I tire of hearing folks repeat “you cannot derive `ought’ from `is’,” because there is an important sense in which most attempts to derive “ought” are built on “is.” Let me explain. An argument for an “ought” is typically built on some set of more basic “obvious” claims that the speaker assumes their audience will accept without argument. Many of those claims have their own supporting arguments somewhere else, but those arguments are also be built on further obvious claims.
Nick, I'm not addressing Hume in the context of his writing, but the wider "you can't derive is from ought" claim tossed about casually as a very general and reliable theorem. Who actually proposed to or went to what effort to derive what when is not so relevant for evaluating the "you can't derive" claim. I'm not talking about efforts to derive or conscious intentions to derive; I'm talking about what people use to support their beliefs when those beliefs are called into question, besides just saying "because."
I'll follow your lead in elaborating the argument further, but renumber for clarity:1.(unexpanded) My meta seemings (about seemings) are reliable indicators of truths about seemings 2.(observation) My moral seemings seem (to me) reliable indicators of moral truth3 (combine 1&2) My moral seemings are reliable indicators of moral truth.4.(observation) It seems to me that harming innocents is wrong5.(combine 3&4) I know that harming innocents is wrong6.(unexpanded) If I know something, that something is true.7.(combine 5&6) Harming innocents is wrong.
My basic claim here is that all of the unexpanded and observation claims (1,2,4,6) are reasonably thought of as "is" claims.
The content of the belief that it seems to me that harming innocents is wrong includes the normative notion of wrongness. Therefore it could be argued that the proposition believed is normative (something similar to this issue was, if I remember rightly, discussed by Hare in his reply to Searle’s ‘How to derive an ought from an is’). You might reject this on the grounds that ‘seems’ is not a factive operator and so having a normative notion within its scope does not make the proposition normative. OK, so take it that both are purely descriptive beliefs. The derivation you are proposing is1.It seems to me that harming innocents is wrong2.My seemings seem to me to be reliable indicators of (moral) truth3.Therefore harming innocents is wrongI doubt whether I or anyone else actually derived their moral beliefs by such reasoning, any more than we reason from sensory seemings to perceptual beliefs. Rather, we have the sensory seemings and absent certain defeaters they cause us to have the perceptual belief, and this process is sometimes misleadingly described as an inference. Similarly my belief that harming innocents is wrong might be caused by it seeming like that.
But perhaps you are proposing that this is how an ethical intuitionist would defend the moral principle that harming innocents is wrong. There is a simple reason why no intuitionist should present this as a derivation of the moral principle: the form of the argument is invalid (and hence {1,2} is not a set of descriptive propositions that entail a normative proposition).
Some ethical intuitionists explain how it is that we know moral truth in terms similar to this (but the second premiss would be that seemings are reliable, not that seemings seem to be reliable). Taken like that, it is missing an intermediate conclusion ‘2b. therefore I know that harming innocents is wrong’. The step from 2b to 3 is valid but the argument as a whole is still invalid. Furthermore, such externalist explanations of knowledge are explanations of how we know what is true, when it is true, so properly formulated it would take 3 as a premiss, and 2b would be the conclusion.
Moreover, such explanations of moral knowledge don’t entail that belief in the conclusion (3) is not a basic belief, (in the sense that basic beliefs are where the regress of justification stops). For example, if intuitionists were then to explain the seeming and the reliability in terms of conceptual competence and the nature of the concepts involved their whole story could be that ‘harming innocents is wrong’ is a self evident proposition.
Finally, you might say, great, the step from 2b to 3 shows that you can derive an ‘ought’ from an ‘is’, since 2b is prima facie a descriptive proposition whilst 3 is normative. First of all, that’s fine, since I was not trying to defend Hume’s critique but point out that it was about entailment of normative propositions rather than knowledge of them. Secondly, however, since knowledge is factive it is less clear that it is a mere quibble to point out that the proposition ‘I know that harming innocents is wrong’ contains a normative notion. The normative notion is within the scope of a factive operator and that looks like a ground for saying it is a normative proposition. Thirdly, setting that aside, 2b is still the wrong kind of proposition to embarrass Hume’s critique. He is criticising people for deriving things like ‘you ought to help Fred’ from ‘Fred is suffering’ alone. That you can derive ‘you ought to help Fred’ from ‘I know you ought to help Fred’ is beside his point.
Nick, I say your ought belief that hitting innocents is wrong is not basic, but derived from your beliefs that it seems to you that hitting is wrong, and that your seemings seem reliable indicators of moral truth. I say both of these are is beliefs? Which of these two beliefs do you say is an ought and not an is belief?
Hi Robin. I think you are conflating two quite separate issues, one is about the relation of logical consequence and the other is the basis on which we know (or justifiably believe, but for brevity I’ll speak of knowledge) ethical propositions. Either of these might be being referred to when speaking of what we can derive and the fact that knowing the conclusion of an argument depends both on the whether we know the premisses and whether the premisses entail the conclusion allows the two to be conflated by talk of derivation.
The issue of knowledge you raise is that of knowing the general ethical principles from which we derive particular ethical conclusions, such as when I infer I ought not hit Johhny because harming innocents is wrong, hitting is a harm, and Johhny is innocent. Whether I know that harming innocents is wrong may well depend on the truth about the nature of ethical intuition. If I don’t know the ethical premiss then I don’t know the conclusion despite the validity of the inference. Perhaps it is only if ethical intuition is reliable that I know the ethical premiss. But the issue over deriving ‘ought’ from ‘is’ is about whether the inference is valid.
The validity depends on the nature of the relation of logical consequence. The denial of deriving an 'ought' from an 'is' is the claim that there are no sets of purely descriptive propositions that entail a normative proposition, alternatively, that any set of propositions that entails a normative proposition contains a normative proposition. Quite clearly, the fact you mention, that whether I know an ethical principle depends on a descriptive truth about the reliability of ethical intution, has nothing to do with whether there are any sets of purely descriptive propositions that entail a normative proposition. Hence that fact is irrelevant to the question of whether we can derive an ‘ought’ from an ‘is’.
I think the problem is that Robin is confusing thought with feeling (or perhaps intuition). I don't believe torture is wrong because I believe "my intuition says torture is wrong." I believe torture is wrong because my intuition says torture is wrong!
If I can even conceive of that intuition being wrong, it's either because I have a conflicting intuition, or because the relevance of the intuition is based on false premises. In the case of conflicting intuitions, my intuition not to torture is weighed against the fact that, if I don't, a thousand people might die in a terrorist attack, and my intuition against torture might be wrong as a guide to action in this case, but not generally. In the case of the false premises, maybe it turns out that the victim of torture is not really a person, but a non-sentient automaton, and if I examine my intuitions, I discover that I'm really only against the torture of sentient beings. Yes, there are tons of "is-es" there, but it ultimately comes to an "ought" (or two).
An argument for an “ought” is typically built on some set of more basic “obvious” claims that the speaker assumes their audience will accept without argument. Many of those claims have their own supporting arguments somewhere else, but those arguments are also be built on further obvious claims.Eventually we end up with with a set of basic supporting claims that seem obvious, but which don’t have much in the way of explicit arguments supporting them. Yes, almost always one of these obvious but not explicitly argued claims is of the “ought” type. So in this sense every “ought” is derived from an “is.”
Did you mean to write something else? What this actually says is that every "ought" is derived from an "ought".
I’m instead pointing out that most every attempt to derive an “ought” is based ultimately on “is” claims about the reliability of our intuitions about such more basic “ought” claims. If we can’t find a coherent way to integrate these “is” claims with the rest of our network of reasonable “is” claims, then we can’t argue coherently for such “ought” claims at all.
And this says it again. I think the logical closure over what you said is this:
Every "ought" claim1. is ungrounded and ultimately meaningless, because every ought claim depends on another ought claim (you argue this twice, but it fails to play any part in the main point you are making), and2. "comes from" "is" claims, because you need to have facts to make an argument.
You have no evidence for the first claim; and the second is irrelevant. Yes, "ought" claims make use of an infrastructure of "is" claims; but that isn't what people mean when they say "ought doesn't come from is". They mean that an "ought" claim can't bottom out entirely in "is" claims. And not only did you not argue with that, you agreed with it.
Now, does that make wikipedia worthless, no; accepting it as straight fact, however, is obviously a bad idea.
Maybe, but by what standard of discernment can, or do, you make that claim? Does this suggest there is an is? From which we derive ought?
To what extent is Wikipedia and assault on 'Authority' or esp the tenure that goes with it, which is spent of lifestyles that cranks up 'relative deprivation' in those who have not secured tenure? Do education systems tend to select with a far too broad an apparatus? There will be may losers, but the winners will be of superior quality, and the 'losers' afterall, have transferable skills - or appetites of entitlement.
I think the problem with wikipedia is that it's based on the idea of crowd-sourcing, or "the average of what most people will say will probably be right". Where i think wikipedia fails is that, in fact, most people dont take the time to weigh in on topics.
Now, does that make wikipedia worthless, no; accepting it as straight fact, however, is obviously a bad idea.
I've not read this one but can imagine the gist. Take a recent article about Wikipedia, people are rightly sceptical but some.. well... I call it Wikiphobia (why the hell not eh?) People have invested so much in moral relativism that they fear the notion that a more concrete social phenomenon is forming (and yet paradoxically, I recently referred to everything as dissolving). A response to this article went: T
here are so many inaccuracies that Wikipedia is to be avoided as a research tool at all costs. For example, look at the entry for Israel. It states that the Israeli Capital is Jerusalem when all the world apart from four countries recognise it as Tel Aviv. If they can't get that right, what hope is there for the rest of their entries.
This unleashed a volley of replies along these lines
Poor example since they do have a link to a footnote which explains the situation properly.
Your post only appears to address how people behave (in a sort of pop anthropological way). You seem to be saying that because people act as if 'ought' can be derived from 'is' that it must be true.
If you're arguing something else, I've missed it either through my own poor reading or the subtlety of your argument.
(Personally, I believe I make mistakes all the time. I admit my life would much easier (though probably shorter) if I believed that the fact that I behaved a certain way meant that that was the right way to behave.
(Well, maybe not shorter. Dr. Pangloss seemed awfully hale for someone hanged, burned, and separated from this best of all possible worlds.))
The point about the ‘is’ ‘ought’ gap is that you cannot get to an ‘ought’ merely on ‘is’ claims.
Robin Hanson presented a reason why this isn't true: in a moral argument, the specifically-ought part will be something like, "Torturing a human is wrong." -- but even that "pure ought" statement is actually based on "My intuition says torture is wrong" and "My intuitions are more often correct than incorrect", both of which are "is" statements.
Constant, my only point was that we have very pragmatic of things that 'look like stealing without being considered evil' (using the meaning of look that I think was used originally).
I agree entirely with you that with a more rigorous definition of 'look' these things are obvious not stealing.
Though to complicate things, maybe the historical (and social I think) distinction that you rightfully bring up only exists in our conscious minds, and so we might debate whether that distinction is 'physically observable.'
So a Martian anthropologist (there's no other profession on Mars except slaving away at thought experiments) can build a robust model to predict how humans will behave that distinguishes between taxes and robbery, and he might being a martianocentric thinker attribute some moral mental states to these two categories, but he wouldn't really know our mental states, whether these categories have different mental states, and even whether we have consciousness (as understood by Martians).
This is when the Martians get worried that their nearest neighbor has been overrun by qualia eating zombies...
The solipsist says: give me just one example of something that exists apart from myself. But he rejects everything offered.
That cuts quite to the heart of it, in my opinion. This has to do with what we are willing to accept as justification. I have a very hard time imagining accepting some 'is' statement as sole justification for an 'ought' statement but it might occur. The reason I can't say definitively "it will never happen" is because I haven't defined for myself what statements I'll accept as justification are/look-like.
Robin's proposed 'is' statement to justify an 'ought' statement doesn't seem like justification to me. I can't make it not be a justification for Robin or for anyone else. I can only try to get them to believe that it is not a justification...usually by showing that it contradicts some other belief that they have.
This is a very messy problem that I'd appreciate enlightenment on, but I recognize is not obviously tied in to is/ought question. I believe though in my gut that it is tied into the is/ought question, but I haven't come up with a clear and cogent argument how.
In your comparison between bananas and evil, you are comparing re-assiging the labels while retaining our original understanding of the subject matter so that it does not get re-assigned with the labels on the one hand (what you envision doing with bananas), with with re-assigning labels while also re-assigning our understanding of the subject matter so that it shifts along with the labels.
That's not a genuine comparison between bananas and evil. It's just a comparison of the two different ways you are treating the terms.
You could just have easily done the opposite: shift the meaning along with the label "banana", so that you end up trying to eat whatever new object it is you are calling a "banana", while on the other hand retaining your original moral understanding of actions while re-assigning the labels "good" and "evil" while this time treating them as empty labels (so that you punish some actions you are now labeling "good", for example).
It's not the term "banana" itself that is empty while the term "evil" is loaded. It's you who (in the hypothetical scenario) have chosen to treat "banana" as empty (as re-assignable without changing our understanding of the things being relabeled) while treating "evil" as loaded.
Nick, I'm not addressing Hume in the context of his writing, but the wider "you can't derive is from ought" claim tossed about casually as a very general and reliable theorem. Who actually proposed to or went to what effort to derive what when is not so relevant for evaluating the "you can't derive" claim. I'm not talking about efforts to derive or conscious intentions to derive; I'm talking about what people use to support their beliefs when those beliefs are called into question, besides just saying "because."
I'll follow your lead in elaborating the argument further, but renumber for clarity:1.(unexpanded) My meta seemings (about seemings) are reliable indicators of truths about seemings 2.(observation) My moral seemings seem (to me) reliable indicators of moral truth3 (combine 1&2) My moral seemings are reliable indicators of moral truth.4.(observation) It seems to me that harming innocents is wrong5.(combine 3&4) I know that harming innocents is wrong6.(unexpanded) If I know something, that something is true.7.(combine 5&6) Harming innocents is wrong.
My basic claim here is that all of the unexpanded and observation claims (1,2,4,6) are reasonably thought of as "is" claims.
The content of the belief that it seems to me that harming innocents is wrong includes the normative notion of wrongness. Therefore it could be argued that the proposition believed is normative (something similar to this issue was, if I remember rightly, discussed by Hare in his reply to Searle’s ‘How to derive an ought from an is’). You might reject this on the grounds that ‘seems’ is not a factive operator and so having a normative notion within its scope does not make the proposition normative. OK, so take it that both are purely descriptive beliefs. The derivation you are proposing is1.It seems to me that harming innocents is wrong2.My seemings seem to me to be reliable indicators of (moral) truth3.Therefore harming innocents is wrongI doubt whether I or anyone else actually derived their moral beliefs by such reasoning, any more than we reason from sensory seemings to perceptual beliefs. Rather, we have the sensory seemings and absent certain defeaters they cause us to have the perceptual belief, and this process is sometimes misleadingly described as an inference. Similarly my belief that harming innocents is wrong might be caused by it seeming like that.
But perhaps you are proposing that this is how an ethical intuitionist would defend the moral principle that harming innocents is wrong. There is a simple reason why no intuitionist should present this as a derivation of the moral principle: the form of the argument is invalid (and hence {1,2} is not a set of descriptive propositions that entail a normative proposition).
Some ethical intuitionists explain how it is that we know moral truth in terms similar to this (but the second premiss would be that seemings are reliable, not that seemings seem to be reliable). Taken like that, it is missing an intermediate conclusion ‘2b. therefore I know that harming innocents is wrong’. The step from 2b to 3 is valid but the argument as a whole is still invalid. Furthermore, such externalist explanations of knowledge are explanations of how we know what is true, when it is true, so properly formulated it would take 3 as a premiss, and 2b would be the conclusion.
Moreover, such explanations of moral knowledge don’t entail that belief in the conclusion (3) is not a basic belief, (in the sense that basic beliefs are where the regress of justification stops). For example, if intuitionists were then to explain the seeming and the reliability in terms of conceptual competence and the nature of the concepts involved their whole story could be that ‘harming innocents is wrong’ is a self evident proposition.
Finally, you might say, great, the step from 2b to 3 shows that you can derive an ‘ought’ from an ‘is’, since 2b is prima facie a descriptive proposition whilst 3 is normative. First of all, that’s fine, since I was not trying to defend Hume’s critique but point out that it was about entailment of normative propositions rather than knowledge of them. Secondly, however, since knowledge is factive it is less clear that it is a mere quibble to point out that the proposition ‘I know that harming innocents is wrong’ contains a normative notion. The normative notion is within the scope of a factive operator and that looks like a ground for saying it is a normative proposition. Thirdly, setting that aside, 2b is still the wrong kind of proposition to embarrass Hume’s critique. He is criticising people for deriving things like ‘you ought to help Fred’ from ‘Fred is suffering’ alone. That you can derive ‘you ought to help Fred’ from ‘I know you ought to help Fred’ is beside his point.
Nick, I say your ought belief that hitting innocents is wrong is not basic, but derived from your beliefs that it seems to you that hitting is wrong, and that your seemings seem reliable indicators of moral truth. I say both of these are is beliefs? Which of these two beliefs do you say is an ought and not an is belief?
Hi Robin. I think you are conflating two quite separate issues, one is about the relation of logical consequence and the other is the basis on which we know (or justifiably believe, but for brevity I’ll speak of knowledge) ethical propositions. Either of these might be being referred to when speaking of what we can derive and the fact that knowing the conclusion of an argument depends both on the whether we know the premisses and whether the premisses entail the conclusion allows the two to be conflated by talk of derivation.
The issue of knowledge you raise is that of knowing the general ethical principles from which we derive particular ethical conclusions, such as when I infer I ought not hit Johhny because harming innocents is wrong, hitting is a harm, and Johhny is innocent. Whether I know that harming innocents is wrong may well depend on the truth about the nature of ethical intuition. If I don’t know the ethical premiss then I don’t know the conclusion despite the validity of the inference. Perhaps it is only if ethical intuition is reliable that I know the ethical premiss. But the issue over deriving ‘ought’ from ‘is’ is about whether the inference is valid.
The validity depends on the nature of the relation of logical consequence. The denial of deriving an 'ought' from an 'is' is the claim that there are no sets of purely descriptive propositions that entail a normative proposition, alternatively, that any set of propositions that entails a normative proposition contains a normative proposition. Quite clearly, the fact you mention, that whether I know an ethical principle depends on a descriptive truth about the reliability of ethical intution, has nothing to do with whether there are any sets of purely descriptive propositions that entail a normative proposition. Hence that fact is irrelevant to the question of whether we can derive an ‘ought’ from an ‘is’.
I think the problem is that Robin is confusing thought with feeling (or perhaps intuition). I don't believe torture is wrong because I believe "my intuition says torture is wrong." I believe torture is wrong because my intuition says torture is wrong!
If I can even conceive of that intuition being wrong, it's either because I have a conflicting intuition, or because the relevance of the intuition is based on false premises. In the case of conflicting intuitions, my intuition not to torture is weighed against the fact that, if I don't, a thousand people might die in a terrorist attack, and my intuition against torture might be wrong as a guide to action in this case, but not generally. In the case of the false premises, maybe it turns out that the victim of torture is not really a person, but a non-sentient automaton, and if I examine my intuitions, I discover that I'm really only against the torture of sentient beings. Yes, there are tons of "is-es" there, but it ultimately comes to an "ought" (or two).
Yes the last word of the first quote is a typo - have fixed.
An argument for an “ought” is typically built on some set of more basic “obvious” claims that the speaker assumes their audience will accept without argument. Many of those claims have their own supporting arguments somewhere else, but those arguments are also be built on further obvious claims.Eventually we end up with with a set of basic supporting claims that seem obvious, but which don’t have much in the way of explicit arguments supporting them. Yes, almost always one of these obvious but not explicitly argued claims is of the “ought” type. So in this sense every “ought” is derived from an “is.”
Did you mean to write something else? What this actually says is that every "ought" is derived from an "ought".
I’m instead pointing out that most every attempt to derive an “ought” is based ultimately on “is” claims about the reliability of our intuitions about such more basic “ought” claims. If we can’t find a coherent way to integrate these “is” claims with the rest of our network of reasonable “is” claims, then we can’t argue coherently for such “ought” claims at all.
And this says it again. I think the logical closure over what you said is this:
Every "ought" claim1. is ungrounded and ultimately meaningless, because every ought claim depends on another ought claim (you argue this twice, but it fails to play any part in the main point you are making), and2. "comes from" "is" claims, because you need to have facts to make an argument.
You have no evidence for the first claim; and the second is irrelevant. Yes, "ought" claims make use of an infrastructure of "is" claims; but that isn't what people mean when they say "ought doesn't come from is". They mean that an "ought" claim can't bottom out entirely in "is" claims. And not only did you not argue with that, you agreed with it.
Now, does that make wikipedia worthless, no; accepting it as straight fact, however, is obviously a bad idea.
Maybe, but by what standard of discernment can, or do, you make that claim? Does this suggest there is an is? From which we derive ought?
To what extent is Wikipedia and assault on 'Authority' or esp the tenure that goes with it, which is spent of lifestyles that cranks up 'relative deprivation' in those who have not secured tenure? Do education systems tend to select with a far too broad an apparatus? There will be may losers, but the winners will be of superior quality, and the 'losers' afterall, have transferable skills - or appetites of entitlement.
I think the problem with wikipedia is that it's based on the idea of crowd-sourcing, or "the average of what most people will say will probably be right". Where i think wikipedia fails is that, in fact, most people dont take the time to weigh in on topics.
Now, does that make wikipedia worthless, no; accepting it as straight fact, however, is obviously a bad idea.
I've not read this one but can imagine the gist. Take a recent article about Wikipedia, people are rightly sceptical but some.. well... I call it Wikiphobia (why the hell not eh?) People have invested so much in moral relativism that they fear the notion that a more concrete social phenomenon is forming (and yet paradoxically, I recently referred to everything as dissolving). A response to this article went: T
here are so many inaccuracies that Wikipedia is to be avoided as a research tool at all costs. For example, look at the entry for Israel. It states that the Israeli Capital is Jerusalem when all the world apart from four countries recognise it as Tel Aviv. If they can't get that right, what hope is there for the rest of their entries.
This unleashed a volley of replies along these lines
Poor example since they do have a link to a footnote which explains the situation properly.
Your post only appears to address how people behave (in a sort of pop anthropological way). You seem to be saying that because people act as if 'ought' can be derived from 'is' that it must be true.
If you're arguing something else, I've missed it either through my own poor reading or the subtlety of your argument.
(Personally, I believe I make mistakes all the time. I admit my life would much easier (though probably shorter) if I believed that the fact that I behaved a certain way meant that that was the right way to behave.
(Well, maybe not shorter. Dr. Pangloss seemed awfully hale for someone hanged, burned, and separated from this best of all possible worlds.))
I agree with the notion that "oughts derive from is". I have a blog post on the idea here:
http://blog.takeonit.com/po...
The point about the ‘is’ ‘ought’ gap is that you cannot get to an ‘ought’ merely on ‘is’ claims.
Robin Hanson presented a reason why this isn't true: in a moral argument, the specifically-ought part will be something like, "Torturing a human is wrong." -- but even that "pure ought" statement is actually based on "My intuition says torture is wrong" and "My intuitions are more often correct than incorrect", both of which are "is" statements.
Constant, my only point was that we have very pragmatic of things that 'look like stealing without being considered evil' (using the meaning of look that I think was used originally).
I agree entirely with you that with a more rigorous definition of 'look' these things are obvious not stealing.
Though to complicate things, maybe the historical (and social I think) distinction that you rightfully bring up only exists in our conscious minds, and so we might debate whether that distinction is 'physically observable.'
So a Martian anthropologist (there's no other profession on Mars except slaving away at thought experiments) can build a robust model to predict how humans will behave that distinguishes between taxes and robbery, and he might being a martianocentric thinker attribute some moral mental states to these two categories, but he wouldn't really know our mental states, whether these categories have different mental states, and even whether we have consciousness (as understood by Martians).
This is when the Martians get worried that their nearest neighbor has been overrun by qualia eating zombies...
The solipsist says: give me just one example of something that exists apart from myself. But he rejects everything offered.
That cuts quite to the heart of it, in my opinion. This has to do with what we are willing to accept as justification. I have a very hard time imagining accepting some 'is' statement as sole justification for an 'ought' statement but it might occur. The reason I can't say definitively "it will never happen" is because I haven't defined for myself what statements I'll accept as justification are/look-like.
Robin's proposed 'is' statement to justify an 'ought' statement doesn't seem like justification to me. I can't make it not be a justification for Robin or for anyone else. I can only try to get them to believe that it is not a justification...usually by showing that it contradicts some other belief that they have.
This is a very messy problem that I'd appreciate enlightenment on, but I recognize is not obviously tied in to is/ought question. I believe though in my gut that it is tied into the is/ought question, but I haven't come up with a clear and cogent argument how.
In your comparison between bananas and evil, you are comparing re-assiging the labels while retaining our original understanding of the subject matter so that it does not get re-assigned with the labels on the one hand (what you envision doing with bananas), with with re-assigning labels while also re-assigning our understanding of the subject matter so that it shifts along with the labels.
That's not a genuine comparison between bananas and evil. It's just a comparison of the two different ways you are treating the terms.
You could just have easily done the opposite: shift the meaning along with the label "banana", so that you end up trying to eat whatever new object it is you are calling a "banana", while on the other hand retaining your original moral understanding of actions while re-assigning the labels "good" and "evil" while this time treating them as empty labels (so that you punish some actions you are now labeling "good", for example).
It's not the term "banana" itself that is empty while the term "evil" is loaded. It's you who (in the hypothetical scenario) have chosen to treat "banana" as empty (as re-assignable without changing our understanding of the things being relabeled) while treating "evil" as loaded.