Footnotes to Plato from the foothills of the Superstition Mountains
Knowledge as Requiring Objective Certainty
Some brief Substacknotes on Butchvarov for Elliot, our resident epistemologist.
7 thoughts on “Knowledge as Requiring Objective Certainty”
I’ve always liked Butchvarov’s criterion of knowledge. But I remember expressing it in a graduate seminar when I was at the University of Toronto, and the other grad students went nuts (I was still an undergrad, so that may be part of the reason for their response).
Years later, I realized that they probably understood me to mean something like that knowledge is only possible for beings who can never be mistaken about anything.
For me, as an externalist epistemologist, the key is that “knowledge” is not synonymous with “indubatibility”.
In other words, it is possible to be in a state of knowledge with regard to a belief, X, but still be in a state of subjective uncertainty with regard to one’s epistemic state with regard to X.
As far as I can see, pared down to its bare bones “impossible to be mistaken about X (at time t)” is more or less substitutable, salve veritate, with “X is true (at time t)”
>>Admittedly, knowledge as impossibility of mistake is a very stringent concept of knowledge. Why should we care to set the bar so high? Why is knowledge in the strict sense important? It is important because there are life and death situations in which one needs to know in order to decide on a course of action.<<
Let’s call this position epistemic infallibilism (EI). In addition to the life and death reason for setting the bar so high, EI (arguably) offers the best solution to the Gettier problem. Gettier cases contain the conjunction of luck and fallibility. If one’s knowledge that p entails the impossibility of error, then one’s vulnerability to that conjunction is eliminated.
EI is a version of epistemic invariantism. I can hear the epistemic contextualist working up a reply.
The contextualist might agree that knowledge entails impossibility of error in high-stakes situations such as life and death matters or other cases in which the truth is of great importance and error is costly. But the contextualist holds that what counts as ‘knowledge’ depends on the epistemic context in which a knowledge claim is made. More specifically, “EC is roughly the view that what is expressed by a knowledge attribution — a claim to the effect that S “knows” that p — depends partly on something in the context of the attributor, … Because such an utterance is context-dependent, so too is whether the attribution is true.”
On this view, ‘know’ is (or at least is similar to) an indexical. Its meaning depends on the context in which it’s used.
The contextualist might say that “Smith knows that p” is false in cases in which the stakes are high, such as life and death situations, or advanced philosophical discussions about brains in vats, since in such high-stakes cases ‘know’ means ‘impossibility of mistake.’ However, the claim that “Smith knows that p,” if made in lower-stakes situations, such as whether the furniture store opens at 10 AM on Saturday, or an informal conversation about what time the game starts, is true because in low-stakes cases ‘know’ does not mean ‘impossibility of error’ but means something less epistemically demanding.
To elaborate on the difference between invariantism and contextualism, consider the following claims:
1. If I don’t know that I’m not a bodiless, immaterial soul, then I don’t know that I have hands.
2. I don’t know that I’m not a bodiless, immaterial soul.
3. I know that I have hands.
The epistemic infallibilist, as an invariantist, holds that ‘know’ is invariant in all cases. Properly used, ‘know’ always means ‘can’t be mistaken.’ The infallibilist accepts (1) and (2) but denies (3).
The contextualist claims to be justified in accepting all three claims, since ‘know’ is not an invariant term. The contextualist agrees that (1) is true. He holds that (2) is true in the high-level context of advanced philosophical discussion, and that (3) is true in the low-level context of a tennis instructor teaching kids to play tennis.
There’s a ‘new’ post on the Gettier cases. Scroll up.
How does epistemic contextualism differ, if it does differ, from the notion that knowledge is a family-resemblance concept like the concept, game. (Wittgenstein). I take Ludwig’s idea to be that there is no essence of knowledge; there are just a number of different, but related, concepts corresponding to different, but related, uses of ‘know’ and cognates. In my phil. lexicon, concepts are mental; essences are extra-mental. Concepts represent essences, more or less adequately. LW might not want to talk about concepts, just words and their senses. But then how distinguish a sense from a concept? For example, ‘Dreieck’ and ‘triangle’ are different words, both at the type-level and the token-level. But they have the same sense (Frege’s *Sinn*) and express the same concept. So how do sense and concept differ?
I like your aporetic triad; it focuses the issue nicely.
According to the SEP article, LW’s notion was a forerunner to contextualism, “setting the stage” for it.
>In my phil. lexicon, concepts are mental; essences are extra-mental.<
I agree.
One concern I have about contextualism is that it seems to be a thesis of semantics rather than epistemology. If contextualists are merely addressing how uses of ‘know’ change meaning in different contexts, then it seems they are doing linguistics/semantics, which is interesting but not strictly relevant to the epistemological debate regarding the nature of knowledge.
The infallibilists and fallibilists are debating the nature of knowledge, such as whether or not knowledge requires infallible justification. The thesis that the meaning of ‘know’ changes in various contexts does not address the debate about the essence of knowledge itself.
That debate presupposes that there is an essence of knowledge. If there is no such thing, as LW seems to have held, then the debate seems unnecessary.
Elliot,
I wonder if we can make a clean cut between the semantic and the epistemic. Frege’s senses do both sorts of job: they function as word senses but also as epistemic intermediaries. What makes ‘Hesperus is Phosphorus’ informative, unlike ‘Hesperus is Hesperus,’ is the fact that the two terms have different senses; but these different word senses also mediate both thinking reference and linguistic reference to one and the same chunk of physical reality, the planet Venus. (I would say that thinking reference is grounded in linguistic reference, but this too is a contested point and was a bone of contention between Chisholm and Sellars.) Frege calls them modes of presentation (Darstellungsweisen) of the physical thing. They are thus epistemic intermediaries and not merely word senses.
One can push this further along the lines of Castaneda and Butch by ‘ontologizing’ these senses and making them literal parts of the physical thing. So for Hector C. the thing is a consubstantiational system of ontological guises in his jargon, whereas for Butch the physical entity is an assemblage of what he calls objects. If we move in this direction, senses cease being epistemic intermediaries.
Or we can move in the opposite direction back toward LW and meaning-as-use and enclosure within language-games.
Three views: senses as merely semantic-linguistic; senses as epistemic intermediaries; senses as constituents of their referents. On the first view, senses have no ontic status; on the second ontic-epistemic status; on the third, solely ontic status.
Or can we get rid of senses entirely? Could we say that the meaning or sense of a proper name is identical to its referent? Wasn’t that J S Mill’s view?
Thanks for the helpful reply. I think that the second view (senses as epistemic intermediaries) is more plausible than the first. I’m not familiar enough with Castaneda’s and Butchvarov’s “ontologizing” view to assess it. What are some good texts I can use to get up to speed on this?
Mill’s position is worth discussing in more detail.
I’ve always liked Butchvarov’s criterion of knowledge. But I remember expressing it in a graduate seminar when I was at the University of Toronto, and the other grad students went nuts (I was still an undergrad, so that may be part of the reason for their response).
Years later, I realized that they probably understood me to mean something like that knowledge is only possible for beings who can never be mistaken about anything.
For me, as an externalist epistemologist, the key is that “knowledge” is not synonymous with “indubatibility”.
In other words, it is possible to be in a state of knowledge with regard to a belief, X, but still be in a state of subjective uncertainty with regard to one’s epistemic state with regard to X.
As far as I can see, pared down to its bare bones “impossible to be mistaken about X (at time t)” is more or less substitutable, salve veritate, with “X is true (at time t)”
Bill, thanks for this interesting post.
>>Admittedly, knowledge as impossibility of mistake is a very stringent concept of knowledge. Why should we care to set the bar so high? Why is knowledge in the strict sense important? It is important because there are life and death situations in which one needs to know in order to decide on a course of action.<<
Let’s call this position epistemic infallibilism (EI). In addition to the life and death reason for setting the bar so high, EI (arguably) offers the best solution to the Gettier problem. Gettier cases contain the conjunction of luck and fallibility. If one’s knowledge that p entails the impossibility of error, then one’s vulnerability to that conjunction is eliminated.
EI is a version of epistemic invariantism. I can hear the epistemic contextualist working up a reply.
The contextualist might agree that knowledge entails impossibility of error in high-stakes situations such as life and death matters or other cases in which the truth is of great importance and error is costly. But the contextualist holds that what counts as ‘knowledge’ depends on the epistemic context in which a knowledge claim is made. More specifically, “EC is roughly the view that what is expressed by a knowledge attribution — a claim to the effect that S “knows” that p — depends partly on something in the context of the attributor, … Because such an utterance is context-dependent, so too is whether the attribution is true.”
https://plato.stanford.edu/entries/contextualism-epistemology/
On this view, ‘know’ is (or at least is similar to) an indexical. Its meaning depends on the context in which it’s used.
The contextualist might say that “Smith knows that p” is false in cases in which the stakes are high, such as life and death situations, or advanced philosophical discussions about brains in vats, since in such high-stakes cases ‘know’ means ‘impossibility of mistake.’ However, the claim that “Smith knows that p,” if made in lower-stakes situations, such as whether the furniture store opens at 10 AM on Saturday, or an informal conversation about what time the game starts, is true because in low-stakes cases ‘know’ does not mean ‘impossibility of error’ but means something less epistemically demanding.
To elaborate on the difference between invariantism and contextualism, consider the following claims:
1. If I don’t know that I’m not a bodiless, immaterial soul, then I don’t know that I have hands.
2. I don’t know that I’m not a bodiless, immaterial soul.
3. I know that I have hands.
The epistemic infallibilist, as an invariantist, holds that ‘know’ is invariant in all cases. Properly used, ‘know’ always means ‘can’t be mistaken.’ The infallibilist accepts (1) and (2) but denies (3).
The contextualist claims to be justified in accepting all three claims, since ‘know’ is not an invariant term. The contextualist agrees that (1) is true. He holds that (2) is true in the high-level context of advanced philosophical discussion, and that (3) is true in the low-level context of a tennis instructor teaching kids to play tennis.
There’s a ‘new’ post on the Gettier cases. Scroll up.
How does epistemic contextualism differ, if it does differ, from the notion that knowledge is a family-resemblance concept like the concept, game. (Wittgenstein). I take Ludwig’s idea to be that there is no essence of knowledge; there are just a number of different, but related, concepts corresponding to different, but related, uses of ‘know’ and cognates. In my phil. lexicon, concepts are mental; essences are extra-mental. Concepts represent essences, more or less adequately. LW might not want to talk about concepts, just words and their senses. But then how distinguish a sense from a concept? For example, ‘Dreieck’ and ‘triangle’ are different words, both at the type-level and the token-level. But they have the same sense (Frege’s *Sinn*) and express the same concept. So how do sense and concept differ?
I like your aporetic triad; it focuses the issue nicely.
According to the SEP article, LW’s notion was a forerunner to contextualism, “setting the stage” for it.
>In my phil. lexicon, concepts are mental; essences are extra-mental.<
I agree.
One concern I have about contextualism is that it seems to be a thesis of semantics rather than epistemology. If contextualists are merely addressing how uses of ‘know’ change meaning in different contexts, then it seems they are doing linguistics/semantics, which is interesting but not strictly relevant to the epistemological debate regarding the nature of knowledge.
The infallibilists and fallibilists are debating the nature of knowledge, such as whether or not knowledge requires infallible justification. The thesis that the meaning of ‘know’ changes in various contexts does not address the debate about the essence of knowledge itself.
That debate presupposes that there is an essence of knowledge. If there is no such thing, as LW seems to have held, then the debate seems unnecessary.
Elliot,
I wonder if we can make a clean cut between the semantic and the epistemic. Frege’s senses do both sorts of job: they function as word senses but also as epistemic intermediaries. What makes ‘Hesperus is Phosphorus’ informative, unlike ‘Hesperus is Hesperus,’ is the fact that the two terms have different senses; but these different word senses also mediate both thinking reference and linguistic reference to one and the same chunk of physical reality, the planet Venus. (I would say that thinking reference is grounded in linguistic reference, but this too is a contested point and was a bone of contention between Chisholm and Sellars.) Frege calls them modes of presentation (Darstellungsweisen) of the physical thing. They are thus epistemic intermediaries and not merely word senses.
One can push this further along the lines of Castaneda and Butch by ‘ontologizing’ these senses and making them literal parts of the physical thing. So for Hector C. the thing is a consubstantiational system of ontological guises in his jargon, whereas for Butch the physical entity is an assemblage of what he calls objects. If we move in this direction, senses cease being epistemic intermediaries.
Or we can move in the opposite direction back toward LW and meaning-as-use and enclosure within language-games.
Three views: senses as merely semantic-linguistic; senses as epistemic intermediaries; senses as constituents of their referents. On the first view, senses have no ontic status; on the second ontic-epistemic status; on the third, solely ontic status.
Or can we get rid of senses entirely? Could we say that the meaning or sense of a proper name is identical to its referent? Wasn’t that J S Mill’s view?
Bill,
Thanks for the helpful reply. I think that the second view (senses as epistemic intermediaries) is more plausible than the first. I’m not familiar enough with Castaneda’s and Butchvarov’s “ontologizing” view to assess it. What are some good texts I can use to get up to speed on this?
Mill’s position is worth discussing in more detail.