Can naturalists believe in free will, even if it is compatibilist free will?
C. S. Lewis wrote:
Thus no thoroughgoing Naturalist believes in free will: for free will would mean that human beings have the power of independent action, the power of doing something more or other than what was involved by the total series of events. And any such separate power of originating events is what the naturalist denies. Spontaneity, originality, action “on its own” is a privilege reserved for “the whole show” which he calls Nature.
The reason Lewis seems to be offering for saying that the Naturalist must deny free will doesn’t seem to mainly be that if Naturalism were true, determinism would be true, but he seems rather to be saying that free will involves a kind of independent agency on the part of persons that would be proscribed given naturalism.
In a footnote, John Beversluis replies as follows:
Some contemporary naturalists, for example, Daniel Dennett, John Searle, Jaegwon Kim, and Keith Parsons, reject determinism not only on the level of microparticles but generally and argue that naturalism is compatible with believing that human beings have free will.
I am not sure about these philosophers, and what kind of free will these people believe in. Students of the free will question know that there are two conceptions of free will: a conception compatible with determinism, and a concept that is incompatible with determinism. Daniel Dennett wrote an entire book, Elbow Room: The Varieties of Free Will Worth Wanting, which is well known as a classic defense of the compatibilism. Parsons, however, has argued the compatibility of libertarian free agency with naturalism.
Eventually, I would like to consider the question of whether a thoroughgoing naturalism is compatible with the incompatibilist or libertarian conception of free will. For the purposes of this discussion, however, I want to concede, for the sake of argument, that compatibilism is true, and I will try to show that it is far from clear that a thoroughgoing naturalism is really compatible with free will.
Compatibilist theories of free will trade on the idea that even if determinism is true, the proximate cause of an action can be one’s desire to perform the action. A compatibilist or soft determinist will say emphasize the fact that if you did something, if it is free in the compatibilist sense, you did what you wanted to do. If, say, you robbed the local Bank of America branch, it is not likely to be true that you wanted to be a law-abiding citizen, but the fickle finger of fate grabbed you by the scruff of the neck and made you commit a crime. No, you robbed the bank because, in the words of Willie Sutton, “That’s where the money is."
But notice what is implied in these kinds of theories. First, in order for this theory to be true, desires have to exist. There are naturalistic theories of mind, eliminativist theories, according to which desires are the posits of “folk psychology” and do not in fact exist. Now eliminativists do maintain that a matured neuroscience will replace the terms of folk psychology with successors, but will can the compatibilist theory be fitted in with a successor? Have eliminativists even addressed this issue?
But suppose we accept the existence of desires. In order for the compatibilist theories to work, the desires have to be causally efficacious. It must be the case that my desire for X can cause my action in pursuit of X. But, of course, naturalistic theories of mind, given their commitment to the causal closure of the physical, inevitably face the specter of epiphenomenalism. That is, even if it is thought that beliefs and desires exist on the hypothesis of naturalism, (which, as I have indicated in a previous post, typically involves a commitment to a causally closed mental-free realm and the bottom of everything), how can it be that my desire can cause anything? In other words, in order for a naturalist to even accept a compatibilist theory of free will, they must solve the problem of mental causation. William Hasker and I have argued that naturalists cannot solve the problem of mental causation, and if I have been right in my discussions here, they cannot consistently even believe in compatibilist free will, much less incompatibilist or libertarian free will.
Labels: Beversluis, freedom and determinism, Naturalism
posted by Victor Reppert @ 10:29 AM
50 Comments
50 Comments:
I think you're attacking a straw man version of naturalism.
Particle physics is not only mental-free, it's also planet-free. That doesn't stop plate tectonics from being causally efficacious. Planets (or any other collective phenomenon I could name) are not "epiphenomenal" just because they supervene upon a deeper physics.
I expect that 99% of naturalists would agree with me that minds are machines, that machines are not epiphenomenal, and that intentionality is not basic. That's a solid naturalist position, even for unconventional naturalists like myself.
DL: Particle physics is not only mental-free, it's also planet-free. That doesn't stop plate tectonics from being causally efficacious. Planets (or any other collective phenomenon I could name) are not "epiphenomenal" just because they supervene upon a deeper physics.
VR: But with planets, it looks as if you have a fairly clean reduction from planet-talk to particle talk, In other words, if you were physically oomniscient, and you knew what a planet was, it would be evident what planetary truths were true. Physical facts don't seem to me to leave planetary facts indeterminate.
When you get to the mental, for various reasons, I don't think the same holds. It you really think the mental reduces to the physical then you believe that the normative reduces to the nonnormative, the intentional reduces to the nonintentional, the the logical reduces to the nonlogical, the purposive reduces to the nonpurposive, the perspectival reduces to the nonperspectival. The machine analogy, in my book, is an extremely popular way to fudge over a whole raft of important distinctions. Many naturallist philosophers are uncomfortable with the machine analogy, which is why they have proposed non-reductive forms of materialism and have to constantly battle the problem of epiphenomenalism. Even eliminativism suggests that the mental-to-physical reductions aren't going to work, he difference here is they blame the reductive failure on the insufficiency of "folk psychology" rather than on any kind of profound ontological problem.
I don't know why you think this is a straw man. I am pointing to problems for a naturalistic account of action that are acknowledged to be problems by naturalistic philosophers from Jaegwon Kim to David Chalmers to Colin McGinn.
What I was pointing out here is that unless the problem of mental causation has a solution, naturalists cannot even claim that we have compatibilist free will, much less libertarian free will. Obviously if you think there are no problems, since you think the machine analogy works just fine, then this shouldn't trouble you. But if it does work just fine, then a lot of naturalistic philosophers have been troubling themselves about nothing.
There's another big problem when you try to compare minds to machines. Saying that a computer has beliefs seems sensible if you're enough of a functionalist. Saying a machine has desires just seems totally insane.
VR: What I was pointing out here is that unless the problem of mental causation has a solution, naturalists cannot even claim that we have compatibilist free will, much less libertarian free will.
Sorry, I was a bit off course there. I think you're probably right on that point, but let's not pretend that naturalism is in retreat or backed against the wall. Alternative naturalist philosophers like Chalmers do not represent the mainstream viewpoint, and I personally think their notoriety is due to the novelty of their views, not the number of their converts. I could be wrong on this point, but I doubt it.
VR: If you really think the mental reduces to the physical then you believe that the normative reduces to the nonnormative, the intentional reduces to the nonintentional, the the logical reduces to the nonlogical, the purposive reduces to the nonpurposive, the perspectival reduces to the nonperspectival.
I do.
I had a good blog conversation with Darek some time ago about some of these issues (the logical, the normative). Darek's attack was based on the claim that, since all rational thinkers agree they ought to think logically, therefore, logical laws cannot rationally be on the same status as natural laws because of the naturalistic fallacy. However, we are not so much agreeing on "laws of thought" as "mores of thought", so the argument really doesn't go anywhere. Assuming rational thinking assumes a morality of thought, so the argument begs the question.
That is, there is only a problem if you think that morality is absolute, and that, for example, humans absolutely ought to think logically. But humans only ought to think logically if they want to make optimal survival use of their faculties, and that's not an absolute.
I have yet to see any compelling argument on the other grounds (perspective, intention, and purpose).
For example, there are arguments that go: X is a member of some set of physical things, X is not about anything (e.g., X = a floppy disc, or X = a sundial, or X = Deep Blue, etc.). Therefore, no physical X can be about anything. Of course, this is a non sequitur.
The deepest flaw in these arguments is that recognition and prediction abilities are necessary for intention (even human intention), so choosing X where X has no relevant capability is a bit like saying that stuffed owls can't fly, therefore owl flight is supernatural.
VR: But if it does work just fine, then a lot of naturalistic philosophers have been troubling themselves about nothing.
Actually, that's exactly what I think. :)
VR: Saying a machine has desires just seems totally insane.
Why? Do dogs or mice have desires? Are they supernatural or do they have dualistic minds? I see no such necessity. Desire is a matter of preference, and I suspect that even earthworms have preferences of some sort.
We haven't really made any machines with desires, but that isn't a proof that it can't be done or that we aren't machines ourselves. So your insanity comment strikes me as an argument from personal incredulity.
Victor, if brain states were identical to mental states would this remove the causal closure road block to mental causation?
Well, there are two types of identity claims. One is token identity, and the other is type identity. Suppose I throw the baseball that Luis Gonzalez hit to win the 2001 World Series and it strikes the window of my house. The ball has the property of being the ball Gonzo hit, but that property is irrelevant to whether it will break the window or not. The window will break, or not break, depending on the physical structure of the window, that of the ball, and the trajectory of the ball. Gonzo's having hit is will be irrelevant. One thought has to cause another thought in virtue of its content. So the mental state can be identical to the brain state, and yet the mental content be causally irrelevant and epiphenomenal.
DL
>>That is, there is only a problem if you think that morality is absolute, and that, for example, humans absolutely ought to think logically. But humans only ought to think logically if they want to make optimal survival use of their faculties, and that's not an absolute.<<
OK, instead of "absolute," let's say that the existence of correct versus incorrect ways of arriving at conclusions cannot be inferred from physical facts. That brain processes in humans can generate more or less adaptive behaviors, on the other hand, can be inferred from physical facts. One cannot be inferred, the other can, therefore they are not equivalent.
I should hardly have to point out that any inference from facts assumes the reality of correct versus incorrect ways of arriving at conclusions, and therefore the reality of such ways cannot itself be inferred.
Do you really mean to say that the arguments you express here do not assume the objective reality of correct ways of arriving at conclusions? If they do not, why should you or any of the rest of us even bother?
Darek,
I should hardly have to point out that any inference from facts assumes the reality of correct versus incorrect ways of arriving at conclusions, and therefore the reality of such ways cannot itself be inferred.
I agree that the rules of rationality cannot be rationally proven, and must be assumed before they can be used.
However, while we believe the rules of rationality, they are just assumptions. They are not knowledge because they aren't (and cannot be) justified rationally.
Nor are the rules of rationality irreducible. Rational inference is just a kind of event that occurs on a rational landscape, i.e., in a world where there is some set of rules and, therefore, a law of non-contradiction.
We rational folk assume that our world is a rational landscape. We come to believe this assumption instinctively, perhaps with psychological reinforcement from all of the experiences in which rationality seems to work. However, the notion of rationality is an abstraction. Sort of like the way we look at instances of cats and define the universal "catness". This abstraction does not justify a claim that since catness is an abstraction beyond any particular cat we have seen (or ever will see), catness denies the possibility that there is a finite number of particular cats and that's all. A Platonist could assume that universals exist independently of particulars, but that is just an aesthetic assumption.
Yet your conclusion is similar: that we cannot possibly live in a particular rational landscape (i.e., our physical world) because we regard the abstraction "rational landscape" to be prior to any information about the rationality of our particular world.
So, Darek, I disagree with you on several grounds.
First, we could not know what a rational landscape was unless we were living in one, and were filled with particular experiences of it. I conclude that the very notion of a rational landscape is derived from the particulars of our local landscape.
Second, the rationality of our landscape is not a law. It is an abstracted property of all lawful worlds. (Catness is not a cat.) It does not have a life of its own just because we thinking machines abstract it above from the particulars (or potential particulars) of our universe.
Finally, we have to remember that we don't know the universe is a rational landscape. We assume it. It is not knowledge because it is not justified belief. Any rational justification would be circular. Indeed, this was our starting point. Rationality is not a discovered law because it's not a law at all.
Doctor Logic, help me here. Logic (it seems to me) is a mental representation of reality and we know that reality is not identical to representations. The mental state has intentionality; it’s hard to see that rocks, brain states, etc are in logical relations to anything. What says you?
Victor, let me understand this first. Suppose we grant that mental and brain states are identical, are we admitting that neuronal structures have propositional content in which case seems to entail that brain states have intentionality?
normajean,
The mental state has intentionality; it’s hard to see that rocks, brain states, etc are in logical relations to anything. What says you?
How do you know that the mental state has intentionality? Suppose you are roused from sleep, and have a thought, but have reason to believe that the thought was confused, and possibly not about anything sensible. There would have to be some criteria by which you would judge the thought sensible. My claim is that you know your thought was about an object if you would recognize the object when you saw it (or thought of it again).
For example, suppose I think of what a rabbit looks like. The thought about a rabbit is not the same as seeing the rabbit. Yet the thing that makes my thought about the rabbit is my ability to recognize the experience about which I was thinking. When I see a rabbit, I will be able to say, "Yes! This is what my thought was about."
So it would seem that intentionality is intimately connected with recognition.
Now, rocks can't predict or recognize, so obviously, they cannot be about other things.
However, brain states are most certainly predictors and recognizers. The brain is one big prediction/recognition machine.
So a brain state can be about an external object when it correlates with the way the brain will recognize that object. And since the brain also reflects back on itself and can recognize internal states, it can also have thoughts about thoughts.
DL, rabbits are contingent aspects of reality; logic is a different kuppajoe whose normative qualities hardly look natural.
normajean,
rabbits are contingent aspects of reality; logic is a different kuppajoe whose normative qualities hardly look natural.
What do you mean by "logic" here?
1) Do you mean that every thinking thing ought to think logically?
I don't see that logic has any normative qualities of its own.
All I see is that there are survival advantages to rational thought on a rational landscape, and our evolved, subjective preferences reflect that desire to survive. We ought to think logically if we think we ought to survive. Neither logic nor survival are absolutely normative.
2) Do you mean that the universe must have no contradictions?
For a world to be intelligible, it must be logical, and that's why you regard logic as being necessary (in contrast with the contingency of the rabbit). So the abstraction of a logical universe is driven by its necessity for intelligibility. But if our world is physical, we already have intelligibility.
It's sort of like you're asking for something sweet, but rejecting my sugar cookies because they are merely one physical instance of sweetness.
For the record, I disagree that we have to simply assume our inference rules are right. That kind of foundationalism seems wrong to me. We can justify inference rules as long as we don't use rule X to justify rule X. E.g., if we observe that people using certain transitions among sentences that lead to false sentences, then we have reason to reject that inter-sentence inference rule.
E.g.,
If A, then B.
B.
Therefore, A.
Why is the above not a good rule? Do you need to understand formal logic to say why it isn't a good rule? If you were interested in truth-preserving transitions among sentences, would you be able to come up with an example that showed why the above was not good?
Propositional logic is not something humans are good at naturally. It has to be learned.
I think the idea that logical reasoning should be axiomatized was a human intellectual watershed, but for a few reasons it makes people think that the mind works via axioms and rules that are somehow written into our brains.
Note this is all getting off the topic of Victor's post. His point is that unless naturalists can get causally efficacious semantic contents, then compatabilism will not work. I think this is an interesting argument. It may even be true (but the antecedent is false).
DL
>>First, we could not know what a rational landscape was unless we were living in one, and were filled with particular experiences of it. I conclude that the very notion of a rational landscape is derived from the particulars of our local landscape.<<
You conclude? But you can have reached this conclusion correctly only on the assumption that there are correct (as opposed to incorrect) ways of reaching conclusions.
You seem to acknowledge that the reality of correct versus incorrect ways of reaching conclusions cannot be inferred. But we have no choice other than to acknowledge it as real in order to continue this discussion. Simply by making arguments you are concurring in the assumption.
Certainly, scientific facts are inferred from empirical evidence. Therefore the existence of correct versus incorrect ways of reaching conlusions is not a scientific fact. You seem both to agree and disagree here.
That which is a fact, but not a scientific fact, cannot be explained in terms of scientific facts. If it could be explained in terms of such facts, we could derive it from them. But we cannot do so.
Therefore, all facts that we both acknowledge as such are not scientific facts or explainable in terms of scientific facts.
BDK:
I am asserting the necessary assumption of correct versus incorrect ways of reaching conclusions. This does not mean, granting that assumption, that particulars about correct versus incorrect ways cannot be inferred. Of course they can. But such inferences about the particulars will always be found to assume correct versus incorrect ways of arriving at conclusions. I dare say that any argument you may respond with will make that assumption, too.
While this subject is a bit off the specific topic of Victor's post, DL raised it and it is on topic for this blog as a whole.
Darek,
Therefore the existence of correct versus incorrect ways of reaching conclusions is not a scientific fact. You seem both to agree and disagree here.
First, the "existence of correct versus incorrect ways of reaching conclusions" is not a fact. It does not have the same status as, say, quantum mechanics. It is an axiomatic assumption, and one that cannot be rationally proven.
Second, you misunderstood my point. My fault, perhaps, for not being more clear.
What I am saying is this. On a particular rational landscape, a rational thinker will (by definition) assume that rationality works despite lacking the ability to prove it, and that thinker will find that assumption bears fruit. This mirrors our own experience.
And all this is without the need for there to be some overarching metaphysical law that all landscapes be rational. All we need to avoid contradicting our assumptions of rationality is to exist on a particular rational landscape. We don't need an ontologically independent Platonic universal for "rational landscape". We just need one particular rational landscape. And physicalism provides for that particular.
You say that we know we are on a rational landscape independent of scientific methods. But we don't know it, we merely assume it. The assumption of rational landscape is necessary for intelligibility, and physicalism provides for intelligibility. There's only a problem for physicalism if a contradiction arises between it and rationality, but you haven't found one.
BDK,
For the record, I disagree that we have to simply assume our inference rules are right.
I think I agree with you. The things that we need to assume are more basic than the example you give. Subject to the constraint that there are no contradictions, there are all sorts of logics we can invent. However, I think that the assumption that we ought to deal in a space of facts without contradictions is fundamental to all possible logics one might invent. The logics are just the ways we might generically relate facts without contradiction.
Note this is all getting off the topic of Victor's post. His point is that unless naturalists can get causally efficacious semantic contents, then compatabilism will not work. I think this is an interesting argument. It may even be true (but the antecedent is false).
Agreed.
DL & BDK
Perhaps not so far off the topic as it seems.
The Argument from Reason is one of three arguments that C. S. Lewis sketched that in my view can be grouped under an umbrella that we could call the Argument from
Normativity. It turns on the necessary assumption of three
norms:
The Norm of Rationality.
My own description, as I have already stated, is that there are correct, as opposed to incorrect, ways to arrive at conclusions.
The Norm of Morality
There are ethically right and wrong ways for intelligent beings to think and behave.
The Norm of Intelligibility
There are intelligible, as opposed to unintelligible, ways for physical processes to proceeed.
None of these norms are testable/verifiable. The first must be assumed for the sake of anaytical thought and discussion. Most of us assume the second in the course of interpersonal relations.
Science assumes the third because it is necessary to the value of the scientific method. Note that Quantum Theory, for example,
although counterintuitive is intelligible to the extent that it is experimentally verifiable. Hume gets credit for showing that this norm, too, must be assumed, although he referred to it as uniformity rather than intelligibility.
Notice that the first two norms need not be adhered to all of the time or even most of the time. We assume that physical processes adhere to the third at least most of the time. The first two norms therefore entail a kind of autonomy of the human mind that the third does not. This is one way to think of the conflict between mechanism and will--an autonomy that may be difficult to precisely describe but that enters into relations between the human mind and the first two norms, yet seems to be excluded from processes subject only to the third norm.
DL
>>First, the "existence of correct versus incorrect ways of reaching conclusions" is not a fact.<<
>>You say that we know we are on a rational landscape independent of scientific methods. But we don't know it, we merely assume it.<<
To say that we assume there are correct versus incorrect ways of arriving at conclusions is just to say that we assume the factuality of the proposition that there are correct versus incorrect ways. (To save my fingers, I'm just going to say "the norm of rationality" for correct and incorrect ways.)
By providing me with what you clearly feel are conclusions correctly arrived at, you are assuming/stipulating the factuality fo the norm of rationality. This reminds me of the observation of A. J. Ayer that to assert that a proposition is true is equivalent simply to asserting the proposition and vice versa.
What is the procedure for arriving at the conclusion 'The computer is on my desk' when I am sitting here looking at the computer on my desk?
Darek,
Notice that the first two norms need not be adhered to all of the time or even most of the time. We assume that physical processes adhere to the third at least most of the time. The first two norms therefore entail a kind of autonomy of the human mind that the third does not.
Not at all. The third norm is a norm held by us, not by the physical world. I can easily violate that norm by believing in supernatural causation. So it's not that the world believes it ought to act lawfully, but that we thinkers think it ought to.
By referencing Ayer, you warm the cockles of my heart. :) However, I do think there's a distinction between an assumption and a fact. IMO, proven theorems, direct experiences, and inductive inferences are facts, but assumptions are not. However we decide to define a fact, there's a definite distinction between knowledge and assumption.
BDK
A conclusion is the result of of a process of inference, I'd say.
If it's merely a perception, it need not be a conclusion.
Is your point that perceptions are not conclusions or that the norm of rationality is open to question? If that is your point is that the norm of rationality is open to question, is it fair to call that a conclusion you have reached?
DL
>>However, I do think there's a distinction between an assumption and a fact. IMO, proven theorems, direct experiences, and inductive inferences are facts, but assumptions are not.<<
Are "proven theorems" proven only on the assumption that there are correct as opposed to incorrect ways of arriving at conclusions? If so, then they are assumptions themselves, aren't they? A proof that depends upon an assumption is no better than the assumption, right? Even an "inductive inference" is the result of a process of inferring (i.e., a conclusion), otherwise it cannot be called an inference. And inferences are dependent upon the norm of rationality.
DL
Sorry, I realize now that I neglected your first point.
To say that we assume the norm of intelligibility is to say we assume that nature adheres to it, not to say that our thought processes adhere to it as nature does. For the sake of science we have no choice but to assume that nature can be investigated systematically; as C. S. Lewis noted, we are repelled by the idea of nature being any other way. We assume, in other words, that nature substantially plays by rules we can understand--or that beings of sufficient intelligence could understand. But this is primarily an assumption about the way nature proceeds rather than the way our thought processes proceed (unless you are Kant!). We assume that nature continues to "play by the rules" even when we loose our power to attend to nature or reason about it.
Darek,
A proof that depends upon an assumption is no better than the assumption, right?
In general, this isn't true. I have extremely high confidence in the theorems of Euclidean geometry, but I don't have the same confidence in the axioms of Euclidean geometry. So I can have very high confidence in the fact that "if A then B" without having as high a confidence in "A".
I admit that the assumption of rationality is a special case. However, I still say it's distinct from knowledge. All justification relies on the assumption of rationality, and so it cannot be justified without our using X to prove X.
But this is primarily an assumption about the way nature proceeds rather than the way our thought processes proceed (unless you are Kant!). We assume that nature continues to "play by the rules" even when we loose our power to attend to nature or reason about it.
I disagree completely on this. In fact, I think the norm of intelligibility is part of the norm of rationality. There's no rationality without the same kind of inductive inference as is required by the sciences.
To highlight the problem more specifically, let's look at the analogues. For the norm of rationality, we think there are logical facts whether we reason to those facts or not. For the norm of intelligibility, we think there are scientifically reachable physical facts whether we scientifically reach them or not. I don't see any significant difference. The only distinction is an artificial/irrelevant one between the way we reach physical facts and the way we reach mental facts.
DB said:
A conclusion is the result of of a process of inference, I'd say.
If it's merely a perception, it need not be a conclusion.
I'm not just talking about seeing my computer on the table, but the judgment that my computer is on the table that I can make when I see my computer on the table (e.g., someone asks "Where is your computer?" and my room was just rearranged, I look around, and see it on the table and answer "On the table."). There is no conscious inference going on, but the judgment, at least apparently, is justified.
I've called such judgments 'conclusions' but perhaps I should just say 'perception-based judgments.' (As opposed to just 'perceptions' as that is so ambiguous in the literature--it can mean so many things, and doesn't always even imply a judgment).
DL
>>In general, this isn't true. I have extremely high confidence in the theorems of Euclidean geometry, but I don't have the same confidence in the axioms of Euclidean geometry. So I can have very high confidence in the fact that "if A then B" without having as high a confidence in "A".<<
When you express your confidence in tested scientific hypotheses or inductive inferences, you are expressing confidence in B, not just in "iff A then B." Obviously, given "iff A then B," our confidence in B itself is a function of our confidence in A, isn't it? If your confidence in tested hypotheses (the equivalent of B) is dependent upon there being correct ways to arrive at conclusions (the equivalent of A), you must endorse the factuality of A at least to the same extent you do that of B and "iff A then B."
"There are correct as opposed to incorrect ways to arrive at conclusions." This is a proposition, and it is intelligible, whether or not its assumption is necessary. It is not like the propositions that the number thirteen weighs 4 oz and that the Moonlight Sonata is the color of ripe mangoes. It is an intelligible proposition and therefore a candidate for facthood. You cannot say, "Well, my definition of fact is a scientific fact. The norm of rationality is not a scientific fact and therefore I refuse to acknowledge it as a fact." The question before us is whether all interesting, intelligible facts are scientific facts. You cannot call into question the factuality of the norm of rationality without at the same time casting doubt upon knowledge--including most scientific knowledge--that presupposes it.
>>To highlight the problem more specifically, let's look at the analogues. For the norm of rationality, we think there are logical facts whether we reason to those facts or not. For the norm of intelligibility, we think there are scientsifically reachable physical facts whether we scientifically reach them or not. I don't see any significant difference.<<
The norms are related, of course. But they are different norms. This is easily seen by their definitions according to process: there is a correct way for the process of arriving at conclusions to proceed; there are intelligible ways for physical processes to proceed. If I analyze a leaf, I always expect to see photosynthesis respecting the norm of intelligibility by obeying the laws of chemistry. If I listen to my young sons explaining their reasons for answering a question a certain way, I am prepared to find that their thinking disregards the norm of rationality fairly often. (And judging by the epidemic of cheating on college exams and terms papers, if nothing else, the norm of morality gets disregarded a good deal of the time in our society.)
Note carefully that while the norm of intelligibility must be assumed, just how this intelligibility is manifested--the particulars--must be inferred from evidence. As Einstein said (sorry for repeating this), the rule-governed character of nature must be assumed, but exactly what the rules are must be determined by the scientific method (I'm paraphrasing).
The reason the above point is critical is that the norm of rationality is often explained as one of the "particulars" of the norm of intelligibility. But it cannot be explained that way because the "particulars" of intelligibility are inferred from empirical evidence and the norm of rationality cannot be so inferred. It truly is a "norm," unavoidably presupposed.
BDK
>>(e.g., someone asks "Where is your computer?" and my room was just rearranged, I look around, and see it on the table and answer "On the table."). There is no conscious inference going on, but the judgment, at least apparently, is justified.<<
Admittedly, there is a fuzzy border between raw percepts and conclusions. If there is any reflection involved, then I think that boundary is crossed. If you don't see your computer and merely glance around, a conclusion is not in view (probably). But if you look for your computer in the usual place and don't see it and then say (or think), "Someone must have moved my computer," the judgment is reasoned and not merely reactive.
Darek,
It is an intelligible proposition and therefore a candidate for facthood.
There's a difference between propositions and facts. Are you saying that confidence in a proposition bestows facthood upon it? Or is it rational (and perhaps experiential) justification that bestows facthood?
Under normal circumstances, we can say
1) If A1, A2, A3, then B1, B2.
2) B1, B2
3) We are more confident in A1, A2, A3.
However, if we number the axioms of rationality in A1, A2, or A3, then we violate the axioms by employing circular logic. We would be using the prniciples of inference to prove the principles of inference.
So, while observing {B1, B2} might lead us to regard {A1, A2, A3} as factual, it cannot rationally lead us to believe that our methods of inference are factual without assuming those methods.
As you say, the assumptions of rationality are special, and it cannot be validated scientifically. But it is also the case then that it cannot be rationally validated by experience whatsoever. I believe this is Hume's point writ large.
The reason the above point is critical is that the norm of rationality is often explained as one of the "particulars" of the norm of intelligibility.
I disagree that this is what people are doing. Cognitive science isn't proving rationality. It's proving that machines can reason according to those assumptions.
The chain of reasoning is like this. First, we assume rationality, i.e., methods of inference, induction, factuality of experience. Second, we create scientific models of the world as mandated by the prior assumption. Third, we find mechanisms that reason to conclusions that are valid according to our basic assumptions. As long as we don't try to use those discovered mechanisms to prove our basic assumptions, everything is good. We will successfully explain how matter can form systems that reason, even if we cannot explain why reason is the way it is.
The reason I kept bringing up perception-based judgments it that I think such judgments are typically justified, but not by any conscious or explicit argument. They are justified because they are produced (in the right way) by a properly functioning visual system which is properly connected to my "belief" box. I could imagine a monkey making almost exclusively perception-based judgments.
I bring this up because I think the emergence of logic owes a lot to such perception-based judgments. Note I am not advocating a hardcore empiricist foundationalism here--I am not claiming that all knowledge is grounded in such judgments, but I think the recent trend away from foundationalist justification and toward coherentism has perhaps made people focus too much on infra-lingual modes of justification. Obviously, the externalists in epistemology are pushing a programme that talks a lot about this.
And, of course, it definitely isn't foundationalist because I base my belief in good perceptual systems partly on my study of the neuroscience and psychology of perceptual systems, and have reason to believe my perceptual system is normal. So one could say there are Quine-Duhem type concerns to address with this whole reliablist, externalist perspective on justification.
On the other hand, I think a monkey could have a perception-based judgment that it couldn't justify, it knows nothing about perceptual systems. It doesn't need to have access to the reasons that its perception-based belief is justified. That doesn't mean such beliefs are not justified!
Again, this has nothing to do with Victor's argument, but was a main thread in these comments I wanted to finish.
Perhaps I could say that my theory of meta-justification is more coherentist, but my theory of justification has reliablist roots (quasi-foundationalist).
DL
>>There's a difference between propositions and facts. Are you saying that confidence in a proposition bestows facthood upon it?<<
No, it doesn't "bestow facthood," but the confidence may be due to the factuality of the proposition.
Some propositions cannot be assigned either true or false designations. "The taste of lemon is louder than the smell of grilled steak," for example, or "The note B-flat is heavier than A-natural." The norm of reason is not like that. True and false mean something with respect to it. This is entirely apart from the issue of its justification.
>>Under normal circumstances, we can say
1) If A1, A2, A3, then B1, B2.
2) B1, B2
3) We are more confident in A1, A2, A3.
However, if we number the axioms of rationality in A1, A2, or A3, then we violate the axioms by employing circular logic. We would be using the prniciples of inference to prove the principles of inference.<<
Good point. But I didn't intend to claim that our confidence in B1, B2justifies our confidence in A.
What I said was that our confidence in B1, B2 is a function--an indicator, if you will--of our confidence in A1, where A1 is the norm of rationality. We can't coherently claim that we have higher confidence in deliberations of science that depend upon the norm of reason than we do in the norm itself.
>>So, while observing {B1, B2} might lead us to regard {A1, A2, A3} as factual, it cannot rationally lead us to believe that our methods of inference are factual without assuming those methods.<<
Again, the way I would put this is that our regarding consequents of the norm of reason as factual is incoherent apart from our regarding the norm as factual or real. There is no question of trying to justify justification.
>>I disagree that this is what people are doing. Cognitive science isn't proving rationality.<<
I didn't say that cognitive science is trying to justify rationality. Some thinkers and researchers are, however, effectively trying to explain rationality in scientific terms. The title of Churchland's _The Engine of Reason, the Seat of the Soul_ says a lot. Or look at Damasio's _Descartes' Error_ (pp. 198-201, for example). I have no beef with the fascinating results of neural research. But the impression that some naturalists convey is that the heart pumps blood, the stomach digests, the liver produces bile and the brain reasons--the difference in the last case being merely the complexity of the function. If you ask a naturalistic cogsci type, "Is reason a biological function?," or "Can science explain our faculty of reason," they will, I think, say yes.
>>As you say, the assumptions of rationality are special, and it cannot be validated scientifically.<<
We seem to agree here. The norm of reason, "There are correct versus incorrect ways of reaching conclusions," is not inferred from empirical evidence, it is not a scientific fact. Now here is something that follows: That which is not a scientific fact/hypothesis cannot be explained in terms of scientific facts/hypotheses. Here's another: Our awareness of that which is not a scientific fact cannot be scientifically described.
>>It's proving that machines can reason according to those assumptions.<<
Well, hold on. Computers produce symbolic results that accord with reason because of our awareness of the norm of reason, not because of their own awareness of it. Let me draw an analogy from another norm, the moral one. Let's say that I am aware that cheating on exams is morally wrong, but I have a student who is so thoroughly amoral that he thinks there is nothing wrong with cheating on exams if the cheater can avoid getting caught. So I set up an exam with heavy penalties for cheating and keep a close eye on the student, prompting him to take the exam honestly. The student's behavior has been modified by the norm of morality, but indirectly, because of my own awareness of the norm as applied to cheating--but not because of his own awareness of it.
Likewise, because of our awareness of the norm of reason, which cannot have a scientific explanation, we produce computers that function a certain way.
But here's something else to consider. If the norm of reason is real, but cannot be scientifically explained, then maybe the norm of morality is real in spite of being scientifically inexplicable. Note, I did not say that apparently moral behavior is inexplicable, but behavior that is motivated by awareness that there are right and wrong ways for humans to behave. Finally, if there is a source for norms like those of rationality, morality, intelligibility, perhaps even aesthetics, then that source lies outside the world that science is capable of describing, i.e., the natural world.
I love reading this back and forth. Any good book recommendations that cover a lot of the bases ur touching?
Good work all!
Two points:
1. Explaining animal cognition scientifically is not the same thing as explaining rationality. We can describe how humans, dogs, monkeys think, the transitions among thoughts or whatever while staying completely neutral about the rationality question.
2. I don't think rationality is an 'assumption' of science. It is one of many empirically tuned, evolving sets of standards about what types of inferences are allowed.
What is rational is something that is learned, that is fought for over historical time. Inference rules evolve, different fallacies recognized or hedged (e.g., affirming the consequent is not typically a fallacy within a Bayesian framework).
I should say that something like consistency or law of the excluded middle seems pretty central (though there are paraconsistent logics that toy with this idea to help deal with liar paradoxes and the like).
Gila Sher says the most interesting things on this topic, for instance in Is logic a theory of the obvious? and Logical Consequence: An epistemic outlook. This stuff is definitely not my specialty, but her work seems to be the most reasonable and has a really nice description of logic as supplying rules for inference that are invariant to permutations of the objects over which the statements apply (and briefly discusses how what counts as invariant to such permutations can change over time).
BDK
>>2. I don't think rationality is an 'assumption' of science. It is one of many empirically tuned, evolving sets of standards about what types of inferences are allowed.<<
Do you mean by this that scientists and philosophers of science do not assume that there are correct as opposed to incorrect ways of arriving at conclusions? I think you have your work cut out for defending that as a conclusion correctly arrived at.
You'll notice that my "norm of rationality" statement does not necessarily rule out alternate logics, but it seems to require that they be capable of being argued for. An argument is an invitation to see that the arguer's conclusions are correctly arrived at.
Mavericks like Wittgenstein and Quine made a case for their views. Gila Sher does the same. So do advocates of paraconsistent logics. Theoretical physicists continue to make a case for their views as being conclusions correctly arrived at.
We will know that the norm of rationality has been challenged when the latest theoretical model of physics is "explained" and "advocated," not by papers containing propositional statements and mathematical formulas, but by free form sculptures and drip paintings.
DB: OK you are saying something less provocative than I originally thought. It seems nothing interesting really follows from the claim that scientists (and everyone that uses language) thinks there are better and worse ways to attempt to justify something. This shouldn't distinguish the naturalists and non-naturalists (for reasons I gave above, there are perfectly naturalistic ways to sculpt specific inference rules).
normajean: I am kind of out of touch with the general literature on this, and anything I'd suggest would probably be outside the mainstream (e.g., Churchland's book 'Engine of Reason Seat of the Soul' is biased toward one naturalistic position). Fodor hasn't written any popularizations. Probably any introductory philosophy of mind book would be good. Actually, come to think of it I liked Jaegewon Kim's introduction, found here.
It is hard to find a good, readable overview of philosophy of mind, and especially one that would touch on the issues we are discussing here as it is considered an "advanced" topic (intersection of philosophy of mind and epistemology).
Oh, and I really liked this introduction to epistemology. Very clear and basic.
BDK
>>It seems nothing interesting really follows from the claim that scientists (and everyone that uses language) thinks there are better and worse ways to attempt to justify something.<<
We disagree here, fundamentally. Unless by "better and worse ways" you mean nothing more than ways that each scientist happens to prefer to others. As Nagel says, we must be talking about something objective that transcends individual or cultural preference. Which brings us to normativity and its problems for naturalism.
Thanks, BDK, very much!
DB: I already gave examples of how this is a lot more trivial than you might think (the affirming the consequent example).
Just restating it, once a language-user is concerned with truth (a property of sentences), there is nothing magical about developing the goal of encouraging intralinguistic transitions that will preserve this property and trying to weed out other patterns. It can be treated as just another property out in the world.
(For those wondering what I mean by 'truth' I just mean weather a sentence is true or false, e.g., 'I am nine feet tall' is false, and 'I am over six feet tall' is true--true statements say of what is, that it is, and what is not that it is not (that's Aristotle's nice definition)).
Certain means to this end employing truth-preserving transitions are certainly better than others, but certain ways of constructing a paper weight are better than others too but this doesn't imply naturalism is false or that we need any metaphysical appendages to build paper weights. So you either need to argue that when the property in question is the truth of a sentence, things become very unique and special, or that building a paper weight has important metaphysical consequences.
I think neither is true, and that is more likely the real disagreement.
BDK
Whether or not conclusions correctly arrived at can be false, it is definitely the case that conclusions incorrectly arrived at can be true. So the question is not simply one of the truth values of statements.
>>Certain means to this end employing truth-preserving transitions are certainly better than others, but certain ways of constructing a paper weight are better than others too but this doesn't imply naturalism is false or that we need any metaphysical appendages to build paper weights<<
Can it be inferred that there are better as opposed to worse ways of building paper weights without assuming that there are? If so, then the analogy is not apt.
Paper weights built not using the principles of paper-weight design can also work well as paper weights. Obviously we want to discover the principles of good paper-weight making, the non-accidental features that make a good paper weight. We can do the same for truth-preserving sequences of sentences.
Can it be inferred that there are better as opposed to worse ways of building paper weights without assuming that there are? If so, then the analogy is not apt.
I can observe that certain inference rules work but others don't (hence my focus on observation judgments above) at least as well as I can observe whether a particular paper weight works. I don't have to assume that there are good inference rules to do this, just as I don't have to assume that there exist good paper weights in order to observe that one paper weight is better than another.
I think we are likely to not agree on this :)
Hi Darek,
To the extent that we can speak of the norm of rationality as a proposition, we could probably say that we have confidence in it. However, that confidence doesn't make it knowledge, and doesn't make it factual. If it is a proposition, it is an axiom necessary for propositions.
And this leads us right back to cognitive science. We are not really aware of the norm of reason in propositional form. It's more like we cannot help ourselves. It's instinctive for us to be rational, or at least partially so. We are aware of consistency and aware of routes to solutions in the same way we are aware of breathing or seeing things.
Is there a norm of breathing? Does it have a truth value? Given the assumption that we ought to breath, lots of conclusions follow, but that doesn't make the norm of breathing a fact or a truth.
The main difference between us and computers is that we are rationally aware of our instincts and sensations. A contemporary computer doesn't know that it sees, it simply sees. It's quite plausible that squirrels are similar - they see, but don't know that they see. Likewise, a contemporary computer doesn't know that it reasons, it just does. So while computers are not yet programmed to be reflective like we are, that's not a proof that they cannot be reflective. You are employing an "All ravens are black" argument against non-black ravens. You are saying that since we have not programmed reflective computers yet, it cannot be done. That's not a compelling argument at all. There's no reason at all to rule out the possibility of reflective thinking machines.
I have a bit more to explore on this topic, but time is limited right now. Back soon!
BDK
>>I can observe that certain inference rules work but others don't<<
I believe that concealed in the term "observe" here is the process of arriving at a conclusion in a way presumed to be correct.
You are right. We are not going to agree.
DL
>>However, that confidence doesn't make it knowledge, and doesn't make it factual.<<
1) Scientific facts are knowledge.
2) Humans know most or all scientific facts by correctly arriving at conclusions.
3) Therefore, "knowledge" of scientific facts qualifies as such only to the extent that there are, in fact, correct as opposed to incorrect ways of arriving at conclusions.
I don't see what is so incomprehensible about this.
Normajean
If you are still following this thread, I honestly don't know of books that take up this subject from precisely this angle. There is an interesting article by Hilary Putnam, "Why Reason Cannot be Naturalized" in a book titled _Realism and Reason_ (1983). And there is C. S. Lewis material, though it doesn't go into great technical depth (as I don't either).
Thanks, Darek, I am watching---You guys are solid!
I believe that concealed in the term "observe" here is the process of arriving at a conclusion in a way presumed to be correct.
Again, we talked about observation judgments above where you seemed to be OK with what I said other than me calling them 'conclusions.' Now you are saying we should say they are conclusions. Fine. I don't care what we call it.
So consider again, upon looking around my office, I notice a computer on my desk, and tell someone:
X: 'There is a computer on my desk' X is true, justified, and something I did not infer consciously. It is because you complained about my calling such observation judgments 'conclusions' that I switched to the more neutral 'observation judgments.'
A child that has no epistemological theory, that has no idea what justification is, could have thought X. But it would be a true thought, justified, and a belief. That is, it knows there is a TV on the table.
Regardless of my argument that a child with no epistemology training can know X, would you agree with the conclusion that a naive child can have such knowledge (i.e., that observation judgments can be knowledge)?
I'm not asking for an analysis of why they are knowledge, but if you think they do not constitute knowledge, then we definitely will not get any further. If you do there might be interesting questions at that point. So, can a philosophically naive child know X in the way I've described?
Hi BDK,
Sorry for joining so late, and I admit I haven't read through all of this post, but, your comment here is interesting:
"A child that has no epistemological theory, that has no idea what justification is, could have thought X. But it would be a true thought, justified, and a belief. That is, it knows there is a TV on the table."
It seems you are suggesting that the child has knowledge in this case. Is that what you are saying? This has long been a standard objection to the view of knowledge as justified true belief, dating back to Plato at least. Namely, the fact that a child (in the case of Plato), or a computer in our age, or an adult who utters a justified sentence in another language but has no understanding of it, proves that these things don't have knowledge when they utter the sentence, since they are just spouting off something they don't understand. It is epistemic luck for them to have said it. In the same way as it is epistemic luck to spin until you are dizzy and fall to the ground while pointing and say 'New York is in that direction', and you are right, that is not knowledge, you just got lucky.
So, are you suggesting that you can have knowledge in these circumstances of uttering these sentences by luck, though you have no understanding of what you are saying, or how the justification follows?
Cheers.
BDK
As I tried to communicate up above somewhere, I don't think this is an all-or-nothing question. A glass of water sits on the table. An ant crawling across the table encounters the glass and goes around it. Does the aunt "know" that the glass is there? Since the ant is sentient in some infinitesimally small way, perhaps in some proportionately meager sense it knows about the cup.
A thirsty child runs into the room, and glances quickly at the glass. Does the child know that the glass is there? Obviously, she does to a greater degree than the ant. Has she "concluded" that the glass is there? Hard to say. But suppose I ask her, "What is that on the table?" and she answers, and then I follow up, "Are you sure that's what it is?" If the child is capable of stopping and evaluating her answer with some care, we can answer with greater assurance that a process of arriving at a conclusion is underway.
I'd say we are well across the border by the time we get to recursion, which was the point at issue. You had brought up paraconsistent logical systems, and I commented that those who advocate them invite us to see their opinions as conclusions correctly arrived at.
Likewise, if you say that you can "observe" the results of different inference rules, can you blame me for understanding you to mean that you can correctly arrive at conclusions about those rules?
That certain laws of nature obtain and that certain physical processes occur in accordance with those laws, to the extent that these are items of human knowledge, are conclusions. And we have confidence in them to to the extent that we consider them to be conclusions correctly arrived at from empirical evidence. But something is clearly wrong with claiming that we are confident that "there are correct and incorrect ways of arriving at conclusions" to the extent that we consider that to be a conclusion correctly arrived at. It's like throwing a rope ladder up in the air and then trying to climb it before it falls.
Rino and DB:
Wow you are making this much more complicated than it should be. My claim is that it is possible for an epistemically naive (i.e., he thankfully knows no epistemology) child to know, based on seeing the TV on the table and judging based on this perception that the TV is on the table, that the TV is on the table.
I'm not talking about an ant. I'm not talking about a kid on acid or with a martian up his butt. I'm talking about a normal kid who walks in, sees the TV on the table, calls his dad and says "Hey the TV is on the table." No conscious inference, just sees it and judges it correctly.
As I already explicitly said he knows the TV is on the table. Just read my previous post as I basically just repeated it deflecting your hems and haws.
You are acting all suspicious and hedgy. Don't worry this isn't like one of those born again christian novelty whoopie cushion arguments where I walk you through trivial seeming conclusions and then spring on you that you have just proven god exists.
BDK
It's also not complicated that "know" has a certain ambiguity in our usage. There are levels of "knowing" from casual, perhaps even subconscious recognitions to studied judgments. Wherever you begin on the scale, you can't bootstrap up to the norm of rationality in a way that is amenable to naturalistic explanation of that norm.
BTW, I should have restated a distinction between analyses of rules of inference and the norm of correct as opposed to incorrect ways of arriving at conclusions. As I said before, just as we must assume that nature is lawlike but must investigate the specific shape of this "lawlikeness," something similar may be said for the norm of rationality versus the elaborations of formal logic.
DB: I've given the example, and you continue to flapperate. Good day sir!
Seriously, this discussion has run out of steam so I'm cutting myself off.
Post a Comment
Subscribe to Post Comments [Atom]
<< Home