On Sellars and the Inconsistent Triad:
- Brassier and Brandom -
I have been continuing my study of Sellars' Empiricism and the Philosophy of Mind, and I think a few crucial points are worth making in relation to Ray Brassier's (2010) presentation of the rejection of the Myth of the Given, at the excellent conference To Have Done With Life.
The point in contention concerns how Sellars responds to a quandary concerning the construal of sense contents. Sellars considers the following inconsistent triad:
A) 'S senses red sense content x' entails 'S non inferentially believes (knows) that x is red.'
B) The ability to sense sense contents is unacquired
C). The capacity to have classificatory belieds of the form 'x is F' is acquired.
In Brandom's study guide to EPM he claims that Sellars drops A, since the latter is a version of the Myth of the Given, i.e. the belief that certain experiential states can serve as foundations for beliefs and epistemic states just by virtue of having them. Now, the interest thing is that Brassier claimed in Zagreb that Sellars endorsed A and C, and refused B instead. With this in mind I was trying to figure out why these discrepancies obtained in their respective readings, and what I thought was going on.
What Brandom argues is that A must fail for Sellars in favor of a new account of thoughts and sensations, at the order of justification and causation, respectively. Thus, he thinks such an account must disassociate epistemic states which concern the intentionality of thoughts and beliefs, from the non-epistemic realm of causes which concerns the causal order of sensations. This is part of what Sellars envisions as an integration of sensa into the scientific image of thought, as opposed to reifying non-conceptual immediacy as a phenomenological bedrock for the underdetermination of all posterior intentional states. Thus, Sellars contends that linguistic intentionality must be predated by a kind of psychological intentionality, i.e. experience is epistemically mute, both at the level of objectual individuation (Husserl) and of holistic integrated comportments (Heidegger). I shall suggest below why this is an issue of particular importance within the contemporary debates concerning epistemology and psychology that have become central in much philosophy of mind, and in particular in the dispute between externalists and internalists. The point to stress, for the moment, is simply that Sellars takes sensible experiences to be non-cognitive in character, and thus refuses to reify them as foundational items for higher order intentionality. If so, then how could Sellars be said to endorse A in any sense, like Brassier stipulates?
Now, I think that what Brassier was trying to convey by saying that Sellars retains A and drops B is the fact that non-inferentially acquired states (i.e. causally triggered states) are not for this reason unacquired, since one can give an evolutionary story for how organisms and species come to be in these sorts of states. These states are not miraculously divined capacities, and that much is certainly also a crucial rejoinder that Sellars makes to the Cartesian account of sensory transparency. Now, the delicate issue here is that non-inferentially acquired states can amount to knowledge if the subject is imbedded within the space of reasons, i.e. if the subject is not just capable of reliably responding to the situation in question, but also knows that he/she is reliable in doing so. This means that for a non-inferentially acquired state to count as a knowledge, the subject must be capable of responding to a hypothetical challenge to his endorsement.
The examples McDowell gives are simple and excellent: one directly comes to observe that one's neighbor is at home when one sees the neighbor's car parked in his parking lot; one does not need to make an inference to arrive at this state. Yet for this state to count as a case of observational knowledge, it must be possible for the us to respond to being challenged about the issue: "couldn't have he taken the bus to work?", in which case I must be able to provide a justification for my claims, or perhaps, revise my commitment and accept that my belief was not a good one. So, to return to the inconsistent triad: the idea behind reading Sellars as endorsing A would be to say that there is a sense in which non-inferential states which serve as observation reports can count as knowledge. Yet one must resist the idea that this entails these capacities are unacquired, since they surely require a period of conditioning and acculturation, i.e. one learns to see that one's neighbor is home on the basis of seeing the car present, and knowing oneself to be reliably reporting in such cases. And even if this capacity required at some stage inferential procedures, the point is that there it is possible to be in a state of knowing that was not arrived at by means of inference: this is what Sellars famously calls language entry transitions, of which perception is the paradigm case.
Indeed, as Brassier has highlighted more recently, this is what Sellars' thinks in terms of the disambiguation between doing things for a reason and things because of reasons. The former does not require intellection or sapience, while the latter requires that one be engrossed in the logical space of reasons, and thus be capable in engaging in meta-discursive talk about the rules one is following. The former case is exhibited uniformly by sentient creatures and machines, and simply play a functional role, i.e. a thermometer may reliably report on the room's temperature, but it lacks knowledge since it cannot know of its reliability or respond to being challenged. This means that although non-inferential capacity to issue observation reports can amount to knowledge, they only do so on condition that these reports/capacities be properly integrated within the capacity to draw inferences from them, i.e. to be able to undertake logical revision of beliefs and states. Thus, not all non-inferentially acquired states are knowledge.
However, I think that what is probably at stake for Brandom in the three claims is the idea that insofar as we are speaking about sensory states qua causally triggered states, then these are in strict Sellarsian terminology not epistemic in any case. I believe that he would insist in that we need to be careful not to conflate perceptual states which involve judgments on the basis of perception to function as language-entry moves, with sensations of sense contents qua non-epistemic causally triggered states. And the idea is plainly that while non-inferential perceptual states can be epistemic insofar as they entail that one judges (that some x is the case), sensations qua purely causal phenomena intractable to the conceptual domain of judgment remains resolutely non-epistemic in character.
Brandom's proposal is to read Sellars along these lines and to say that the capacity to bear sense contents is correspondingly unacquired in the simple sense that such a capacity not gained by virtue of the process of conditioning that we associate with knowledge-acquisition, i.e. hypothesis building, projectibility, etc. So, he thinks that Sellars endorses B in the inconsistent triad. Brandom thinks that Sellars' sense of non-acquisition is narrower than the sense in which Brassier reads him, i.e. he does not take unacquired states to imply that the capacity to have the appropriate causal onsets requires a complex story in the development of the organism's growth, and more broadly its evolutionary history within a species. For in this latter sense 'acquired' appears almost trivially equivalent to 'having appeared in time for some reason', and he thinks that can't possibly be of philosophical interest.
With this distinction in mind, the disavowal of A seems to run from the disambiguation between sense/perception. If sensation is epistemically mute, then we must distinguish perceptual beliefs qua non-inferential states, and which are already of facts- propositionally articulated by universal subsumption, from the sensory apprehension of particulars. So the idea would be that Sellars drops A in order to distinguish sensation from perception, deflating sense-datum theories, and running with the idea, proposed by (C), that the capacity to be in epistemic states is acquired by acculturation and a process of conditioning which itself might be inferential, without for this reason vitiating the non-inferential character of the states it comprises.
Perhaps a route out of this dilemma is to claim that Sellars actually endorses (A), (B) and (C) in qualified manners, by reshuffling the terms, albeit he seeks to rescue crucial points from the three claims in their usual form.
From (A) he seeks to rescue the idea that perceptual states may constitute non-inferential knowings; but he will distinguish conceptually enveloped perceptual judgments from non-conceptual and so non-epistemic sensory states.
From (B) he seeks to rescue the idea that sensory states are causally acquired capacities while accepting that they are non-conditioned by cultural learning in the sense defined above; so he will distinguish causally triggered sensory states from causally triggered epistemic states.
From (C) he seeks to rescue the idea that epistemic states require cultural conditioning and so are acquired in that sense; but he will disambiguate between the evolutionary rationality of sensory acquisition that explains causally triggering non-epistemic states, and the causal triggering of non-inferentially acquired beliefs as epistemic states that function as language-entry transitions within the logical space of reasons.
Thus, both sensations and perceptual judgments are causally triggered in a broad sense, and so acquired, i.e. the evolutionary sense in which nothing happens ex nihilo. We learn to enter a language much in the same way as we are causally delivered into sensory states, we develop capacities by virtue of our evolutionary history, and our causal interactions with the environment. The crucial difference is that classificatory beliefs require conditioning within the sensus communis and social order of language, and sapience marks the transition from merely causally triggered states to the epistemic capacities associated with entering a conceptual economy. But the capacity to have epistemic states, while causally acquired and non-inferential like sensibility, involves a narrower sense of conditioning. We move from merely causal non-inferential state of being to non-inferential knowledge when we move from sentience to sapience, from pure mechanical process to knowledge.
Thus, we should insist, in that proposition (C) is endorsed, since perceptual judgments are acquired causally, both in the way that sense contents are, but also in the narrower sense in which they require conditioning into language.
So the tricky word here is 'acquisition', since it is used in three ways:
a) Generally causal: in this sense both classificatory beliefs and sense contents are acquired (C, holds, but B fails)
b) Narrowly/Culturally causal: in this sense only classificatory beliefs are acquired (both B and C hold)
c) Non-inferential: In this sense both classificatory beliefs and sense contents are unacquired (C fails, B holds)
The trick concerns disentangling these three senses and realizing that while perceptual knowings of the sort required for classificatory belief are also causally triggered, they require a narrow sense of cultural acculturation which distinguishes the sentient strata of species-relative general causality, from full blown conceptual rationality and knowledge. By the same token, sensations are generally acquired in the causal sense, but unacquired in the cultural sense. Both are non-inferentially triggered, and so causally derived in the general sense, but ony knowings involve acculturation and social conditioning in their acquisition, and are so acquired also in the narrow sense. I think this goes back to what Tripplet and Devries call attention to by disambiguating between non-inferential knowing, and independent knowledge. The former simply means that a given piece of knowledge did not result from an inferential procedure. The latter implies the claim that there are knowings of given facts that may be had independently of any other knowings. Sellars endorses the possibility of the former, but denies the latter. Knowledge can be non-inferentially triggered in perception, but in any case, if it is to count as knowledge, one needs to be able to integrate it within a nexus of other beliefs that suffice to render such states justified, i.e. embedding in the space of reasons.
I think this might be why also Sellars does not flatly say which of the three propositions he would drop; he reworks the triad by amplifying the descriptive scope.
II - Some Brief Considerations About Sellarsianism
The question that obviously suggests itself is whether in fact it is fair to construe perceptual representation as involving something like conceptual/linguistic judgment. This is the heart of Tyler Burge's contentions against Sellars, and I think they map onto the tricky debacle between externalists and internalists.
On the one hand, it might seem as if Burge's contention that Sellars over-intellectualizes perception by enveloping it conceptually is really a terminological issue. This is because Sellars is perfectly aware that organisms have a pre-conceptual capacity to react to their environments in reliable or unreliable ways; this is the whole point of RDRD's as a condition for sentience, as Brandom emphasizes. However, by reserving the term perception for the conceptual achievement knit to judgments, Sellars seeks to render it epistemically productive as part of what characterizes sapient activity, and full blown linguistic rationality. Thus, while Sellars reserves a role for sensation simpliter, the latter is plainly non-epistemic in itself, or more precisely is indifferent to the distinction between reasons and causes, i.e. the latter is only a distinction possible from within the normative space which adjudicates claims on the basis of their justification and inferential role. Only creatures embedded in the logical space of reasons could ever articulate knowledge. One could thus just say that perception in Burge's sense would fall within sensation in Sellars' terminology, without there being a substantial disagreement at stake. However, there is a more profound non-terminological issue at stake, and this is the crux of the debate.
Whereas intentionality remains, for Sellars, restricted to the epistemic domain of inferentially articulated beliefs and reasons, Burge ascribes intentional content to non-conceptual, perceptual states. Sellars warned against this because it became the gulf for versions of the Myth according to which experiential primitives could serve as justifications for conceptually articulates beliefs. However, Burge dislodges intentionality from knowledge; sentient creatures have intentional states without the necessity for conceptual rationality. Thus, although Burge agrees with Sellars in that the primitive ability to have experiences would not be conceptual and so would be not-epistemic, it is nevertheless representational insofar as perceptual states have conditions for veridicality which explain the conditions for success or failure in the performance of a perceptual faculty in relation to its environment. The result is that while for Sellars the domain of sensibilia is delegated to the scientific study of the neurophysiology of the individual, bereft of any appeals to intentionality, Burge follows the nativist lead in thinking rather that perceptual psychology is key to understanding perception as a representational capacity. The latter option becomes then that of the possibility of ascribing intentional states and representational capacities to non-sapient creatures, or whether we cannot do this lest we make a category mistake. The question is whether something like representational states, of the sort that host conditions for veridicality, can be built into the Sellarsian account of sensing as a purely causal-mechanical functional process void of epistemic valence. I think the stakes here are far from clear.
This is, I think, an incredibly interesting juncture of the debate, since it constitutes a crucial dividing line within naturalist approaches to the philosophy of mind today.