The (very) idea of applying CSs to reality may itself be suspect. It implies that “there is an uninterpreted reality, something outside all schemes and science” (189, “The Very Idea…”). Perhaps we are born into CSs, so there is never any cognitive application of a particular – or any – CS. This is not in itself a denial of CSs. Being born into CSs clearly means that CSs exist. However, Donald Davidson does indeed deny the “very idea of a conceptual scheme”:

…if we cannot intelligibly say that schemes are different, neither can we say that they are one. (198)

Davidson’s position could be represented this way:

Interpreted reality Ö Interpreted reality

rather than

CSs ® uninterpreted reality

There being an always and already interpreted reality doesn’t stop us re-interpreting reality. However, it does stop us interpreting an unconceptualised reality. But surely there must have been a time, for all of us, when reality was indeed uninterpreted. That is, in early childhood. Well, for a start, we could come into the world with Kantian or quasi-Kantian concepts such as event, cause and effect, object, spatial configuration, temporal sequence, etc. (And Davidson has hinted, though not in his 19?, that he has leanings towards Kantianism in a Pearcian way.) And with these basic concepts in place, things non-Kantian soon fall into place. That is, we don’t really interpret new phenomena. Sometimes we were simply told what they were. And we, of course, took such explanations, usually, as the truth. And even when we weren’t given explanations or a names for new phenomena, which must have happened, we didn’t really interpret or infer. We saw floating seeds as fairies. Objects as animals and animals as objects. New things as – related to – old things. Cats as little dogs and dogs as big cats. And so on.

But still, one may ask, there must be some interpretation. Yes, but only in terms of already acquired concepts. And that wouldn’t be an example of “uninterpreted reality”. The cat may be interpreted under the concept [small dog], but it is not interpreted ex nihilo – from “uninterpreted reality”. What if we came across alien objects or properties? (Apart from the fact that we don’t come across genuinely alien properties and objects as such.) An alien object or property must come under certain basic Kantian concepts: [object], [thing], [entity] [event. [attribute], etc. And then, in addition, non-Kantian concepts like [small], [brown], [wet], [fast] etc. In any case, how alien can an alien property or object be? A round square would be truly alien I suppose, but it wouldn’t exist even in another possible world. Of course, a golden mountain is far from being alien. Take Armstrong’s position:

Each alien property must have its own nature…But these natures…are not to be found in the space-time world…Suppose orange to have never been instantiated. Would it not then have been alien? But, even so, could we not ‘fix’ it as the property between red and yellow…Orange would not be totally alien (orange is a colour)… (187)

So, on this reading, alien properties and objects are not bona fide aliens:

…the possible is determined by the actual. So the actual universals set the limit, a limit given by the totality of their recombination… (187)

So it would be up to a brainy philosopher like David Lewis to invent a bona fide alien. And, of course, it must live in another possible world.

The epistemic points about the conceptualised and interpreted world is given an ontological underpinning by Armstrong’s position on this-worldly things and their attributes. What we do with unfamiliar phenomena is we “fix” them according to what we do know. Borrowing from Armstrong’s example, if we hadn’t come across orange before, we would “fix” is between the concepts [red] and [yellow]. And, of course, orange would fall under the concept [colour]. Even the alien aliens we watch in our Hollywood sci-fi films have had their properties “fixed” by what we do know. The alien in Alien was a “combinatorial” being. This possible alien was “determined by the actual” (e.g. fangs, scales, saliva, a tail etc). (Perhaps this is one of the reasons why Fodor believes that all concepts are in fact innate (1975, pg 370).

Despite all the above, the problem is that Davidson, in his (19 ), doesn’t argue very much for this position against scheme/content “dualism”, although he does argue strongly against other aspects of “conceptual relativism”, etc. That’s why I tried to fill in the blanks. I will need to go elsewhere for Davidson’s own arguments.

Let’s start with how Davidson himself defines a CS:

Conceptual schemes, we are told, are ways of organising experience; they are systems of categories that give form to the data of sensation… (1)

Davidson’s ironic or sarcastic “we are told” hints at the possibility that we have been told incorrectly. He doesn’t think, that is, that this happens to the “data of sensation”. CSs don’t “give form” to experience – it comes with “form”. Clearly he is thinking here of contingent CSs, not the Kantian micro CSs I hinted at earlier. In contemporary Kantian terms, the brain “organises experience”, but the mind doesn’t. Or at least the mind doesn’t in a cognitive sense. Any “form” that is given “to the data of sensation” is done by the brain, not the cognitive mind. (There is no prima facie reason why the brain, not the “transcendental ego” or mind, can’t be Kantian or at least quasi-Kantian.)

(Clearly this is an account of CSs from the perspective of an American analytic philosopher. Hence his reference to the “data of sensation”. It seems that one of his main targets is Quine. So his concept of a CS may not seem too familiar to those outside this tradition.)

Davidson is asking us whether or not we can make an epistemic or even ontological distinction between CS and “content”. Take the “dualism” of sensory data and interpretation. We can say that  when “Galileo saw the moons of Jupiter through his telescope…the impact on his retina was hard…even though its consequences were…different for different communities” (81). This is a distinction between sensory data (or even “the Given”) and interpretation. Can we make such a distinction?  Davidson said that

causation is not under a description, but explanation is.

So the retinal stimulation doesn’t have factual status, for two reasons. One, facts are “hybrid entities” and our assertions depend on “our antecedent choice of response to such stimuli”.

So Davidson was putting us firmly in touch with the world by stressing causation. We cannot escape causal forces. It’s what we make of these causal forces, both with our minds and sense modalities, which matters. To take a preliminary sceptical position, Causal forces can be interpreted or “explained” in just about any way. We can’t do anything about the causal forces. That is, the sensory data’s impact on the brain and mind. But to say the same about interpretation and explanation would be ridiculous. And yet that’s the logical conclusion of the metaphysical realist’s, for one, position. But all this is dangerously unDavidsonian and dualist. All this talk about sensory data and – then - interpretation.

What is there between causal forces and our beliefs? There are theories, languages and our concepts. They come between the hardness of the given, of sensory data, and what we know to be true. But all this is, according to Davidson, is scheme/content “dualism”. There is, in fact, no sensory data without concepts, theory and language.

We need to choose, or non-cognitively use, a form of description set by our CSs. Things can’t choose their own descriptions or the correct CS to use. This is the conceptualist’s position.

For Davidson, beliefs and their causes are as one. He says that “beliefs are…identified, directly and indirectly, by their causes” (“A Coherence Theory…”, pp. 318-319). Causes “fix the content of our beliefs”. Davidson actually talks of “causal truth”.

Prima facie, the statement that “beliefs are…identified…by their causes” seems a little odd. How can a belief and a cause, or a belief and an object or property, be the same? There must be some temporal and cognitive distinction between the two. In fact it seems like a metaphysical realist’s position. That is, there is an algorithmic relation between causes/objects/events and beliefs. One is a faithful representation of the other. However, to say that the beliefs and their causes are as one, or identifiable, doesn’t itself mean that the beliefs faithfully represents the objects/causes. It’s just that there’s no interpretative gap between causes/objects and beliefs. The naïve realist believes that the belief can only be of one kind if in proper contact with the object/cause. The metaphysical realist believes that in principle we could faithfully reflect the object/cause. Davidson may just be saying that there is no interpretation, though the concepts we have are contingent and possibly false. However, he does say that our beliefs are mostly true.

But there are two – amongst others – possible ways of thinking about this causal story. The first is offered by Sellars. He thinks that “it takes a long time…for those causal links to whip us properly into correspondent shape, and in the meantime we may be talking about what doesn’t exist”. He then goes on: “Davidson, on the other hand, simply ignores such problematics.” He thinks that such causal links can do nothing but whip us into the right “correspondent shapes”. Why? Because we are language users and language users are part of a community and a community can only share a language if it is being whipped into the same correct shapes by the same causal phenomena. Davidson therefore extricates himself, he thinks, from both conceptual relativism and the “very idea of a conceptual scheme”.

There is a strong conclusion to Davidson’s line of thought. It is that “everybody has always talked about mostly real things, and has made mostly true statements” (159)

We probably are being “whipped” by the same causal forces. But I don’t see how it can be said that we are being “whipped” into the correct shapes. Sellars appears to be correct. In principle, we might be massively wrong about what there is and the way it is. Davidson’s strong and possibly justified acceptance of the “hardness” of retinal stimulation doesn’t entail in itself beliefs in “mostly real things”. The causation is necessary, perhaps, but our concepts and beliefs aren’t.

We could share a language, in fact we obviously do, and be massively wrong about the world and the way it is. Sharing a language doesn’t seem to entail “mostly true statements” in and of itself. Sharing a language may mean that we are interacting with the same casual forces in same way and with the same mental and brain architectures and the same sensory receptors. But what about a possible Universal Evil Demon? What about massive error? What if we are all brains in vats? What if our concepts are way off mark? What about all the rest of the sceptical stratagems? Are they all false? Davidson surely can’t be playing a numbers game. Communities, no matter how large, have been massively wrong in the past (about the flat earth, about the non-human nature of Jewish people, about demons and devils etc). This begs the question: What are Davidson’s “true statements”? He can’t be talking about the true statements of logic and mathematics because they don’t involve “causal forces” or any causal intercourse. Perhaps it is very basic Kantian concepts and beliefs he is referring to like those about objects, causation itself, temporal sequence and the nature of space.

So Davidson doesn’t think that that we have CSs on the one hand, and the uninterpreted world, or sensory data, on the other. What if the two are as one? What if we couldn’t have sensory data without concepts? They may be conceptually separable (think of old-style phenomenalists turning sense-data into objects and vice versa), but not ontologically so.

So sensory data always come with labels attached. We can, however, change the labels.

Davidson sees language and concepts as causal product of the world. Does that mean he thinks that language does not organise the world - it is only a product of the world?  Why can’t language organise the world and be a product of the world? It is just as much part of the naturalistic framework as anything else, and not something that guarantees “conditions of describability”.

I would accept its naturalistic status too. It is precisely because it is natural product of the world that it may be wrong about it. Even a Darwinian accepts that nature can make mistakes. Naturalism, in and of itself, doesn’t insure accuracy of representation or conceptual organisation, etc. Here’s Plantinga on the subject:

…the relevant faculties may be functioning properly…but nevertheless not in a way that leads to truth, to the formation of true beliefs. But then proper function in a right environment is not sufficient for warrant…Not all aspects of the design of our cognitive faculties need be aimed at the production of true belief; some might be such as to conduce to survival, or relief from suffering, or the possibility of loyalty, or inclination to have more children, and so on. (453)

However, the transcendentalists about language or mind, or the metaphysical realist, hasn’t, prima facie, any more reason than a Davidsonian naturalist to believe in the truth or accuracy of representation, concepts and language itself. Even the innatist about concepts, or an old-fashioned Rationalist, doesn’t by being so guarantee himself that the mind and language are “the mirrors of nature”. Such people have simply taken the mind, its language and concepts, out of nature without thereby guaranteeing faithful intercourse with the world. The non-material mind may have as difficult relationship with the material world as the material mind/brain has with non-material platonic universals or abstract objects generally

Perhaps Davidson’s epistemological criticisms of Quine will help clarify things.

Quine “makes interpretation depend on patterns of sensory stimulation”. Davidson (161), on the other hand, makes it depend on “external events and objects”. There is, therefore, no application of a CS to an “uninterpreted reality”. Davidson, obviously, doesn’t deny that there are in fact sensory stimulations; he says only that the sensations, which make up events and objects, are already interpreted. Or, more correctly, there is no cognitive interpretation such that “x is F”. There is no separation between sensory data and object/event at the epistemic level, though there is at the causal and, perhaps, ontological level. We can see a cat as a cat. We don’t infer that a cat is a cat. Of course, the sensory data that emanates from the cat must come before the belief that “That is a cat”, but this is a causal not an epistemic antecedent.

Quine’s position

A cat Õsensory stimulationsÕ.rational inferencesY”That is a cat.”

Davidson’s position

A catÖ“That is a cat.”

We don’t experience sensory data or stimulations or sense-data, we experience objects and events right from the very start. This is not true only of the Folk; it was true, also, of old-style phenomenalists. And therefore Quine too. Even if the phenomenalists had their philosophical hats on, they would have still experienced objects and events not sense-data. What, in fact, does “sensory data” actually mean (in the sense of it being prior to our cognitions of objects and events)? Even if we try our hardest, we can’t get away from objects and events. This is a quasi-Kantian position that we conceptualise and categorise objects and events a priori. We don’t make sense of objects “from scattered sense events” (Quine). We see them. Therefore there is no application of a CS, or anything else for that matter, according to Davidson.

What of objects and events we have never experienced before? In these situations we apply concepts and categories which we already have, but, perhaps, only provisionally. For example, an unknown object or event is still an object or event of some kind. Such objects and events still fall under pre-existing concepts. Later, we may apply criteria of identity to the already conceptualised objects and events. And, at a lower level, the new object or event may still contain properties (non-alien) that we are already familiar with. For example, the object may be brown, small, shiny, wet, etc., and the event may be short, causal etc (see Armstrong and Lewis). (Much of this may be partly a result of what Goodman calls “categorial conservatism”- which is a psychological, rather than Kantian or philosophical point.)

The experiences we have, or, more technically, the sensory data we receive, according to Davidson, have already been contaminated by our beliefs and concepts. In a sense, or even literally, we don’t confront sensory data, we confront beliefs and concepts by virtue of the fact we confront individuated objects and events. This means that pure metaphysical realist experience – experience untouched by concepts, theories etc – is not possible. We confront, in fact, sensory data that has already become evidence. That is, sensory data touched by theory and concepts. There is never an epistemic time before sensory data is so untouched. This means that evidence, not sensory data, makes beliefs true, according to Davidson. It is often thought that evidence is a consequence of sensory data or experience. In reality, the experience is the evidence or it comes with the evidence.

But is Davidson right about Quine? Did Quine really claim that we don’t perceive objects and events; we actually see “scattered sense-events” and from these infer or posit the existence of objects and events? It is of course the case that non-cognitively, that is non-epistemically,  “scattered sense-events” are synthesised, in a Kantian way if you are a Kantian or quasi-Kantian. But wouldn’t this be a subject for the neuroscientist and not the philosopher? Consequently, Quine, even though he is an epistemological naturalist and is, therefore, committed to the findings of the hard sciences, surely doesn’t mean this. He is, after all, a philosopher (if a naturalist one).

So is Davidson simply making the obvious point that we don’t always – or ever – apply our CS to “uninterpreted reality”? But why shouldn’t a theorist or philosopher, or even a philosophical person outside the Academy, do so?

If Quine, or old-style phenomenalists, really did infer objects and events from “scattered sense-events” or sense-data, that is, posit objects and events (as “convenient myths”), they would have been unable to function cognitively. More correctly, the contingent mind would have been doing the things that the brain should and does do automatically.

Quine could have asked why we see objects and events and not “scattered sense-events” or sensory data. That is, how or why is it all put together? This is, of course, a valid philosophical question. But he could still not have inferred or even posited objects and events from the sensory data. Of course there are some experiences that aren’t “object-involving” (Davies, 310) or event-involving. For example, when we are dazed and confused or when we have just rubbed our eyes. But Quine couldn’t have been talking about such cognitive or sensory malfunctionings. And there are cases when we are cognitively and sensorily functioning but are still, as it were, confused. This still wouldn’t be a case of seeing “scattered sense-events”. It would be a case of seeing unfamiliar objects and events. So Quine couldn’t have been talking about these examples either.

Quine could have asked if we were justified in believing object A to be object A. So he could have worked in the contrary direction: from objects to sensory data. His appeal to sensory stimulations gave him, perhaps, a hook into the world. But even working backwards this enterprise may not be possible. I see a cat and ask myself: Is that truly a cat? Therefore I analyse the sensory stimulations that gave rise to my positing of a cat… Or do I really analyse the sensory stimulations? What does that mean? How do I stop my sensory stimulations being “object-involving”? Quine could have said: OK, it is still “object-involving”, but not necessarily cat-involving. However, if he had said that it would no longer be an appeal to naked and pure sensory stimulations. So we would be back to a Davidsonian, I think, position. (See Fred Dretske’s Knowledge and the Flow of Information, 1981 – page 822 and his Seeing and Knowing, 1969, 652)

There are, indeed, cognitive inferences in experience, but not from sensory data to objects and events, but from objects to other (perhaps the correct) objects. This is the case with the famous elliptical penny example of the phenomenalists. For example, I can see what I think is a bloody corpse in the woods and infer, without a change in the sensations (see Peacocke), that it is in fact a felled tree. This is an inference from an object (a corpse) to an object (a felled tree), not from sensory data to an object. Even in the most bizarre hallucinations, or non-hallucinations, the inferences would still be from object to object not sensory data to object.

I am a Kantian in this sense. Some concepts are not – or never – cognitively applied, at least not by people who already have fairly determinate concepts (this last clause is about contingent concepts, and is therefore non-Kantian point). Adults hardly ever – or never – apply the concept [cat] to a cat or [crash] to a crash, unless they are cognitively malfunctioning or their sensory receptors are in disarray. Indeed if we applied concepts to everything we would be put in a state of cognitive disarray. The innateness of Kantian concepts, and the acquired “innateness”, as it were, of contingent concepts, is given a Darwinian twist. This needs to be the case for us to survive. If we applied the concept [crash] to a crash, we could well become the victim of a crash. Similarly, if we cognitively applied concepts to everything, it would take us a day to get out of bed.

This is not to say, of course, that new concepts never come into being or are never created. For example, take this almost-new concept, the concept [lad]. The concept [lad] has certain conceptual criteria of identity: [a lad drinks copious pints of lager], [wears short-sleeved shirts in winter], [has a very low voice], [likes girlies and football], etc. I have created a new concept [lad] (based, perhaps, on older concepts). However, the concept has a real extension (the Set of Lads). Five weeks ago I may not have had this concept [lad], which doesn’t mean, however, that it is a false or unreal concept. So I, at one point, cognitively applied the new concept to young men of a certain kind. The predicate “is a lad”, or the concept [lad], as I’ve said, has conceptual criteria of identity, which itself implies the cognitive application of the concept. Therefore it is a thoroughly non-Kantian contingent concept.

Of course I needn’t have, I didn’t, sit down and do a bit of conceptual analysis of lads. I didn’t need to. However, I now have specific conceptual criteria of ladhood that were once a little vague and indeterminate. I have sententialised the concept [lad]. (Incidentally, a mental image of the archetypal lad is an essential part of my lad concept. See my “Non-Conceptual Experience” earlier.)

In the future, the concept [lad] may not be applied in the same way. I will simply non-cognitively see a lad as a lad. Similarly, the concept may change some – but not all – of its conceptual criteria of identity. (Is this lad essentialism?)

(Is there now the universal Lad in the platonic realm drinking lager? Is He instantiated here on earth by particular lads drinking lager?)

Again, we can change our concepts. We can even adopt a contradictory one of the same object. I believed that S was a [good man] at t. But then I believed that S was a [bad man] at t-1.

So Davidson may simply be denying that we cognitively apply concepts to particular objects and events. Alternatively, it may be conceptual schemes that he is saying that we don’t adopt or apply. He may be saying that just as we don’t cognitively apply concepts to objects and events (for reasons of hard causal interaction), perhaps we don’t cognitively apply and adopt a CS and apply it to objects and events. All this is pure speculation. As I said, Davidson doesn’t offer enough argumentation in his “The Very Idea of a Conceptual Scheme” about scheme/content “dualism”.

In any case, philosophers, theorists and philosophical members of the Folk do cognitively adopt CSs. And so cognitively apply concepts. This much is obvious to me. So what does Davidson mean?