To Know Is Not To Describe

Streetlight

Among the many claims made in Wilfred Sellars' "Empiricism and Philosophy of Mind", is this one, which I think is indispensable: that knowledge is not a matter of description. That is, to say that 'this is red' is not (or not only) to provide a description about a state of affairs (Sellars: "In characterizing an episode or a state as that of knowing, we are not giving an empirical description of that episode or state"). On the face of it, this might seem to be a pretty provocative claim (what could knowledge about something be if not a description of it?), so I wanna explore what it really means and how Sellars gets there.

He begins with a thought experiment. It goes like this: imagine a world where electric lighting has not yet been invented. In this world, all colors appear exactly as they are. Green looks green in natural light, and so on. With the invention of electric lighting however, colors can now appear to look other than what they are. A tie salesman, John, who is used to calling colors 'as he sees them', is intitally quite confused by this. In his shop, when asked by a customer what color a particular (green) tie is, John says: 'it's blue'. He says this, because it looks blue to him in the electric light.

Upon hearing this, his colleague, Bob, quickly takes John aside and tells him that the tie is green, and that John should say that it is. John protests and says that it looks blue. But Bob, being a more competent wielder of language, points out that that's not how color-talk works: we simply don't talk of things changing color depending on the lighting. The tie, Bob says, is green, even though it looks blue. After a few days, John gets the hang of this way of talking. John has learnt a new way of talking.

So what does this thought experiment show? At the very least, it is meant to indicate the plausibility of the claim that one cannot simply 'read off' a claim of knowledge from a state-of-affairs. Something else, or rather, something over and above the state of affairs is needed; this 'something over and above' being not just what is being talked about (the color), but also how it is being talked about. Importantly, this 'how' involves what Sellars refers to as a normative dimension of knowledge claims, an 'ought-to-say' over and above a mere 'is'. I'll not labour over this point too much, and place the emphasis instead on negative claim (knowledge <> description) and see where a discussion might go.

I might do more than one thread on the EPM essay, which is a incredibly rich with ideas. Here's one of them.

Ying

raza

Essentially anything you see is everything else you don't see.

apokrisis

The tie, Bob says, is green, even though it looks blue. After a few days, John gets the hang of this way of talking. John has learnt a new way of talking. — StreetlightX

And interestingly, almost a Platonic way of talking. Conception sees beyond mere appearance.

Or more pragmatically, the mind seeks out a way to impose stability on the flux of experience. It is trying to describe the essence that endures.

And even more pragmatically - accepting now that experience is a phenomenal umwelt organised by a mediating system of signs - we do end up knowing only our concrete symbols of things. Green is something we can reliably recognise, no matter what its current guise.

Importantly, this 'how' involves what Sellars refers to as a normative dimension of knowledge claims, an 'ought-to-say' over and above a mere 'is'. — StreetlightX

Again, the pragmatic story. Knowledge - in the linguistically-structured sense particular to humans - is what a community sharing a language, a system of linguistic sign, would come to agree the meaning of in the long run. To say "green" would have a socially-constrained interpretation. And to be a functioning part of a linguistic community is to be able to participate as a native in its linguistic habits.

So yes. There is a normative ought that emerges at the sociocultural level. But it is a soft "ought" in being just a constraint. There is still an irreducible freedom or creativity about what any particular speech act might actually mean.

That too is a key aspect of semiosis. Essences or generalisations are limiting but not completely restrictive.

Pseudonym

↪StreetlightX

So, I seem to be missing something here. The tie seems green in natural light, it seems blue under electric light, the shopkeeper tells the assistant that the correct use of colour language is to use the term which is appropriate under natural light.

These just seem like three knowledge as description claims.

1. The colour of tie under electric light is blue (a description of the state under electric light)
2. The colour of the tie under natural light is green (second-hand knowledge, but a description nonetheless as it describes how the tie would look under natural light)
3. The common choice of which colour term to use in discourse so that we understand each other is the one which describes it under natural light (a description of the way colour terms are applied)

So what am I missing here?

Streetlight

So what am I missing here? — Pseudonym

Normativity.

Pseudonym

↪StreetlightX

There's no normative knowledge claim here that I can see. Bob is telling John how to describe the tie entirely contextually (in his language use), the normative claim is about language use, it's not about John's knowledge of the tie's colour, John knows two things about the tie's colour; that is appears blue under electric light and that it appears green under natural light. The normative claim is not about which constitutes knowledge, but about which is appropriate to use in discourse.

Streetlight

The normative claim is not about which constitutes knowledge, but about which is appropriate to use in discourse. — Pseudonym

An odd distinction.

Pseudonym

↪StreetlightX

Really, how? One can know of an object more than one of its properties. When we are asked as a witness to "describe the thief", we do not talk about his internal anger or his two-leggedness despite the fact that these are two things we 'know' about the thief. We also know that in context, those two properties are not fit for discourse, the first because it is transient, the second because it is non-specifying. It doesn't alter their knowledge status, I still 'know' the theif has two legs, it's just not appropriate for me to declare that in this particular discourse.

The example you gave is only different if you accept the tie's colour as a single property, but I see no reason why anyone would do that. There are two properties - its colour under natural light and its colour under electric light. It's merely linguistic convention that when the lighting is unspecified, we're to presume it's natural light.

Streetlight

There are two properties - its colour under natural light and its colour under electric light. — Pseudonym

No one is asking about 'properties'. Just the color of a tie.

Pseudonym

↪StreetlightX

Is its colour not a property?

Streetlight

↪Pseudonym

Maybe, maybe not. But one knows or does not know the color of the tie. Does one know it's 'properties'? Seems like a bunch of philosophical sky-castling. A non-issue pretending to be an issue.

Pseudonym

But one knows or does not know the color of the tie. — StreetlightX

Yes, but only if we define 'colour' as "that sense most people get on seeing the object under natural light". All Bob is doing is correcting John on his use of the word 'colour' in that context. Obviously 'colour' can't refer to our private meaning (the sense we happen to get of it at any one time) otherwise we'd be into private languages.

Streetlight

All Bob is doing is... — Pseudonym

What else would you have him do?

Pseudonym

↪StreetlightX

It's not what else I'd have him do, it's what he's not doing. He's not in any way implying that John's knowledge about the colour of the tie is wrong because he merely describes it. It is the speech act that Bob is correcting, so where does the contention that knowledge is not description arise from?

Streetlight

↪Pseudonym

But John's knowledge of the color of the tie is - or was - wrong. The tie is not blue.

Pseudonym

↪StreetlightX

No, John knows two things about the tie, its colour under the electric shop light and it's colour under natural light, both of which he is right about.

Streetlight

↪Pseudonym

But he was only asked about the color of the tie simpliciter. And he was wrong.

Pseudonym

↪StreetlightX

Yes, but only because he was wrong about the meaning of the word colour in this context. Think about it, if he genuinely was asked about the colour of the tie simpliciter without any ambiguity about the meaning of the words then his answer would have been exactly the same as his answer to the question "how would you describe a normal-sighted person's sensation of colour when exposed to this tie under natural light?", because that's what "what colour is this tie?" actually means. But his answer to that question would have been "I don't know" (having never seen the tie under natural light). So it is his understanding of the meaning of the question which is wrong, not his knowledge of the tie's of colour.

Streetlight

So it is his understanding of the meaning of the question which is wrong, not his knowledge of the tie's of colour. — Pseudonym

An odd distinction.

Pseudonym

↪StreetlightX

"What's the best way to get from Paris to London?"

"Well, you should take the train to Marseille, then along to Cannes and finally knit your way back through the Jura until you get to Calais, Dover then London"

"But that would take me ages!"

"Sorry, by 'best' I thought you meant most pleasant, evidently you meant quickest"

How does the interlocutor's knowledge of the most pleasant or the quickest way to London change at all in that exchange? All that changes is their understanding of the meaning of the word 'best', in that context,specifically to which of those two knowledge items it refers.

Streetlight

↪Pseudonym

The conversation is disanalogous. John was wrong about the color of the tie. He did not know what color it was. One wonders if your interlocutor knows or does not know the quickest way to London from Paris. But this is not part of your conversation.

Pseudonym

John was wrong about the color of the tie. — StreetlightX

No, I maintain the "the colour of the tie" is simply an ambiguous property and therefore can be in multiple states (both green and blue). There is a linguistic convention about which of those states one reports when asked "what is the colour of this tie" without any caveats. I could ask the question "what is the colour of this tie under this light?" and be easily understood. The correct answer would be different to the answer to the question "what is the colour of this tie under natural light?" and yet both questions refer to the single property 'the colour of this tie'.

Its like asking "what is the speed of this car?" the answer is different depending on how much fuel is being allowed into the engine.

If a sentence yields unintuitive epistemological consequences one should first look to see if the fault lies with the sentence. This is basic Wittgenstein.

Streetlight

No, I maintain the "the colour of the tie" is simply an ambiguous property — Pseudonym

It takes a philosopher to say this. So much the worse for philosophers.

Aleksander Kvam

excuse me, if this sounds stupid. but how would this work with something that isnt as easily evident, as the consept of God. many discribe "God" differently im sure and to some, like me(but not that it matters), find no meaning or fully "understandable" description to it at all. like someone being colourblind from birth being told that the tie is red.

Aaron R

↪StreetlightX

I think the title of your post obscures the meaning of Sellars's statement, which is really that to attribute knowledge to an agent is not to provide an empirical description of that agent. In other words, knowing is not reducible to an empirical state.

Streetlight

↪Aaron R

I have to admit I've been puzzling a bit over Sellars' wording in the bit I've quoted re: "an episode or a state... of knowing", which, it's true, does seem to refer to a state of an agent. But I kinda wanna have my cake and eat it too, and say that that this claim is itself tributary to the co-extensive claim that facts themselves are irreducible to 'sheer observation'; and that, as Sellars will put in various ways, all 'observational knowledge' (of the form "X is a reliable symptom of Y") presuposes a nexus of concepts (the 'logical space of reason') which allows it to be acknowledged as a fact:

"For the point is specifically that observational knowledge of any particular fact, e.g. that this is green, presupposes that one knows general facts of the form X is a reliable symptom of Y. And to admit this requires an abandonment of the traditional empiricist idea that observational knowledge “stands on its own feet.”" My own attempt to paraphrase this was to say that "one cannot simply 'read off' a claim of knowledge from a state-of-affairs". If this is the case - and let me know if you think I've erred - then the issue is more of a a minor one in which I've simply picked the wrong quote to shore up a (right) reconstruction. But of course I'm still working though this so happy to get feedback.

Marchesk

For the point is specifically that observational knowledge of any particular fact, e.g. that this is green, presupposes that one knows general facts of the form X is a reliable symptom of Y. And to admit this requires an abandonment of the traditional empiricist idea that observational knowledge “stands on its own feet.”" My own attempt to paraphrase this was to say that "one cannot simply 'read off' a claim of knowledge from a state-of-affairs". — StreetlightX

Seems rather obvious on the face of it. Did empiricists really believe that the raw sensory datum could be used to construct an understanding of the world? On what basis?

Aaron R

↪StreetlightX

I think you're heading down the right track. There's two separate but related lines of questioning at play here. First there is the question of whether the achievement of knowing can be analyzed in purely causal/empirical terms. Second is the question of whether empirical description forms a realm of discourse that is semantically independent of all others. Sellars answers "no" to both questions. In the first case this is because he thinks there is an ineliminable normative aspect to knowing. To know is not just to engage in certain causal relations in certain reliable ways, but also (in some sense) to do what ought to be done. In the second case this is because empirical description presupposes an understanding of various modal and normative concepts that cannot be reduced to mere descriptions. In both cases normativity plays a key role in explaining Sellars's denial, but for slightly different reasons.

To Know Is Not To Describe

Welcome to The Philosophy Forum!

Categories

More Discussions