Epistemology’s False Dichotomy

 Consider the following quote by William Hasker: 

“Epistemo­logical externalism has its greatest plausibility in cases where the warrant for our beliefs depends crucially on matters not accessible to reflection­, for instance, on the proper functioning of our sensory capacities. Rational inference, in contrast, is the paradigmatic example of a situation in which the factors relevant to warrant are accessible to reflection; for this reason, examples based on rational insight have always formed the prime ex­amples for internalist epistemologies.”

Hasker’s comment highlights a key distinction between internalism and externalism in epistemology. For many internalists, warrant must be something that is subjectively available to the believer. It is something the believer can reflect upon and that allows them to make sense of why they form the beliefs that they do (and why they are right to do so). By contrast, for many externalists, a belief can be warranted whether or not a subject is able to recognize or reflect upon what warrants their belief. 

Insofar as these positions are understood as competing views about the conditions under which beliefs are warranted, their merits can be straightforwardly compared with one another. However, it is not clear to me that this is how we should understand these positions. 

Consider Descartes’s reflections on doubt. When Descartes realized that, for all he could subjectively establish, many of his beliefs might have been false, he was plagued by doubt. This sent him in search of a method that would enable him to overcome his doubt and rest his beliefs on a secure foundation. In Descartes’s case, what was at issue was not whether his beliefs were warranted or not. For all he knew, his beliefs may very well have been warranted if externalism is true. Rather, Descartes wondered how he could possibly maintain his beliefs in light of his inability to subjectively establish their truth. It would be of no benefit to Descartes to suggest that his beliefs might be warranted by being the results of reliable cognitive faculties. If we cannot maintain beliefs in the absence of subjectively accessible warrant, the possibility of inaccessible warrant won’t provide any assistance in the Cartesian struggle against doubt. 

In my opinion, the case of Descartes provides a key insight into the debate between internalists and externalists in epistemology: The reason why internalists see externalism as a non-starter is because, unless the warrant for one’s beliefs is subjectively accessible, one will not be able to maintain those beliefs at all. Insofar as a subject realizes that they cannot recognize any basis for thinking that their beliefs are true, they will simply refrain from holding those beliefs. This truth, according to internalists, is self-evident upon reflection to any rational thinker.

In light of this internalist position, the externalist insistence on the possibility of subjectively inaccessible warrant is worthless as a response to reflective doubt. The internalist basis for rejecting externalism is not simply the fact that externalist standards for warrant are too low. Rather, it is the fact externalist standards for warrant apply to beliefs that are not possible to maintain upon reflection. The disagreement between internalists and externalists is not about the conditions under which a belief has some commonly agreed upon status called “warrant”. Internalists are concerned with warrant in the sense that subjects can have upon reflection and externalists are not. To this extent, internalists and externalists are talking past one another. 

Addendum: 

Externalists often oppose this kind of internalist position on the grounds that our judgements simply have to bottom out in beliefs that lack subjectively accessible warrant. According to these externalists, perceptual judgements and other basic beliefs cannot be formed on the basis of further subjectively accessible reasons. Consequently, unless we wish to deny the possibility of perceptual judgements that can withstand rational reflection, we must accept the that beliefs can be maintained with subjectively inaccessible warrant.

However, this worry rests on the assumption that beliefs are the only source of subjectively accessible warrant. Internalists can reject this assumption and recognize perceptual experiences, in addition to beliefs, as conceptual states capable of providing subjectively accessible warrant. This enables the internalist to insist that subjectively accessible warrant is necessary for maintaining beliefs upon reflection while also recognizing the possibility of basic beliefs.

An argument from reason against mind-body physicalism

When one believes something on the basis of certain reasons, one takes those reasons to explain why one’s belief is true. For example, if I believe that Socrates is a mortal because all men are mortal and Socrates is a man, I do not simply think that my belief is causally explained by those reasons. Rather, I also take those reasons, insofar as they are the causes of my belief, to explain why my belief that Socrates is a mortal is true. If I did not take those reasons to explain why my belief that Socrates is a mortal is true, I would not believe that Socrates is a mortal at all. 

If the same reasons that cause my beliefs also explain why my beliefs are true, then the content of my beliefs cannot depend on anything other than the character of the reasons that cause them. If they depended on anything further, then the fact that certain reasons caused my beliefs would not also be sufficient to explain the truth of those beliefs. Instead, an explanation of the truth of my beliefs would depend upon the nature of the further factor that influenced the content of my beliefs in addition to the reasons that caused me to form them. This would prevent me from from ever recognizing my beliefs as true by recognizing my reasons for forming them. 

If mind-body physicalism is true, then our beliefs are physical phenomena that are governed by the laws of nature and that stand in causal relations with other physical phenomena. These laws determine the causal relations that obtain between our beliefs and the causes that explain them. Furthermore, these laws are not sensitive to rational relations of implication or entailment. If this is the case, then the truth of our beliefs does not simply depend on the causes of those beliefs. Rather, it also depends on the natural laws of our universe. Specifically, it depends on whether the natural laws of our universe entail causal relations that ensure the truth of beliefs that are caused by the appropriate reasons. This is precisely the kind of further factor that prevents a subject from ever recognizing their beliefs as true on the basis of their reasons for forming them. In order to know whether my beliefs were true, in this case, I would first have to determine whether the laws of nature were of the appropriate sort. But any belief about the laws of nature would itself have to be recognized as true on the basis of the reasons that explained it. This is impossible unless I already take the contents of my beliefs to be explained solely in terms of the character of the reasons that cause them. Consequently, a belief in mind-body physicalism precludes the rational possibility of recognizing the truth of our beliefs, including our belief in mind-body physicalism.

In short, if the reasons that cause our beliefs also explain their truth, then the contents of our beliefs do not depend on anything other than the character of the reasons that cause them. But if mind-body physicalism is true, then the contents of our beliefs do depend on something other than the character of the reasons that cause them. So if mind-body physicalism is true, then it is not the case that the reasons that cause our beliefs also explain their truth. The truth of this last consequent is incompatible with the possibility of rational warrant for one’s beliefs. Therefore, mind-body physicalism is incompatible with rational warrant for one’s beliefs, including one’s belief in mind-body physicalism.

Addendum:

While the argument above is an objection to mind-body physicalism alone, related (though more controversial) considerations will also apply to anti-physicalist alternatives. 

If the mind is a physical or a non-physical substance, then there will necessarily be some truths that it cannot know. In particular, it will be incapable of knowing its own nature precisely because the truth of any belief that a mind has about its own nature will depend upon that nature itself. In order to know that the reasons that cause my beliefs also explain their truth, I must have some understanding of the nature of belief and the mind. In particular, I must know that beliefs are such as to be caused by reasons that ensure their truth. But this knowledge is precisely what is unavailable to me insofar as the mind is a physical or non-physical substance. I have no basis for trusting any belief that I form about the nature of a mental substance because the recognition of any trustworthy basis would depend upon knowledge of the very mental nature that I am trying to grasp. If I believe that a mental substance normally forms true beliefs, I can only trust this judgment insofar as I know that my beliefs are responsive to reality. But this is precisely what I need to establish in the first place. 

I can only know the nature of my mind insofar as I know that this nature is simply what i think it to be. Insofar as it is anything other than what I think it to be, its nature lies beyond me and I cannot know that beliefs are such as to be responsive to objective reality. I do not know the mind to be a physical or a non-physical substance in thought. I do not know it to be any substance at all. Therefore, if the mind is a physical or a non-physical substance, I cannot know its nature and I cannot know my beliefs to be warranted. But I do know my beliefs to be warranted because I know that the mind is no physical or non-physical substance.

Comments On Physicalism And Vagueness

- Comment 1:

Some philosophers have suggested that our difficulties with conceiving of borderline cases of consciousness stem from our temptation to accept a paradoxical demand. Specifically, when we try to conceive of a borderline case of consciousness, they claim that we try to conceive of what it would be like to be in such a state. However, if there is something it is like to be in a certain state, that state just is a determinate case of consciousness. So, according to this proposal, we are trying to conceive of a borderline case of consciousness in a way that prevents us from actually doing so, whether or not they are possible. But this proposal mischaracterizes the relevant situation. When we attempt to conceive of a borderline case of consciousness, we try to imagine a relation to (or a state that is related to) a borderline case of phenomenal character. This is not an attempt to conceive of a borderline case of consciousness in terms of what it is like. Therefore, the attempt does not present any paradoxical demand. On a related note, it may be true that, if we were in such a borderline state, we would not be able to determinately recognize it as such (because we would not be clearly conscious). But one can surely recognize the character of a state that one conceives of, even if one could not recognize the character of that state while the state was actually occurring.

- Comment 2:

According to some philosophers, in order to competently grasp a given concept, we must be able to conceive of determinate cases where the referent of that concept is present or absent and, if the concept is vague, borderline cases as well. Since we seem to be unable to conceive of relevant borderline cases for our Consciousness concept (which we seem to competently grasp), should we conclude that consciousness is sharp? I don’t think so. Consider the case of the concept Life. In the past, when people took our Life concept to be sharp, it is plausible to suppose that they would not take any conceivable situation to involve a borderline case of life. However, it seems wrong to suggest that they did not competently grasp the relevant concept (if there is, in fact, such a thing as competence with respect to our “grasp” of concepts). Rather, these past subjects simply came to make discoveries about the nature of life that led them to realize that borderline cases of life are possible. Their prior inability to recognize borderline cases of life was caused by false beliefs about life, not conceptual incompetence. We could similarly imagine a subject who came to hold strange beliefs about the sufficient conditions for life that specified sharp boundaries for its presence. It seems that this subject would also take borderline cases of life to be inconceivable. Nevertheless, despite their strange beliefs, there is no reason to suppose that the subject would lack conceptual competence instead of simply having misleading beliefs. Consequently, we should deny that conceptual competence with vague concepts implies an ability to conceive of relevant borderline cases (or at least cases that one would recognize as borderline).


- Comment 3:

Normally, if a concept is vague, we can conceive of situations in which a borderline case of its referent is present. What explains why this is true? Without offering a general account, the following hypothesis suggests itself: When we have a competent grasp of a given empirical concept, we can conceive of the sorts of situations in which the evidence that leads us to apply that concept is present. If an empirical concept is vague, it will normally be the case that the evidence that leads us to apply that concept can come in greater or lesser degrees. Consequently, if we competently grasp a vague empirical concept, we should normally be able to conceive of a situation in which the relevant evidence is present to such a degree that we recognize a borderline case of that concept’s referent. This proposal provides a satisfying explanation of why we view the inconceivability of borderline cases as evidence for their metaphysical impossibility: If we can conceive of the sorts of situations in which the evidence that leads us to apply a vague empirical concept is present, in most if not all cases, that evidence will come in greater or lesser degrees. So if we are unable to conceive of such evidence in greater or lesser degrees that would lead us to recognize a borderline case, this is a defeasible reason to conclude that the concept is sharp. However, this principle runs into trouble when we consider the case of consciousness.

When we conceive of conscious experiences, we apply our Consciousness concept on the basis of the phenomenal character of our experience. Crucially, we do not use this concept to refer to the phenomenal character of our experience. Rather, when we recognize the phenomenal character of our experience, this is our evidential basis for concluding that we are having some kind of conscious experience. In this sense, our introspective judgments about consciousness are displaced insofar as they are made on the basis of something we are conscious of rather than an awareness of consciousness itself. Unlike ordinary cases, however, it is not obvious that we should expect this evidential basis to come in greater or lesser degrees. For example, suppose that strong property representationalism is true. In this case, the phenomenal character of an experience will be an abstract property complex that the experience represents. Can an abstract property complex be present in greater or lesser degrees? It is hard to even make sense of this suggestion. First, a property complex is an abstract object, so the degree to which it is present cannot vary between situations. Second, abstract objects are plausibly fundamental in a sense that precludes being present to a certain degree at all. Perhaps it could be argued that, when there is a borderline case of the representation relation, an experience only makes a property complex somewhat present for the subject. In this scenario, it might be claimed, we have a borderline case of a property complex. However, a borderline presentation of a property complex does not entail any presentation of a borderline property complex. As argued above, there are no possible borderline property complexes to be presented. Consequently, according to at least one theory of phenomenal character, we should not expect our evidential basis for applying our Consciousness concept to come in greater or lesser degrees. Most importantly, this would be true even if borderline cases of consciousness are, in fact, possible. Because the absence of variation by degrees in our evidential basis for applying our Consciousness concept does not imply that consciousness itself is sharp, our inability to conceive of situations where we recognize a borderline case of consciousness is not a good reason to conclude that borderline cases of consciousness are metaphysically impossible.

- Comment 4:

According to the explanatory proposal offered above for the conceivability of borderline cases, when we competently grasp a vague concept, we should normally be able to conceive of situations in which our evidence for applying that concept is present in greater or lesser degrees. According to many non-representationalist theories of consciousness defended by contemporary physicalists, our Phenomenal Character concept is vague. Consequently, these proposals suggest that we should be able to conceive of situations where our evidence for phenomenal character is present to greater or lesser degrees. Tye, Antony, and others have persuasively argued that this is not the case. Furthermore, unlike in the case of consciousness, the absence of variation by degrees in our evidential basis for applying our Phenomenal Character concept is not compatible with its being vague. This is because, if we apply our Phenomenal Character concept on the basis of our direct acquaintance with phenomenal character itself (or some underlying physical property) and phenomenal character admits borderline cases, we should be able to imagine situations where our evidence establishes the presence of phenomenal character to a greater or lesser extent. Therefore, if we endorse the explanatory proposal described above, we should conclude that phenomenal character is sharp and reject any physicalist proposal that suggests otherwise (this would include the views proposed by David Papineau and Ned Block among others). As noted previously, this implication does not threaten strong property representationalism.

Conceivability, Zombies, And Vagueness

Among philosophers, it is often suggested that there is some link between conceivability and metaphysical possibility. Most commonly, this suggestion takes the form of the following principle: If P is conceivable, then P is metaphysically possible. This principle is sometimes qualified so as to specify that the link in question is defeasible. In other words, if P is conceivable, this provides defeasible evidence for the claim that P is metaphysically possible. 

A stronger form of this alleged link between conceivability and possibility has been proposed along the following lines: If P is conceivable, then P is metaphysically possible. Furthermore, if P is not conceivable, then P is metaphysically impossible. The latter half of this stronger principle can also be qualified in terms of defeasible evidence.

These principles underlie two key arguments against physicalism in the philosophy of mind:

1. It is conceivable that all of the actual physical facts obtain in the absence of any mental facts.

2. If it is conceivable that all of the actual physical facts obtain in the absence of any mental facts, then it is possible that all of the actual physical facts obtain in the absence of any mental facts.

3. Therefore, it is possible that all of the actual physical facts obtain in the absence of any mental facts.

This is the so-called “zombie argument” against physicalism. 

The second argument runs as follows:

1. If mental states/properties are physical states/properties, then it is possible that there are borderline cases of mental states/properties.

2. It is inconceivable that there are borderline cases of mental states/properties.

3. If it is inconceivable that there are borderline cases of mental states/properties, then it is not possible that there are borderline cases of mental states/properties.

4. Therefore, it is not possible that there are borderline cases of mental states/properties.

5. Therefore, mental states/properties are not physical states/properties. 

This is the so-called “vagueness argument” against physicalism. 

Premise 2 of the zombie argument and premise 3 of the vagueness argument are both supported by the conceivability principles described above. Consequently, in order to evaluate these arguments, we should evaluate the conceivability principles that support them. 

Turning first to the weaker conceivability principle, one potential motivation for accepting that conceivability is evidence for metaphysical possibility is its intuitive plausibility. It seems to us that what we conceive of being the case is what is metaphysically possible. However, how should we understand problematic cases such as the apparent conceivability of a world where water is not composed of H20 (it has been persuasively argued by Saul Kripke that such a world is metaphysically impossible)? In such cases, Kripke and others have suggested, what we really take to be conceivable is a world where some watery stuff is not composed of H20, rather than a world where water is not composed of H20. Consequently, the apparent contingency of water’s identity with H20 can be explained away in terms of a world where the contingent properties we normally associate with water are not instantiated by H20. According to this explanation, conceivability is a genuine guide to metaphysical possibility and metaphysically impossible worlds are inconceivable. 

However, what explains why we take certain possibilities to be conceivable at all? Without offering a general account, the following hypothesis suggests itself: When we normally apply two empirical concepts on the basis of independent evidence (where the evidence that leads us to apply one concept is not necessarily associated with evidence that leads us to apply the other concept in the same situation), we can conceive (or take ourselves to conceive) of a world in which the evidence that leads us to apply one concept is present in the absence of the evidence that leads us to apply the other. For example, since our H20 concept is applied on the basis of H20’s presence while our water concept is applied on the basis of some watery stuff’s presence, we can conceive of a world where some watery stuff is present in the absence of H20. Likewise I can conceive of a world where a journalist at the Daily Planet is present in the absence of a flying superhero with a red cape. This latter case gives rise to the apparent contingency of Clark Kent’s identity with Superman. 

The hypothesis described above provides a satisfying explanation for why we take conceivability to be a guide to metaphysical possibility: If the evidence that leads us to apply one empirical concept is necessarily associated with the evidence that leads us to apply another concept, in most cases, we will apply both concepts in the same epistemic situations. Consequently, if we recognize that we do not apply both concepts in the same epistemic situations, this is a good reason to conclude that their referents are only contingently correlated. However, this principle runs into trouble when we consider the case of mind-brain identity. 

When we apply our C-fiber concept and our pain concept, we do so on the basis of different evidence. We apply our C-fiber concept on the basis of the presence of C-fibers. By contrast, we apply our pain concept, directly, on the basis of our experience of pain and, indirectly, on the basis of the behavior of other organisms. According to the hypothesis under consideration, this will entail that we can conceive of a world where pain is present in the absence of C-fibers. But unlike ordinary cases where we apply our empirical concepts, our C-fiber and Pain concepts are applied on the basis of two radically distinct forms of epistemic access to reality. Specifically, our Pain concept is applied on the basis of direct experiential acquaintance while our C-fiber concept is not. For this reason, even if the evidential bases for applying these concepts were necessarily associated, we would still not expect them to be applied in the same situation. This is because, in one situation, we may be directly acquainted with pain without being in a situation where we would apply our C-fiber concept. Likewise, we could be in a situation where we would apply our C-fiber concept without being directly acquainted with pain. Because such cases are possible, the fact that we do not apply our Pain and C-fiber concepts in all of the same situations is not a good reason to conclude that their referents are only contingently correlated. Consequently, if the hypothesis presented above is accepted, we have a good reason for rejecting premise 2 of the zombie argument. 

Turning now to the stronger conceivability principle, one motivation for accepting that inconceivability is a guide to the metaphysical impossibility of borderline cases is the simply fact that, normally, if a concept is vague, we can conceive of situations in which a relevant borderline case of its referent is present. What explains why this is true? Without offering a general account, the following hypothesis suggests itself: When we have a competent grasp of a given empirical concept, we can conceive of the sorts of situations in which the evidence that leads us to apply that concept is present. If an empirical concept is vague, it will normally be the case that the evidence that leads us to apply that concept can come in greater or lesser degrees. Consequently, if we competently grasp a vague empirical concept, we should be able to conceive of a situation in which the relevant evidence is present to such a degree that we recognize a borderline case of that concept’s referent. 

The hypothesis described above provides a satisfying explanation of why we take the inconceivability of borderline cases as evidence for their metaphysical impossibility: If we can conceive of the sorts of situations in which the evidence that leads us to apply a vague empirical concept is present, in most cases, we will be able to conceive of situations in which that evidence is present to greater or lesser degrees. If we are unable to do so, this is a good reason to conclude that a given concept is sharp. However, this principle runs into trouble when we consider the case of consciousness. 

When we conceive of conscious experiences, we apply our “Consciousness” concept on the basis of the phenomenal character of our experience. Crucially, we do not use this concept to refer to the phenomenal character of our experience. Rather, when we recognize the phenomenal character of our experience, this is our evidential basis for concluding that we are having a kind of conscious experience. Unlike ordinary cases where we apply empirical concepts, however, it is not obvious that we should expect this evidential basis to come in greater or lesser degrees. For example, suppose that strong representationalism is true. In this case, the phenomenal character of an experience would be an abstract property complex that the experience represents. Can an abstract property complex be present to greater or lesser degrees? It is hard to even make sense of this suggestion. First, a property complex is an abstract object, so the degree to which it is present cannot vary between situations. Second, abstract objects are plausibly fundamental in a sense that precludes being present to a certain degree at all. Consequently, according to at least one theory of phenomenal character, our evidential basis for applying our Consciousness concept will not come in greater or lesser degrees. Most importantly, this could be true even if borderline cases of consciousness are, in fact, possible. Because the sharpness of our evidential basis for applying our Consciousness concept does not imply that consciousness itself is sharp, our inability to conceive of situations where we recognize a borderline case of consciousness is not a good reason to conclude that borderline cases of consciousness are metaphysically impossible. Consequently, if the hypothesis presented above is accepted, we should reject premise 3 of the vagueness argument against physicalism. 

In this post, I have provided two hypotheses that aim to explain when and why we are able to conceive of certain situations being actual. I have argued that, if these hypotheses are accepted, we have good reasons for rejecting key premises that occur in the zombie and vagueness arguments against physicalism. Whether these hypotheses are ultimately endorsed will depend upon wider considerations relating to modal epistemic theorizing, but they strike me as plausible and explanatorily powerful. For this reason, I believe physicalists have firm grounds for rejecting the zombie and vagueness arguments against physicalism.

Addendum:

According to the second explanatory hypotheses presented above, when we competently grasp a vague empirical concept, we should normally be able to conceive of situations in which our evidence for applying that concept is present in greater or lesser degrees. According to many non-representationalist theories of consciousness defended by physicalists, our Phenomenal Character concept is vague. Consequently, this hypothesis implies that we should be able to conceive of situations where phenomenal character is present to greater or lesser degrees. Tye, Antony, and others have persuasively argued that this is not the case. Furthermore, unlike the case of consciousness, a sharp evidential basis for applying our Phenomenal Character concept is not compatible with its being vague. This is because, if we apply our Phenomenal Character concept on the basis of our direct acquaintance with phenomenal character itself and phenomenal character admits of borderline cases, we should be able to imagine situations where the relevant evidence is present to greater or lesser degrees. Therefore, if we endorse the hypothesis under consideration, we should conclude that phenomenal character is sharp and reject any physicalist proposal that suggests otherwise (this would include the views proposed by David Papineau and Ned Block among others).