Sellars, Wilfrid Stalker (1912–89)

3. Epistemological perspectives

Sellars’ key nominalistic strategies could succeed only if the notion of a linguistic item’s having a semantic or inferential function could itself be explicated without recourse to irreducibly platonistic or mentalistic conceptions. One of the greatest strengths of his systematic philosophy is the exquisite care with which he proceeded to locate the normative conceptual order within the causal order and to interpret the modes of causality exercised by linguistic rules. For Sellars, inference itself is always a normative affair, a matter of the judgments one ought to or is entitled to make. He defuses the circularity which threatens such an account by arguing that our knowledge of what implies and follows from various claims is, in the first instance, a practical ability to discriminate, that is, to respond differentially to, good and bad inferences. Rule-governed linguistic behaviour develops out of multiple repertoires of ‘pattern-governed behaviour’, behaviour which exhibits a pattern because the propensity to produce behaviour belonging to that pattern has been selectively reinforced and contrary propensities selectively extinguished (see language, social nature of; Meaning and rule-following). The pattern-governed behaviour characteristic of language includes ‘language-entry transitions’, propensities to respond to non-linguistic states of affairs (such as sensory stimulations) with appropriate linguistic activity; ‘language-departure transitions’, propensities to respond to a subset of linguistic representings (for example, such first-person future-tensed conduct-ascriptions as ‘I shall now raise my hand’) with appropriate corresponding behaviour; and ‘intra-linguistic moves’, propensities to respond to linguistic representings with further linguistic episodes (only) in patterns corresponding to valid theoretical and practical inferences. Linguistic roles or functions, Sellars suggested, are ultimately individuated in terms of the structures of positive and negative uniformities generated in the natural order by such pattern-governed activities.

In the Kantian tradition Sellars insisted that, in contrast to the mere capacity to be sensorily affected by external objects, perception of how things are requires not only systematic differential response dispositions but also the ability to respond to sensory stimulation with a judgment, that is, the endorsement of a claim (see Perception, epistemic issues in). Sellars went on, however, to propose that reports of how things look or seem, rather than employing ‘more primitive’ concepts, result instead from withholding these characteristic endorsements. This account enabled him to explain the incorrigibility of ‘seems’ judgments that Cartesianism takes as its fundamental datum. Their incorrigibility is simply a matter of their tentativeness; a ‘seems’ judgment expresses a perceptual ascription without endorsing it. It follows that ‘seems’ judgments do not express a special class of immediate cognitions. Applying the concept ‘looks red’ requires the same mastery of inferential articulations, the same inferential ‘know how’, as does applying the concept ‘is red’.

Sellars’ analysis of the Cartesian incorrigibility of perceptual ‘seemings’ is one strand of the philosophical dialectic most frequently associated with his name, his comprehensive critique of the ‘Myth of the Given’. Basic to this critique is his insistence on the irreducibly normative character of epistemic discourse. Characterizing an episode or state in epistemic terms is not giving an empirical description of it but rather placing it within a social framework of justifications, of having and being able to give reasons for what one says. All knowledge that something is the case – all ‘subsumption of particulars under universals’ – presupposes intersubjective learning and concept formation. It follows that the ability to be (epistemically) aware of a sort of thing rests upon a prior command of the concept of that sort of thing and cannot account for it – and this holds equally true for concepts pertaining to ‘inner episodes’. The first-person reporting role of such concepts, a use Cartesians interpret as evidencing the ‘privacy’ of the mental and one’s ‘privileged access’ to one’s own mental states, is necessarily built upon and presupposes their intersubjective status.

The idea that a language necessarily contains a stratum of claims any of which, on various occasions, can occur either as a report (an unmediated causally evoked response) or as the conclusion of an inference (mediated by other linguistic-conceptual representations) suggests the possibility of a further stratum of purely inferential claims. That is how Sellars proposes that we understand theoretical claims, that is, as claims that one can de facto become entitled to endorse only as the conclusion of an inference. On this view, the difference between theoretical objects and observables is not ontological but methodological. It is not a matter of the kind of thing our claims are about; it is a matter of how we come to be entitled to make them – and, as our technological sophistication increases, this can change. Once it is recognized that there is no epistemically privileged stratum of judgments with which to respond to one’s sensory environment, and that such a ‘language-entry’ response is what ‘observation’ is, it follows that nothing is in principle unobservable. Noninferential and purely inferential claims can both be about the same kind of object (see Observation).

Sellars’ holistic view of justification implied that the reasonableness of accepting even first principles is a matter of the availability of good arguments warranting their acceptance. Since it is definitive of first principles that they cannot be derived by sound theoretical reasonings from still more basic premises, Sellars located the requisite arguments in a stratum of ‘vindications’, practical reasonings in which the adopting of specific principles (an epistemic conduct) is demonstrated to contribute to the realization of non-arbitrary epistemic ends. From the ontological point of view, however, such practical cognitions – intentions and volitions – are simply species of occurrent thinkings whose unique conduct-structuring functional role is to be understood in terms of their special (‘language-exit’) relationships to non-linguistic behaviour (see Justification, epistemic).

For Sellars, it is finally practical thinking which lies at the centre of the framework of persons as such (see Persons). To think of an entity as a person is to think of it as actually or potentially a member of a community, and it is the most general common intentions of a community – its ‘we-intentions’ – that fundamentally define the structure of norms and values, crucially including those which make meaningful discourse and rationality itself possible, in terms of which the conducts of its members come to be appraised. A genuinely synoptic vision of man-in-the-world, Sellars concluded, can therefore be achieved only by enriching the scientific image with the language of individual and shared intentions.