This seems to presuppose that all non-natural facts require NR above the functions of ordinary physical mind. Unless there is NR, then we can’t use deductive reasoning to conclude transitivity or that 1+1=2. But this seems to presuppose that only non-physical mind can perceive non-physical facts and not sufficiently deal with the fact that the functions of the physical mind can use deduction to perceive non-physical facts.
Computers wouldn’t have acquaintance knowledge without conscious experience, but they clearly have deductive powers that would allow them to infer non-natural facts. Especially those necessary facts you list which would be true in all possible worlds, that physicalism would provide no reason to doubt. Physicalism is perfectly compatible with the ontology and epistemology of non-physical facts given the deductive functions that it explains. But let's say it doesn’t that non-natural facts require NR to be justified, what would NR need to have to give it the powers to perceive non-natural facts?
The claim is that it would be a coincidence if we had knowledge of the non-physical facts when they by definition don't cause things in the physical world. Some might ground facts about the physical world, but it's not clear how a mechanistic system could know that those are grounded by the non-physical facts, and many don't--e.g. facts about possible worlds.
Non-physical facts don't need to have direct mind-independent causal properties to be known. They just need to have sufficient explanations for a belief in them to be justified. If physical mechanisms can provide these explanations, then their belief in non-physical facts would be justified. These beliefs in non-physical facts would then inform our perception of physical facts, so non-physical facts can be causal in that sense.
If we say that time exist, we don't mean it to be something physical, but a mental property generated by physical functions to be imposed on our understanding of physical things.
How can they have explanatory properties? How can the fact that pain is bad explain any facts about atoms? Or the fact that married bachelors can't exist in other possible worlds.
> Humans have the ability to reason in a way that enables us to be in some way acquainted with the truth in a way that no algorithm designed purely by natural selection could ever be.
This sounds like a bold objective claim, that there is something beyond natural selection that made humans reason the way we do. However then you clarify:
> Evolution selects for the physical systems that give rise to rationality because they are adaptive. Being able to reason well and grasp significant truths is adaptive, so it’s selected for.
So... what's the disagreement then? Reductive materialists say: we are reasoning mechanisms produced by natural selection. NR say: No, we have special non-mechanistic way to reason also produced by the natural selection. This seems to be a purely semantical confusion. Both positions seem to be in agreement on the objective matter whether our reasoning properties are produced by natural selection or not. The disagreement about the usage of terms "special" and "mechanism". But if we taboo these words then there is simply nothing to argue about. Is there?
> Humans have intuitions about a lot of domains.
All these domains can be mostly reduced to reasoning about "which conclusions follow from which premises". And then there are some small number of the initial premises that intuitively feel "right".
> On such an account, natural selection is solely responsible for imbuing us with our beliefs, wholly unrelated to their truth.
What do you mean wholly unrelated to the truth? Truth is the relation of correspondence between the world and the belief about this world. Being able to correctly reason about the world in a truth preserving manner is adaptive, therefore we have the relation between whether our "if-then" reasoning appears correct to us and it's actual correctness, through natural selection.
The only category that falls short here are ethical and aesthetic axioms. But that only means that the most naive form of moral realism is false, not that logic, mathematics, epistemology, or other ways to reason about ethics are completely unjustified for us.
> The same point can be made with reference to any of our other beliefs. For instance, take our belief that the prime minister is not a prime number.
We do not believe that prime minister not being a prime number is some kind of fundamental law of the universe. It's just the way we draw category borders. We do not even need any selection pressures here, we can easily re-classify things in such a way that prime minister is a prime number. For example, through defining "Prime minister" as being equal to 2.
> NR has a good explanation of that fact—through our ability to reason, we can simply see that the prime minister isn’t a prime number.
That's not an explanation at all. How do we know that our ability to reason is actually correlated with the truth, if we do not appeal to natural selection? How does NR allows us to evade the 8-ball scenario? Like, imagine a world where humans definetely had some intuitions which they couldn't possibly developped through natural selection, that there is definetely some "special ability to reason". And one of such intuitions is: "all of these special intuitions are true". Can we just trust it?
This is really odd - the brain evolved a lot of very useful stuff, like intuitions about social behavior, quantities, the reliability of induction and non-contradiction, because it's useful. Is it the best possible model of the world, is it metaphysically transcendent, or is it just a bag of tricks that work pretty well for us? Who knows? Maybe I don't have the right kind of brain to ever know!
I think it's true that it's wrong to torture babies and the sun will rise tomorrow. If you don't think induction is true but merely has been useful well, want to bet?
I agree it's true that torturing babies is wrong, in the same sense we probably agree spoilt meat smells foul and heights can be scary.
I don't believe classical mechanics is correct, but there aren't many circumstances where I'd bet against it. The rule of thumb that the future will resemble the past seems to work very well, and I'm the kind of creature who who's evolved a mind that intuits that as a universal truth. But if our mental model of the world turns out to be "true" in some deeper sense than "historically reliable" I don't see what that could be other than a coincidence.
It seems like there are two questions lurking here. One is whether our beliefs about non-natural facts are explained by those facts. It seems to me that the answer on either NR or ~NR is just no, if non-natural facts are causally inert. It’s not clear to me how non-natural facts are supposed to explain our beliefs on NR.
The other question is whether our beliefs about non-natural facts are reliable in some sort of probabilistic sense, like the sense captured in your premise 1. NR seems to guarantee this by basically stipulating that our non-mechanistic reasoning faculty is reliable. But stipulating reliability isn’t the same as explaining it. This stipulation doesn’t seem like a very deep advantage to me—at least without a positive reason to think our beliefs couldn’t be reliable on ~NR.
Stipulating reliability is explaining it if the basic explanation is that there is some basic faculty that enables reliability. Our beliefs about non-natural facts can be explained by the non-natural facts on NR, just as the fact that 11 is odd can explain why 11 apples can't be divided into two even groups.
I can understand how eleven being odd explains why eleven apples can’t be evenly divided into two groups, but I’m not sure that example helps me understand how non-natural facts can explain our beliefs on NR. I guess you could say that given that the non-mechanistic reasoning faculty is reliable, the truth of some non-natural fact F increases the probability that we believe that F (assuming you can make sense of the idea of probabilities conditional on the denial of F where F refers to facts like eleven being odd). But then it’s open to opponents of NR to offer their own accounts of the reliability of our mechanistic reasoning processes and argue that they can give similar explanations.
> "I think all the premises are hard to argue against."
Premise 4 is false. Intuitions aren't reasons; rather, they aim (sometimes successfully, sometimes not) to put us in touch with intrinsically credible propositions. It's those propositions (e.g. "pain is bad", and "badness isn't a physical property") that are good reasons for believing other things (e.g. that there are non-physical properties).
> "You do not get to stipulate that the process that leads to our beliefs is wildly unreliable, but then also that you get non-inferential justification!"
Who ever claimed otherwise? My response to Street's moral lottery is precisely to show how we can have *reliable* moral beliefs despite moral properties being causally inefficacious. The key is to individuate "processes" by their substantive starting points: two processes may be structurally similar while only one of the two is actually reliable, due to their different substantive starting points (e.g. whether starting from the assumption that pain is good or that pain is bad).
> "A computer can’t have direct acquaintance with any facts!"
Are you assuming that a computer can't have a mind? If the psycho-physical bridging laws connect computational states to phenomenal ones, then I don't see any in-principle barrier here. The key thing is just that direct acquaintance requires phenomenal consciousness. Once you've got that, I don't see why you'd further need your consciousness to magically push atoms around.
> "this seems to still call out for explanation."
Look up a few paragraphs to where you gave an argument for the conclusion that there's "an easy explanation" here. (The magic 8-ball analogy is terrible because the 8-ball involves a chancy mechanism, and so could not reliably track any necessary truths. Brain mechanisms need not be chancy in this way. So it's perfectly possible for brains to reliably track necessary truths, *so long as they're set up in the right way.* To check whether they are so set up, we just need to look at (1) what are the truths? and (2) Can we make sense of why belief in *those propositions* would result from natural selection etc.?)
> "The falsity of NR is often assumed but rarely argued for."
Here's an argument:
1. Zombies are possible.
2. Zombies would say the same things we do, as a result of purely mechanistic processes.
3. Given (2), the things we say are explicable as a result of purely mechanistic processes.
//Premise 4 is false. Intuitions aren't reasons; rather, they aim (sometimes successfully, sometimes not) to put us in touch with intrinsically credible propositions. It's those propositions (e.g. "pain is bad", and "badness isn't a physical property") that are good reasons for believing other things (e.g. that there are non-physical properties).//
I think that intuitions are reasons precisely because they aim to put us in touch with intrinsically credible propositions. When I say intuitions give us reasons to believe things, I mean roughly in a nearby possible world where we don't have any of those intuitions, we wouldn't have reasons to believe the propositions.
//Who ever claimed otherwise? My response to Street's moral lottery is precisely to show how we can have *reliable* moral beliefs despite moral properties being causally inefficacious. The key is to individuate "processes" by their substantive starting points: two processes may be structurally similar while only one of the two is actually reliable, due to their different substantive starting points (e.g. whether starting from the assumption that pain is good or that pain is bad).//
Yeah, reliable was probably the wrong term. The intuition is that if the reason you believe things has nothing to do with their truth, and instead is the output of a purely mechanistic process, then that doesn't enable justification--see the 8-ball example above.
//Are you assuming that a computer can't have a mind? If the psycho-physical bridging laws connect computational states to phenomenal ones, then I don't see any in-principle barrier here.//
No, I'm saying that if computers don't have minds and are purely mechanical, then they can't have minds. I'm open to the notion that the psychophysical laws would enable computers to have minds, but I claim that if they don't, then they couldn't have anything like beliefs or thoughts about anything. If the story of how you come to believe something is the same as the story of how a zombie does, but you have an epiphenomenal consciousness that just represents the information stored in the zombies' brain, then if the zombies' brain can't accurately represent information that merits non-natural beliefs, then you can't have justification for such beliefs.
//Look up a few paragraphs to where you gave an argument for the conclusion that there's "an easy explanation" here. (The magic 8-ball analogy is terrible because the 8-ball involves a chancy mechanism, and so could not reliably track any necessary truths. Brain mechanisms need not be chancy in this way. So it's perfectly possible for brains to reliably track necessary truths, *so long as they're set up in the right way.* To check whether they are so set up, we just need to look at (1) what are the truths? and (2) Can we make sense of why belief in *those propositions* would result from natural selection etc.?)//
I don't think the magic 8-ball explanation is terrible because it involves a chancy mechanism! I mean, suppose that the 8-ball didn't result from a chancy mechanism--it produced predictable outputs based on the inputs. Suppose you can even tell some just-so-story of how the 8-ball would come to say the particular things that are true. It still would seem like a massive coincidence. Like, take the case of mathematical facts or our belief that modal facts are true in all possible worlds. What's the explanation of why those would be adaptive?
//Here's an argument:
1. Zombies are possible.
2. Zombies would say the same things we do, as a result of purely mechanistic processes.
3. Given (2), the things we say are explicable as a result of purely mechanistic processes.
C. So NR is false.//
I think I might grant every premise. But even if the things we say are *explicable* as the result of a mechanistic process, that doesn't mean that we are just a mechanistic process. You could make a mechanism that mirrors the physical behavior of a non-physical mind, with very complicated physical laws, but that doesn't mean the brain has no non-mechanical faculty for recognizing truths. If you think that the brain grasps the badness of pain and then believes pain is bad because it is bad, you could still make a zombie copy that behaves identically.
Interesting points! Thanks for the detailed commentary!
One thing I'd be curious about Richard; your view seems to make all debunking arguments have no force. After all, you seem to pretty much grant the debunkers causal story but say that it lacks relevance. If this is true, then your view would rule out any targeted debunking. Is this correct?
This seems to presuppose that all non-natural facts require NR above the functions of ordinary physical mind. Unless there is NR, then we can’t use deductive reasoning to conclude transitivity or that 1+1=2. But this seems to presuppose that only non-physical mind can perceive non-physical facts and not sufficiently deal with the fact that the functions of the physical mind can use deduction to perceive non-physical facts.
Computers wouldn’t have acquaintance knowledge without conscious experience, but they clearly have deductive powers that would allow them to infer non-natural facts. Especially those necessary facts you list which would be true in all possible worlds, that physicalism would provide no reason to doubt. Physicalism is perfectly compatible with the ontology and epistemology of non-physical facts given the deductive functions that it explains. But let's say it doesn’t that non-natural facts require NR to be justified, what would NR need to have to give it the powers to perceive non-natural facts?
The claim is that it would be a coincidence if we had knowledge of the non-physical facts when they by definition don't cause things in the physical world. Some might ground facts about the physical world, but it's not clear how a mechanistic system could know that those are grounded by the non-physical facts, and many don't--e.g. facts about possible worlds.
Non-physical facts don't need to have direct mind-independent causal properties to be known. They just need to have sufficient explanations for a belief in them to be justified. If physical mechanisms can provide these explanations, then their belief in non-physical facts would be justified. These beliefs in non-physical facts would then inform our perception of physical facts, so non-physical facts can be causal in that sense.
If we say that time exist, we don't mean it to be something physical, but a mental property generated by physical functions to be imposed on our understanding of physical things.
How can they have explanatory properties? How can the fact that pain is bad explain any facts about atoms? Or the fact that married bachelors can't exist in other possible worlds.
> Humans have the ability to reason in a way that enables us to be in some way acquainted with the truth in a way that no algorithm designed purely by natural selection could ever be.
This sounds like a bold objective claim, that there is something beyond natural selection that made humans reason the way we do. However then you clarify:
> Evolution selects for the physical systems that give rise to rationality because they are adaptive. Being able to reason well and grasp significant truths is adaptive, so it’s selected for.
So... what's the disagreement then? Reductive materialists say: we are reasoning mechanisms produced by natural selection. NR say: No, we have special non-mechanistic way to reason also produced by the natural selection. This seems to be a purely semantical confusion. Both positions seem to be in agreement on the objective matter whether our reasoning properties are produced by natural selection or not. The disagreement about the usage of terms "special" and "mechanism". But if we taboo these words then there is simply nothing to argue about. Is there?
> Humans have intuitions about a lot of domains.
All these domains can be mostly reduced to reasoning about "which conclusions follow from which premises". And then there are some small number of the initial premises that intuitively feel "right".
> On such an account, natural selection is solely responsible for imbuing us with our beliefs, wholly unrelated to their truth.
What do you mean wholly unrelated to the truth? Truth is the relation of correspondence between the world and the belief about this world. Being able to correctly reason about the world in a truth preserving manner is adaptive, therefore we have the relation between whether our "if-then" reasoning appears correct to us and it's actual correctness, through natural selection.
The only category that falls short here are ethical and aesthetic axioms. But that only means that the most naive form of moral realism is false, not that logic, mathematics, epistemology, or other ways to reason about ethics are completely unjustified for us.
> The same point can be made with reference to any of our other beliefs. For instance, take our belief that the prime minister is not a prime number.
We do not believe that prime minister not being a prime number is some kind of fundamental law of the universe. It's just the way we draw category borders. We do not even need any selection pressures here, we can easily re-classify things in such a way that prime minister is a prime number. For example, through defining "Prime minister" as being equal to 2.
> NR has a good explanation of that fact—through our ability to reason, we can simply see that the prime minister isn’t a prime number.
That's not an explanation at all. How do we know that our ability to reason is actually correlated with the truth, if we do not appeal to natural selection? How does NR allows us to evade the 8-ball scenario? Like, imagine a world where humans definetely had some intuitions which they couldn't possibly developped through natural selection, that there is definetely some "special ability to reason". And one of such intuitions is: "all of these special intuitions are true". Can we just trust it?
This is really odd - the brain evolved a lot of very useful stuff, like intuitions about social behavior, quantities, the reliability of induction and non-contradiction, because it's useful. Is it the best possible model of the world, is it metaphysically transcendent, or is it just a bag of tricks that work pretty well for us? Who knows? Maybe I don't have the right kind of brain to ever know!
I think it's true that it's wrong to torture babies and the sun will rise tomorrow. If you don't think induction is true but merely has been useful well, want to bet?
I agree it's true that torturing babies is wrong, in the same sense we probably agree spoilt meat smells foul and heights can be scary.
I don't believe classical mechanics is correct, but there aren't many circumstances where I'd bet against it. The rule of thumb that the future will resemble the past seems to work very well, and I'm the kind of creature who who's evolved a mind that intuits that as a universal truth. But if our mental model of the world turns out to be "true" in some deeper sense than "historically reliable" I don't see what that could be other than a coincidence.
It seems like there are two questions lurking here. One is whether our beliefs about non-natural facts are explained by those facts. It seems to me that the answer on either NR or ~NR is just no, if non-natural facts are causally inert. It’s not clear to me how non-natural facts are supposed to explain our beliefs on NR.
The other question is whether our beliefs about non-natural facts are reliable in some sort of probabilistic sense, like the sense captured in your premise 1. NR seems to guarantee this by basically stipulating that our non-mechanistic reasoning faculty is reliable. But stipulating reliability isn’t the same as explaining it. This stipulation doesn’t seem like a very deep advantage to me—at least without a positive reason to think our beliefs couldn’t be reliable on ~NR.
Stipulating reliability is explaining it if the basic explanation is that there is some basic faculty that enables reliability. Our beliefs about non-natural facts can be explained by the non-natural facts on NR, just as the fact that 11 is odd can explain why 11 apples can't be divided into two even groups.
I can understand how eleven being odd explains why eleven apples can’t be evenly divided into two groups, but I’m not sure that example helps me understand how non-natural facts can explain our beliefs on NR. I guess you could say that given that the non-mechanistic reasoning faculty is reliable, the truth of some non-natural fact F increases the probability that we believe that F (assuming you can make sense of the idea of probabilities conditional on the denial of F where F refers to facts like eleven being odd). But then it’s open to opponents of NR to offer their own accounts of the reliability of our mechanistic reasoning processes and argue that they can give similar explanations.
I'm not sure what the mystery is supposed to be. We have some basic reasoning faculty that allows us to grasp various nonnatural facts.
> "I think all the premises are hard to argue against."
Premise 4 is false. Intuitions aren't reasons; rather, they aim (sometimes successfully, sometimes not) to put us in touch with intrinsically credible propositions. It's those propositions (e.g. "pain is bad", and "badness isn't a physical property") that are good reasons for believing other things (e.g. that there are non-physical properties).
> "You do not get to stipulate that the process that leads to our beliefs is wildly unreliable, but then also that you get non-inferential justification!"
Who ever claimed otherwise? My response to Street's moral lottery is precisely to show how we can have *reliable* moral beliefs despite moral properties being causally inefficacious. The key is to individuate "processes" by their substantive starting points: two processes may be structurally similar while only one of the two is actually reliable, due to their different substantive starting points (e.g. whether starting from the assumption that pain is good or that pain is bad).
> "A computer can’t have direct acquaintance with any facts!"
Are you assuming that a computer can't have a mind? If the psycho-physical bridging laws connect computational states to phenomenal ones, then I don't see any in-principle barrier here. The key thing is just that direct acquaintance requires phenomenal consciousness. Once you've got that, I don't see why you'd further need your consciousness to magically push atoms around.
> "this seems to still call out for explanation."
Look up a few paragraphs to where you gave an argument for the conclusion that there's "an easy explanation" here. (The magic 8-ball analogy is terrible because the 8-ball involves a chancy mechanism, and so could not reliably track any necessary truths. Brain mechanisms need not be chancy in this way. So it's perfectly possible for brains to reliably track necessary truths, *so long as they're set up in the right way.* To check whether they are so set up, we just need to look at (1) what are the truths? and (2) Can we make sense of why belief in *those propositions* would result from natural selection etc.?)
> "The falsity of NR is often assumed but rarely argued for."
Here's an argument:
1. Zombies are possible.
2. Zombies would say the same things we do, as a result of purely mechanistic processes.
3. Given (2), the things we say are explicable as a result of purely mechanistic processes.
C. So NR is false.
Thanks for the reply!
//Premise 4 is false. Intuitions aren't reasons; rather, they aim (sometimes successfully, sometimes not) to put us in touch with intrinsically credible propositions. It's those propositions (e.g. "pain is bad", and "badness isn't a physical property") that are good reasons for believing other things (e.g. that there are non-physical properties).//
I think that intuitions are reasons precisely because they aim to put us in touch with intrinsically credible propositions. When I say intuitions give us reasons to believe things, I mean roughly in a nearby possible world where we don't have any of those intuitions, we wouldn't have reasons to believe the propositions.
//Who ever claimed otherwise? My response to Street's moral lottery is precisely to show how we can have *reliable* moral beliefs despite moral properties being causally inefficacious. The key is to individuate "processes" by their substantive starting points: two processes may be structurally similar while only one of the two is actually reliable, due to their different substantive starting points (e.g. whether starting from the assumption that pain is good or that pain is bad).//
Yeah, reliable was probably the wrong term. The intuition is that if the reason you believe things has nothing to do with their truth, and instead is the output of a purely mechanistic process, then that doesn't enable justification--see the 8-ball example above.
//Are you assuming that a computer can't have a mind? If the psycho-physical bridging laws connect computational states to phenomenal ones, then I don't see any in-principle barrier here.//
No, I'm saying that if computers don't have minds and are purely mechanical, then they can't have minds. I'm open to the notion that the psychophysical laws would enable computers to have minds, but I claim that if they don't, then they couldn't have anything like beliefs or thoughts about anything. If the story of how you come to believe something is the same as the story of how a zombie does, but you have an epiphenomenal consciousness that just represents the information stored in the zombies' brain, then if the zombies' brain can't accurately represent information that merits non-natural beliefs, then you can't have justification for such beliefs.
//Look up a few paragraphs to where you gave an argument for the conclusion that there's "an easy explanation" here. (The magic 8-ball analogy is terrible because the 8-ball involves a chancy mechanism, and so could not reliably track any necessary truths. Brain mechanisms need not be chancy in this way. So it's perfectly possible for brains to reliably track necessary truths, *so long as they're set up in the right way.* To check whether they are so set up, we just need to look at (1) what are the truths? and (2) Can we make sense of why belief in *those propositions* would result from natural selection etc.?)//
I don't think the magic 8-ball explanation is terrible because it involves a chancy mechanism! I mean, suppose that the 8-ball didn't result from a chancy mechanism--it produced predictable outputs based on the inputs. Suppose you can even tell some just-so-story of how the 8-ball would come to say the particular things that are true. It still would seem like a massive coincidence. Like, take the case of mathematical facts or our belief that modal facts are true in all possible worlds. What's the explanation of why those would be adaptive?
//Here's an argument:
1. Zombies are possible.
2. Zombies would say the same things we do, as a result of purely mechanistic processes.
3. Given (2), the things we say are explicable as a result of purely mechanistic processes.
C. So NR is false.//
I think I might grant every premise. But even if the things we say are *explicable* as the result of a mechanistic process, that doesn't mean that we are just a mechanistic process. You could make a mechanism that mirrors the physical behavior of a non-physical mind, with very complicated physical laws, but that doesn't mean the brain has no non-mechanical faculty for recognizing truths. If you think that the brain grasps the badness of pain and then believes pain is bad because it is bad, you could still make a zombie copy that behaves identically.
Interesting points! Thanks for the detailed commentary!
One thing I'd be curious about Richard; your view seems to make all debunking arguments have no force. After all, you seem to pretty much grant the debunkers causal story but say that it lacks relevance. If this is true, then your view would rule out any targeted debunking. Is this correct?
good work on physicalism.