erecting elaborate structures on the basis of meagre observations. Thus does he strive to build a world conforming to his chosen concepts and obeying his universal laws’ (p. 15); (iv) deform ing o r reshaping: ‘The physicist smooths out the simplest rough curve that fits all his data’ (p. 16). Needless to say, Goodman did not set out to produce a classification of functions of language.
•'‘^Gooding, ‘Thought in action’, p. 139.
'^^Gooding, Experiment and the making o f meaning, p. 186.
succeeded in realising continuous assisted rotation of the crank wire, he was faced with the task of building upon three related elements in his experience: (i) practical knowledge of the experimental conditions under which the crank would swing past the magnet; (ii) mental models of that phenomenon and of the phenomenon of unassisted rotation he sought; and (iii) a mental model of the apparatus that could realise unassisted rotation. By 4 September 1821 these elements had been developed and had converged
with the desired effect. Faraday had achieved mutual compatibility of his construals of phenomena, his models of the manipulations and workings of (actual and conjectural) experimental apparatus, and his observations and interpretations.^"*^
According to Gooding’s argument, Faraday’s pursuit of circular ‘natural motion’ shows that convergence is not easily achieved (and that it needn’t come at all, or may be quickly dissolved), and is always the result of revisions of construals and material arrangements, and accumulation of theoretical and observational skills, in the course of making experience intelhgible. Experience begins to become intelligible when models and practices yield phenomena that are stable and reproducible and satisfy tentative expectations about outcomes.
Convergence in Faraday’s experiments of September 1821 came when he achieved rotation of a crank wire while moving a magnet to and fro. The result suggested ways of simplifying the configuration of wire and magnet and producing an orderly, seemingly non-artefactual natural phenomenon. It also clarified the reasons behind earlier unexpected phenomena, such as the continuous sideways motion of the wire, and inspired a tentative distinction between electric current and conducting matter. Is it reasonable to believe that when convergence occurs scientists possibly learn something about the way the world is? Does the term ‘convergence’ imply that experimentation is
'^^Gooding’s construals, models, and interpretations correspond closely to Pickering’s own triumvirate of material procedures, instrumental models, and phenomenal models; ‘By material procedure, I refer to experimental action in the material world: setting up the apparatus, running and monitoring it in the laboratory. The instrumental model expresses the experimenter’s conceptual understanding of how the apparatus functions, and is central to the design, performance and interpretation of the experim ent... The third element, which endows experimental findings with meaning and significance, is a pheno menal model, a conceptual understanding of whatever aspect of the phenomenal world is under investigation’ ( ‘Living in the material world’, p. 277). For fact-production to occur, the three must
hang together and reinforce one another. Pickering’s term for this achieved relation is 'coherence' (for
an example, see ibid., pp. 286-287).
'"‘^See Faraday’s compact rotation device in Experiment and the making o f meaning, p. 152. For other
a process converging to something besides a certain sense of order, reproducibility, and control? What room, if any, is there here for realism?
Gooding’s response is in two parts. First, he argues that Faraday’s systematic reconstruction of his experience, conceptual and practical, clearly shows that he engaged factors beyond his control. His command over his apparatus, his construals and expectations only partially determined his experience. Emerging interpretations were a consequence of a continuing engagement with nature. For Gooding this much realism or empiricism is inescapable. Contact with nature is demonstrable early on in the lives of theories, in the course of what might be called a ‘proto-theoretical’ stage. It is manifested in the recalcitrance of nature to be manipulated and construed in certain ways.^"*®
Gooding is keen to point out, however, that engagement with nature need not compel assent about empirical matters. Both Faraday and Ampère, for example, engaged nature but were led to different theoretical interpretations. As for the natural facts that emerge from engagement with nature, Gooding comments on the considerable effort needed, both in the accumulation of skills and the fine-tuning of experimental apparatus, to elicit and distinguish very specific sought-after outcomes from the large number of potential outcomes that engagements with nature reveal:
During a repetition of [trials with a replica of Faraday’s early rotation device] we found it very difficult to obtain rotations even with the ‘correct’ configu ration. We made trial and error adjustments to the depth of immersion of the wire, to balance the effect of the viscosity of the mercury (which inhibits the motion) against the effect of allowing the wire too much freedom of motion. If not immersed deeply enough, it will oscillate like a pendulum.
Gooding distances himself from realists who maintain that engagement occurs when theories are confronted with decisive experimental results, for one can show that experimental results need not be treated as theoretically decisive. At the same time,
'^^Gooding, Experiment and the making o f meaning, pp. 179-180.
' “^^Davy and Biot elicited phenomena by using very similar experimental setups, deriving from Oersted’s experiments (ibid., p. 50). Thus, experimenters with different theoretical views, in different parts of the world, elicited very similar novel phenomena which they at first construed in very similar ways. The ‘engagement’ argument runs as follows: the fact that people in different places with different ideas using different languages can elicit similar phenomena (and construe them similarly) shows that there is a factor, nature, that is independent of their cultural circumstances and explains (through its pre-given make-up) the coincidence of their accounts (for more on this argument, see section 2.3.3 of this thesis).
Gooding maintains, one can hardly deny that empirical constraints do often operate decisively at the level of (what I have called) proto-theoretical experimentation. The empiricists’ desire to locate engagement with nature in corroboration or falsification has meant a neglect of its presence in the early processes of articulation: ‘the empirical basis that really constrains theories is found in experimentation itself, not in observation reports of experimental results’. R h e t o r i c and traditions of narrative writing are responsible for the placement (or misplacement) of the empirical base in observation reports, and for the notion of an unmediated compulsory force of nature.
Having thus indicated and secured a place for empirical constraints in his scheme, Gooding goes on to attack the realist metaphor of gradual approximation of theories to reality, or convergent r e a l i s m . H e argues that the metaphor of gradual approximation derives from the consideration of polished scientific accounts — a metaphor, in Kuhn’s words, ‘conditioned by science texts’ — in which the testing and matching of predicted and observed values has been given rhetorical priority, and from which most of the process of developing and defending theories has been edited out.^^^ In other words (and in the context of Part 2 this is becoming a familiar theme), the appearance of correspondence of words to things, and of the convergence of theoretical
'^°Ibid., p. 193. Herein lies the difference between Gooding’s account and Quine’s ‘ontological relativity’. In ‘Two dogmas of empiricism’ {From a logical point o f view, pp. 20-46), Quine talks of the raw experience upon which we impose order by means of a conceptual scheme. Our conceptual scheme must accommodate this experience, but is under-determined by it. We impose order on raw experience by cutting it up into objects. There are different ways of doing this, different objects that can be posited. In ordinary common sense we posit physical objects — but even here we can conceive of alternative schemes, which cut experience up into different objects — rabbits, rabbit stages, rabbit- bushes, etc. And of course we might choose to posit other objects: atoms, sense-data, numbers, etc., in addition, or instead of the ordinary objects of our conceptual scheme. Which theory (which ontology) we choose depends, for Quine, solely on the explanatory value its objects hold fo r us. This emphasis on explanation (and theory) completely fails to acknowledge Gooding’s level of practice (and direct constraint). Putnam approaches the matter from the same angle, with a similar result: ‘Intemalism does not deny that there are experiential inputs to knowledge; knowledge is not a story
with no constraints except internal coherence; but it does deny that there are any inputs which are not
themselves to some extent shaped by our concepts, by the vocabulary we use to report and describe
them, or any inputs which admit of only one description, independent of all conceptual choices. ... The
very inputs upon which our knowledge is based are conceptually contaminated’ {Reason, truth and
history, p. 54).
*^'For two varieties of convergent realism, see Experiment and the making o f meaning, p. 185.
propositions on real structures, is a made image that has had a powerful but unwar ranted influence in philosophy. Gooding recommends that convergent realism be replaced by ‘asymptotic realism’, an image meant to preserve (as I explain below) the central thesis of his analysis, viz. that what is disclosed about nature is made, not discovered ready-made. Alternative pathways in Faraday’s researches, defined by different sets of preferences and different end results, and evident in Gooding’s maps of experimental cognitive practice, demonstrate that practices do not converge (as convergent realists might wish) to an outcome that is given, or ‘there’ to be discovered.
Gooding depicts asymptotic realism as the approach of two curves, a and b, having the jc,y coordinates of a Cartesian system as asymptotes (in his diagram they converge upwards along either side of the positive y values). Curve a is the curve of representations and theoretical practice; b is the curve of phenomena that scientists make
stable (and seemingly real) by manipulating and construing nature (both curves belong
to world The curves represent different aspects of the engagements of human agency with nature (world B), even though each one is a compound of both agency and recalcitrant nature. Their mutual embodiment of empirical constraints means that they need not (and most often do not) converge dramatically in practice. But when they do converge, a very real psychological sense of correspondence is inevitable, which then finds expression in conventional objectifying rhetoric (discussed in Part 2.1). Gooding thus displaces the ‘metaphysical convergence’ of convergent realism with his own notion of procedural (hands on, psychological/rhetorical) convergence. As he explains,
when the mutual approach becomes sufficiently close and sufficiently enduring, convergence is taken to be correspondence between representations in the conceptual world to real things in the material world.
'^^See ibid., p. 154.
'^'’Pickering writes: ‘The interaction between the instrumental and phenomenal images ... hints at a potential anarchy in the empirical base of science, with individual experimenters free to produce whatever data suit their fancy. One can, however, see that such disquiet is unnecessary, if one is willing to regard science as a social enterprise. Experimenters have to argue for acceptance of their results within their community, and when they do, the instrumental and phenomenal images are part of the currency of exchange. Experiments are performed and evaluated within a socially sustained matrix of commitments — beliefs and practices — and this matrix severely constrains the acceptable constructions an experimenter can place upon his esoteric experience. The empirical base of science is ‘softer’ than it is often conceived to be, but it is not anarchic’ (‘The hunting of the quark’, p. 235).
Experiment and the making o f meaning, p. 187. Cf. ‘asymptotic convergence’ in Honner, The description o f nature, p. 21.
It is at that point, in other words, that practicing scientists become satisfied that their talk and thought has engaged the world and revealed pre-existing structures. Pheno mena that were at first made ‘easy to see’ and then ‘evident’ finally are seen to exist ‘out there’. The role of human intervention is gradually removed from accounts altogether, and phenomena appear as independently g i v e n . C o r r e s p o n d e n c e is the psychological and rhetorical upshot of the mutual approach of the products of experimental practice. But although the correspondence of representations to entities and properties is a made relationship, it is not arbitrary. It is grounded in the resistances of practice and constrained by the mutual pursuit of consensus.
Gooding’s asymptotic realism thus accounts for an appearance of language/world fit without relying on the language/world disjunction that convergent realism presupposes. But is asymptotic realism really realism! Gooding argues that the direct and analysable relation between world A and world B is one of made convergence, not natural correspondence. A realistic role for experiment is secured by the empirical constraints to which the processes of experimentation are subject. Theory, with its origins in experiment, is a product of these constraints despite the fact that it is empirically