(Perhaps too, the sense of 'content' in 'propositional content' is analogous? We might say that a proposition expresses some propositional content (we can understand why one wouldn't want to reduce the propositional content of some proposition to that particular proposition, or a relation whereby a proposition expresses some propositional content) — csalisbury

The analogy is appealing.

We say various subjects on various occasions undergo various “propositional attitudes” with respect to the same “propositional content.” On each such occasion there’s a relation of the content and the attitude; in at least some cases the attitude is or involves an “act” like doubting, affirming, hoping, and so on. Do we also say that the propositional content is supposed, or by many supposed, to have the form of a fact?

Likewise, we might expect that various subjects on various occasions undergo various “intuitions” (acts of awareness with “sensible character”) with respect to the same “sense content”. On each such occasion there’s a relation of the content and an act like sensing or imagining. Does the sense-datum theorist say (or does Sellars’ argument drive the sense-datum theorist to say) that the sense content has the form of a fact?

How are sense contents related to propositional contents? If a sense content occurs in “an act of intuition with sensible character”, can it also occur as a propositional content in a thought without (the relevant sort of) “sensible character” -- i.e., when if ever is the propositional content in such a thought also a sense content? On the other hand, is a sense content always a propositional content -- all sense contents are propositional contents, but not all propositional contents are sense contents?

Are Sellars and the sense-data theorists at all concerned with what I’ve called “sensible character” here, what I might call qualia or phenomenal characteristics paradigmatically associated with perceptual experience? I’ve got the idea that those influenced by Sellars tend to differ along these lines. McDowell seems more interested in phenomenology, while Brandom and Rorty seem, like Davidson, wary of subjectivity.

You've really analyzed the hell out of this paper. Nice work, man! Like Calisbury, I'm short on time otherwise I'd participate a little more. — Aaron R

Thanks. I'm afraid it's a clumsy, discursive way of working through a text. But it's my favorite way of traversing the hermeneutic circle.

I won't be able to maintain this work rate for long. Hopefully each of us pushes the thing along when he has some time for it.

The essay isn't going anywhere!

Thanks. I'm afraid it's a clumsy, discursive way of working through a text. But it's my favorite way of traversing the hermeneutic circle. — Cabbage Farmer

Nothing wrong with that. I imagine this is the way most people work through a difficult text, whether they realize or not.

You asked about historical context back on page 7 (I think), and so I wanted to add some historical notes that might be of interest:

EPM is probably best be understood as a response to theories put forward by the likes of William James, Bertrand Russell, C.D. Broad, C.I. Lewis, G.E. Moore, A.J. Ayer, H.H. Price, and Hans Reichenbach (among others) in the early part of the 20th century. He probably also had philosophers as diverse as Rudolph Carnap and Edmund Husserl in view as well.

There are obviously substantial theoretical differences in the positions of each of the aforementioned thinkers (insofar as it even makes sense to say that a philosopher ever has a stable position throughout the course of their careers). Riechenbach, for instance, argued that physical objects, not sense-impressions, are directly given in experience. According to Riechenbach, sense-impressions are mere abstracta that are never "seen" by anyone, much less seen directly (see Experience and Prediction, 1938). C.I. Lewis and H.H. Price, by contrast, tended to argue that it is colors, shapes, textures, etc. that are directly given in experience, and that beliefs about physical objects (and even physical objects themselves) are inferentially constructed out of these (See Lewis' Mind and World Order, and Price's Perception). Consider the following quotation from Price's Perception:

When I see a tomato there is much that I can doubt. I can doubt whether it is a tomato that I am seeing, and not a cleverly painted piece of wax. I can doubt whether there is any material thing there at all. [...] One thing however that I cannot doubt: that there exists a red patch of a round and somewhat bulgy shape [...] directly present to my consciousness. [...] And when I say that it is directly present to my consciousness, I mean that my consciousness of it is not reached by inference [...]. This peculiar and ultimate manner of being present to consciousness is called being given and that which is thus present is called a datum. — Price

Russel and Moore also tended to analyze physical objects as reducible to directly apprehended sense-data (e.g. secondary qualities).

William James was somewhat unique in maintaining that pretty much everything except the "entirety" or "totality" of experience is directly apprehended. In his view, sense-data (or "percepts") are no more or no less "given" than the relations (both causal and inferential) that bind them together (see A World of Pure Experience, 1904).

The common thread running through the diverse positions of all of these thinkers is this notion of a directly apprehended datum, or "given". EPM does not take issue with the concept of direct apprehension per se, but rather with that the way that the concept of direct apprehension is used to justify foundationalist epistemologies in which the objects or contents of direct apprehension are taken to be incorrigibly or indubitably known purely on the basis of their being directly apprehended.

Thanks for the background! Super helpful

We should have you rewrite EPM, because you write far more clearly than Sellars.

Here’s an extended summary regarding Sellars's analysis of "Looks" talk. I don't spend much time justifying my interpretation or explicitly engaging with the text (and I gloss some things), but I'd be happy to dig deeper if anyone wants to.

I think it was @Calisbury who somewhere said that Sellars employs a kind-of "anstoss" logic in regards to the concept of veridicality, and that this under-girds his analysis of the “is/looks” distinction. I think that’s basically correct. In the “Looks” sections Sellars is responding to the Pricean line of argument that goes something like this (very roughly):

1. When I (ostensibly) see an apple, there are many things I can doubt about my experience, like that the object is really an apple, or that there is a material object present at all.

2. However, there are some things I can't doubt, such as that there is a round red patch present to my awareness.

3. The fact that I can't doubt these beliefs implies that they need not be justified on the basis of any other beliefs (e.g. they are not inferentially mediated).

4. Therefore, it is things like round red patches that are the direct objects of apprehension, and it is only my beliefs about such things (i.e. beliefs about how things "appear") that could (and should) serve as the indubitable foundational (non-inferential) basis of all empirical knowledge.

Sellars basically thinks that this gets things ass-backwards. He doesn’t exactly deny that beliefs about the appearance of round red patches in a person’s visual field can serve as evidence for the existence of, say, an apple on a desk, but he will claim that they provide (at most) very weak evidence when taken on their own, but that this is not their primary function anyway.

Sellars’s alternative analysis is basically this: the primary function of claims about the way things “appear” or “look” or “seem” is to provide a mechanism for withdrawing from claims about the way things really are. So to say “the object over there looks red”, or “It looks like there is a red object over there” is to withdraw further and further away from a claim about how things really are, namely, that “there is a red object over there”. The notion of “looking” or “appearing” (e.g. looks red) is conceptually dependent on the more fundamental notion of “being” (e.g. is red).

If we are inclined to agree with this analysis, then it becomes hard to deny that claims about appearances can't really provide an adequate evidential basis for claims about how things really are. That’s because each and every claim about how things appear is nothing more than the withdrawal from some companion claim about how things really are. As such, it’s a mistake to think that we do or ought to start from claims about how things appear and then inferentially move to claims about how things really are. Instead, we start from claims about how things really are, and then inferentially move to claims about how things merely “appear” only upon being confronted by evidence that suggests we were mistaken about how things really are.

So, if claims about appearances can’t serve as an adequate evidential basis for empirical knowledge of how things really are, then what can? For Sellars, the only adequate justification for such claims is to be found in other empirical claims about how things really are. In particular, claims about the context of observation ought to (and do) serve as the primary justificatory backdrop for our first-order observational claims. To justify a claim like “this necktie is green” we ought to appeal to facts about the situation in which we find ourselves – that is, to claims about our historical reliability at distinguishing colors, about the lighting conditions, about whether we are afflicted with color-blindness, about whether we are wearing green-tinted contact lenses, about whether we are currently on drugs that effect our ability to distinguish colors, about whether someone is likely to be playing a practical joke on us, etc. Again, the ultimate appeal to empirical claims as opposed to claims about appearances is hardly surprising given Sellars’s analysis of the looks/is distinction. For Sellars, to vacate all claims about what is the case is equivalent with vacating all claims to knowledge itself, and so a substratum consisting purely of beliefs about appearances is therefore simply not a viable starting point for empirical knowledge.

As an aside, it’s interesting to consider how Sellars ultimately deals with the charge of begging the question against skepticism (this is only cursorily covered in EPM. See the articles “More on Giveness and Explanatory Coherence” and “Kant’s Theory of Experience” for details). If empirical claims about what is the case are to be justified by other empirical claims about what is the case then it appears that we’re caught in a vicious circle. Sellars will make a Kantian-style transcendental argument to the effect that the linguistic practices that are the pre-condition for the formulation of the skeptic’s challenge (or any challenge) are only intelligible against the backdrop of the very empirical knowledge that the skeptic is challenging the legitimacy of. Not that this approach is original to Sellars (though he puts an interesting “linguistic” twist on it), but it is note-worthy in that he is one of the only philosophers in his milieu (outside of Strawson) to explicitly employ the machinery of transcendental argumentation in his work.

Can you develop the Hegelian point?

I'm never sure what's meant by "qualities" and "universals." I recall Kant speaks of "sensible qualities" -- is that the right ballpark?

I can think of two sorts of "mediation" to watch out for: mediation by inference and mediation by concepts. I expect the given to play the role of foil in this narrative, while Sellars develops an argument supporting a view in which nothing is given in noninferential knowledge acquisition without some contribution from concepts -- or however he puts his version of the Hegelian theme developed in our time by Sellarsians like Brandom and McDowell.

I'm thinking of Hegel's analysis of perception in Section 2 of Phenomenology of Spirit. I just took a look back at the text and I guess he actually uses the word 'property' (or, more exactly, Eigenschaft, which can be also be translated as quality.)

In any case, (my interpretation of) Hegel's point is that, in speaking of a 'quality' or 'property,' we're speaking about something repeatable, something that could (at least in principle) also be a quality or property of something else. For Hegel, this means that we're dealing with universals and I think he's right. But 'universal' is a loaded term, and I don't want to drag things too far afield. In any case, the 'mediation' I mentioned is exactly the second sort of mediation you mention, a mediation by concepts. To understand a quality or property as a quality or property, we must have recourse to the conceptual.

I hope that makes sense. A lot of these terms (concept, universal, property, quality) kind of bleed into one another.

Nothing wrong with that. I imagine this is the way most people work through a difficult text, whether they realize or not. — Aaron R

It does seem to belong to the process of interpretation whether we notice it or not. I enjoy building it into my method, and I like to pretend this improves my results. For instance, by breaking the text into small parts and working through them in a format that resembles conversation.

The practice is in keeping with a characterization of interpretation as a conversation with the text, and also in keeping with ancient customs of philosophical dialogue and philosophical commentaries, with ordinary human conversation, and with our behavior in forums like this one here.

But it takes too damn long.

You asked about historical context back on page 7 (I think), and so I wanted to add some historical notes that might be of interest — Aaron R

Thanks so much! I was hoping someone would fill in that blank.

EPM is probably best be understood as a response to theories put forward by the likes of William James, Bertrand Russell, C.D. Broad, C.I. Lewis, G.E. Moore, A.J. Ayer, H.H. Price, and Hans Reichenbach (among others) in the early part of the 20th century. He probably also had philosophers as diverse as Rudolph Carnap and Edmund Husserl in view as well — Aaron R

Quite a cast of characters. Which would you say figure most prominently in, or had most influence in the field in the decades preceding, EPM? Or, which if any are closest to the typical sense-datum theory Sellars takes aim at in the essay?

Riechenbach, for instance, argued that physical objects, not sense-impressions, are directly given in experience. According to Riechenbach, sense-impressions are mere abstracta that are never "seen" by anyone, much less seen directly (see Experience and Prediction, 1938). — Aaron R

I'm not sure I've heard of Riechenbach. The view that it's physical objects, not sense-impressions, which are "seen" sounds remarkably fresh.

So he buries sense-impressions behind experience.... Are the impressions mere theoretical constructs for him, or do these constructs perhaps refer to terms that play some causal or functional role in the organization of experience, or does he reject all such theories....

C.I. Lewis and H.H. Price, by contrast, tended to argue that it is colors, shapes, textures, etc. that are directly given in experience, and that beliefs about physical objects (and even physical objects themselves) are inferentially constructed out of these (See Lewis' Mind and World Order, and Price's Perception). — Aaron R

This sounds perhaps more problematic. Does it mean that some "beliefs" about "given" colors are also "given"? I recall Sellars' distinction between sensing particulars and sensing facts.

How and where does "inference" occur in their models? Surely not in experience: We don't run around the world ceaselessly making conscious inferential judgments about physical objects on the basis of given colors and shapes; but arguably we have (or many of us have) countlessly many "beliefs" about local physical objects on the basis of current perception.

If there's something like "inferential judgment" at play in all our perceptually grounded beliefs about physical objects in the environment, it must be a sort of "inference" that occurs behind the scenes of conscious awareness.

Perhaps this is one way to interpret, or to correct, strange old views in our tradition concerning "judgments of sensation": The synthesis of "objects" from the most basic elements of sensation is achieved primarily by nonconscious cognitive processes that inform conscious perceptual experience. The application of the language of "inference" and "judgment" to such nonconscious spontaneous processes, excusable in days gone by, is merely a sort of analogy. However, when nonconscious processes of sensory cognition result in perceptual experience, we may begin to speak of perceptual content, and this content is not only open to inferential judgment, but also arguably charged with meaning, with conceptual content, and thus full of implications.

Of course the conceptual content, the "beliefs", and the "implications" of perceptual content, would depend in part on the conceptual stance and character of the perceiver, as well as on what's "given" to perception by sensation. Particular conscious acts of inference on the basis of perception occur within this constant shifting context.

I suppose the hard problem of perceptual content, along these lines, is to clearly distinguish what's "given" to perception by sensation.

Consider the following quotation from Price's Perception:

When I see a tomato there is much that I can doubt. I can doubt whether it is a tomato that I am seeing, and not a cleverly painted piece of wax. I can doubt whether there is any material thing there at all. [...] One thing however that I cannot doubt: that there exists a red patch of a round and somewhat bulgy shape [...] directly present to my consciousness. [...] And when I say that it is directly present to my consciousness, I mean that my consciousness of it is not reached by inference [...]. This peculiar and ultimate manner of being present to consciousness is called being given and that which is thus present is called a datum. — Price — Aaron R

A beautiful passage. What's Price's spin on the phrase "reached by inference"? It could mean i) "achieved by a process of inference", or far more weakly, ii) "open to revision by a process of inference". A huge difference. As I suggested earlier, I think (i) is just wrong. If we run with (ii), then we have a conception of "presence to consciousness" or "givenness" according to which the datum that is given in perceptual experience is indubitable. It's not merely that the datum is indubitable; it seems Price has here carved out the sense-datum as whatever's indubitable in perception. An admirable exercise in skepticism. Good show.

Perhaps we can call this "datum" the sense-content component of perceptual content, and say that perceptual content includes both sense-content and conceptual content?

Is Price the source among analytic philosophers of talk about "direct presence" and "presence to consciousness", or did the usage precede him? I believe the first time I encountered the term, or the first time it stuck, I was reading Anscombe's "First Person".

Russel and Moore also tended to analyze physical objects as reducible to directly apprehended sense-data (e.g. secondary qualities). — Aaron R

They say the physical objects themselves, or merely our experience and knowledge of physical objects, are thus reducible?

William James was somewhat unique in maintaining that pretty much everything except the "entirety" or "totality" of experience is directly apprehended. In his view, sense-data (or "percepts") are no more or no less "given" than the relations (both causal and inferential) that bind them together (see A World of Pure Experience, 1904). — Aaron R

It makes sense to say some relations are ordinarily given in perception: I don't perceive two objects and then infer that one is to the left of the other; I see that it's so. Likewise, I don't feel the pot and and feel some heat and infer that the pot is hot, I feel that it's hot. I don't see the glass fall to the floor and infer that falling to the floor caused the glass to shatter... I believe falling was the cause of shattering when I see it, without making any inference....

But each of those claims is "open to doubt", and hence not "directly present to consciousness" or "given" in the sense carved out by Price. Or at most the first one, about relative spatial position, is given in Price's sense. (on my reading of that little passage)

The common thread running through the diverse positions of all of these thinkers is this notion of a directly apprehended datum, or "given". — Aaron R

So the family of terms includes "direct apprehension", "direct presence (to consciousness)", "direct givenness", "givenness", and "immediacy". And on the object side, "percept", "datum", "sense-impression", "sense-datum", "sense content", and I don't know what.

EPM does not take issue with the concept of direct apprehension per se, but rather with that the way that the concept of direct apprehension is used to justify foundationalist epistemologies in which the objects or contents of direct apprehension are taken to be incorrigibly or indubitably known purely on the basis of their being directly apprehended. — Aaron R

I'm inclined to say there is something indubitable in experience, along the lines I just drew out from Price, and to characterize what's there indubitable as "appearance". I'd say, further, that this indubitable appearance does provide a sort of anchor or foundation for the beliefs and knowledge of each rational sentient agent, from one moment to the next. But not the sort of foundation that could possibly ground anything like certain knowledge beyond the appearance of the present moment.

Of course knowledge is not the same thing as certain and indubitable knowledge.

I suppose the "foundationalist epistemologies" in question are in search of something more impressive than mere knowledge?

In any case, (my interpretation of) Hegel's point is that, in speaking of a 'quality' or 'property,' we're speaking about something repeatable, something that could (at least in principle) also be a quality or property of something else. For Hegel, this means that we're dealing with universals and I think he's right. — csalisbury

It often occurs to me that repeatability is essential somehow to the conceptual, but I've never managed to satisfy myself unpacking that "somehow".

Frege and Russell instruct us in formulating concepts of particulars. These are repeatable across various occasions, in different propositional contexts, in utterances by various speakers. The object corresponding to a concept of a particular may be varied in thought (e.g. in ignorance of the relevant facts or in counterfactual exercises), and thus the concept is "repeated" in application to a range of possible objects. Accordingly, it seems there is a generic character even to concepts of particulars. But I'm wary of conflating this generic and repeatable character of concepts with a presumption that concepts always correspond to something like "general terms" in the ordinary sense, man in general as opposed to this or that man.

But 'universal' is a loaded term, and I don't want to drag things too far afield. In any case, the 'mediation' I mentioned is exactly the second sort of mediation you mention, a mediation by concepts. To understand a quality or property as a quality or property, we must have recourse to the conceptual. — csalisbury

We who speak here aim to understand the quality as a quality, and to account for the sense in which a perceiver who says "This apple is red" says so in virtue of his own grasp of the "qualities" that appear to him. But to speak of "qualities" this way is to have already conceptualized the "content" of perception.

I'm inclined to say, on the basis of introspection, that ordinary perceptual experience is always already conceptualized. That doesn't entail that there is no determination of perceptual content by the "receptive" powers of sense perception; and it rather seems there is such a determining contribution from sensation in perceptual experience.

Accordingly, to characterize "sense qualities" as something like general concepts is arguably to jump ahead of a tricky discourse about the receptive character of a perceptual experience strongly determined by sensation. A discourse Rorty, for instance, seems eager to evade at any price.


Consider the quality as in the first place, not a conceptualized appearance, but an activity of the organism or mind involving an appearance of a particular type.

I move my arm to wave; I move it again to make "the same" gesture on a second occasion. I call it "the same". Perhaps you're observing me and agree it's "the same". Perhaps we turn to details, fleshing out criteria of similarity, declaring when to count a gesture as a wave, when to count a wave as this type of wave, and how to handle borderline cases.... However that exercise turns out, it's two different events in the world that we're considering, in terms of a similarity that we assign, on the basis of objective characteristics that we deem relevant and assess from a certain point of view, with a certain purpose, with a certain degree of accuracy and precision....

Likewise, we might say two visual appearings of or involving a quality of red are the same with respect to this quality on the basis of objective characteristics that we deem relevant and assess from a certain point of view, with a certain purpose, with a certain degree of accuracy and precision....

Along these lines, one trick would be to distinguish the "objective characteristics" occurring in two events from the conceptualization of characteristics by virtue of which two instances are called alike, the same type, two of a kind, for instance, "red".

It's not that the concept red appears to me; but rather what appears to me I immediately recognize as red.

I hope that makes sense. A lot of these terms (concept, universal, property, quality) kind of bleed into one another. — csalisbury

So much philosophical activity must be spent sorting out such terms as they occur in the speech of others and in one's own usage. The difficulty of that task, the brevity of life, and the growth of the record of discourse over time, give a strong motive for plain speaking.

Here’s an extended summary regarding Sellars's analysis of "Looks" talk. I don't spend much time justifying my interpretation or explicitly engaging with the text (and I gloss some things), but I'd be happy to dig deeper if anyone wants to.

I think it was Calisbury who somewhere said that Sellars employs a kind-of "anstoss" logic in regards to the concept of veridicality, and that this under-girds his analysis of the “is/looks” distinction. — Aaron R

Would anyone care to say something about “anstoss logic”? What does this phrase mean?

I think that’s basically correct. In the “Looks” sections Sellars is responding to the Pricean line of argument that goes something like this (very roughly):

1. When I (ostensibly) see an apple, there are many things I can doubt about my experience, like that the object is really an apple, or that there is a material object present at all.

2. However, there are some things I can't doubt, such as that there is a round red patch present to my awareness. — Aaron R

Or perhaps less boldly:

2’: … such as that there seems to be a round red patch present to my awareness.

Or even less:

2’’: … such as that there seems to be a seeming-round seeming-red seeming-patch present to my awareness.

Are such revisions in keeping with Price? Or would he deem the “seemings” redundant or inappropriate here, perhaps in keeping with a view that “the presence of a round red patch” is not the sort of thing that “seems”, and is instead in every respect immune to doubt?

3. The fact that I can't doubt these beliefs implies that they need not be justified on the basis of any other beliefs (e.g. they are not inferentially mediated). — Aaron R

There’s something intuitively plausible about this, once we massage it the right way. I don’t ordinarily justify my belief that I am seemingly seeing a seemingly red seeming object on any other grounds than that it appears so to me in the ordinary first-person point of view; ordinarily there are no other grounds available for such claims.

However, putting it the way Price has here, one might ask him, what justifies your belief that there are such things as “patches”? And it may turn out that this is only a loose manner of speaking.

4. Therefore, it is things like round red patches that are the direct objects of apprehension, and it is only my beliefs about such things (i.e. beliefs about how things "appear") that could (and should) serve as the indubitable foundational (non-inferential) basis of all empirical knowledge. — Aaron R

See my concerns about “seeming” and “patches” above. And my previous and subsequent mutterings about epistemological “certainty” and “foundations”.

Sellars basically thinks that this gets things ass-backwards. He doesn’t exactly deny that beliefs about the appearance of round red patches in a person’s visual field can serve as evidence for the existence of, say, an apple on a desk, but he will claim that they provide (at most) very weak evidence when taken on their own, but that this is not their primary function anyway. — Aaron R

I hope to find myself aligned with Sellars in criticizing the strength of the warrant provided by claims like “I seem to see a seeming-red seeming-apple” for claims like “There is a red apple there”. We might say the first claim provides a prima facie reason for belief in the second claim, but nothing close to a foundation for “certain knowledge” that the second claim is true.

Likewise, however, the dubitability that burdens the more objective claim does not seem to transfer to the more subjective claim. If Price is one of those epistemologists with anxieties about “certainty”, who hopes to derive perfectly certain empirically grounded beliefs about apples and tables from the beliefs about seemings we may carve out as an epistemic layer entailed by experience, of course he’s barking up the wrong tree.

That doesn’t make it nonsense to speak about the epistemic role of that deep layer of belief based on appearances, which serves as a safety net for us to fall back on gracefully, precisely because it persists while we ignore it.

Sellars’s alternative analysis is basically this: the primary function of claims about the way things “appear” or “look” or “seem” is to provide a mechanism for withdrawing from claims about the way things really are. — Aaron R

Like Wittgenstein in On Certainty: One always forgets the expression “I thought I knew”.

Who forgets such an expression? Only a befuddled epistemologist like Moore, who seems to play the role of buffoon in Wittgenstein’s opera.

So to say “the object over there looks red”, or “It looks like there is a red object over there” is to withdraw further and further away from a claim about how things really are, namely, that “there is a red object over there”. The notion of “looking” or “appearing” (e.g. looks red) is conceptually dependent on the more fundamental notion of “being” (e.g. is red). — Aaron R

It’s a withdrawal from claims about how things really are over there, where there seems to be a red apple. But this doesn’t entail another withdrawal, from all claims about how things really are in the world: For this appearance, and this seeming, may be said to be part of the world. And it’s not clear that it may be coherently denied that the appearance and seeming are part of the world.

Accordingly, the appearance and the seeming are among the “facts” we may aim to piece together in each case. For instance by asking: How is it the case, how does it happen, that things appear thus and so (to me, now)?

In ordinary, happy cases, one fair answer is: Because there’s an apple there, and your eyes are open and aimed that way. The frequency with which seeming-seeings of seeming-apples turn out to be, to all appearances and “to a practical certainty”, genuine seeings of genuine apples, also serves as justification for the belief that “There is an apple there”, and lets the seeming-seeing of a seeming-apple stand as a noninferentially acquired defeasible warrant for the latter belief.

A more developed variation on this theme could haul into the account an empirically grounded story about light and the light-relative properties of physical objects; about retinas and cones and optic nerves; about perceptual processes in cognition.

None of this amounts to absolute “theoretical certainty” that there really is a real apple there -- a certainty we never attain, even as we bite and chew and swallow.

If we are inclined to agree with this analysis, then it becomes hard to deny that claims about appearances can't really provide an adequate evidential basis for claims about how things really are. That’s because each and every claim about how things appear is nothing more than the withdrawal from some companion claim about how things really are. — Aaron R

What shall we count as “adequate” evidential basis for claims about how things really are? I would argue along lines just indicated above, that the seeming-seeing of a seeming-apple does provide adequate evidential basis for claims about real apples, though not for claims that such claims are certain claims. In cases in which new evidence comes to light, that the seeming-apple was not in fact an apple, or that the seeming-seeing was not a seeing, we amend the record of discourse by adding new alleged facts (including claims about the seeming course of seemings) and revise or withdraw from our previous claim about what’s over there in the world where (it seems) it had seemed there was a seeming-apple.

Let’s not forget the role of expressions like “I thought I knew”. I thought there was an apple there, but it turned out to be a lump of wax. That doesn’t mean the initial claim was unjustified, unwarranted, and groundless -- only that it turned out to be incorrect; or rather that the speaker turned out to have reason to correct it. This new reason, or new judgment, is as fallible in principle as the first which it amends. But it seems such claims are correct when they happen to be so, and stand uncorrected until there’s reason for correction.

As such, it’s a mistake to think that we do or ought to start from claims about how things appear and then inferentially move to claims about how things really are. Instead, we start from claims about how things really are, and then inferentially move to claims about how things merely “appear” only upon being confronted by evidence that suggests we were mistaken about how things really are. — Aaron R

Who’s doing the “moving” here, and for what purpose? Arguably neither route will get us more “certainty” in the claims we make about what’s “over there”. It’s true that ordinary speech for ordinary purposes tends to follow the more direct route, speaking about “what’s over there” without wasting time on appearance-talk except in special cases, as when one is prompted by events to revise his own considered view about “what he thought he knew.”

The efficiency of that fine custom of, as it were, directly addressing things in the world as they appear to us, and revising only when it matters -- instead of constantly referring to the mediations of experience and conceivable doubts of reason -- gives us no reason to suppose those mediations vanish whenever there’s harmony between the seeming and the fact. It seems rather that when they appear in agreement, ordinarily we focus on the main track, according to our purpose, only falling back on the other, and to the task of realignment, when they fall out of whack.


I don’t mean to imply, by speaking this way, that the “appearance” is something like a color patch in my head, ontologically isolable from the “causal chain” that is -- I want to say -- identical to that appearance.

We may analyze a causal chain into parts, or happen upon one link at a time, and then in ignorant conjecture or counterfactual hypothesis, characterize that one link as part of indefinitely many conceivable causal chains. Some such stories seem to line up with the facts better than others. Yet others seem entirely off base.

And the facts keep coming in, or so it seems.

So, if claims about appearances can’t serve as an adequate evidential basis for empirical knowledge of how things really are, then what can? For Sellars, the only adequate justification for such claims is to be found in other empirical claims about how things really are. In particular, claims about the context of observation ought to (and do) serve as the primary justificatory backdrop for our first-order observational claims. To justify a claim like “this necktie is green” we ought to appeal to facts about the situation in which we find ourselves – that is, to claims about our historical reliability at distinguishing colors, about the lighting conditions, about whether we are afflicted with color-blindness, about whether we are wearing green-tinted contact lenses, about whether we are currently on drugs that effect our ability to distinguish colors, about whether someone is likely to be playing a practical joke on us, etc. — Aaron R

I strongly agree, objective claims about how things really are make good justifications for other objective claims about how things really are, and should be perhaps required in order to count such claims as “justified”. So far as I can make out, sophisticated theories about light and vision add little in this regard to a common-sense grasp of how seeing works.

Moreover, general knowledge about light and the historical reliability of judgments of color does not inform me of my present circumstances at all, unless experience informs me of my present circumstances in such a way as to warrant the application of thoughts about light and color discrimination to my thoughts about present circumstances. There’s no theory that tells me whether my eyes are open or closed right now, for instance, or whether it’s dark or light in here, or where the proximate light sources are in my vicinity of the world and what color of light they seem in my estimation to emit. Ordinarily I acquire such information noninferentially by using my eyes to see. It seems that any general theory of vision I may have acquired secondhand over the years, has accrued through centuries determined in part by processes involving the same sort of basis for judgment in others, who used their eyes in about the same way that I do on each particular occasion.

Again, the ultimate appeal to empirical claims as opposed to claims about appearances is hardly surprising given Sellars’s analysis of the looks/is distinction. For Sellars, to vacate all claims about what is the case is equivalent with vacating all claims to knowledge itself, and so a substratum consisting purely of beliefs about appearances is therefore simply not a viable starting point for empirical knowledge. — Aaron R

The main thrust of this point seems reasonable, but there’s arguably an unwarranted and undesirable implication, that “appearances” are not part of the empirical world, and that our accounts of appearances are not accounts of the empirical world.

I suggest, to the contrary, that “having an appearance” or “being appeared to” is a sort of knowledge of the empirical world -- a most fundamental sort -- that can be analyzed and expressed in terms of appearance-talk that coincides with matters of fact, whether or not the subject understands that or how the appearance coincides with other matters of fact.

The appearance is part of reality, that’s how it coincides. It is itself a matter of fact related to other matters of fact in the world. That’s how perceptual experience binds our thoughts to nature and opens the world to each perceiver.

As an aside, it’s interesting to consider how Sellars ultimately deals with the charge of begging the question against skepticism (this is only cursorily covered in EPM. See the articles “More on Giveness and Explanatory Coherence” and “Kant’s Theory of Experience” for details). If empirical claims about what is the case are to be justified by other empirical claims about what is the case then it appears that we’re caught in a vicious circle. Sellars will make a Kantian-style transcendental argument to the effect that the linguistic practices that are the pre-condition for the formulation of the skeptic’s challenge (or any challenge) are only intelligible against the backdrop of the very empirical knowledge that the skeptic is challenging the legitimacy of. Not that this approach is original to Sellars (though he puts an interesting “linguistic” twist on it), but it is note-worthy in that he is one of the only philosophers in his milieu (outside of Strawson) to explicitly employ the machinery of transcendental argumentation in his work. — Aaron R

I’m wary of transcendental arguments. For one thing, it seems they’re always missing the point in conversation with a skeptic who’s inclined to gesture at an infinite horizon of conceivable possibilities when we ask a question like “How possible?” On what grounds, in whose terms, for what purpose, do we limit the frame?

I consider myself a sort of skeptic, but I don’t worry much about circles. Language is prior to sophisticated epistemological arguments. Perceptual experience is prior to sophisticated empirical theories. Where do such self-evident, practically tautological, claims leave our conversation about the logic of appearances and its role in observational judgments?

One concern I have with the view you’ve attributed to Sellars, is that it may leave a gap between general empirical beliefs and particular perceptual occasions. I mean specifically, with respect to handling claims about whether and how appearances can “provide an adequate evidential basis for claims about how things really are”; and about “the primary justificatory backdrop for our first-order observational claims”. To unpack such baggage the wrong way, may be to fall into some version of the “frictionless spinning in a void” associated with coherentism.

By my way of reckoning, it makes more sense to emphasize the cooperation on each occasion of general beliefs and beliefs based on current perception. The essential role of the more general beliefs, and the ordinary tendency to gloss over appearance-talk in happy cases, doesn’t mean there isn’t a deep layer of belief that’s analyzable or even best described in terms of appearance-talk, playing an equally essential role in our acquisition of noninferential knowledge of present circumstances.

There’s no way I can keep up with you @Cabbage Farmer, but here’s a start at some answers to your questions:

Quite a cast of characters. Which would you say figure most prominently in, or had most influence in the field in the decades preceding, EPM? Or, which if any are closest to the typical sense-datum theory Sellars takes aim at in the essay? — Cabbage Farmer

I’d say that C.I. Lewis and H.H. Price figured very prominently, though the others mentioned are not far behind. As an aside, it’s interesting to consider that Sellars studied under C.I. Lewis while completing graduate studies Harvard.

I'm not sure I've heard of Riechenbach. The view that it's physical objects, not sense-impressions, which are "seen" sounds remarkably fresh.

So he buries sense-impressions behind experience.... Are the impressions mere theoretical constructs for him, or do these constructs perhaps refer to terms that play some causal or functional role in the organization of experience, or does he reject all such theories.... — Cabbage Farmer

In Chapter 2 of Experience and Prediction Reichenbach writes:

I cannot admit that impressions have the character of observable facts. What I observe are things, not impressions. I see tables, and houses, and thermometers, and trees, and men, and the sun, and many other things in the sphere of crude physical objects; but I have never seen my impression of these things […] I do not say I doubt the existence of my impressions. I believe that there are impressions, but I have never sensed them. When I consider the question in an unprejudiced manner I find that I infer the existence of my impressions. — Reichenbach

How are they inferred? He claims that they are inferred in order to explain the difference between appearance and reality:

To explain this difference, I introduce the distinction between the physical thing and my impression of that thing. I say that usually there are both physical things and impressions within me but that sometimes there are impressions only […] we need this assumption to explain that in the case of the confused world one of the two worlds, the external world, is dropped. The distinction between the world of things and the world of impressions or representation is therefore the result of epistemological reflection. — Reichenbach

He goes on to talk about “the abstract character of impressions” as part and parcel of a “duplicity theory” of perception. And this is where, as seen through the lens of EPM, Reichenbach’s account starts to go off the rails. We can already see in the quotes above the seeds of an ambiguity between impressions qua objects and impressions qua facts. Reichenbach says that he doesn’t “see” his impressions and also says that impressions are “not observable facts”. And yet he’s invoking impressions in order to explain discrepancies in what is seen.

Things get really bad in chapter 3 where Reichenbach seems to firmly place impressions into the category of “immediate existence”:

Imagine we are taking a walk at dusk through a lonely moor; we see before us at some distance a man in the road […] we do not doubt the man’s reality. We walk farther and discover it is not a man but a juniper bush. […] We shall say that both the man and the bush have immediate existence at the moments we see them. — Reichenbach

So here we seem to have the paradigm case of an impression. We think we see a man, but it’s really just juniper bush so we invoke the concept of an impression to explain the mistake. The man is inferred to be an impression, but here we are told that the “man” also has immediate existence. So what’s immediate existence?

We may regard immediate existence as a concept known to everybody. If somebody does not understand us, we put him into a certain situation and pronounce the term, thus accustoming him to the term and the situation seen by him. […] If a child asks “what is a knife” we take a knife and show it to the child. — Reichenbach

In other words, immediate existence is the existence of whatever is directly apprehended or, if you will, known by acquaintance. But wait…this directly contradicts Reichenbach’s earlier claim that impressions are indirect, theoretical entities. These are exactly the kinds of confusions that Sellars was responding to in EPM.

...

Well, I didn't make much progress here, but (as usual) I'm out of time. I'll try to respond with more later.

More thoughts:

This sounds perhaps more problematic. Does it mean that some "beliefs" about "given" colors are also "given"? I recall Sellars' distinction between sensing particulars and sensing facts. — Cabbage Farmer

Right. Both Lewis and Price (like Reichenbach) tend to elide the distinction between sensing particulars and sensing facts.

How and where does "inference" occur in their models? Surely not in experience: We don't run around the world ceaselessly making conscious inferential judgments about physical objects on the basis of given colors and shapes; but arguably we have (or many of us have) countlessly many "beliefs" about local physical objects on the basis of current perception.

If there's something like "inferential judgment" at play in all our perceptually grounded beliefs about physical objects in the environment, it must be a sort of "inference" that occurs behind the scenes of conscious awareness.

Perhaps this is one way to interpret, or to correct, strange old views in our tradition concerning "judgments of sensation": The synthesis of "objects" from the most basic elements of sensation is achieved primarily by nonconscious cognitive processes that inform conscious perceptual experience. The application of the language of "inference" and "judgment" to such nonconscious spontaneous processes, excusable in days gone by, is merely a sort of analogy. However, when nonconscious processes of sensory cognition result in perceptual experience, we may begin to speak of perceptual content, and this content is not only open to inferential judgment, but also arguably charged with meaning, with conceptual content, and thus full of implications.

Of course the conceptual content, the "beliefs", and the "implications" of perceptual content, would depend in part on the conceptual stance and character of the perceiver, as well as on what's "given" to perception by sensation. Particular conscious acts of inference on the basis of perception occur within this constant shifting context.

I suppose the hard problem of perceptual content, along these lines, is to clearly distinguish what's "given" to perception by sensation. — Cabbage Farmer

These are exactly the kinds of questions that Lewis and Price were wrestling with. For Lewis, the “given” was opposed primarily to the “concept”, and the hallmark of conceptuality was logical form. By implication, the given qua given has no logical form, it has no inferential implications and it does not constitute empirical knowledge. So Lewis’s epistemology is quite Kantian in nature. The mind applies concepts to the sensory given and it is the application of concepts that license inferences.

And yet, Lewis seems to recognize the tension that results from taking this kind of position. For what is the epistemic status of the given before concepts are applied, and what is the nature of the cognitive process by which concepts are applied to it? Is the given completely formless and ineffable, or does it exhibit some form of structure. If the latter, then how are we to understand the structure of the given if not in conceptual terms?

Here’s Lewis writing on the given in Chapter 2 of Mind and World Order

There is, in all experience, that element which we are aware that we do not create by thinking and cannot, in general, displace or alter. As a first approximation, we may designate it as "the sensuous."

[…]

At the moment, I have a fountain pen in my hand. When I so describe this item of my present experience, I make use of terms whose meaning I have learned. Correlatively I abstract this item from the total field of my present consciousness and relate it to what is not just now present in ways which I have learned and which reflect modes of action which I have acquired. […] what I refer to as "the given" in this experience is, in broad terms, qualitatively no different than it would be if I were an infant or an ignorant savage.

[…]

The distinction between this element of interpretation and the given is emphasized by the fact that the latter is what remains unaltered, no matter what our interests, no matter how we think or conceive.

[…]

While we can thus isolate the element of the given by these criteria of its unalterability and its character as sensuous feel or quality, we cannot describe any particular given as such , because in describing it, in whatever fashion, we qualify it by bringing it under some category or other, select from it, emphasize aspects of it, and relate it in particular and avoidable ways. — Lewis

The given is essentially being defined here as an “invariant” in experience – its structure does not change despite being emenable to multiple classifications dependent on the interests and background knowledge of the agent. The process by which concepts are applied is described as a process of “abstraction” and that’s where things start to get murky insofar as Lewis wants to claim that the structure of the given itself determines what classifications are or are not applicable in a given context:

I can apprehend this thing (given) as pen or rubber or cylinder, but I cannot, by taking thought, discover it as paper or soft or cubical. — Lewis

The underlying tension is becoming more apparent now. If the given is not conceptually structured, and if the application of concepts is solely the province of the agent, then how is it the case that the given can nonetheless constrain conceptual classification? How is the case that this non-conceptual given simply cannot be conceptually classified as "paper" or "soft" or "cubical"? How is that possible?

Reading through the chapter it becomes clear that Lewis wants the given to pull double-duty. He wants it to be non-conceptual and yet he wants it to have enough epistemic authority to act as a constraint on thought. He wants it to be the concrete basis of all experience, and yet also abstract enough to exhibit a repeatable structure.

And this is where (per Csalisbury) the Hegel connection comes into play. Sellars leverages aspects of Hegel’s dialectic in the Sense Certainty chapter to expose an ambiguity between the non-conceptuality of the act of sensation vs. the non-conceptuality of the content of sensation. Lewis elides the distinction by treating the given as the concrete correlate of direct apprehension while yet investing it with enough epistemic authority to act as a constraint on conceptual thought.

Sellars, like Hegel, essentially argues that insofar as the structure of the sensory “given” is determinate enough to warrant some classifications (“pen”, “cylinder”, etc.) but not others (“paper”, “soft”, etc.), it must be considered to be conceptual in nature for the simple reason that classifications have inferential implications. For instance, to say that some aspect of the given simply cannot be classified as “soft” implies that claims like “this object is soft” cannot be true and, by extension, that various other claims implied by that claim cannot be true (and so on).

So returning to your original question, inference plays an ambiguous role in Lewis’s epistemology insofar as the epistemic status of given is ambiguous. Does the “abstraction” process count as a form of inference? It almost seems like it has to insofar as it is a process by which certain classifications are determined to be applicable and others are not. But how can inference occur in the absence of concepts? It can’t, which seems to imply that the “given” is conceptually structured after all (or that there is no such thing as the given after all)

As a side note, some of the details of Price’s epistemology differ from Lewis’s, but many of the same of questions arise with regard to it.

Chipping away...

Would anyone care to say something about “anstoss logic”? What does this phrase mean? — Cabbage Farmer

The anstoss concept goes back to Fichte. The self posits the "not-self" in order to posit the "self". The anstoss is the spontaneous impulse that moves the self toward such a posit. The underlying logic is a variation on transcendental reasoning in general, where something is demonstrated to be a pre-condition for something else.

Similar reasoning undergirds Sellars's analysis of the looks/is distinction in EPM:

The point I wish to stress at this time, however, is that the concept of
looking green, the ability to recognize that something looks green,
presupposes the concept of being green, and that the latter concept involves
the ability to tell what colors objects have by looking at them -- which, in
turn, involves knowing in what circumstances to place an object if one
wishes to ascertain its color by looking at it. — Sellars

Understanding what it means to say that something merely looks green requires as a pre-condition understanding what it is to say that something really is green, or so Sellars argues.

Are such revisions in keeping with Price? Or would he deem the “seemings” redundant or inappropriate here, perhaps in keeping with a view that “the presence of a round red patch” is not the sort of thing that “seems”, and is instead in every respect immune to doubt? — Cabbage Farmer

I don't know that he addresses this specific question. Unfortunately, I no longer have access to Price's book and my notes don't mention anything about this.

There’s something intuitively plausible about this, once we massage it the right way. I don’t ordinarily justify my belief that I am seemingly seeing a seemingly red seeming object on any other grounds than that it appears so to me in the ordinary first-person point of view; ordinarily there are no other grounds available for such claims.

However, putting it the way Price has here, one might ask him, what justifies your belief that there are such things as “patches”? And it may turn out that this is only a loose manner of speaking. — Cabbage Farmer

Right, and that's essentially what Sellars is asking: what justifies your belief that there are such things as round red patches? Sellars doesn't believe that such patches exist, and EPM is essentially a critical examination of the notion that such things do exist, or perhaps more accurately, that such things are seen.

At the end of EPM he rehearses the myth of genius Jones, a fictional ancestor who explains perceptual mistakes on the model of "inner replicas" of physical objects. The main difference from sense-datum theory is that these replicas are understood to be "states" that are had by the observer rather than as particulars that are observed by the observer. In other words, there is nothing that is literally red and round in the world that is the object of observation when someone has an hallucination of an apple, but rather the observer comes to have an internal state that is "somehow analogous" to a red round patch.

The "somehow" is never explained in EPM, and Sellars actually ends up backing away from the notion that sense-impressions are internal states of the observer when he explicates his theory of absolute processes in the much later Carrus Lectures.

There’s no way I can keep up with you Cabbage Farmer, but here’s a start at some answers to your questions: — Aaron R

I can't keep up with my own ramblings, or with anyone else's for that matter. If we keep putting one foot in front of the other, the conversation takes care of itself.

I’d say that C.I. Lewis and H.H. Price figured very prominently, though the others mentioned are not far behind. As an aside, it’s interesting to consider that Sellars studied under C.I. Lewis while completing graduate studies Harvard. — Aaron R

That's good to know, I might have gone straight to Ayer.

I cannot admit that impressions have the character of observable facts. What I observe are things, not impressions. I see tables, and houses, and thermometers, and trees, and men, and the sun, and many other things in the sphere of crude physical objects; but I have never seen my impression of these things […] I do not say I doubt the existence of my impressions. I believe that there are impressions, but I have never sensed them. When I consider the question in an unprejudiced manner I find that I infer the existence of my impressions. — Reichenbach

Perhaps an important distinction lies hidden here. Recalling Sellars' distinction between "sensing particulars" and "sensing facts", I might deny the claim that I can see my own impressions, but affirm the claim that I observe (the fact) that I have current impressions. Accordingly, we might say I observe the impressions, or say at least that I make observational judgments about the impressions, without "sensing" them.

When I look up at the sky, is it the sun I see, or only a bit of its light? I like Brandom along these lines, on the observation of mu mesons, or muons, along with their traces.

Reichenbach's general belief that there are such things as impressions is part of a theory, or model, or discourse about perceptual experience -- a logos he picked up at second hand before kicking the tires himself. Say he confirms the theory to his own satisfaction, and thus acquires the general belief, by way of some inference, and thus ceases to "doubt the existence" of impressions in general, wherever there is perception like ours. It's the theory that positions him to make the inference, on any particular occasion:

If I am sensing right now, then there are impressions in me; and
I am sensing right now;

Therefore, there are impressions in me.

In fact, he doesn't have to bother making this inference on each separate occasion, once he's acquired the relevant general belief; much as I don't have to infer that gravity draws a glass to the floor when I drop it, or that the bright spot in the sky is a massive ball of gas. Inference may play a role in the initial formation of such beliefs, and in the formation of the concepts associated with the beliefs; but once the story is told, most of us who take it on, do so without bothering over what inferences and evidence informed the story in the beginning.

So long as the general belief is correct enough in the relevant ways, and so long as I have applied it correctly enough on a particular occasion, the observational judgment I make in terms of or in keeping with the belief "inherits" the correctness of the general belief -- of the story, and of whatever chain of inferences and observations may have produced the story, whether they were inferences and observations in my head or in heads before mine.

To explain this difference, I introduce the distinction between the physical thing and my impression of that thing. — Reichenbach

We shouldn't suppose this mere distinction informs us about what sort of thing the "impression" is. For instance, we needn't suppose that, whenever there is an impression, there is a one-to-one correspondence between the impression and a physical thing of which it is an impression.

I say that usually there are both physical things and impressions within me but that sometimes there are impressions only […] — Reichenbach

I suppose he doesn't mean: Usually there are physical things within me and impressions within me, but sometimes impressions only.

So I suppose he means something like: Usually the impressions within me correspond (in some relevant way) to physical things (that are not or need not be within me), but sometimes the impressions within me do not correspond to any physical thing (in the relevant way).

In the latter case, this might be a formulation of a distinction similar to the distinction I draw between perception and imagination: the difference between hearing a melody and repeating the same melody in your head; between hearing a sentence uttered and repeating the same sentence in your head; between seeing an apple and hallucinating or dreaming an apple.

One might argue that such imaginings, even at their most "vivid and forceful", hardly deserve the name "impressions". On the other hand, the name seems as suitable for perceptual illusions and other misperceptions as it does for ordinary perceptions.

we need this assumption to explain that in the case of the confused world one of the two worlds, the external world, is dropped. The distinction between the world of things and the world of impressions or representation is therefore the result of epistemological reflection. — Reichenbach

It seems to me we need something like a distinction between sense-perception and imagination just to make sense of the fact that we dream, hallucinate, and imagine -- though how we conceptualize such phenomena, and what judgments we make in light of them, seems to vary from one person and one cultural context to another.

I have the impression that 20th-century analytic philosophers, especially in the shadow of the spooky behaviorist tendency we might trace through Ryle and Quine, tended to neglect such phenomena, and to perhaps quietly lump false judgments made on the basis of imaginings into the same account as false judgments made on the basis of misperceptions. In any case, it's often hard to tell how to map a term like Reichenbach's "impression" here onto a distinction like the one I draw between perceiving and imagining.

Perhaps the neatest way is to say that imaginings are among our impressions. If I mistake my hallucination for perception, and judge that there's an apple on the table on the basis of that hallucination, the source of the confusion is that I have mistaken hallucinating for perceiving, or in other words, I have incorrectly taken an instance of hallucinating as an instance of perceiving. I may correct my own error when I move to the table and aim to fetch the apple; discover that nothing's there, though it still looks like there's an apple; and now correctly judge that I am and have been hallucinating, despite the persistent vision-like appearance of an apple on the table.

The hallucinating is part of the world, just like dreams and less vivid and forceful imaginings are part of the world. Until we have adequate understanding of these phenomena, we might be confused by them and led to make false judgments in light of them; but as we develop adequate understanding, we're more likely to avoid error. In this respect, our concept of hallucination is just like other empirical concepts.

If we're willing to say that each perceiving and each imagining grasped in introspection is correlated in a relevant way with a bit of neural activity in the world; then we should be willing to say that, once we're equipped with the right sort of concepts and general beliefs, we make more or less reliable observational judgments on the basis of introspection about those bits of neural activity, and that these, too, are things in the world of which we have "impressions" whenever we're aware of our own perceiving or imagining, no matter how otherwise ignorant or confused we may be about the relevant facts; much as we make reliable observational judgments about the sun on the basis of visual perception, no matter whether we think it's a god in a chariot or a massive ball of gas.

Along those lines, we might say "the confused world" is the one in which I judge incorrectly, on the basis of current hallucination, that I am perceiving an apple and that this apple is on the table. And we might say the harmoniously conceived, or well sorted, world is the one in which I judge correctly, on the basis of current hallucination, that I am hallucinating and that there is no apple on the table.

Reichenbach, however, seems perhaps to speak as though the confusion is somehow or other contained in the impression, and now he slides from "impression" to "representation". But what sort of thing is the impression supposed to be, if the error in judgment is already contained in it? I can stare at a mirage for hours, and change my mind a thousand times while it remains the same mirage. It looks the same, the world appears to me in the same way, while I cycle through various judgments on the basis of that one appearance. The appearance doesn't tell me how to judge, and I don't tell it how to appear. The judging is up to me, and the appearing is up to the appearance, though I can turn my head and get past it.

Along these lines, I make a three-part epistemological distinction: In the first place there is the fact of the matter, the way things are in the world in fact. In the second place there is the appearance, the fact of how things appear to me (e.g., how things "look to me"). In the third place there is the seeming, the fact of how I take things in fact to be, in part on the basis of appearances. In making active judgments about how things seem, in part on the basis of appearances, I change the seeming, but not the appearances. Normally I don't need to make such judgments in order for things to seem to me one way or another; ordinarily I resort to such judging only when salient features of my experience seem uncertain, confused, or otherwise inadequate to form the basis of a reasonable judgment I have some interest in making. The judgment amends or completes or suspends the seeming.

If I'm going to use the word "impression" in telling that story, the impression belongs with the appearance, not with the seeming.

Note that both the appearance and the seeming are parts of the world, are matters of fact.

He goes on to talk about “the abstract character of impressions” as part and parcel of a “duplicity theory” of perception. And this is where, as seen through the lens of EPM, Reichenbach’s account starts to go off the rails. We can already see in the quotes above the seeds of an ambiguity between impressions qua objects and impressions qua facts. Reichenbach says that he doesn’t “see” his impressions and also says that impressions are “not observable facts”. And yet he’s invoking impressions in order to explain discrepancies in what is seen. — Aaron R

Based on your discussion and the passages you've cited, his discourse strikes me as rather confused, and it's perhaps exceptionally hard to tell what sort of thing he supposes an impression to be.

Things get really bad in chapter 3 where Reichenbach seems to firmly place impressions into the category of “immediate existence”: — Aaron R

Imagine we are taking a walk at dusk through a lonely moor; we see before us at some distance a man in the road […] we do not doubt the man’s reality. We walk farther and discover it is not a man but a juniper bush. […] We shall say that both the man and the bush have immediate existence at the moments we see them. — "Reichenbach

So here we seem to have the paradigm case of an impression. We think we see a man, but it’s really just juniper bush so we invoke the concept of an impression to explain the mistake. The man is inferred to be an impression, but here we are told that the “man” also has immediate existence. So what’s immediate existence? — Aaron R

That's the most pressing question here, for sure.

On the side: Doesn't Reichenbach call the juniper bush an "impression", as well as the man? I've interpreted the passages you've cited this way: It's not that, having fallen short of the mark, Reichenbach infers that the man was an impression, while persisting in the belief that the bush is a physical object but not an impression. Rather, in seeking to understand such cases, to analyze the ways in which we fall short of the mark and wind up in confusion, Reichenbach infers that he has impressions within him, both when he sees correctly and when he sees confusedly. In happy cases there is both physical thing and impression; in unhappy cases, impression alone; and in this one case he judges that his first impression was an impression without a corresponding physical thing.

Of course, in the case at hand, it’s not accurate to say there’s no physical object corresponding in a relevant way to the confused perception of a man. It’s the same physical thing, the bush, corresponding to perception and judgment in the happy case and in the breakdown case. (It’s not clear in the example whether the breakdown is a misperception or a mistaken judgment on the basis of perception, or somehow both).

I’m not sure how Reichenbach intends to sort it all out.

We may regard immediate existence as a concept known to everybody. If somebody does not understand us, we put him into a certain situation and pronounce the term, thus accustoming him to the term and the situation seen by him. […] If a child asks “what is a knife” we take a knife and show it to the child. — "Reichenbach

This strikes me as perhaps the most uninformative sort of definition conceivable. I don't mean the definition of the knife in the example, which is an ordinary ostensive definition. Rather, this attempt to characterize "immediate existence".

Is there any other clue from Reichenbach? Is there supposed to be an analogous situation into which one may be put, where he will observe the meaning of "immediate existence"? Are we supposed to grasp the sense of this term in the context of misperceptions and mistaken judgments on the basis of perception, like the mistaking of bush for man?

It might seem less strange if he'd said "immediate experience", as in: At first I see what appears to be a man, but soon I see that it's a bush.

But what could it mean to say "both the man and the bush have immediate existence at the moments we see them"?

Maybe something like this: At the moment we mistake the bush for a man, we have an idea of a man there whom we're seeing, and the idea of existence belongs to that idea of a man.

Even so, it seems tortured use, to call this "immediate existence", instead of an immediate idea of existence -- i.e., an idea of existence (and an idea of an existing man) that has not yet been mediated and overruled by a sound view of the facts. Perhaps this strained usage is another symptom of the same confusion we've noted in Reichenbach, who seems to smear altogether the seeming, the appearing, and the facts and objects that seem and appear.

In other words, immediate existence is the existence of whatever is directly apprehended or, if you will, known by acquaintance. But wait…this directly contradicts Reichenbach’s earlier claim that impressions are indirect, theoretical entities. These are exactly the kinds of confusions that Sellars was responding to in EPM. — Aaron R

The view seems entirely confused, at least as we've reviewed it here. I hope the sense-datum theorists turn out to have a better champion. For it would seem shameful for a whole generation of experts to be taken in by such a jumble.

My turn to take a break. I look forward to working through the rest of your recent comments.

Right. Both Lewis and Price (like Reichenbach) tend to elide the distinction between sensing particulars and sensing facts. — Aaron R

This distinction keeps rearing its head in our conversation.

These are exactly the kinds of questions that Lewis and Price were wrestling with. For Lewis, the "given" was opposed primarily to the "concept", and the hallmark of conceptuality was logical form. By implication, the given qua given has no logical form, it has no inferential implications and it does not constitute empirical knowledge. — Aaron R

Does Lewis locate the given within experience, or prior to experience? How should we coordinate the terms "given", "sense-datum", and "direct apprehension" with respect to his account?

I'd say nothing is given without a perceiver. A perceiver already has a stock of conceptual capacities, expectations, general beliefs... with inferential structure and "logical form". The given qua sensation-dervied constraint has no fixed logical form, but in each particular context of perceptual experience, it does have conceptual and inferential implications, and it does open the way to empirical knowledge.

So Lewis’s epistemology is quite Kantian in nature. The mind applies concepts to the sensory given and it is the application of concepts that license inferences. — Aaron R

If it makes sense to say "the mind applies concepts to the sensory given", it makes sense to say "the mind applies the sensory given to concepts". The standing repertoire of conceptual capacities and the current intake of sensation are applied together in perceptual experience on each occasion.

Without that cooperation of sensation and conceptualization, there would be no license for the inferences we draw on the basis of experience. We would make no such inferences.

And yet, Lewis seems to recognize the tension that results from taking this kind of position. For what is the epistemic status of the given before concepts are applied, and what is the nature of the cognitive process by which concepts are applied to it? — Aaron R

"Before" concepts are applied, there is no experience, and no appearance, but only, we say, sensory and perceptual processes in preconscious cognition. It seems that the contribution from sensation is determined prior to its manifestation in experience, for instance to reflect impingements of the world on sense receptors; and that sensory and perceptual processes in preconscious cognition are normally more or less insulated from direct interference by conceptually driven processes, so that what appears in experience is radically determined by sensation.

The contribution of sensation to experience seems organized, in the first place, according to a "deep structure" in keeping with principles we associate with our concepts of space and time, light and dark, hot and cold, loud and quiet, and so on, across a range of sensory modalities, which provide a fundamental framework for perceptual experience on each occasion and for empirical knowledge over time. On each occasion, the "synthesis of the manifold of sensation" according to these fixed basic principles provides an integrated perceptual context within which individual phenomena are spontaneously differentiated. That differentiation is in the first place the result of automatic and involuntary cognitive processes, not of any conscious act of judgment. In being thus differentiated within the organized perceptual field, phenomena are conceptualized, and this conceptualization is likewise, in the first place, automatic and involuntary.

Perceptual experience thus constituted -- as spontaneous and involuntary conceptualization of a predetermined contribution from sensation -- provides the context in which spontaneous and voluntary thoughts and judgments may review appearances, to consider, for instance, whether things will continue to appear, upon further inspection, as they now appear, or whether things are in fact as they seem, to all appearances, to be.

To all appearances, it seems neither sort of spontaneous conceptual activity has the power to reconstitute the predetermined contribution of sensation. The fundamental epistemic role of that contribution is characterizable in terms of appearance-talk: I know it's a juniper bush, but in the shadows, from here, it looks just like a man.

Appearances persist -- the look and sound of things persist -- despite shifts in our conceptual attitudes toward particular appearances, despite exercises of our capacity to conceptually reframe a whole perceptual occasion, despite wholesale reorganization of our thoughts about appearances in general.

Is the given completely formless and ineffable, or does it exhibit some form of structure. If the latter, then how are we to understand the structure of the given if not in conceptual terms? — Aaron R

Empirical science continues to develop its account of sensation and sensory perception, and of all cognition, in light of ongoing empirical investigation. That account is one way for us to characterize the contribution of sensation to perceptual experience, or the "structure of the given" -- perhaps already the most informative way, despite the primitive status of the science.

All empirical science is a rigorous extension of ordinary empirical knowledge, and has an essentially phenomenological character. The most sophisticated scientific accounts of perception are continuations of investigations each of us makes by noticing differences in the way things appear when we squint, or close our eyes, or turn out the lights.

A phenomenological account that relies on the unaided introspection of each interlocutor is said to run into deep trouble, for it’s impossible to describe anything -- including preconceptually determined features of one’s own experience -- without employing concepts.

I want to say we blow this difficulty out of proportion in such discourses, by taking for granted an unwarranted and incoherent separation of empirical investigation and phenomenology, and by conflating speculative "metaphysical" questions about the whole world as it appears to us, with phenomenological and empirical questions about animal perception.

Here’s Lewis writing on the given in Chapter 2 of Mind and World Order — Aaron R

These passages from Lewis make a good impression, and seem less overburdened than those you’ve supplied from Reichenbach and Price.

There is, in all experience, that element which we are aware that we do not create by thinking and cannot, in general, displace or alter. As a first approximation, we may designate it as "the sensuous." — Lewis

We don’t say we "create" our limbs when we use them. Do we ordinarily say we "create" the voluntary motions of our limbs? We produce, perform, discharge… voluntary motions of our limbs by moving them, and only some "elements of experience" by thinking. Likewise, we produce sounds by moving around. When "create" is appropriate here, it’s as another way to characterize such production.

The crucial difference Lewis gestures at aligns with the old Stoic discourse about what is and what is not "up to us" or "in our power". Epictetus speaks as if we have as much responsibility for our emotions and desires as for our thoughts, but less for our bodies, which can be altered or restrained without our consent. Lewis speaks here as if we "create" our thoughts and their content from whole cloth without any hindrance, but are constrained in our encounters with "the sensuous". We might refine the truth in such views by distinguishing various classes of perceptual modality (or basis for noninferential judgment)

exteroception -- proprioception -- interoception -- affect -- imagination -- thought

and accounting for the various constraints on spontaneous activity associated with each of the "modes" or "bases" thus distinguished.

At the moment, I have a fountain pen in my hand. When I so describe this item of my present experience, I make use of terms whose meaning I have learned. Correlatively I abstract this item from the total field of my present consciousness and relate it to what is not just now present in ways which I have learned and which reflect modes of action which I have acquired. […] what I refer to as "the given" in this experience is, in broad terms, qualitatively no different than it would be if I were an infant or an ignorant savage. — Lewis

I suppose we might substitute "I abstract this item from the total field of my present consciousness..." for "I bring a whole field of conceptual relations to bear on this one item in my present field of perceptual experience." The latter seems more apt for many cases; the former perhaps for cases in which we stop paying attention to the perceptual field, and become lost in thoughts about fountain pens, or about this fountain pen, or about what in this vision of a fountain pen is "given".

The passage leans heavily on the phrase "in broad terms". The trouble is to make those terms explicit and perhaps to narrow them down. There’s also an implication, perhaps along pragmatic lines, that the whole conceptual field should be analyzed primarily as reflecting "modes of action I have acquired".

The distinction between this element of interpretation and the given is emphasized by the fact that the latter is what remains unaltered, no matter what our interests, no matter how we think or conceive. — Lewis

The crucial difference.

While we can thus isolate the element of the given by these criteria of its unalterability and its character as sensuous feel or quality, we cannot describe any particular given as such , because in describing it, in whatever fashion, we qualify it by bringing it under some category or other, select from it, emphasize aspects of it, and relate it in particular and avoidable ways. — Lewis

The notorious difficulty.

The given is essentially being defined here as an "invariant" in experience – its structure does not change despite being emenable to multiple classifications dependent on the interests and background knowledge of the agent. The process by which concepts are applied is described as a process of "abstraction" and that’s where things start to get murky insofar as Lewis wants to claim that the structure of the given itself determines what classifications are or are not applicable in a given context: — Aaron R

I can apprehend this thing (given) as pen or rubber or cylinder, but I cannot, by taking thought, discover it as paper or soft or cubical. — Lewis

The thought must be that what's "abstracted" -- the product of the abstraction, not the material it's abstracted from -- has a conceptual structure determined in part by the sensuous given. At each moment, there is a fixed structure to the given, and a fixed structure to my concepts, and merely by focusing on any element in the perceptual field, I spontaneously "abstract", draw off, a conceptual structure determined by the given in the context of my conceptual frame. The mind applies the sensuous given to its concepts, as it applies its concepts to the sensuous given.

I suppose my account of "involuntary conceptualization of the predetermined contribution from sensation" is designed to perform a theoretical role similar to that of Lewis' "abstraction".

The underlying tension is becoming more apparent now. If the given is not conceptually structured, and if the application of concepts is solely the province of the agent, then how is it the case that the given can nonetheless constrain conceptual classification? How is the case that this non-conceptual given simply cannot be conceptually classified as "paper" or "soft" or "cubical"? How is that possible? — Aaron R

I assume "cannot" here has practical limits, and does not entail logical impossibility. Looking at this stone, and feeling it with my hands, I find that I cannot sincerely apply the predicate of softness to this stone. I acknowledge that my judgment may be flawed in principle, that my senses may be deceiving me, etc.; but to all appearances, the stone is not soft.

It seems no more puzzling to me that an experience of a stone should constrain the concepts we sincerely apply to the stone, than that the stone should constrain the concepts we sincerely apply to it.

Here is a stone, here is a haptic appearance of the stone, here is a concept associated with the English word "soft" that is ordinarily inapplicable in such contexts.

Concepts like these are naturally coordinated with experiences like these because they are produced and refined and applied in the context of experiences like these. There is something it's like to feel a hard thing; having an experience of that sort is what rules out the application of the concept "soft" to the thing that feels hard. Not as an eternal truth or an incorrigible assertion, not on the basis of certain knowledge about a stone, but in the ordinary way, provisionally and corrigibly, on the basis of current experience.

Reading through the chapter it becomes clear that Lewis wants the given to pull double-duty. He wants it to be non-conceptual and yet he wants it to have enough epistemic authority to act as a constraint on thought. He wants it to be the concrete basis of all experience, and yet also abstract enough to exhibit a repeatable structure. — Aaron R

To all appearances, there's repeating structure in i) the things themselves as they appear to us, the stones and pens that we perceive and otherwise use; ii) the physical contexts of sensation, including for instance light and air; iii) the body of the perceiver, including physiological processes of sensation and perception; and iv) the conceptual frame of each perceiver, which tends to be fairly stable over time, especially with respect to ordinary empirical concepts that compel us to recognize appearances of, for instance, stones and pens whether we want to or not.

To rule out an objective account of sensation -- a phenomenologically grounded empirical account of the physical and physiological context of perception -- is to lose touch with the preconceptual structure of the contribution of sensation to experience.

Approaching the problem this way, as a pseudoproblem, it seems an error to suppose that the preconceptualized contribution of sensation can be isolated or "directly apprehended" in experience, from an artificially circumscribed first-person point of view, in abstraction from correlate empirical accounts of the objects and context and processes of sensation, as if experience were a groundless picture-show instead of one part of a whole world. We understand sensation by understanding ourselves as natural beings, as perceiving animals in a physical world; or we don't understand sensation.

And this is where (per Csalisbury) the Hegel connection comes into play. Sellars leverages aspects of Hegel’s dialectic in the Sense Certainty chapter to expose an ambiguity between the non-conceptuality of the act of sensation vs. the non-conceptuality of the content of sensation. — Aaron R

What is this distinction between "act" and "content"?

Lewis elides the distinction by treating the given as the concrete correlate of direct apprehension while yet investing it with enough epistemic authority to act as a constraint on conceptual thought. — Aaron R

What does "concrete correlate of direct apprehension" mean?

Sellars, like Hegel, essentially argues that insofar as the structure of the sensory "given" is determinate enough to warrant some classifications ("pen", "cylinder", etc.) but not others ("paper", "soft", etc.), it must be considered to be conceptual in nature for the simple reason that classifications have inferential implications. — Aaron R

To me this seems like bending over backward. Where do the classifications come from? They develop in individuals and in culture on the basis of experience. It's the regularity of the world as it appears to us that produces concepts and that warrants the application of concepts in and on the basis of perception.

There's no reason to insist that the structure of a stone "must be conceptual in nature" in order to account for the fact that we apply concepts to stones or that the predicate "being a stone" has conceptual implications.

Likewise, there's no reason to insist that the structure of the given qua given -- the structure of the contribution to experience from sensation -- is "conceptual" or has "logical form", merely in order to account for the fact that we apply concepts to stones on the basis of perceptual experience of stones.

All we need is the notion that the relevant concepts are coordinated with the structure of sensation and with features of the objective world that appears to us through the mediation of sensation. The world has structure, sensation has structure, concepts have a "logical form" that tends to be more or less coordinated with that structure.

For instance, to say that some aspect of the given simply cannot be classified as "soft" implies that claims like "this object is soft" cannot be true and, by extension, that various other claims implied by that claim cannot be true (and so on). — Aaron R

I assume in this context "cannot" has practical limits, and doesn't entail logical impossibility. Looking at this stone, and feeling it with my hands, I find that I cannot sincerely apply the predicate of softness to this stone, though I acknowledge that this judgment may be flawed in principle, that my senses may be deceiving me, etc. To all appearances, the stone is not soft.

This does not imply that the stone is conceptual, or that the stone's hardness is conceptual. It only implies that I am correct to apply the concepts "stone" and "hard" to this thing.

The reason I am correct to call this thing a hard stone, is that it is a hard stone, a fact that I grasp one way or another in perceptual experience, like other intelligent animals, thanks to the faithful contribution of sensation; and that I grasp in one specific way when I call this thing a hard stone.

So returning to your original question, inference plays an ambiguous role in Lewis’s epistemology insofar as the epistemic status of given is ambiguous. Does the "abstraction" process count as a form of inference? It almost seems like it has to insofar as it is a process by which certain classifications are determined to be applicable and others are not. But how can inference occur in the absence of concepts? It can’t, which seems to imply that the "given" is conceptually structured after all (or that there is no such thing as the given after all) — Aaron R

I say it can't be an inference, on phenomenological grounds.

The thought may be that the contribution of sensation "triggers" automatic conceptualization in perceptual experience. In special cases, such provisional conceptualization seems inadequate for judgment or other action, leading us to take a closer look, to check whether things seem to be as they have appeared, to wonder whether things are as they seem to be. Until the automatic, tentative conceptualization is suspended or overruled by such conceptual activity, it has conceptual implications for the perceiver. The initial conceptualization and its implications, like the considered view achieved by careful judgment, are determined by the conceptual stance of the perceiver as well as by the contribution of sensation.

How does the automatic and provisional conceptualization take place? What are the rules by which it proceeds? It seems to me this is an empirical question that cannot be adequately answered on the basis of ordinary introspection, because this front line of conceptualization is not constituted by conscious activity, but rather constitutes the perceptual experience that is central to conscious awareness. Introspection reveals to us that perceptual experience is always already conceptualized, and that sense-perception is radically determined by a contribution from sensation; but introspection is insufficient to inform a fine-grained view of the individual mechanisms that produce such experience.

As a side note, some of the details of Price’s epistemology differ from Lewis’s, but many of the same of questions arise with regard to it. — Aaron R

Any remarkable differences between Price and Lewis?

So far I like Lewis best. I don't find the same muddle in these passages that we noted in Price and Reichenbach. Perhaps there are other passages that bring out similar problems in Lewis more clearly?

I'd say the "tension" you attribute to his account seems largely dissipated, at least with respect to these passages, if we're allowed to shift the frame of the conversation along the lines I've suggested: i) by rejecting an arbitrary and arguably incoherent separation of phenomenology and empirical investigation; ii) by developing an empirical account of perception (especially the contribution of sensation) that coordinates first-person introspective reports with third-person observations; and iii) by outsourcing conversations about "what really is", aside from the world as it appears to us, to a more abstract and general region of discourse.

So far as I can tell, that shift in frame is enough to open the way for us to speak of concepts as naturally coordinated with perceptual experiences and perceptual objects, as produced and applied on the basis of experience.

To all appearances, that natural cooperation is the original context of conceptual activity, which is constrained to reflect the objective environment of an animal perceiver through the mediation of sensation, and thus to coordinate the activity of an organism with salient features of its environment. Human language and culture render us more sophisticated conceptualizers, and open for us an infinitely greater range of action, knowledge, fantasy, and confusion. There's no end to the variety of conceptual stances we may take on the basis of perceptual experience. But that experience, and any coherent conceptualization of it, remains radically constrained by the preconceptual structure of the contribution from sensation, no less for us than for nonhuman animal perceivers.

Short of a full-fledged empirical account of the preconceptual structure of sensation for each animal perceiver, we may locate the fundamental constraint, the essential epistemic role, of perceptual experience by tracing the limits of appearance-talk.

The anstoss concept goes back to Fichte. The self posits the "not-self" in order to posit the "self". The anstoss is the spontaneous impulse that moves the self toward such a posit. The underlying logic is a variation on transcendental reasoning in general, where something is demonstrated to be a pre-condition for something else. — Aaron R

Does transcendental reasoning demonstrate that it must be an "impulse" that posits not-self and self? And how do the relevant transcendental arguments define "impulse"?

It seems there's a basic self-awareness that emerges in nature along with sentient animals. We might expect, along the lines of the Kantian's "unity of apperception", that there is a reflexive aspect to all consciousness, that animal consciousness involves reflexive awareness, awareness of oneself.

I see no reason to suppose there is a special impulse for such awareness, or that some special act of positing is required. It seems rather part of the form of consciousness, of minds like ours, to distinguish between ourselves and things outside ourselves. Our minds are organized in keeping with the natural organization of our bodies. Every impulse we have arises in the context of this organization, but I'm not sure what theoretical need there is to posit a special impulse of not-self positing.

Similar reasoning undergirds Sellars's analysis of the looks/is distinction in EPM: — Aaron R

I've been wondering when we'd get back to Sellars. I've been enjoying the preliminary survey of sense-datum theorists in the meantime. I'd say it's a precondition for understanding Sellars' essay.

The point I wish to stress at this time, however, is that the concept of looking green, the ability to recognize that something looks green, presupposes the concept of being green, and that the latter concept involves the ability to tell what colors objects have by looking at them -- which, in turn, involves knowing in what circumstances to place an object if one wishes to ascertain its color by looking at it. — Sellars

There's a fine circle.

Does Sellars mean to suggest that the concept of looking green is identical with the ability to recognize that something looks green, while the concept of being green merely involves the ability to tell what color objects have by looking?

"The concept of looking green, [which is to say,] the ability to recognize that something looks green, presupposes the concept of being green, and that latter concept involves the ability to tell what colors objects have by looking at them...."

In any case, it's an interesting alignment of the terms "ability" and "concept".

Understanding what it means to say that something merely looks green requires as a pre-condition understanding what it is to say that something really is green, or so Sellars argues. — Aaron R

I'd like to hear more of this argument.

For starters, I'd want to clear up the various sorts of "concept" we might distinguish in this connection. I might have a more or less refined ability to tell what colors objects have by looking at them, without having thought about it much, without having any sophisticated theories about color and light, without having acquired a relevant repertoire of "looks" talk. I'm happy to equate that ability with a "concept", but this simple recognitional capacity leaves plenty of room for variation in one's other concepts associated with judgments of color.

Does Sellars say that this "ability" -- the ability to tell what colors objects have by looking at them -- is the precondition of the concept of "being green" (and I suppose of "having color"), and that the concept of "being green" is the precondition of the concept of merely "looking green"?

The ability at the beginning of this chain is not the same as either concept. For in the original ability, there is no distinction between "looking green" and "being green". We might say that "looking green" and "being green" are the same concept in the early stages of color-concept acquisition, or perhaps that to begin with there is no concept of "[merely] looking green", and no correlate concept of "being [truly] green."

These concepts are distinguished from each other as we learn, on the basis of experience, that there's more at play in making correct judgments of color than seemed to be at play at first glance. One way of refining usage and judgment to reflect this new insight into the nature of things is to distinguish between the color things "[truly] have" (in ordinary circumstances) and the color things merely "appear to have" (in extraordinary circumstances); though I'm not sure this is the best way to sort out the facts about color.

It's only after one loses innocence about color judgments that the concept of "being [truly] green" is associated with sophisticated techniques like those Sellars mentions: techniques that apply hard-earned knowledge of "what circumstances to place an object if one wishes to ascertain its [true] color by looking at it."

The concept of "being [truly] green" that's associated with such sophisticated knowledge and techniques is not the original concept of "being green" that is a precondition for the distinction between "[merely] looking green" and "being [truly] green". The precondition is the simple concept of "being green", prior to the loss of innocence that motivates the distinction between "[merely] looking green" and "being [truly] green".

That simple concept does not depend on the ability to tell what color things "truly have" by looking at them -- for that's a more sophisticated concept that requires knowledge of the way that apparent color varies along with circumstances. The original, unrefined concept depends merely on the ability to innocently tell the (apparent) color of things by looking at them.


Having sorted out some of the relevant concepts along these lines, I'll agree that the ability to innocently report the colors things (appear to) have is the precondition for an unrefined concept of things "having color" or "being colored"; and that both this ability and the corresponding concept are refined when we lose innocence and begin to acquire knowledge about the ways that the apparent colors of things vary along with circumstances.

Perhaps enduring philosophical confusion about color judgments makes this example ill-suited to an epistemological discussion about appearances. Why muddy such murky waters?

Right, and that's essentially what Sellars is asking: what justifies your belief that there are such things as round red patches? Sellars doesn't believe that such patches exist, and EPM is essentially a critical examination of the notion that such things do exist, or perhaps more accurately, that such things are seen. — Aaron R

What could justify a belief in such "patches"?

Does he leave room for the sense-datum theorist to posit such things as "patches" as theoretical constructs that are not "seen" or otherwise "sensed"? In that case, we might expect that some ways of talking about "patches" or other "sense-data" would be harmless enough, or even consistent with other useful ways of speaking about the contribution of sensation to perceptual experience -- for instance, perhaps along the lines of the cognitive scientist's "information" or "representation".

At the end of EPM he rehearses the myth of genius Jones, a fictional ancestor who explains perceptual mistakes on the model of "inner replicas" of physical objects. The main difference from sense-datum theory is that these replicas are understood to be "states" that are had by the observer rather than as particulars that are observed by the observer. — Aaron R

This seems like the sort of modelling I had in mind just now.

In other words, there is nothing that is literally red and round in the world that is the object of observation when someone has an hallucination of an apple, but rather the observer comes to have an internal state that is "somehow analogous" to a red round patch. — Aaron R

I think the analogy is best made between the visual perception of an apple and the vision-like hallucination of an apple. The similarity is between the two "states", not between the real apple and its imaginary correlate.

The "somehow" is never explained in EPM, and Sellars actually ends up backing away from the notion that sense-impressions are internal states of the observer when he explicates his theory of absolute processes in the much later Carrus Lectures — Aaron R

Is his view in EPM that it makes sense to talk about sense-impressions as internal states of the observer? What concept of "sense-impression" does he employ in this connection?

And what's the change of direction with "absolute processes" in the Carrus Lectures?

Hi @Cabbage Farmer. You're putting me to shame here with the sheer volume of your replies. Sorry I'm not keeping up, but I'll go ahead pick up where I left off:

That doesn’t make it nonsense to speak about the epistemic role of that deep layer of belief based on appearances, which serves as a safety net for us to fall back on gracefully, precisely because it persists while we ignore it. — Cabbage Farmer

Sure, and I don’t think Sellars would say that it’s nonsense to talk about the epistemic role of such claims - in fact, that is exactly what he is interested in talking about!

It’s a withdrawal from claims about how things really are over there, where there seems to be a red apple. But this doesn’t entail another withdrawal, from all claims about how things really are in the world: For this appearance, and this seeming, may be said to be part of the world. And it’s not clear that it may be coherently denied that the appearance and seeming are part of the world. — Cabbage Farmer

Yes, and Sellars acknowledges that in making such claims we are still stating facts about the world – namely, facts of the form “x looks F to S”, etc. Sellars doesn’t deny that “appearances” exist per se, but he doesn’t think that sense-data exist. That is, he doesn’t think that appearances are little colored-and-shaped particulars nor that reality is fundamentally constructed out of such things.

Accordingly, the appearance and the seeming are among the “facts” we may aim to piece together in each case. For instance by asking: How is it the case, how does it happen, that things appear thus and so (to me, now)? — Cabbage Farmer

Sure, but we have to be careful in referring to appearances as “facts”, because the term is ambiguous.

By “fact”, we can mean something that has propositional form (e.g. “x is F”) or we can mean some kind of particular process or object in the world (e.g. trees, rocks, etc.). For Sellars, propositions are essentially functional roles in linguistic practice. In other words, whenever someone utters a claim, the abstract meaning of that claim is determined by the functional role that the claim plays within the “game of giving and asking for reasons”. In contrast, particulars such as trees and rocks are not essentially functional roles within our language games. Particular trees and rocks can certainly have roles within our language games, but neither their essence nor existence is dependent on those roles.

And that’s why Sellars harps on properly making a distinction between sensing particulars and sensing facts. Sellars maintains that we sense particulars, but perceive facts, and he thinks that sense-datum theories run into difficulties in part because they don’t properly distinguish between the two. Falling into the myth of the given is thinking that there is something that is sensed directly that is both an essentially extra-linguistic particular and an essentially linguistic entity such as a proposition. In Sellars book, this is illegitimate insofar as it posits something that is an instance of two incompatible types.

In ordinary, happy cases, one fair answer is: Because there’s an apple there, and your eyes are open and aimed that way. The frequency with which seeming-seeings of seeming-apples turn out to be, to all appearances and “to a practical certainty”, genuine seeings of genuine apples, also serves as justification for the belief that “There is an apple there”, and lets the seeming-seeing of a seeming-apple stand as a noninferentially acquired defeasible warrant for the latter belief.

A more developed variation on this theme could haul into the account an empirically grounded story about light and the light-relative properties of physical objects; about retinas and cones and optic nerves; about perceptual processes in cognition.

None of this amounts to absolute “theoretical certainty” that there really is a real apple there -- a certainty we never attain, even as we bite and chew and swallow. — Cabbage Farmer

So, my take on this is that Sellars wouldn’t deny that “looks” claims can provide evidence for “is” claims under the right circumstances. If 99.99% of the time that Jones utters claims of the form “x looks F” it turns out that the corresponding “x is F” claim is also true, then there’s a sense is which Jones's uttering “x looks F” can be considered as evidence for the claim “x is F”, ceteris paribus. But even in this case, the evidence provided by Jones’s “looks” claims is parasitic on the evidence provided by other more fundamental “is” claims. They only provide evidence insofar as Jones really is 99.99% accurate. And that’s really the point that I take Sellars to be making here - there’s no layer of “looks” claims that can serve as an independent evidential foundation (whether certain or probable) for all of our “is” claims.

Continuing on…

What shall we count as “adequate” evidential basis for claims about how things really are? I would argue along lines just indicated above, that the seeming-seeing of a seeming-apple does provide adequate evidential basis for claims about real apples, though not for claims that such claims are certain claims. In cases in which new evidence comes to light, that the seeming-apple was not in fact an apple, or that the seeming-seeing was not a seeing, we amend the record of discourse by adding new alleged facts (including claims about the seeming course of seemings) and revise or withdraw from our previous claim about what’s over there in the world where (it seems) it had seemed there was a seeming-apple. — Cabbage Farmer

Again, the evidence provided by “looks” claims seems to be parasitic upon more fundamental “is” claims. Jones’s “looks” claim is counted as evidence only because Jones’s reliability has already been established. You couldn’t establish that reliability on the basis of other “looks” claims alone because "reliable" in this case just means "an agreement between how things look and how things are".

Let’s not forget the role of expressions like “I thought I knew”. I thought there was an apple there, but it turned out to be a lump of wax. That doesn’t mean the initial claim was unjustified, unwarranted, and groundless -- only that it turned out to be incorrect; or rather that the speaker turned out to have reason to correct it. This new reason, or new judgment, is as fallible in principle as the first which it amends. But it seems such claims are correct when they happen to be so, and stand uncorrected until there’s reason for correction. — Cabbage Farmer

Yep, agreed.

Who’s doing the “moving” here, and for what purpose? Arguably neither route will get us more “certainty” in the claims we make about what’s “over there”. It’s true that ordinary speech for ordinary purposes tends to follow the more direct route, speaking about “what’s over there” without wasting time on appearance-talk except in special cases, as when one is prompted by events to revise his own considered view about “what he thought he knew.” — Cabbage Farmer

The rational agent is doing the “moving”, and for the purpose of avoiding full accountability for the “is” claim.

The efficiency of that fine custom of, as it were, directly addressing things in the world as they appear to us, and revising only when it matters -- instead of constantly referring to the mediations of experience and conceivable doubts of reason -- gives us no reason to suppose those mediations vanish whenever there’s harmony between the seeming and the fact. It seems rather that when they appear in agreement, ordinarily we focus on the main track, according to our purpose, only falling back on the other, and to the task of realignment, when they fall out of whack.


I don’t mean to imply, by speaking this way, that the “appearance” is something like a color patch in my head, ontologically isolable from the “causal chain” that is -- I want to say -- identical to that appearance.

We may analyze a causal chain into parts, or happen upon one link at a time, and then in ignorant conjecture or counterfactual hypothesis, characterize that one link as part of indefinitely many conceivable causal chains. Some such stories seem to line up with the facts better than others. Yet others seem entirely off base.

And the facts keep coming in, or so it seems. — Cabbage Farmer

I don’t think that Sellars would say that the mediations of experience and reason vanish when there is a harmony between seeming and fact. Sellars actually defended a version of the correspondence theory of truth based on something that he called a “picturing” relation, which he understood to be something like a homomorphism between physical events within the agent and other physical events not originating within the agent.

I strongly agree, objective claims about how things really are make good justifications for other objective claims about how things really are, and should be perhaps required in order to count such claims as “justified”. So far as I can make out, sophisticated theories about light and vision add little in this regard to a common-sense grasp of how seeing works.

Moreover, general knowledge about light and the historical reliability of judgments of color does not inform me of my present circumstances at all, unless experience informs me of my present circumstances in such a way as to warrant the application of thoughts about light and color discrimination to my thoughts about present circumstances. There’s no theory that tells me whether my eyes are open or closed right now, for instance, or whether it’s dark or light in here, or where the proximate light sources are in my vicinity of the world and what color of light they seem in my estimation to emit. Ordinarily I acquire such information noninferentially by using my eyes to see. It seems that any general theory of vision I may have acquired secondhand over the years, has accrued through centuries determined in part by processes involving the same sort of basis for judgment in others, who used their eyes in about the same way that I do on each particular occasion. — Cabbage Farmer

Right, theoretical knowledge is generally not a replacement for the deliverances of perception, but in cases where I suspect my perceptions to be in error my background knowledge about my current circumstances could inferentially justify a belief that directly contradicts my non-inferentially elicited observational beliefs. For instance, if I take myself to know that the room in which I am standing is currently drenched in blue light, I might come to suspect that my non-inferentially elicited judgment “this tie is purple” to be in error. I may come - on the basis of my theoretical knowledge of light, color and vision - to conclude “this tie is red” instead.

The main thrust of this point seems reasonable, but there’s arguably an unwarranted and undesirable implication, that “appearances” are not part of the empirical world, and that our accounts of appearances are not accounts of the empirical world. — Cabbage Farmer

Sellars believes in the existence of what he calls “sensa”, but he thinks that sensa are epistemically inert events that occur in the objective world. So there is a sense in which Sellars would allow that claims about sensa are indeed legitimate empirical claims. That said, sensa are conceived by Sellars to be the direct cause of our non-inferentially elicited beliefs, but not direct evidence for those beliefs. Again, this is Sellars’s attempt to avoid what he takes to be an illicit equivocation between the ontological and epistemological dimensions of perception.

I suggest, to the contrary, that “having an appearance” or “being appeared to” is a sort of knowledge of the empirical world -- a most fundamental sort -- that can be analyzed and expressed in terms of appearance-talk that coincides with matters of fact, whether or not the subject understands that or how the appearance coincides with other matters of fact.

The appearance is part of reality, that’s how it coincides. It is itself a matter of fact related to other matters of fact in the world. That’s how perceptual experience binds our thoughts to nature and opens the world to each perceiver. — Cabbage Farmer

Again, Sellars does not deny that we can have knowledge of appearances, or that appearances are part of the objective world. He only denies that such knowledge is epistemically “fundamental” for reasons already rehearsed above.

One concern I have with the view you’ve attributed to Sellars, is that it may leave a gap between general empirical beliefs and particular perceptual occasions. I mean specifically, with respect to handling claims about whether and how appearances can “provide an adequate evidential basis for claims about how things really are”; and about “the primary justificatory backdrop for our first-order observational claims”. To unpack such baggage the wrong way, may be to fall into some version of the “frictionless spinning in a void” associated with coherentism.

By my way of reckoning, it makes more sense to emphasize the cooperation on each occasion of general beliefs and beliefs based on current perception. The essential role of the more general beliefs, and the ordinary tendency to gloss over appearance-talk in happy cases, doesn’t mean there isn’t a deep layer of belief that’s analyzable or even best described in terms of appearance-talk, playing an equally essential role in our acquisition of noninferential knowledge of present circumstances. — Cabbage Farmer

Sellars maintains that non-inferentially elicited observation claims provide the foundation for empirical knowledge. However, he thinks that these are “is” claims (“x is F”), not “looks” claims (“x looks F to S”). Furthermore, he doesn’t think that these are indubitable. For Sellars, the non-inferentially elicited observation claims that make up the foundation of empirical knowledge can be overturned by inferentially elicited claims (see the example of the purple tie from above). He maintains that, in some cases, it is this very fact that makes the recognition of perceptual error possible.

Ok,stopping for a while.

Perhaps an important distinction lies hidden here. Recalling Sellars' distinction between "sensing particulars" and "sensing facts", I might deny the claim that I can see my own impressions, but affirm the claim that I observe (the fact) that I have current impressions. Accordingly, we might say I observe the impressions, or say at least that I make observational judgments about the impressions, without "sensing" them. — Cabbage Farmer

In my opinion Reichenbach does not make clear what he means when he says he doesn’t “sense” his impressions. On the one hand he says that he never “sees” an impression, but on the other hand he seems to say that all seeing is mediated by impressions such that impressions provide the “sense- content” of all of our seeings. So there is clearly a sense in which it is correct to say that he sees his impressions since they literally provide the raw data out of which all of his perceptions are constructed.

Reichenbach's general belief that there are such things as impressions is part of a theory, or model, or discourse about perceptual experience -- a logos he picked up at second hand before kicking the tires himself. Say he confirms the theory to his own satisfaction, and thus acquires the general belief, by way of some inference, and thus ceases to "doubt the existence" of impressions in general, wherever there is perception like ours. It's the theory that positions him to make the inference, on any particular occasion:

If I am sensing right now, then there are impressions in me; and
I am sensing right now;

Therefore, there are impressions in me. — Cabbage Farmer

He posits impressions as part of a theory, but he posits them as the sensual basis of all experience. So there’s a bit of a tension here, in my opinion, because he wants to say that impressions are theoretical, inferentially mediated entities and yet the role that they play within his theory of perception makes them the direct object of sense perception, which then seems to imply that they are not inferentially mediated entities after all.

In fact, he doesn't have to bother making this inference on each separate occasion, once he's acquired the relevant general belief; much as I don't have to infer that gravity draws a glass to the floor when I drop it, or that the bright spot in the sky is a massive ball of gas. Inference may play a role in the initial formation of such beliefs, and in the formation of the concepts associated with the beliefs; but once the story is told, most of us who take it on, do so without bothering over what inferences and evidence informed the story in the beginning. — Cabbage Farmer

Sure. Once the inference becomes “automatic”, the entities postulated via that inference can become the object of direct observation claims. Sellars makes pretty much the same argument in a couple of his own papers (Scientific Realism or Irenic Instrumentalism and Is Scientific Realism Tenable).

We shouldn't suppose this mere distinction informs us about what sort of thing the "impression" is. For instance, we needn't suppose that, whenever there is an impression, there is a one-to-one correspondence between the impression and a physical thing of which it is an impression. — Cabbage Farmer

Right.

I suppose he doesn't mean: Usually there are physical things within me and impressions within me, but sometimes impressions only. — Cabbage Farmer

That’s my take as well.

It seems to me we need something like a distinction between sense-perception and imagination just to make sense of the fact that we dream, hallucinate, and imagine -- though how we conceptualize such phenomena, and what judgments we make in light of them, seems to vary from one person and one cultural context to another.

I have the impression that 20th-century analytic philosophers, especially in the shadow of the spooky behaviorist tendency we might trace through Ryle and Quine, tended to neglect such phenomena, and to perhaps quietly lump false judgments made on the basis of imaginings into the same account as false judgments made on the basis of misperceptions. In any case, it's often hard to tell how to map a term like Reichenbach's "impression" here onto a distinction like the one I draw between perceiving and imagining. — Cabbage Farmer

My sense is that Reichenbach would lump imaginings in with impressions.

Perhaps the neatest way is to say that imaginings are among our impressions. If I mistake my hallucination for perception, and judge that there's an apple on the table on the basis of that hallucination, the source of the confusion is that I have mistaken hallucinating for perceiving, or in other words, I have incorrectly taken an instance of hallucinating as an instance of perceiving. I may correct my own error when I move to the table and aim to fetch the apple; discover that nothing's there, though it still looks like there's an apple; and now correctly judge that I am and have been hallucinating, despite the persistent vision-like appearance of an apple on the table. — Cabbage Farmer

Well, this is where we run into that whole “sensing facts” vs. “sensing particulars” conundrum again. Are impressions the direct cause of our observation claims or the direct justification for those claims? According to Sellars, claiming that they’re both is to fall prey to the Myth.

Also, for Sellars, as we have seen, perceptions can be true or false. So my impression is that he would say that hallucinations are just false perceptions. Of course, he’s talking about hallucinations qua propositional content. Hallucinations qua sensual contents are sensa – that is, objective events occurring “in” the perceiver. A similar story could be told about imagination, although the structure of imagination qua propositional contents is different. The claim “this unicorn is white” is true enough while I imagine a white unicorn, though the fact that I wouldn’t put forward that claim as a claim in earnest about an actually existing animal in my immediate environment is what distinguishes the two.

Reichenbach, however, seems perhaps to speak as though the confusion is somehow or other contained in the impression, and now he slides from "impression" to "representation". — Cabbage Farmer

Right. Again, apparently eliding the distinction between the causal and epistemological dimensions sensory-perception.

But what sort of thing is the impression supposed to be, if the error in judgment is already contained in it? I can stare at a mirage for hours, and change my mind a thousand times while it remains the same mirage. It looks the same, the world appears to me in the same way, while I cycle through various judgments on the basis of that one appearance. The appearance doesn't tell me how to judge, and I don't tell it how to appear. The judging is up to me, and the appearing is up to the appearance, though I can turn my head and get past it. — Cabbage Farmer

Yep.

Along these lines, I make a three-part epistemological distinction: In the first place there is the fact of the matter, the way things are in the world in fact. In the second place there is the appearance, the fact of how things appear to me (e.g., how things "look to me"). In the third place there is the seeming, the fact of how I take things in fact to be, in part on the basis of appearances. In making active judgments about how things seem, in part on the basis of appearances, I change the seeming, but not the appearances. Normally I don't need to make such judgments in order for things to seem to me one way or another; ordinarily I resort to such judging only when salient features of my experience seem uncertain, confused, or otherwise inadequate to form the basis of a reasonable judgment I have some interest in making. The judgment amends or completes or suspends the seeming. — Cabbage Farmer

I’m turning into a broken record here, but again we seem to confront this ambiguity in the word “fact”. When we use the word “fact” are we referring to something that is propositionally structured? Or to ask the same question in a different way, are we referring to something that could serve as justification for our beliefs? Or are we referring to non-propositionally structured states of the world – that is, things that could cause our beliefs but not justify them?