Kuhnian paradigms

Thomas Kuhn’s The Structure of Scientific Revolutions systematised a thesis which had previously been only put forward informally in the earlier theories of authors such as Koyre, Polanyi and Feyerband. The proposition that science proceeds not as an accumulation of facts which serve to inductively corroborate some theory, the classical notion of science, but as a series of paradigms which instead determine what sorts of facts may be produced, was succinctly and compellingly elaborated here. Since its publication, it has tremendous influence in the history of science and, arguably, even greater appeal through the ‘trickle-down’ dissemination of key terms into the broader cultural lexicon. In the practice of historical and social studies of science, rival models such as Actor-Network Theory now hold sway. Nevertheless Kuhn’s thesis is a pivotal moment in the widespread recognition that the movement of science is marked not only by quantitative epistemic growth but also, and more foundationally, through qualitative epistemic shifts. These qualitative shifts are described by two key concepts, which warrant further exploration: paradigms and incommensurability.

Paradigms are, for Kuhn, the epistemological conditions under which the normal enterprise of science operates. They are the ‘ways of knowing’ which make possible the posing and solving of puzzles: ‘The existence of this strong network of commitments—conceptual, theoretical, instrumental and methodological—is a principle source of the metaphor that relates normal science to puzzle-solving’ (Kuhn 1970, p. 42). While practitioners of science exhibit a kind of ‘know-how’ imparted by a paradigm, this does not necessarily constitute ‘knowing-that’ the paradigm exists; paradigms need not be, and only infrequently are, explicit: ‘Scientists work from models acquired through education and through subsequent exposure to the literature often without quite knowing or needing to know what characteristics have given these models the status of community paradigms’ (Kuhn 1970, p. 46). Paradigms are also the grounding conditions of the scientific practices they make possible, and cannot be made explicit as form of rules, statements or axioms: ‘Paradigms may be prior to, more binding, and more complete than any set of rules for research that could be unequivocally abstracted from them’ (Kuhn 1970, p. 46). As Kuhn notes elsewhere, paradigms have a very broad extensional definition—seeming to mean a set of both implied and explicit epistemic commitments; a meta-theoretical ground on which more granular theories and experimental practice can be constructed; a process of socialisation—what scientists do to become scientists; and an historically, spatially or epistemically bounded field, separated from others before, after or around it. In fact each of these definitional facets permit further generous extensions in the application of ‘paradigm’ (or its eventual cognate partner employed here) (Kuhn 2002).

On the one hand, following its popular usage there seems little reason to limit the use of paradigm to scientific forms of knowledge only. Other disciplinary fields—academic and industrial—seem to meet some if not all of the qualifying features of paradigms. Management practices, for instance, operate within broad cultural conceptual models, which seem to undergo more or less radical transitions when they lose their efficacy for problem solving. Moreover, often the distinction between a paradigm of rigorous science and, perhaps, a broader paradigm of the science and its applications are difficult to distinguish, particularly as science itself frequently bifurcates into sub-disciplinary and inter-disciplinary fields—ontology matching, introduced in earlier chapters, is a good example of a self-contained field that borrows approaches from beyond its computational science borders. Paradigms applied to clear historical revolutions—the Copernican turn in cosmology, Newton’s laws of physics, Maxwell’s thermodynamic equations, Darwin’s theory of evolution, to take some of Kuhn’s examples—show comparatively harsh lines of theoretical demarcation, and make a compelling narrative of different ways of seeing and knowing experienced through various paradigmatic lens. In the twilight world of less spectacular transitions—but no less revolutionary, for those who go through them—such lines are more blurry. Knowledge systems constitute ideal candidate expressions of more fine-grained paradigms, and these paradigms need considerably more elastic definition.

Similarly, for Kuhn—as for Foucault, further below—paradigms are invariably expressed in historical terms; paradigms succeed each other. Yet at a more granular level, different spheres of knowledge frequently coincide, compete, intersect and merge. Kuhn allows, for example, for a physicist and a chemist to provide different definitions of a molecule, based on the different paradigms they operate under (Kuhn 1970). However, even this example assumes too rigid a distinction around an agent’s paradigmatic engagement—as though one can be either a physicist or a chemist, but not both. Paradigms operating at a useful enough level of granularity to be applied to knowledge systems need to be the sorts of structures which agents can belong to severally, and for limited durations—enough to allow self-conscious and self-reflexive examination of the key commitments belonging to one or another co-existing paradigm. This is not an arbitrary mandate imposed here, but rather a feature of the kinds of analysis which take place in the translation of knowledge systems. System analysts, at least on an implicit, phenomenological level, frequently operate within a cognitive geometry of rotating paradigms—figuring out how to describe a concept first one way, then another; how to classify an object one way, and then another; and so on. Kuhn’s paradigms all but cover these scenarios, but need to be extended down towards the ‘micro-paradigmatic’ level to be useful here. I therefore look to carry across the insights of Kuhn’s definition of ‘paradigm’, as a general set of epistemological commitments and practices, into the theoretical dimensions of the framework, while looking to refine it into a concept more applicable to more fine-grained knowledge structures.

Kuhn also introduces the even more critical term—in the context of the present study—of ‘incommensurability’. This is used in the The Structure of Scientific Revolutions to describe how rival paradigms hold ‘incommensurable ways of seeing the world and of practicing science in it’ (Kuhn 1970, p. 4). As discussed below, incommensurability is a controversial term. Incommensurable paradigms suggest, literally, that propositions belonging to one are untranslatable into those of the other. By extension, these propositions are true or false only within the frame of reference of the current paradigm. Kuhn first depicts incommensurability in optical rather than linguistic metaphors. In a key chapter titled ‘Revolutions as changes of world view’, he works carefully to avoid the most obvious accusations of relativism:

Kuhn emphasises, over and over, how different paradigms cause scientists literally to see objects differently. The constellation of concepts which compose a given paradigm structure the observational work of science. Echoing Sellars’ critique of empiricism in particular, Kuhn writes: ‘The operations and measurements that a scientist undertakes in the laboratory are not “the given” of experience but rather “the collected with difficulty”’ (Kuhn 1970, p. 126).

Moreover empirical data is experienced against a paradigmatic backdrop which provides a pre-existing construct for how the data can be organised and interpreted:

As these visual cues suggest, Kuhn wants perspectives to be incommensurable to the extent that perceived objects are organised in different conceptual configurations and constellations. In a later postscript (written after, and in response to, Davidson’s critique discussed below), Kuhn asserts this incommensurability is not merely terminological, and ‘they cannot be resolved simply by stipulating the definitions of troublesome terms’ (Kuhn 1970). There is, on the one hand, no immediate language capable of neutralising these conceptual and perspectival differences: ‘The claim that two theories are incommensurable is then the claim that there is no language, neutral or otherwise, into which both theories, conceived as sets of sentences, can be translated without residue or loss’ (Kuhn 2002, p. 36).

Yet, on the other, there must be a way out of the solipsistic world of a paradigm. There is both the direct ‘stimuli’ of experience, and a broader communal culture that agents under incommensurable paradigms share. This makes possible, not a neutral language, but translation between two paradigmatically committed languages: ‘Briefly put, what the participants in a communication breakdown can do is recognize each other as members of different language communities and then become translators’ (Kuhn 1970, p. 202).

Translation is then the vehicle which begins the process of persuasion and conversion to move from one paradigm to another. However, it is frequently insufficient: ‘To translate a theory or world view into one’s own language is not to make it one’s own. For that one must go native, discover that one is thinking and working in, not simply translating out of, a language that was previously foreign’ (Kuhn 1970, p. 204, my emphasis).

Paradigms are more, then, than use of a language—they are ways of knowing, ‘thinking and working’. Until scientists are thinking and acting as though the world operates under Einstein’s theory of relativity, they are not operating within the paradigm of relativity.

It is not clear, however, that these clarifications are adequate. In spite of Kuhn’s caveats, the edges or boundaries of paradigms seem immutably hard, and nothing explains quite how, except by a process of gradual elision, an individual agent’s views shift to the point of belonging to another paradigm. Yet this introduces the pain of regress to an infinity of intermediate paradigms along the way, unacceptably diluting the concept altogether. One of the steps through this difficulty is to suggest that paradigm are structures of considerably greater elasticity than Kuhn wants to allow—that ‘scientific revolutions’ represent simply quantitatively greater shifts than other, more fine-grained changes in perspectival position. Paradigmatic incommensurability then becomes a question of degree rather than kind—a measure of ‘semantic heterogeneity’ rather than a quality of conceptual structures.

Foucauldian epistemes

Foucault’s notion of episteme is more difficult to trace and delineate than Kuhnian paradigms. As Hacking (2002) notes, Foucault rarely lays out explicit definitions of concepts, nor holds on to them for long. In one of his more programmatic declarations, The Order of Things introduces epistemes indirectly as a means for describing epistemological affinities between different strands of intellectual production. To characterise the sense of the modern episteme, Foucault describes the rise of the classificatory disciplines—biology, economics and grammar—in the late eighteenth century:

In this and other descriptions, several qualities of the episteme emerge: it is historical and, unlike a paradigm, situated across rather than within particular knowledge fields. Here, an episteme appears resolutely structural—a broad, temporal swathe of beliefs and practices which constitute both our subjectivity and our concepts. For Foucault, as for Kuhn, these structures affect both ways of seeing and speaking: ‘What came surreptitiously into being between the age of the theatre and that of the catalogue was not the desire for knowledge, but a new way of connecting things both to the eye and to discourse. A new way of making history’ (Foucault 1970, p. 143).

Later, in the more programmatic text of The Archaeology of Knowledge, Foucault explicitly defines an episteme:

Further on, he elaborates:

In this later explication, the structural features of an episteme appear looser, more elastic. Moreover, an episteme is directed less towards the conceptual, and more towards the discursive properties of knowledge formation. An episteme here seems disembodied of the ‘us’ which figured heavily in the description above. Rather it operates as a sort of content-less grid, a network of rules which permits certain kinds of discourse to emerge and be treated as ‘science’ or ‘knowledge’. Naturally there remain human agents in the background, enacting discursive practices, reading, writing and interpreting. However, these are extraneous to the construction of an episteme—what matters are the discursive ‘relations’, ‘rules’ and ‘regularities’. These in turn stand in various relationships to other kinds of non-discursive practices, particularly those employed in the administration of power. For Foucault, clinics, asylums and prisons are the sites par excellence, where power-driven practices, discursive and otherwise, intersect—and where an analysis of these practices might lead to their revision or erosion. Since ‘science’, as a privileged form of epistemic practice, invests particular statements with a peculiar vindicating force for the exercise of power, making explicit the conditions that particularise and legitimate them also exposes them to critique: ‘In the enigma of scientific discourse, what the analysis of the episteme questions is not its right to be a science, but the fact that it exists’ (Foucault 2002, p. 212).

Elsewhere in interviews Foucault has also emphasised the constraining character of an episteme:

In relation to Kuhnian paradigms, then, epistemes clearly have points of affinity and divergence. Among the more notable distinctions:

There are, however, features common to paradigms and epistemes: they are largely unconscious for those who engage them, and require particular interpretive tools or methods to discover them; they are not purely conceptual constructs, but also are exhibited within social practices (discursive and otherwise); both reflect social features of a culture or community, rather than an individualistic psychological state of mind; and both involve some form of ‘orientation’ towards the objects under description—through orientation must be understood in an active sense, as actively constituting rather than merely passively observing those objects. These features carry over to the description of a more fine-grained construct in the framework ahead. Both also offer the problem of how statements can be translated across epistemic or paradigmatic boundaries—a theme returned to below.

Quinean conceptual schemes

Quine’s notion of conceptual schemes fits within the generously extended family, which also includes Kuhnian paradigms and Foucauldian epistemes. As with epistemes, it is difficult to get a definitive view as to what a conceptual scheme is. In a late interview Quine admits: ‘The only meaning I attached to it is a vague one. Namely, the conceptual scheme would be the more abstract general structure of one’s overall theory’ (Quine 1992).

Echoing a famous Quinean sentiment, that science is ‘self-conscious common sense’, conceptual schemes embrace theories of both scientific and everyday varieties (Quine 1964). While conceptual schemes might connote a sort of cognitive structure, for Quine, they are equally ‘cultural posits’. As theories, ultimately they have an explanatory function:

Conceptual schemes of all sorts are irretrievably bound to a language, for Quine: ‘Conceptualization on any considerable scale is inseparable from language’ (Quine 1964, p. 3). He is keen to show that linguistic translation generally, and the special kind of conceptual scheme inherent in scientific theory specifically, suffer the same fate of indeterminacy. For any set of sensory stimuli, there are many more or less meritorious theories to account for them; analogously, for any source language, there are many more or less adequate translations into a target language. For Quine, the idea of a direct ‘sense-datum language’ capable of reporting just the facts is one of the two infamous ‘dogmas of empiricism’, which he attributes to the more naive leanings of positivism (the other dogma is that there can be a strict separation between analytic and synthetic statements) (Quine 1980). Empiricism instead must be rehabituated to the a priori conceptual structures capable of organising perceptions, or else suffer a form of circularity: ‘small wonder that the quest for sense data should be guided by the same sort of knowledge that prompts it’ (Quine 1964, p. 2).

Unlike the rigid characterisation of paradigms, conceptual schemes admit of ongoing partial and ‘self-conscious’ revision. Scientific discovery differs in degree rather than kind from other kinds of belief, precisely because it accepts revisionary evidence. Rather than the relativism of pluralised conceptual schemes, though, Quine thinks it is possible to be epistemically committed to a set of beliefs when none better present themselves:

The problem of under-determination does not just relate to the relationship of data to theory—it equally applies to translation of one theory to another. Multiple viable translations are always possible: ‘manuals for translating one language into another can be set up in divergent ways, all compatible with the totality of speech dispositions, yet incompatible with one another’ (Quine 1980, p. 27). Here it is not that the languages or theories are untranslatable; rather, they are translatable in many different ways. And it is the translations themselves which are incommensurable—‘incompatible’, in Quine’s words—with each other. He emphasises this point with a lengthy thought experiment—what happens when a linguist is faced with some unknown language presented by an informant? The preconditions of the experiment are that there is no interpreter, and the language is not a ‘kindred’ one for the linguist (bears no common cognates or derivations). The linguist is presented with a word, ‘gavagai’, in the presence of a rabbit. The linguist then employs a steady process of elimination, with the aim of deriving the translation ‘gavagai = rabbit’. One of the eventually excluded possibilities is that of equivalence based on stimulus meaning, the sense derived from pure experience of the rabbit-phenomenon: perhaps the word ‘gavagai’ relates to this perception and not the actual animal? (Quine 1964, pp. 31–35). Another is that of occasion sentences, which refer to a particular circumstance of perceiving the rabbit—is the perceiver in a state of ‘gavagai’, then, on remarking of the rabbit? Or is the rabbit instead in a particular state indicated by ‘gavagai’, for example, that of being in position to be shot? (Quine 1964, pp. 35–41). Yet another possibility is that of observational sentences: does ‘gavagai’ relate instead to the phenomenological happening of an observation of a rabbit? (Quine 1964, pp. 41–46). Even after these possibilities are systematically eliminated under controlled circumstances, there is no guarantee of synonymity: ‘Who knows but what the objects to which this term applies are not rabbits after all, but mere stages, or brief temporal segments, of rabbits?’ (Quine 1964, p. 51). Even if the stimulus meaning—the responses of the informant hearing ‘gavagai’ on sight of the rabbit—is consistent with the linguist’s interpretation of the word ‘rabbit’, Quine argues that the extra-stimuli requirements for understanding the use of a term differ radically: ‘the whole apparatus [for using a term] are interdependent, and the very notion of term is as provincial to our culture as are those associated devices… Occasion sentences and stimulus meaning are general coin; terms and reference are local to our conceptual scheme’ (Quine 1964, p. 53). There is ultimately no recourse to sense data or an ideal language for devising a singular mapping between sentences or theory-fragments:

As noted earlier, conceptual schemes seem broadly contiguous with other structural notions of paradigms and epistemes. Quine’s critique of positivism is derived from a logical rather than historical analysis, and consequently specific kinds of historical and cultural ‘posits’ are argued for analytically rather than demonstrated through cases. Consequently their definition is considerably abstract; nevertheless some concrete characteristics can be identified:

There are several implications of Quine’s analysis for the treatment of knowledge systems: first, it admits the possibility both of multiple, equally valid knowledge systems in a given domain, and also, of multiple ways of aligning or matching these systems. Second, such systems do not describe the ‘sense data’, or objects belonging to a domain, neutrally, but bring to bear the background cultural ‘posits’ inherited by their authors. Third, these ‘posits’ invariably undergo revision—hence schemes should not be seen in rigid terms like the systems themselves, but as flexible networks which ‘cohere’ into reified form in systems. Hence uncovering background concepts is necessarily a more heuristic and interpretive exercise.

These three accounts—Kuhnian paradigms, Foucauldian epistemes and Quinean conceptual schemes—are now ‘classic’ structural treatments of knowledge systems in the broad sense, and deserve detailed coverage. The theoretical underpinnings of the ontology commensurability framework build on these treatments, but rely on more recent work of Habermas, Brandom and Gardenfors, which adds both greater descriptive precision about what schemes are, and causal suggestiveness for how it is that multiple schemes arise. In the meantime, the specifically structural character of conceptual schemes and its analogues was to receive significant critical attention in the 1970s; two of these critical lines are reviewed in the next section.

‘On the very idea’…

Davidson explicitly responds to the positions espoused by Kuhn and Quine, suggesting they represent various forms of conceptual relativism. Bracketing Whorfian languages and Kuhnian paradigms under the general heading of Quinean ‘conceptual schemes’, Davidson offers a basic outline of the conceptual relativist’s position:

The inherent paradox involved in such as position is that explaining how conceptual schemes differ presupposes an underlying common language—and corresponding conceptual scheme. The act of description shows how the terms of both schemes can be mapped onto a higher level set of terms, the language of the description itself: ‘Different points of view make sense, but only if there is a common co-ordinate system on which to plot them; yet the existence of a common system belies the claim of dramatic incomparability’ (Davidson 2006, p. 197).

Davidson distinguishes between two ‘failures of translatability’—complete and partial (Davidson 2006, p. 198). Complete untranslatability means no substantial translation between two languages; partial means some sentences can be translated, some cannot. In going after complete untranslatability, Davidson further distinguishes between the following two scenarios: the first, where speakers talk about different worlds using the same language; the second, where speakers talk about the same world using different languages. The first case makes the mistake of exercising one of Quine’s dogmas of empiricism, relying on the distinction between analytic and synthetic truths. The second case, Davidson argues, commits an analogous though different sin of relying on a distinction between ‘scheme’ and ‘content’. This so-called ‘third dogma’—extending the two Quinean dogmas discussed above—is one which Quine himself is guilty of, as are Kuhn and other purveyors of conceptual relativism.

Davidson proceeds to argue that this dogma assumes conceptual schemes are seen as either ‘organising’ or ‘fitting in’ with some ‘content’—in turn, such content is either ‘reality’ or ‘experience’. Taken actively, schemes must organise pluralities of worldly or experiential content. But such pluralities must, for Davidson, consist of observable ‘individualities’, which in the final resort can be shown and demonstrated to other linguistic speakers, even when languages differ. In such cases, ‘a language that organizes such entities must be a language very like our own’ (Davidson 2006, p. 203). Taken passively, schemes must generate true statements of the content they purport to represent: ‘the point is that for a theory to fit or face up to the totality of possible sensory evidence is for that theory to be true’ (Davidson 2006, p. 204). For the thesis of complete untranslatability, it must be possible for two schemes both to be ‘largely true but untranslatable’ (Davidson 2006, p. 205). Davidson then invokes Tarski’s Convention T to demonstrate that on ‘our best intuition as to how the concept of truth is used’, translation is in fact an ‘essential notion’:

Understanding that a sentence is true in another language involves being able to translate that sentence, or a description of it, into a native language. Consequently, a theory of schemes which are mutually and completely untranslatable (or incommensurable), yet which together contain large sets of true sentences, must lapse into incoherence.

The case for partial untranslatability (or incommensurability) rests on this first result. Davidson invokes what he terms a ‘principle of charity’—we understand speakers of other languages (and scientific theories or conceptual schemes) by ‘knowing or assuming a great deal about the speaker’s beliefs’ (Davidson 2006, p. 207). Accordingly, ‘we make maximum sense of the words and thoughts of others when we interpret in a way which optimizes agreement (this includes room, as we said, for explicable error, i.e. differences of opinion)’ (Davidson 2006, p. 207). On this view, substantial agreement needs to precede even the possibility of disagreement; rival conceptual schemes are deflated for local ‘differences of opinion’. Less otiose, for Davidson partial untranslatability is nonetheless equally incoherent, and as a final consequence, the third dogma of empiricism—that of scheme and reality—can at this point be happily dispensed with: ‘In giving up the dualism of scheme and world, we do not give up the world, but reestablish unmediated touch with the familiar objects whose antics make our sentences and opinions true or false’ (Davidson 2006, p. 208).

The incommensurability of madness

Derrida’s early critique of Foucault, written in 1963, is lengthy and complex, pivoting in part on a close reading and interpretation of Descartes, and of Foucault’s reading of Descartes (Derrida 2001). In broad terms it mirrors the broad structure of the critique Davidson levies at Kuhn and Quine; it is also worth noting that this critique is directed towards an early variant of Foucault’s ‘structuralist’ thought, which is revised heavily by the time of the works cited above. Rather than follow the argument in depth, I extract some salient lines of critique to demonstrate the analogy with Davidson; points which also go to the heart of the structuralist program.

In Madness and Civilization, Foucault attempted to mark an epistemic break in the eighteenth century in the dialogue between Reason and Madness (Foucault 1965). During the Middle Ages and the Renaissance, Foucault argues, Madness could be publicly paraded—literally and discursively—as evidenced in the texts of Rabelais, Shakespeare and Cervantes. The Enlightenment marks the breakdown of this dialogue into two separate monologues; or rather, as madness became increasingly silenced through the general institutionalisation of rationality and the specific institutions of psychiatry, into the monologue of Reason and the silence of Madness. Foucault’s project is to provide a history of that silence, to be the voice through which, retrospectively and belatedly, Madness can once again speak. Transposed to more convenient vocabulary, Reason and Madness are historically incommensurable, riven apart by the rationalist episteme since the Enlightenment.

Derrida’s overt object of criticism is whether this project is possible and coherent—or whether, rather, in attempting to liberate Madness from its Enlightenment constraints, Foucault merely repeats the constraining gestures of Reason: ‘Would not the archaeology of silence by the most efficacious and subtle restoration, the repetition, in the most irreducibly ambiguous meaning of the word, of the act perpetrated against madness—and be so at the very moment when this act is denounced?’ (Derrida 2001, p. 41).

And later: ‘Thus, not an expediency, but a different and more ambitious design, one that should lead to a praise of reason… but this time of a reason more profound than that which opposes and determines itself in a historically determined conflict [with madness]’ (Derrida 2001, p. 51).

More obliquely, Derrida also critiques the historical structures Foucault erects to describe the broad epistemic transitions from a dialogical—if strained—relationship between Reason and Madness towards an exclusionary one marked by the advent of the Enlightenment. At one point Derrida directs attention to the apparent arbitrariness with which Foucault cites Descartes as an exemplar of a new, emerging and ominously silencing attitude towards Madness, precisely as Reason is receiving its definitive articulation: ‘It is an example as sample, and not as model’ (Derrida 2001, p. 51). What epistemic status, then, does this ‘sample’ have in a history of madness? Its representativeness of Enlightenment constructions of madness, at least, is for Derrida highly questionable.

Later, describing Foucault’s project, Derrida writes: ‘But I wonder whether, when one is concerned with history (and Foucault wants to write a history), a strict structuralism is possible, and, especially, whether, if only for the sake of order and within the order of its own descriptions, such a study can avoid all etiological questions’ (Derrida 2001, p. 52).

Derrida wants to ask what relationship a singular passage from Descartes—later to be analysed in considerable depth—has to the broad historical structure Foucault seeks to account for: ‘Is this “act of force” described in the dimension of theoretical knowledge and metaphysics [that of Descartes’ Meditations], a symptom, a cause, a language?’ (Derrida 2001, p. 53). If an example does not function as either cause or effect, but simply counts as a kind of suggestive evidence of the existence of a structure, what motivates its selection—and not others, in particular potential counter-examples? Although at some remove from Davidson’s line of argument, there is a common concern with the relationship of the ‘totality’ of the structure to its parts—’individualities’ for Davidson, the unspecified ‘exemplarity’ of Descartes for Derrida. A lack of methodological specificity regarding the role of exemplary parts is met by an equivalent concern voiced earlier: that different historical periods have complex, shifting and overlapping trajectories regarding the conceptual delineations they make. Consequently it is difficult to voice the history of a notion—’Madness’—which itself has undergone considerable transformation over time: ‘Foucault, in rejecting the psychiatric or philosophical material that has always emprisoned the mad, winds up employing—inevitably—a popular and equivocal notion of madness, taken from an unverifiable source’ (Derrida 2001, p. 49).

Just as, for Davidson, the attempt to separate content and scheme collapses—and along with it, the problem of schematic incommensurability—so, for Derrida, the various cases of conceptual and historical structures—between, respectively, the concepts of Madness and History, and medieval and classical periods of treatment of madness—are compromised. On the one hand the conceptual, structural opposition between Reason and Madness is shown to be more complicated—that trying to speak the history of a singularised entity called ‘Madness’ risks objectifying it as an object of an historicising and alienating Reason all over again. On the other hand, the historical ‘structures’ in which these concepts figure are shown to be less stable, and less demonstrable by way of exemplary cases, than Foucault’s clear delineations might suggest.

Resurrecting structures

Davidson and Derrida’s critiques are of the more strenuous variety directed towards the implied relativist tendencies in any talk of conceptual structure, in either abstract or concrete historical terms. It seems anachronistic, then, to commit to untranslatable ‘ontological’ entities such as paradigms, epistemes and schemes. Yet, as discussed below, geometric metaphorisations of cultural and cognitive structure continue to emerge in more recent theories. One way of avoiding the types of traps Davidson and Derrida might lay out for prospective structuralist tendencies of this sort would be to suggest such theories adopt the kinds of tempering characteristics at least implied in Foucault and Quine—to suggest that conceptual schemes are elastic, supple sorts of historical and cultural objects, without clear and rigid boundaries or demarcations. As if to help avoid this impasse, at one stage Davidson makes a crucial elision—caricaturing the relativist’s position, he notes: ‘the test of [schematic] difference remains failure or difficulty of translation’ (Davidson 2006, p. 202, my emphasis). While Kuhn stresses complete untranslatability, Quine in particular wants to acknowledge that translation is a rough-and-ready, more or less inexact and partial process, where ‘difficulty’ need not necessarily elide into ‘failure’. This arguably accords well with everyday intuitions about translation, even between the ‘micro-languages’ of various organisational and cultural settings, as well as, more concretely, between the various orientations adopted within knowledge systems. And it ought to be possible to continue to think ‘schematically’—that is, retaining the language of conceptual schemes and structures—just so long as those schemes are treated in a suitable elastic sense. The question of commensurability can then be posed in many-valued degrees rather than two-valued kind—as a question of ‘how’ rather than ‘whether’ two schemes are commensurable. Moreover, schemes can be happily readopted having been denuded of any palpable and reified ontological form—and instead be treated as convenient bundles of particular concepts, beliefs, statements and practices of some more or less aligned group of actors. Schemes and structures, in a rehabilitated and analytical rather than ontological form, can serve a practical purpose in describing both the tacit and explicit statements of systems, and the notion of commensurability can further serve a derivatively useful function as a measure of the ‘difficulty’ translating one set of schematic commitments into another.

Constructionists tend to maintain that classifications are not determined by how the world is, but are convenient ways in which to represent it. They maintain that the world does not come quietly wrapped up in facts. Facts are the consequences of ways in which we represent the world. The constructionist vision here is splendidly old-fashioned. It is a species of nominalism. It is countered by a strong sense that the world has an inherent structure that we discover (Hacking 1999, p. 33).

‘Relativism’ is the view that every belief on a certain topic, or perhaps about any topic, is as good as every other. No-one holds this view. Except for the occasional cooperative freshman, one cannot find anybody who says that two incompatible opinions on an important topic are equally good. The philosophers who get called ‘relativists’ are those who say that the grounds for choosing between such opinions are less algorithmic than had been thought’ (Rorty 1982, p. 166).

The constructionist maintains a contingency thesis. In the case of physics, (a) physics (theoretical, experimental, material) could have developed in, for example, a nonquarky way, and, by the detailed standards that would have evolved with this alternative physics, could have been as successful as recent physics has been by its detailed standards (Hacking 1999, pp. 78–79).

Even if we have not got things right, it is at least possible that the world is so structured. The whole point of inquiry is to find out about the world. The facts are there, arranged as they are, no matter how we describe them. To think otherwise is not to respect the universe but to suffer from hubris… (Hacking 1999, p. 83).

Spatialising concepts

In Conceptual Spaces, Gardenfors (2000) develops a theory of conceptual representation in the cognitive science tradition developed by Rosch, Lakoff and others (Lakoff and Johnson 1980; Medin 1989; Rosch 1975), surveyed earlier in Chapter 3, ‘The meaning of meaning’. Gardenfors develops a ‘conceptual framework’, a constellation of concepts in which ‘concept’ itself figures prominently. In the first part of the book, Gardenfors presents a framework comprising:

The resulting framework is pragmatic and ‘instrumentalist’; the ‘ontological status’ of conceptual spaces is less relevant than that ‘we can do things with them’ (Gardenfors 2000, p. 31). Specifically, the framework ought to have ‘testable empirical consequences’ and, further, to provide a useful knowledge representation model for ‘constructing artificial systems’ (Gardenfors 2000, p. 31). One advantage of the use of geometric metaphors to describe conceptual arrangements is that it is possible to calculate approximate quantifications of semantic distance between individual concepts and concept clusters. However, the mathematisation of conceptual structures is to be taken as a heuristic rather than deterministic model—for Gardenfors, ‘we constantly learn new concepts and adjust old ones in the light of new experiences’ (Gardenfors 2000, p. 102). In light of this ever-changing configuration of concepts, any calculation of semantic proximity or distance is likely to be at best accurate at a point in time, although statistically—across time and users of conceptual clusters and relations—there may well be computable aggregate tendencies.

The arrangement of concepts and properties within conceptual spaces and domains depends on a coordinating principle of similarity:

Dimensions form the basis against which similarity is assessed—a single dimension for properties, multiple dimensions for concepts. Conceptual similarity for Gardenfors is intrinsically a cognitive and theoretical notion, however, which can consequently be varied as different dimensional properties are found to be more or less salient:

In the latter part of the book, Gardenfors then shows how his framework can be applied to traditional problems of semantics, induction and computational knowledge representation and reasoning (Gardenfors 2000). In particular he emphasises the relationship of conceptual structures to broader spheres of human action and practice. In what is an avowedly ‘pragmatist account’, meaning is put to the service of use within these spheres—though it is not equivalent to it. Unlike conventional semantics, the kind of ‘conceptual semantics’ Gardenfors espouses works down from social practice to fine-grained linguistics utterances: ‘actions are seen as the most basic entities; pragmatics consists of the rules for linguistic actions; semantics is conventionalized pragmatics… and finally syntax adds markers to help disambiguate when the context does not suffice’ (Gardenfors 2000, p. 185).

The pragmatist elements of this account fits well with the analysis of language Brandom undertakes, while the social orientation begins to bring concepts out of mind and language and into the intersubjective domain theorised by Habermas—points of accord succinctly encapsulated in the following quote: ‘In brief, I claim that there is no linguistic meaning that cannot be described by cognitive structures together with sociolinguistic power structures’ (Gardenfors 2000, p. 201). Applied to knowledge systems, Gardenfors supplies a convenient ‘first tier’ description of the kind of entity which includes the explicit conceptualisation of the system itself, and the tacit commitments which stand behind it. ‘Conceptual spaces’, standing here for Quine’s ‘conceptual schemes’, are mentalist metaphors for describing at least part of what it is that a knowledge system represents. The remaining sections add further descriptive tiers on which the framework of the study can be mounted.

Practicing with concepts

Brandom develops a contemporary account of linguistic practices grounded in the pragmatist tradition of Sellars and Rorty. Unlike Rorty, for whom all kinds of linguistic utterance were of equivalent functional significance, Brandom privileges propositional assertions as ‘fundamental speech acts’, without which other speech acts—commands, interrogatives, exclamations—would not be thinkable (Brandom 2000a). Assertions are, for Brandom, tokens in a ‘game of giving and asking for reasons’, nodal components in a vast articulated web of inferences which constitute discursive practice. The primary role of assertions is not, as a correspondence theory of truth would have it, to represent an actual state of affairs, but rather to express, or, in Brandom’s more canonical expression, to make ‘explicit what is implicit’. In a pivotal passage, he continues: ‘This can be understood in a pragmatist sense of turning something we can initially only do into something we can say: codifying some sort of knowing how in the form of a knowing that’ (Brandom 2000a, p. 8).

Where Gardenfors’ primary linguistic unit of analysis is the word, for Brandom it is the sentential structure that provides the key to ‘knowing that’, to assertion making. An atomistic orientation towards concept-use might make it appear that concepts are accumulated, one after another. For Brandom, contra Davidson, the scheme necessarily precedes the individuated concept:

Brandom’s implied broad swipe here is directed towards a whole semantically formalist tradition whose origin he locates in the work of the later Frege (his reading of early Frege is considerably more commensurate with the inferentialist, expressivist and pragmatist line Brandom himself adopts). Demonstrating how sentences, and sub-sentential devices such as anaphora, primarily function to relate concepts within an inferentialist network of reasons takes Brandom much of the 741 pages of his landmark Making It Explicit (Brandom 1994). At the heart of Brandom’s enterprise is an attempt to reconcile the rigour associated with this formalist tradition with a more appropriate philosophically holistic orientation, which sees assertional speech acts within a broad tapestry of human action and ‘social practice’ generally. This has clear resonance with this particular project; although Brandom addresses neither the question of translation nor the question of knowledge systems specifically, several inferences can be drawn from his analysis:

A further note relates to the specific treatment of structure in Brandom’s work. The entire practice of making, interpreting and reasoning with assertions stands within what he terms an ‘I-thou deontic score-keeping’ relation. This, for Brandom, is ‘the fundamental social structure’ (Brandom 1994). This base structure operates like a simplified, idealised model, in which two interlocutors are locked into a game, metaphorically tabulating each others’ reasons offered for actions, practices, commitments, beliefs and attitudes. This theoretically endless activity does not yet offer an account for how some series of disagreements might grow into schemes which are incommensurable. To explain this—without falling prey again to Davidsonian lines of critique—requires a shift in registers, from what appears fundamentally a psychological intersubjective scenario—between two well-intentioned agents—to a sociological one—between two cultures, whose intentions are never quite irreducible to those of the agents who represent them. To make this shift the next section draws on an essential Habermasian distinction, between lifeworld and system, which offers explanation for how more fundamental rifts in the social tapestry might occur.

Socialising concepts

Where Gardenfors and Brandom acknowledge the role played by the social sphere in structuring conceptual arrangements, neither provide an account of what sorts of structure are germane to this sphere itself, nor what might cause rival conceptualisations to emerge. Kuhn and Foucault had developed explanatory theories of sorts but, at least in the case of Kuhn, these theories were limited to a particular domain of the social—the scientific domain. While no encompassing causal theory might adequately account for all variations in cultural conceptualisations—or, less abstractly, differences in how cultures see the world—a theory which at least makes perspicuous some common lines of demarcation would be helpful. Foucault’s later analysis of ‘micro-power’ goes some way in this direction, yet he consciously abjures any abstract generalisable theorisation (Foucault 1980). Bourdieu’s elaboration of ‘habitus’ is similarly useful at an intra-cultural level (Bourdieu 1990), but is not directed oriented towards an explanation of the sorts of inter-cultural differences which might arise, particularly within the ‘networked societies’ engaged in information system development and use (Castells 1996).

Habermas is sometimes taken as being either a theorist of ‘incommensurability’ (Latour 1993) or, at others, its exact opposite: a naïve advocate for an idealised ‘communicative rationality’ directed towards utopian understanding (Flyvbjerg 1998). The interpretation offered here suggests that he represents neither of these extremes, but rather a Kantian rationalism despondent—on the one hand—at the over-systematisation and objectification of modernity, yet conciliatory—on the other—towards the potentials of dialogue and communication for redressing this trend. As with the other theorists encountered here, there is insufficient scope for any kind of thorough treatment of Habermas’ full theoretical apparatus. Instead I focus attention on a pivotal conceptual opposition between system and lifeworld, outlined in The Theory of Communicative Action (Habermas 1987).

For Habermas, the Kantian trichotomy of instrumental, ethical and aesthetic rationality is ontologically primary categories of modernity. These broadly correspond to objective, inter-subjective and subjective spheres of individual experience. Habermas inherits the critical lines of Weber, Lukacs and the Frankfurt school towards post-Enlightenment reason, which has missed its potential to act as a liberating tool. Instead it has been co-opted within specifically modern configurations and systems of power and oppressive administration (Adorno 2001; Horkheimer and Adorno 2002). Rationalisation has been operationalised as an instrumental process within all spheres of human experience— everything has been subjected to systematised logic. Even individual subjectivity, what for Kant ought to remain the inviolable sanctuary of private experience, has been externalised, publicised and rendered transparent to the machinations of modern systems—through, for example, the various concrete vehicles of the media, the professed wisdoms of popular psychology, the endless commodification of art, and the cult of celebrity. For Habermas, as for critical theory, this outgrowth of hyper-rationalisation has a corresponding corrosive and pathological effect on the ‘lifeworld’—the phenomenological horizon experienced by individual subjects. Paradoxically, the domination of a singular form of rationality has also led to a fracturation and destabilisation of a social world into multiple systems. Such systems—at a macro level these include legal, economic, scientific and political systems—operate according to the internal dynamics of their particularist ends, and remain only loosely, if at all, coordinated within a social whole. Accordingly, conceptual schemes are segregated in a profound way within the system spheres in which they are engaged. Habermas describes this development:

However, this historical diversion is not irrevocable within Habermas’ schema; the very conditions that effect the outgrowth of a particular form of rationalisation can also serve to corral it within its proper sphere of operation—that of scientific knowledge of the world. Provocations from the subjective sphere of experience, such as various inflections of Romanticism, are insufficient for this containment and merely serve to buttress the over-extended reach of systemic reason. Rather it is in the intersubjective sphere, where human agents engage in communication and dialogue, where reason can be directed towards not the achievement of specified functional ends, but the formation of social consensus, that functional ends can be re-evaluated within the context of society as a whole. The derivation of consensus through the pure consideration of better reasons—a never fully realised process, but nonetheless operating as a counterfactual ideal—acts as a mediating force between the private wants of subjective selves and the oppressive operations of hyper-rational systems of modernity. For Habermas communication offers the potential to arrest ‘the uncoupling of the system and lifeworld’ and return from ‘the threshold at which the mediatization of the lifeworld turns into its colonization’ (Habermas 1987).

Habermas’ analysis can be seen to supply the missing detail to Brandom’s reference to the ‘the social’, which is posited as an ‘unexplained explainer’ in his account. The connection between Habermas and Brandom is not seamless, as a recent exchange attests. Despite many points of intersection, they differ precisely over the question of whether Kantian trichotomy precedes or is instead subject to the role of logic and inference—Brandom insists, contra Habermas, that specific domains sit downstream from the primordial experience of ‘asking for and giving reasons’ (Brandom 2000b). Broadly, though, Habermas can be seen as having developed an important and encompassing explanatory account of how specific schemes can be incommensurable. In spite of how language appears, even in the work of Brandom, to be an undifferentiated tool for establishing lines of inference between co-operative agents, social systems have in the course of modernity increasingly operated according to local teleological programs, which, through the operationalisation of specific language games and jargon, serve to blunt language’s more incisive communicative potentials. Within the ‘iron cages’ of technocratic institutions, inter-system ‘interfacing’, using perfunctory rational procedures, has replaced genuine intersubjective dialogue. Within these differential ends and parametric conditions, unique morphologies of organisational cultures generate different conceptualisations of common entities. Even conceptualisations which are externalised and globalised—in the form of technology standards, for example—are typically adopted via rationalising fiat, either via conformance to de jure fiat or recognition of de facto network externalities—rather than because of an internally deliberated conclusion brought about by the force of better argument.

Unlike the accounts examined above, it is not a foregone conclusion that the conceptualisations produced within these spheres be radically incommensurable, however—only that it is possible to diagnose the potential causes, along lines of different cultural ends, procedures and intentions, when they are. Moreover the efficacy of idealised dialogue, of the kind both Habermas and Brandom are happy to countenance, and towards which actual communication constantly strives, can assuage the rougher edges of translation in practice. Brought back down to the technical domain of knowledge systems—and in lieu of any active participation between the cultures responsible for them—the role of the analyst is to ferret out both the points of differentiation and the potential conciliatory paths between them. This involves, practically, identification of salient dimensions against which such points and paths can be plotted, and a corresponding process of interrogation of the cultures responsible for the systems under review.

Habermas, then, does not endorse a romantic yearning for an overarching metaphysics, a stable social order or a single governing conceptual scheme. Nor does he champion endless devolution into more granular, localised and ultimately untranslatable systems of meaning. His aim is rather to recuperate the promise of Kantian rationality by recalibrating the obsession of modernity with instrumental reason by emphasing the equivalence of ethical and aesthetic spheres. In practice this allows systems to proliferate in their respective manifold differences, but never so far as to negate the potentials of translatability and commensurability completely. Further consideration of the greater Habermasian project would take this discussion too far afield; here it suffices to provide a sociological and historical explanation of causative factors in the incommensurability of conceptual schemes, and thus serves to connect up Brandom’s pragmatist analysis of linguistic meaning and Gardenfors’ analysis of conceptual spaces to a broad historical context. Together these connections—linking up the linguistic, the cognitive and the social—develop an altogether more fluid and elastic conception of ‘structure’, one avowedly informed by materialist and pragmatist concerns, than those advanced by the earlier generation of theorists discussed above. A path has now been prepared for the description of such a rehabilitated structure, as it relates more directly to knowledge systems and the cultures responsible for them.

Adorno, T.W. The Culture Industry: Selected Essays on Mass Culture. Abingdon: Routledge; 2001.

Austin, J.L. How to Do Things With Words. Abingdon: Routledge; 1998.

Bourdieu, P. The Logic of Practice. Stanford: Stanford University Press; 1990.

Brandom, R. Making It Explicit. Cambridge, MA: Harvard University Press; 1994.

Brandom, R. Articulating Reasons: An Introduction to Inferentialism. Cambridge, MA: Harvard University Press; 2000.

Brandom, R. Facts, Norms, and Normative Facts: A Reply to Habermas. European Journal of Philosophy. 2000; 8:356–374.

Castells, M. The Rise of the Network Society. Cambridge, MA: Blackwell; 1996.

Davidson, D. The Essential Davidson. Oxford: Oxford University Press; 2006.

Derrida, J., Writing and Difference A. Bass (tr.). Routledge, Abingdon, 2001.

Flyvbjerg, B. Habermas and Foucault: Thinkers for Civil Society? British Journal of Sociology. 1998; 49:210–233.

Foucault, M. Madness and Civilization; A History of Insanity in the Age of Reason. New York: Pantheon Books; 1965.

Foucault, M. The Order of Things: An Archaeology of the Human Sciences. New York: Vintage Books; 1970.

Foucault, M., Power/Knowledge: Selected Interviews and Other Writings 1972–1977. C. Gordon (tr.). Pantheon Books, New York, 1980.

Foucault, M. Archaeology of Knowledge. Abingdon: Routledge; 2002.

Gardenfors, P. Conceptual Spaces. Cambridge, MA: MIT Press; 2000.

Habermas, J., The Theory of Communicative Action T. McCarthy (tr.). Beacon Press, Boston, 1987.

Habermas, J., The Structural Transformation of the Public Sphere T. Burger (tr.). MIT Press, Cambridge, MA, 1989.

Hacking, I. The Social Construction of What?. Cambridge, MA: Harvard University Press; 1999.

Hacking, I. Historical Ontology. Cambridge, MA: Harvard University Press; 2002.

Horkheimer, M., Adorno, T.W., Dialectic of Enlightenment: Philosophical Fragments E. Jephcott (tr.). Stanford University Press, Stanford, 2002.

Kuhn, T.S. The Structure of Scientific Revolutions. Chicago, IL: University of Chicago Press; 1970.

Kuhn, T.S., The Road Since Structure: Philosophical Essays 1970-1993with an autobiographical interview. Chicago, IL: University of Chicago Press, 2002.

Lakoff, G., Johnson, M. Metaphors We Live By. Chicago, IL: University of Chicago Press; 1980.

Latour, B. We Have Never Been Modern. Cambridge, MA: Harvard University Press; 1993.

Medin, D.L. Concepts and Conceptual Structure. American Psychologist. 1989; 44:1469–1481.

Quine, W.V.O. Word and Object. Cambridge, MA: MIT Press; 1964.

Quine, W.V.O. From a Logical Point of View. Cambridge, MA: Harvard University Press; 1980.

Quine, W.V.O. Interview Between W.V. Quine and Yasuhiko Tomida. http://www.wvquine.org/quine-tomida.html, 1992. [(accessed 18 January 2010)].

Pinker, S. The Language Instinct. New York: HarperCollins; 1995.

Rorty, R. Consequences of Pragmatism. Minneapolis, MN: University of Minnesota Press; 1982.

Rosch, E. Cognitive Representations of Semantic Categories. Journal of Experimental Psychology: General. 1975; 104:192–233.

Sellars, W. Empiricism and the Philosophy of Mind. Cambridge, MA: Harvard University Press; 1997.

Wittgenstein, L. Tractatus Logico-Philosophicus. Mineola, IA: Dover Publications; 1921.

Wittgenstein, L. Philosophical Investigations, 3, Oxford: Basil Blackwell, 1967. [(first published 1953)].