ON DENYING A PRESUPPOSITION OF SELLARS ’ PROBLEM : A DEFENSE OF PROPOSITIONALISM

SÍNTESE – Há uma profunda divisão entre duas concepções fundamentais na epistemologia ao longo dos últimos trinta a quarenta anos. Alguns rotulam essa divisão como sendo aquela entre internalistas e externalistas, e essa caracterização pode, mesmo, ser exata, conforme alguma explicação dessa distinção. Eu abordarei a divisão por um ângulo diferente, dado que uma abordagem melhor é conceber a divisão como surgindo de uma compreensão do Problema de Sellars. O meu interesse é em posturas que recusam uma pressuposição crucial na formulação do Problema de Sellars. Recusar essa pressuposição, como veremos, é uma das respostas mais comuns ao problema. Argumentarei, no entanto, que tal recusa é simplesmente insustentável, tanto na teoria da justificação quanto na teoria do conhecimento. PALAVRAS-CHAVE – Plantinga. Problema de Sellars. Internalismo. Externalismo. Knowledge. Justification. ABSTRACT – There is a great divide between two approaches to epistemology over the past thirty to forty years. Some label the divide that between internalists and externalists, and that characterization may be accurate on some account of the distinction. I will pursue the divide from a different direction, since a better approach is to think of the divide as arising from an understanding of Sellars’ Problem. My interest is in positions that deny a crucial presupposition in the formulation of Sellars’ Problem. Denying this presupposition, as we will see, is one of the most common responses to the problem. I will argue, however, that this denial is simply untenable, both in the theory of justification and in the theory of knowledge.

There is a great divide between two approaches to epistemology over the past thirty to forty years.Some label the divide that between internalists and externalists, and that characterization may be accurate on some account of the distinction.I will pursue the divide from a different direction, in part because the literature on the distinction between internalism and externalism has become a mess, and I don't want to clean up the mess here.* Department of Philosophy -University of Missouri, Columbia.
A better approach is to think of the divide as arising from an understanding of Sellars' Problem. 1 Though not discussed as often as the internalism/externalism distinction, there are enough varieties of characterizations of this problem that I will start with my own specific characterization.Before doing so, however, I want to distance myself from having to defend the claim that my specific characterization is the proper interpretation of the Sellarsian corpus.I'm interested in the problem itself, not whether the precise formulation I give is one that can be found in Sellars' writings.
In discussing this problem and the lessons we can learn from a proper appreciation of it, I will focus significantly on the proper function theory of knowledge proposed by Alvin Plantinga. 2As will be clear, however, my interest is in a broad range of positions represented by this proper function theory, positions that deny a crucial presupposition in the formulation of Sellars' Problem.Denying this presupposition, as we will see, is one of the most common responses to the problem.I will argue, however, that this denial is simply untenable, both in the theory of justification and in the theory of knowledge.The lesson of our investigation will not be a solution to Sellars' Problem, but a reinvigorated stance that the problem must be faced head-on.
So, what is Sellars' Problem?In a nutshell, the Sellars' Problem concerns the relationship between experience and belief.Sellars wanted to know how experience could justify belief, and there are two options here.Either experiences have semantic (propositional, informational) content or they do not.If they do not, they cannot make the truth of a belief intelligible in the way justification is supposed to make the truth of a claim understood and seen intellectually.They can cause one to hold a belief, we may assume, but the relationship between experience and belief would not be a rational one.It would be more like having a pill that induced certain kinds of thoughts: taking the pill causes the thoughts, but can't be used to explain why the thoughts are true or likely to be true.So, it would appear, if experiences are going to help justify what we believe, they will need to have propositional content.Yet, if these experiences do have propositional content, that content can be either appropriate or inappropriate to the context in question, in which case such contents are themselves epistemically evaluable, and a regress problem looms.The Sellars' Problem is, thus, a problem arising from the question of whether experiences have semantic content.* Department of Philosophy -University of Missouri, Columbia.
The locus classicus of this problem can be found in Wilfrid Sellars, "Empiricism and the Philosophy of Mind", in The Foundations of Science and the Concepts of Psychoanalysis, Minnesota Studies in the Philosophy of Science, Vol.I, ed. by H. Feigl and M. Scriven (Minneapolis, 1956). 2 Plantinga's theory is presented and developed over the course of three volumes: Warrant and Proper Function, (Oxford, 1993); Warrant: The Current Debate, (Oxford, 1993); and Warranted Christian Belief, (Oxford, 2000).The primary source for the view, however, is the first volume just mentioned.
Some quite common approaches to the problem are utter failures.It won't help, for example, to say that experience doesn't justify belief, but only prods belief.If we suppose that there are some beliefs that are directly caused by experience, the question remains whether those beliefs are justified, and if so, how.The standard reply for those who deny that experience justifies belief is coherentist in nature: they are justified by their relation to an entire system of beliefs.This reply avoids Sellars' problem, but only at the expense of creating another.For if such beliefs are justified solely by their relation to a system of beliefs, they'd still be so justified in the absence of the experiences or in the presence of quite different, but same-prompting, experiences.
It will not do, either, to say that it is only our beliefs about our experiences that form the beginning point for justification.If I believe that my experience is of a red apple, that may justify my believing that there is a red apple on the table, but only if my original belief is appropriate to my experience.Besides, such an account is psychologically implausible-most of us don't normally form beliefs about the character or content of our experiences-but even apart from that point, the view simply doesn't address Sellars' question about how experience makes certain beliefs appropriate in a given context.
These approaches, common as they are in the history of epistemology, give cause for concern.Their failure calls for a reflective distancing of ourselves from the fray of epistemological theorizing.They call us to meta-theory to try to figure out what can be done to assuage our discomfort in the face of Sellars' problem.
The meta-theoretical options that have developed can be described as follows.The first option is to call attention to, and then jettison, a presupposition of Sellars' problem.That presupposition is that the notion of justification (or whatever notion is thought to close the gap between true belief and knowledge) is connected with the notions of intelligibility, rational insight, and the cognitively understood, seen, or grasped.The problem as described asks for an explanation as to how experience justifies belief in the sense of making the truth of the belief intelligible, seen, or understood; an explanation as to how a particular experience yields a rational grasp of the truth of the particular belief that it justifies.On such an approach, this presupposition is mistaken; it is better to characterize the cognitive machine more on the model of inputs and outputs, where the outputs are beliefs and the inputs can be whatever the particular theorist wishes to hold generates epistemic value for beliefs (including experiences).Intelligibility and understanding are replaced by notions such as reliability of the mechanism, proper functioning of the system, and production by systems that get us to the truth at least most of the time.
The other approach refuses to give up on the notions of intelligibility and understanding in the theory of justification, hoping to find some way in which experience can play a role in justification without succumbing to Sellars' problem.The difference between this approach and the above approach is this: the present approach tries to solve Sellars' problem, granting that it is a problem that must be solved, while the above approach denies a presupposition of the problem and thereby holds that there is no problem here to be solved.The latter approach holds that what is thought of by the former camp as a distinctive feature of rational beings-the sense of intelligibility and understanding when rationality is displayed-is nothing more than an occasional (or perhaps more general) artifact of cognition.Even if this aspect were a universal accompaniment of displays of rationality, it is a mere epiphenomenon of the material which it is the job of epistemological theory to characterize properly.
My procedure here will be to examine this problem focusing especially on the epistemological theory of Alvin Plantinga.It is essential to Plantinga's philosophy that the presupposition in question is false, for his account of how belief in God is properly basic and thereby warranted depends crucially on the model of inputs and outputs, without any attending need to display the intelligibility of the connection between experience and belief.Plantinga holds, quite rightly, that belief in God can be prompted in any number of ways: seeing the majesty of the mighty Tetons, the delicate beauty of a flower; or by experiencing a feeling of thankfulness for things going well, or simply in the face of the experience of the joy of life itself, etc.What matters is not whether these experiences give rational insight into the truth of the claim that God exists, but whether the input/output relationship between these experiences and the belief that God exists is reliable and a display of a system that is functioning properly.Examining his theory will thus provide a good test case for determining whether jettisoning the presupposition in question is the right approach to take in addressing Sellars' Problem.
Other contemporary approaches side with Plantinga on this issue.Plantinga's theory is a version of reliabilism, a theory which eschews the presupposition of Sellars' problem.Reliabilist theories are characterized by two factors.The first involves some way of sorting mechanisms or processes into kinds.The second is a requirement that the input/output relationship have as outputs beliefs that are likely to be true.The kind of likelihood in question can vary from a simple frequency approach to a more sophisticated albeit mysterious propensity view.
This categorization is not in the least surprising, but it is surprising to note that some versions of foundationalism and some versions of coherentism deny the presupposition as well.Consider the following version of foundationalism.It claims that all justified belief traces to experience, but denies that experiences justify basic beliefs in virtue of some shared content.Instead, experiences have no content, they are merely the causal originators of basic beliefs, and justify those basic beliefs because they are their causes.Such a theory must deny that our experiences make intelligible to us the beliefs that they cause, since justification for basic beliefs would remain the same on such a theory of the relationship between experiences and beliefs were the relationship randomly varied so that different experiences caused different beliefs.
A similar issue plagues certain versions of coherentism.Some versions of coherentism insist that the role played by experience is merely a causal one, but that it is an important role, since it is imagined to be crucial to the success of coherentism that some beliefs are spontaneous responses to the "prick of sense", in Lehrer's memorable terminology. 3As I think of coherentists, they have the strongest interest in rendering the justification relation so that it preserves the rational intelligibility of the justification relation and so that the presence of this relation depends upon some kind of rational insight relating the grounds of the belief.But the form of coherentism that simply limits experience to playing only some causal role in the story of justification, simply so that some elements in the system of beliefs count as spontaneous belief, will have a hard time explaining why it matters which elements in the system of beliefs are the spontaneous ones.The only way to do so is to insist that there are special argument forms needed to justify spontaneous beliefs that are not used in the account of the justification of non-spontaneous beliefs, and even with this emendation, the theory will still be susceptible to the objection that the experiences could vary wildly so long as they still cause the same spontaneous beliefs.
Here's one way to look at these versions of foundationalism and coherentism.Those who develop foundationalist and coherentist theories of the last two types seem to be imaginatively limited regarding the kinds of cognitive beings there might be.If we assume that any possible cognitive being is pretty much like us, having links between experience and belief pretty much like us, it becomes easier to see how such theories can end up looking plausible to their defenders.The problem is the lack of imagination.Cognitive beings can have radically different sensory inputs than we have, and perhaps no sensory inputs at all; and the connections between sensory inputs that are similar to ours can be wired to beliefs in ways radically different than ours.In some such cases, perhaps these differences won't make a difference to the justificatory status of their belief, but in other cases, these differences will make a dramatic difference.In some cases, that is, such a radically abnormal link between experience and belief may still result in justification, but in other cases, it will not.It is crucial to the development of a complete epistemology to consider the implications of one's theory both for normal human beings and for beings quite different from us.I suspect that the attraction for these inadequate versions of foundationalism and coherentism is a simple failure to consider the implications of these theories for radically abnormal possibilities.
This point would seem to play into the hands of those denying the presupposition of Sellars' problem, but that conclusion would be premature.First, cognitive beings devoid of sensory input and who rely on rational intuition and inference alone are an easy case, since rational intuition, on its face, delivers the needed intelligibility feature that is the heart of the presupposition in question.That leaves only beings for whom the initial inputs into the system of beliefs is on the basis of something other than rational intuition, and there is no difficulty using the language of sensory inputs to characterize such beings.Once we get this far, however, Sellars' problem can be stated and the presupposition in question either embraced or denied, so it is a mistake to think that the more general perspective requires denying the presupposition of Sellars' problem solely on the basis of what we have seen to this point.
To reiterate what was said earlier, my goal is to show that there is no future in denying this presupposition of Sellars' problem.I wish here to press two different complaints against such theories.Such theories can be either theories of knowledge or theories of justification, and my complaints fall into these two categories.First, I will argue that theories of justification that deny the presupposition cannot hope to succeed, and then I will argue that theories of knowledge that deny the presupposition suffer from the same malady.

The denial of the presupposition of Sellars' problem in the theory of justification
A central feature of justification is that it can be truly predicated both of beliefs and propositions that are not believed.We can correctly claim that my belief that I have two children is justified, and we can correctly claim that George Bush is justified in believing that going to war was a mistake, even though he does not in fact believe that claim (readers not persuaded of the truth of what I claim here about the President are free to substitute other examples where people refuse to believe what is obvious).I term the former kind of justification "doxastic justification", and the latter "propositional justification." 4 It is worth noting that this feature has an analogue in the arena of action, though it is easy to miss.We can speak of actions that are justified, and refer both to tokens, which of course are actually performed, or to types, which need not have any actual instances.Thus, one can speak of being justified in driving 70 miles per hour on a given stretch of road even though one is going only 55.In such a case, there is an untokened action type that is justified, and as such is an analogue of the kind of justification one can have for a belief that one does not hold.
A complete theory of justification will account for both types of justification, explaining one in terms of the other or claiming that they are irreducible to each other.The irreducibility thesis is hard to sustain, however, since it is fairly easy to see how to characterize doxastic justification in terms of propositional: doxastic justification is simply a function of propositional justification plus proper basing: where the belief is held because of, or on the basis of, the features that justify it.Such an account not only undermines the irreducibility claim but also presents a challenge.The account takes propositional justification as basic and shows how to understand doxastic justification in terms of it; we can call the resulting position "propositionalism."Doxasticism takes a contrary position, according to which doxastic justification is the basic sort, and propositional justification should be understood in terms of it.
As the above account shows, it is easy to see how to be a propositionalist; it is much harder to see how to be a doxasticist.For it is not always true that when one concept can be understood in terms of another, the roles can be reversed.My goal is to show the problems for doxasticism because of its connection to the approach to Sellars' Problem that denies the presupposition of it on which we are focusing.I will first show the defects of doxasticism, and then show the connection this discussion has to Sellar's Problem and this presupposition.
Plantinga's is a paradigm case of a theory that denies the presupposition of Sellars' problem.It is also a paradigm case of doxasticism.Plantinga says of his theory, According to the central and paradigmatic core of our notion of warrant (so I say) a belief B has warrant for you if and only if (1) the cognitive faculties involved in the production of B are functioning properly [...] (2) your cognitive environment is sufficiently similar to the one for which your cognitive faculties are designed; (3) [...] the design plan governing the production of the belief in question involves, as purpose or function, the production of true beliefs [...]; and (4) the design plan is a good one: that is, there is a high statistical or objective probability that a belief produced in accordance with the relevant segment of the design plan in that sort of environment is true.

5
The crucial element in this quote is right up front: according to Plantinga, the "central and paradigmatic core" of warrant is where "a belief B has warrant for you."Talk of beliefs here is not some causal language that could be replaced with talk of the content of a hypothetical belief, i.e., a proposition, for clause (1) of the above account makes clear that B is an actual belief of yours.Plantinga thus endorses doxasticism with regard to the concept of warrant, but he is also aware that the terms of epistemic appraisal apply to things other than beliefs.He notes that we appraise not only a person's beliefs "but also her skepticisms or (to use another Chisholmian term) her withholdings, her refrainings from belief.An unduly credulous person may believe what she ought not; an unduly skeptical (or cynical) person may fail to believe what she ought." 6he question of the relationship between these kinds of epistemic appraisal deserves attention, and the same issues involved with doxasticism regarding justification come into play here as well.Warrant, as well as whatever other term of epistemic appraisal one prefers, can be a feature of both beliefs and propositions, and it is easy to see how to characterize doxastic warrant in terms of propositional warrant.That approach, however, is not Plantinga's, for he holds that the fundamental notion of warrant is doxastic warrant.
There is a way of objecting to what I've said, and will say below, about Plantinga that I want to forestall before proceeding further.In the early stages of developing his theory, Plantinga at times talks as if the term 'warrant' is meant merely as a placeholder for whatever turns true belief into knowledge.When viewed in this way, one might insist that Plantinga only has a theory of knowledge, and that any at-tempt to characterize his theory of warrant so that it can be classified with theories of justification and other terms of epistemic appraisal simply ignores the stipulative character of his use of the term warrant.It is not obvious to me that, even granting this point, there will be anything problematic in what I say about Plantinga's theory of warrant, but it is worth pointing out that Plantinga's mature account of warrant cannot be stipulatively identified with what turns true belief into knowledge.Though Plantinga talked this way in the early stages of developing his account, his mature theory treats warrant as that quantity which, when one has enough of it and when one's cognitive mini-environment is favorable, one's true belief counts as knowledge.This mature account makes warrant necessary, though not sufficient, for turning true belief into knowledge.Moreover, warrant is now specified to be a property that comes in degrees.These two features are all that is needed to classify Plantinga's theory of warrant together with theories of justification and other terms of epistemic appraisal.
Plantinga is not alone in endorsing doxasticism, nor is he alone in ignoring the question of the relationship between doxastic and propositional justification or warrant (or whatever epistemic term of art is preferred here). 7Of the major defenders of theories of epistemic appraisal who count as doxasticists, Alvin Goldman is the only one who has devoted serious attention to the relationship between doxastic and propositional uses of terms of epistemic appraisal. 8he problem is that doxasticism cannot succeed.In order for doxasticism to succeed, it will have to talk about a hypothetical belief when attempting to characterize propositional justification for the content of a possible belief.It will then have to claim that this hypothetical belief would, or could, be held justifiably.The "could" reading is much too broad-nearly any belief could be held justifiably-so the obvious path is to talk in terms of what beliefs would be justified if they were added to one's total corpus of beliefs.In addition, the doxasticist will want to include among the circumstances in which the belief is added that it is added in such a way that it is properly based.
Such a view faces problems from two different directions.The first concerns individuals who are cognitively admirable, so admirable that they simply wouldn't believe a claim if it were not justified for them.At the extreme is God, who has such a property essentially.Even among ordinary humans, however, there is the possibility of such an individual.Given such admirability, the approach just outlined must conclude that such individuals have propositional warrant for absolutely every claim, including having warrant for any claim p and its denial ~p.Since having warrant for contradictory claims is impossible, doxasticism is in trouble.The natural reply notices that for such individuals, one's body of evidence will need to shift in certain cases in order for an added belief to be justified.It would be philosophical suicide, though, for a doxasticist to adopt this characterization of the case, for doing so admits that justificatory status is a function of evidence, and the existence of evidence confirming a claim doesn't require believing it.A better strategy is to try to use a surrogate for the concept of evidence, such as talking about a person's total cognitive state or noetic complex changing.Using this surrogate language, the doxasticist can say that a proposition is justified for a person if and only if, were the person to believe that proposition while remaining in the same total cognitive state, the belief would be justified.This is not quite right, of course: adding the belief can't be done without changing the total cognitive state of the person.So, the idea would be to add the belief while altering the noetic complex as little as possible to accommodate the added belief.
The effort here is noteworthy, but the results are less than one might hope.First, the added requirement works at best only for ordinary cognitive agents.If a being is essentially cognitively admirable, the additional requirement won't help in the least.In the case of God, it doesn't help at all: any belief God might add to his store of beliefs could only be added when he'd be justified in believing the claim in question.Moreover, even for ordinary humans, it might not help.Some epistemologists have held that some beliefs fall into a kind of belief regarding which we cannot be wrong, about which we are infallible.I am not convinced they are right if we are talking about introspective beliefs, but suppose they are: that is, suppose that some of our beliefs are infallible and justified because of it.Then there will be a range of propositions which are such that if we were to believe them, altering our noetic condition no more than is necessary to accommodate the belief, we'd be justified in believing them.On the theory in question, we are essentially cognitively admirable about this range of propositions.The version of doxasticism cannot accommodate this theoretical possibility.
The problem of cognitive admirability is only one of the problems faced by doxasticism, however.The other problem is an extension of the known fact that belief itself sometimes creates evidence for itself.For example, you are more likely to succeed at certain tasks if you believe that you will succeed.The extension of this idea is that belief itself can also count against a claim that you have good evidence for, as when self-doubts cause people to have less evidence than they would have otherwise for thinking they'll succeed.In the starkest cases, belief itself can completely undermine evidence for its content, as when early math students know that they've never squared any number greater than five, and so have evidence that they've never considered the proposition that nine squared is eighty-one.In such a case, this proposition can be justified for them, but if they were to add belief in this claim to their total cognitive state, the added belief would destroy whatever evidence they had for the claim prior to believing it.It would destroy this evidence because, once they come to believe the claim, it will be obvious that they've considered the claim that nine squared is eighty-one.
Such cases block any attempt to understand propositional justification in terms of what beliefs would be justified if they were believed.The proper conclusion to draw is that doxasticism is false, and that propositionalism is true: that the fundamental epistemic relation is that between evidence and propositions supported or confirmed by that evidence.
The point of this discussion of doxasticism and propositionalism is the way in which these positions connect with Sellars' Problem.First, if one is a propositionalist, one will not be able to deny the presupposition of Sellars' Problem that is in focus here.If that is correct, the only hope for those who wish to deny this presupposition is to escape propositionalism, and as we've just seen, that can't be done.So, if propositionalists can't deny this presupposition, one cannot escape Sellars' Problem in the way envisioned by Plantinga and other doxasticists.
The crucial step in this argument is the claim that propositionalists cannot deny the presupposition in question.To defend this step, I'll first give an intuitive argument for it, and then examine how a propositionalist might try to explain the connection between belief and experience without any commitment to the presupposition in question.So, first, here's the intuition argument: the notion of evidence involves the idea of making things evident, and what is made evident to me is what I come to see as obvious, intelligible, and understood.It is something regarding which I've now gained insight on the basis of that which has made it evident to me.Since characterizing the relationship of one thing propositionally justifying another is most plausibly interpreted in terms of the first thing providing evidence for the second, propositionalism is committed to the presupposition of Sellars' Problem that Plantinga and other doxasticists wish to deny.
We can get to this same conclusion by investigating the options of theory development open to propositionalists.For propositionalists having any hope of solving Sellars' problem, evidence will come either in the form of other believed propositions or in terms of experience.If experiences have content, then that content provides resources to help explain how these experiences justify certain propositions in a way that makes the connection between the two intelligible to cognitive agents who have the experience and thereby come to believe the related propositions.Of course, that is not to say that the story will be easy or simple to describe, nor is it to say that any extant stories by propositionalists have given an adequate explanation of the link between experience and belief.My point is only this: if experiences have content, then we are well on our way to a solution to Sellars' Problem.
Suppose, however, that experience has no semantic or representational content.In virtue of what does the experience count as evidence for the related proposition?One answer commits one to doxasticism: that an experience is evidence for a proposition just when anyone who believes that proposition upon having that experience has a justified belief, perhaps when the connection is a reliable one (and perhaps satisfies other conditions as well, such as a proper function condition).As we've seen already, however, doxasticism cannot succeed, and in any case, we are at present trying to discover what a propositionalist might say on the supposition that experiences have no content.So how could a propositionalist explain how an experience with no content counts as evidence for the supported proposition?
I think the answer is that there is no such approach available, but to see this, we need to consider one approach to this problem developed by epistemologists who display at least some sympathy for propositionalism.I will not endeavor to determine whether they are propositionalists in the end.Instead, I only want to show that their approach cannot be used to reject the presupposition of Sellars' Problem we are considering here.Some who are sympathetic to propositionalism adopt an account at this point that appeals only to recognition skills: the experience counts as evidence for the proposition because the cognitive agent has learned to recognize the truth of that proposition on the basis of that experience. 9The general form of explanation here is in terms of know-how.Instead of positing an intelligible connection between the experience and the belief, the present approach treats the connection on the model of practical rather than theoretical knowledge.I want to argue, however, that such an approach cannot be satisfactory without providing the resources needed by propositionalism for retaining the presupposition in question here of Sellars' Problem.
In light of the possibilities of manipulation by evil demons and the like, the practical approach here cannot assume that the process by which stable dispositions to form beliefs in the presence of sensory stimuli is a process that generates know-how, or involves the development of recognition skills, where these latter two notions are understood in a way that requires epistemic success in the process.To learn to recognize colors, for example, implies the development of a skill that is present only when one has learned to distinguish accurately one color from another.But to develop stable dispositions in the presence of sensory stimuli involving colors only requires some sort of systematicity in one's belief formations in the presence of such stimuli, whether or not such systematicity involves accuracy.
The practical approach thus divides into two camps, those who plump for some sort of reliability or accuracy at the end of the process and those who do not.Those who plump for accuracy, I will say, understand the practical approach in terms of the development of know-how on the part of the cognitive agent.It is fair to characterize those who adopt this theory as joining hands with their reliabilist cousins in denying the presupposition of Sellars' problem, for intelligibility, rational insight, and that which is understood, seen, or grasped intellectually plays little or no role in this theory.Instead, these features of the presupposition in question come into play only after the fundamental work of developing the required know-how has been achieved.
What of the weaker version of the practical approach, on which the question is only a matter of developing stable dispositions in the face of sensory stimuli?On this version, once such dispositions are in place, the story of epistemic justification can be told in terms of these dispositions: upon having a certain experience, the individual in question has evidence in favor of a belief just in case that individual has a disposition to connect experiences of that kind with that belief.Here, again, there is no question of intelligibility and its ilk in the story of the relationship between experience and these basic beliefs.Instead, there is only the matter of what kinds of stable dispositions have developed.
As a result, this form of the practical model joins in the rejection of the presupposition of Sellars' problem as well.We might put this point as follows.The practical model, whether or not of reliabilist persuasion, appears to have no need of the presupposition in question.For when a discussion of basic beliefs occurs, the concept of justification employed is that which connects experiences to beliefs, not experiences with a given content to a propositional content that might or might not be believed.On the know-how model, there is only a disposition to come to a particular belief in the presence of a certain experience.The experience has no feature in itself that can make evident the content of the belief that is formed, and because there is no content that can play such a role, there is no intelligible connection between experiences of a certain kind and the beliefs that result from them.
We can show, however, that without endorsing the presupposition in question, the practical knowledge model is bound to fail.Suppose I have a sensory experience for which I have a stable disposition connecting such an experience with believing p. Suppose also that in the present case I do not believe p, perhaps because I have a strong desire for p to be false.So, we assume, the experience justifies p for me, but I don't believe p. Defenders of the practical knowledge model have to explain such justification by recourse to a hypothetical belief that I do not have: had I formed the belief it would have been justified (because of the practical skills I've developed in the process of cognitive maturation).
This approach suffers at the hands of the problem of cognitive admirability as much as doxasticism does: cognitively admirable individuals satisfy the counterfactual claim in question regarding lots of propositions that are not presently justified for them, and cognitively ideal agents who can't fail to believe what's justified turn out to have justification for every proposition (since no proposition is essentially such that there couldn't be a learning process that forms a disposition to believe such a claim in the presence of some sensory stimulation or other).
Defenders of this view will, of course, begin tinkering with the counterfactual I used to characterize that position, in hopes of finding one that avoids the problem I mention.Here I have a challenge, however.All counterfactual accounts must face the issue of the conditional fallacy described by Robert Shope in his 1979 paper in the Journal of Philosophy.In the paper, Shope describes the character of the fallacy, and gives a recipe for constructing counterexamples to counterfactual accounts.So my challenge is this: before being satisfied with any particular tinkering with the account, find an argument for the conclusion that Shope's recipe cannot be applied to the new account.Such an argument is a rational requirement on any counterfactual account in philosophy, and it is astounding how common it is to find philosophers still relying on counterfactual accounts without presenting such an argument and without showing any awareness of Shope's results.
In addition, the practical knowledge model has trouble with the fact that belief itself sometimes creates or undermines evidence for it.Suppose you've developed a stable disposition for forming the belief that p in the presence of sensory stimulation S. Let S be Hume, and suppose Hume has become obsessed with the missing shade of blue, so much so that he has developed the following stable disposition: every sensory stimulation disposes him toward two beliefs, one about the particular color he's experiencing and the other that the object in question is not the missing shade of blue.This disposition is quite reliable, we may assume.But now, after years of experience and conditioning to reinforce this disposition, Hume finds the missing shade of blue.He is disposed to believe that the object is blue and that it is not the missing shade of blue, for that is what our description of his stable dispositions tells us.He believes, however, that the object is blue and that it is the missing shade of blue.Both beliefs are justified, and had he not believed that this is the missing shade of blue, that proposition would still have been justified for him by the experience in question.But for the practical knowledge model, there is the problem of the disposition in question not to form this belief.If that is the only disposition in question, then the experience can't be evidence that this is the missing shade of blue.
So perhaps there is another disposition: the disposition that has developed to discriminate one color from another.The problem now is that the practical knowledge model involves too many dispositions: there are dispositions that underlie ascriptions of justification both to p and to ~p.But that can't be right.So the practical knowledge model will have to incorporate some feature that allows one of the dispositions to trump the other one, without endorsing the obvious explanation that it is the content of the experience of the missing shade of blue itself that makes the claim that this is the missing shade of blue justified, in spite of the competing disposition to believe that this is not the missing shade of blue.
The natural move for the defender of the practical knowledge model to make here is to appeal to the qualitative character of the experiences in question to avoid this problem.The idea here is that experience has a qualitative character to it, even though it lacks propositional or semantic content.A standard example of such is the sensation of pain, which has a qualitative character, but apparently no propositional or semantic content.The idea here, applied to the Hume example above, is that Hume's perceptual states have a qualitative character to them, and believing that the object is the missing shade of blue accords with that qualitative character.Presumably, Hume's prior experiences that justified the claim that the objects of his experience were not the missing shade of blue also had a qualitative character to them.The difference here is that the qualitative character of the present experience is different and does not include the qualitative character that justifies believing that the object in question is not the missing shade of blue.
If the practical knowledge model is developed along these lines, however, it will no longer be in a position to deny the presupposition of Sellars' Problem under discussion here.The qualitative character of the experience involves informational content of some sort.The experience of pain, as well as the experience of the missing shade of blue, is an informational state, where the information can be transferred by the human system in question to a belief state that also involves that same informational content.Qualitative character just is informational content detected by the individual in question.If it weren't, it could help explain the Hume example.For another quality possessed by his experience is that of occurring at a particular distance from the North Pole, and surely no beliefs involving that qualitative feature of the experience would normally be justified by the experience alone.The qualitative character of the experience is a character that involves qualities or properties, ones detected by the cognitive system in question.Hume's belief that the object is the missing shade of blue is a belief constituted in part by some of these very same qualities or properties.As a result, the appeal to qualitative character cannot be something other than some type of informational content that is capable of being transferred by the human system from the modality of experience to the modality of belief.
I think the motivation to deny informational content to experience is driven by the thought that the only content there could be for experience would be conceptual content.That would create additional problems, since it would imply that infants don't have any experience without having concepts, and thus that some concepts would have to be innate.But experiential content need not be conceptual in order to involve informational content.The right question to ask here concerns the best explanation for how beliefs come to have the precise propositional character they have on the basis of the experiences of the organism, and here the answer seems compelling: it is on the basis of the transfer of informational content from one subsystem of the organism to another. 10 submit, then, that there is no avoiding the presupposition of Sellars' problem within the theory of justification.The conclusion to draw is that an adequate theory of justification will have to be propositional in character, and will thus need to account for how experience makes intelligible the beliefs that it causes to be formed.Nothing less can be adequate.It might be, however, that within the theory of knowledge, something less can be adequate, so I turn in the next section to consider this possibility.

The denial of the presupposition of Sellars' problem in the theory of knowledge
We turn, then, to the prospects for a denial of the presupposition in the theory of knowledge.We abandon the attempt to characterize the nature of justification in the process of completing our theory of knowledge; instead, we simply intend to characterize the difference between true belief and knowledge with reference to a theory that refuses to endorse the presupposition of Sellars' problem.
A primary exponent of such an approach is Alvin Plantinga, especially with the early versions of his proper functionalism, though there are other versions of the view as well, a primary example being the views of Armstrong and Dretske.11Here I will focus on Plantinga's version of the approach, since the requirements of the other views are clearly too strong.Armstrong's view requires a piece of knowledge to be the product of some lawlike regularity, and Dretske's requires production by information that makes it a certainty (i.e., a probability of 1) that the belief is correct.Both requirements are too strong, the first because knowledge can occur even in indeterministic worlds.The second view is too strong as well, since our subjective probability should accord with the objective probability of truth, but if our degree of belief in what we know has a probability of 1, then it will never be rational to abandon belief in what we know.Since misleading evidence has such power, we shouldn't accept quite so strong an account of knowledge.
Plantinga's theory fares better on these points.So suppose we use the theory as an account of that which bridges the gap between true belief and knowledge, to see whether the presupposition of Sellars' problem can be avoided.
Plantinga's account of this property, which I will term 'warrant', with the understanding that the term is now being used merely as a placeholder for the value X in the equation "true belief + X = knowledge", involves the following concepts.First, a belief is warranted only if it is produced by cognitive faculties that are functioning properly.In order to be functioning properly, these faculties must be operating according to their design plan.Other features of Plantinga's view tie the notion of warrant more closely to truth.One feature is that the design plan of the faculty in question is one that is aimed at the production of true belief, and it must also be true that the design plan is a good one, i.e., that there is a high objective probability that a belief produced in accord with that design plan is true.Another feature is that the system in question must be functioning in a favorable environment, one that is decently well-suited for the successful operation of these faculties.
These features need to be supplemented in order to deal with variants of the Gettier problem that has plagued the theory of knowledge since Gettier's original article in 1963. 12Plantinga proposes to handle these problems by requiring that the minienvironment M with respect to the belief B and the exercise E of a cognitive faculty be favorable; for short, that MBE is favorable.
Maxienvironments are what a design plan covers: whether the environment has light and air, whether it is full of radioactive waves or not, and the like.Any such maxienvironment covers a range of minienvironments: just add a further feature to a maxienvironment, one that the design plan does not address, and you get a minienvironment.A specific minienvironment includes all the features not specified in the maxienvironment.13   As just noted, the crucial notion here is what it is for MBE to be favorable.Plantinga has tried several accounts of this notion, and, though it is a slight detour from the main thread of the argument here, it is instructive to note how his most recent account fails.His latest attempt, in Warranted Christian Belief, distinguishes those parts of the MBE that are detectable to S from those that are not.The conjunction of all the detectable parts he calls the DMBE.He then defines favorability of an MBE as follows: MBE is favorable just if there is no state of affairs S included in MBE but not in DMBE such that the objective probability of B with respect to the conjunction of DMBE and S falls below r, 14 where r is a reasonably high probability, at a minimum higher than .5.This account fails to explain a variant of one of the standard cases in the literature on the Gettier problem.That example is found in the literature on the defeasibility approach to knowledge which involves a purported pair of twins, Tom and Buck.As the case goes, you see Tom steal a book from the library, but Tom's psychotic mother invents a story which she tells the police that it was really Tom's twin brother Buck who stole the book, hoping thereby to save Tom from his predicament.You do not hear the story told to the police, and the police do not need to take the story seriously-they have a large file of such stories from Tom's mother, aimed at protecting her son.They've even heard this very concoction before.
In such a case, her story doesn't undermine your knowledge that Tom stole the book.But one feature of the total situation is that Tom's mother says that an identical twin stole the book and that Tom was out of the country.Given everything that you can detect about your situation together with the state of affairs of Tom's mother's report gives a quite low probability to your belief.How low?That is hard to say, but if one were to add the information that Tom's mother said what she did, the reasonable thing to do would be to take back one's claim that Tom stole the book, indicating that the probability need not be reasonably high, so we can add that in fact the probability is not reasonably high and still have a possible situation.
The problem raised by the Tom/Buck case is just the problem of misleading defeaters, and it is easy to see how Plantinga's latest account of favorable minienvironments succumbs to it.What makes a defeater misleading is the existence of further, veridical information that overrides the misleading information in question, and Plantinga's account above ignores the way in which further features can override in this way.His account, in short, fails to attend to the possibility of misleading pockets of information, insisting instead that all such information must be non-misleading.
In any case, we can ignore this problem for present concerns.Our primary concern is with the package that identifies knowledge with true belief plus some factor that involves proper function by faculties aimed at truth a good at achieving it, in a cooperative environment where the relevant mini-environment remains favorable.Yet, once we have seen in the last section the fundamental problems facing denials of the presupposition in Sellars' problem, it is not hard to see how to transpose them into the present key, however.If we deny that there needs to be any relation of appropriateness between experience and belief, but only other relationships such as the ones just enumerated, then there are no barriers to counting as knowledge even the wildest and most implausible of such relationships.Plantinga, in criticizing coherentism, imagines a mountain climber having a coherent system of beliefs together with the experiences appropriate to the mountain-climbing environment, who is struck by some malady that freezes the system of beliefs in question.Upon being taken down from the mountain and placed in an opera, the belief system doesn't change, but the class of experiences clearly does.Plantinga takes this example as a clear refutation of coherentism.It is not, of course, since it is an attack directed only at one version of coherentism, the kind that denies the presupposition of Sellars' problem.Once we see this point, however, it is easy to see that the objection is only as good an objection to coherentism as it is against Plantinga's own proposal.For it is clearly possible that the connections between experience and belief that characterize the mountain climber are truth-conducive and in accord with the design plan of the cognitive system in question.Moreover, the other features of Plantinga's account can be met as well, with the possible exception of the issue of the proper functioning of the defeater system.
One might appeal to the defeater system to save the theory from this problem, but I think such an appeal only highlights the problem.Note first that a properly functioning defeater system will not ordinarily undermine truthconducive input/output relationships, so if the particular connection imagined in the preceding paragraph is a regularity for the kind of world in which the events occur, we need a special reason for thinking that the defeater system is properly functioning only if it undermines this connection.The only way I can see to get that result is to say that the defeater system functions so as to honor evidential connections; it kicks in whenever the input/output connections do not honor the evidential connections in question.That requires, however, that the defeater system kick in when no evidence is present as well as when the evidence confirms some contrary of the proposition in question.
If the defeater system is characterized in a way that preserves this point and thereby allows the theory to survive such counterexamples, we have all the information in place to provide a complete theory of evidence.Once we have a complete theory of evidence, however, we have done everything necessary to sustain the presupposition of Sellars' Problem, as I argued in the last section.The proper conclusion to draw is that nothing is gained by focusing on the theory of knowledge itself rather than on the theory of justification, for the same issues reappear even if in a slightly different guise.To solve them, one will need a properly functioning defeater system, and to characterize such a system, one will need a complete theory of evidence, thereby providing the requisite materials to sustain the presupposition of Sellars' problem that had been hoped to be avoided.

Conclusion
Sellars' problem is among the deepest and most interesting problems of epistemology, along with such perennial problems as the regress problem and the problem of the criterion.Plantinga's theory of knowledge partakes of the approach to this problem by denying a presupposition of its formulation, a presupposition to the effect that the relationship between experience and belief must fall within the general category of that which is intelligible to the intellect.Plantinga's theory is also a version of doxasticism, and the connections between this approach and the denial of this presupposition are strong ones.In the end, however, such approaches are problematic, and give no satisfying account of knowledge, justification, warrant, or any other surrogate for such.The proper conclusion here is that Sellars' Problem deserves a real solution, not some attempted circumvention of it.

7
Others committed to doxasticism include: D. M. Armstrong, Belief, Truth and Knowledge, (Cambridge, 1973); Alvin Goldman, Epistemology and Cognition (Cambridge, Mass., 1985); Robert Nozick, Philosophical Explanations, (Cambridge, Mass., 1981); Ernest Sosa, Knowledge in Perspective: Selected Essays in Epistemology,(Oxford, 1991).For a discussion of the approach of these and other authors, see Jonathan L. Kvanvig, The Intellectual Virtues and the Life of the Mind: On the Place of the Virtues in Contemporary Epistemology, (Savage, Maryland, 1992).