Nondoxastic Foundations Beth Williams Department of Philosophy University of Arkansas
Nondoxastic foundationalists make the claim that perceptual states justify certain of our beliefs that we call basic. Basic beliefs must be justified, because they serve as the terminators of an otherwise infinite regress of justification. If perceptual states are incapable of justifying basic beliefs, then the nondoxastic foundationalist will be left with the unsavory conclusion that basic beliefs justify themselves. As it stands, selfjustification of basic beliefs is in much worse shape than nondoxastic justification. Therefore, if nondoxastic justification is our only option and if explanations of nondoxastic justification fail, then foundationalists will not be able to account for the justification of physical object beliefs like “There’s a yellow notepad before me.” If our physical object beliefs cannot be justified, then no belief that we infer from our physical object beliefs will be justified. In order to prevent a skeptical scenario of this magnitude, a plausible account of nondoxastic justification must be given. Part of giving a successful account involves rebutting the arguments against nondoxastic justification. Thus, I will devote the first part of this paper to explaining the issues and dispensing with objections to nondoxastic justification. The last part of the paper will focus on Matthias Steup’s positive theory of nondoxastic justification, which he calls unrestricted foundationalism.
1. The Regress Argument
2 The regress argument provides the motivation for foundationalism. Suppose you ask someone his justification for believing A. He answers that B is his justification for believing A. You then ask him his justification is for believing B. He replies that his justification for believing B is C. You ask him his justification for believing C, knowing that more than likely he will say his justification for C is D. Theoretically this line of question and answer could go on forever, but if it did we would never reach that point at which justification comes to an end. Because we believe that justification does come to an end, we must specify the point in the chain of justification at which the regress ends. There are four possible results (Bonjour 1978): 1) The branch of justification terminates in a basic belief. Foundationalists think an infinite regress of justification is unacceptable and that it is impossible for the justification of a belief to go on infinitely. For that reason, foundationalists opt for a structural theory of justification that situates basic beliefs in the foundation and non-basic beliefs in the superstructure. 2) The branch terminates in an unjustified belief. In order for a belief to be inferentially justified, it must receive its justification either from a basic belief or from another justified belief. Nobody wants to claim that unjustified beliefs provide the stopping point of the regress, because then none of our beliefs would be justified. 3) The branch extends infinitely. This option is untenable, seeing as it does not account for justification at all. 4) The branch circles back upon itself.
3 This option appeals to coherentists. Coherentism suffers from many intractable problems, and thus has fallen out of favor in contemporary epistemology.1 In sum, since possibilities 2-4 are unpalatable and incapable of yielding justification, we must accept that basic beliefs are the only things fit to end chains of justification.
2. Basic Beliefs Foundationalism is a theory about the structure of justification. Basic beliefs play a central explanatory role in foundationalism because the set of basic beliefs constitutes the foundation of justification. The idea is that some beliefs are basic and all other beliefs derive their justification from those basic beliefs. The derivative, non-basic beliefs form the superstructure of knowledge. According to Matthias Steup (1996), basic beliefs have three generally accepted characteristics. (i) They are non-inferential, which means that we do not infer them from any other beliefs. If I see that my window is broken and form the belief that my window is broken, then my belief that my window is broken is a non-inferential belief. My belief that I have been burglarized, however, is an inferential belief, because it is one that I might reasonably infer from my belief that my window is broken. (ii) Basic beliefs are justified. Suppose that I am sitting around the house brooding over my windows. A sudden bout of capriciousness seizes me and I form the belief that my window is broken. This belief is non-inferential, but is also unjustified, because it is formed on the basis of a whim. Suppose I infer from this unjustified belief that I have been burglarized. Then, the belief that I have been burglarized is unjustified because it is
The isolation objection was the main problem that felled coherentism. See Steup (1996) for a clear formulation of the objection.
4 inferred from the unjustified belief that my window is broken. The point is that if the belief that p is inferred from the belief that q, and q is unjustified, then p is unjustified too. Consequently, basic beliefs must be justified if they are going to justify nonbasic beliefs. There is some contention over the third characteristic of basic beliefs, which specifies their source of justification. In “The Myth of the Given”, Roderick Chisholm (1964) identifies four possible sources of justification for basic beliefs. For our purposes, the two most important sources are (a) doxastic—the basic belief justifies itself, or (b) nondoxastic—the basic belief is justified by experience or observation, neither of which can be said to be justified or unjustified. Doxastic foundationalists fancy the former possibility. They attest that basic beliefs are doxastically justified, which is to say that they justify themselves, otherwise they could not stop a regress. Problems with this construal of the third characteristic are manifold. A short passage from Steup (1996) illuminates one such problem: Any allegedly self-justifying belief is either justified in the first place or it is not. If it is, then it can’t justify itself because it is already justified, just as one can’t bake a loaf of bread that is already baked, write a letter that is already written, or melt an ice cube that is already melted. And if the belief is not already justified, then it can’t justify itself because a belief that doesn’t have justification can’t transmit justification onto any other beliefs or itself, just as somebody who doesn’t have any money can’t give any money to anyone, including himself. We are to conclude, therefore, that self-justification is impossible (100). Those who do not want to be committed to either a doxastic or nondoxastic characterization of the source of justification of basic beliefs typically claim that basic beliefs are noninferentially justified. People of this ilk have no trouble accepting either (a) or (b). They will say that a candidate basic belief is justified (the third characteristic implies the second) and one’s justification for believing it is not derived from one’s
5 evidence or justification for holding other beliefs. Since doxastic justification is really no justification, people who claim that basic beliefs are noninferentially justified also claim that (iii) basic beliefs are nondoxastically justified.
3. Nondoxastic Sources of Justification If basic beliefs cannot justify themselves and if they cannot receive their justification from other beliefs (because then they would not be basic anymore), then they must receive their justification from nondoxastic sources. There are four such sources, three of which are experiential and one a priori. Suppose S forms the noninferential belief that p, where p = this pen is blue. P is nondoxastically justified because it is grounded in (1) perception. For Bonjour (1978) there is a three-way distinction that is involved in S’s believing that p. First, there is the belief that the pen is blue. Second, there is the state of affairs (there being a blue pen) that is the object of the belief that p. Third, there is S’s “immediate apprehension” of the state of affairs, i.e., S’s being appeared to bluely (and penly). This view is often referred to as representationalism or indirect realism. According to this view, what justifies S in believing p is the experience of being appeared to bluely. There is, however, another view, which is called direct realism by its proponents and naïve realism by its critics. Contra the representationalist, the direct realist holds that what justifies S in believing p is the state of affairs that is the object of the belief and not the “immediate apprehension” of the state of affairs. Although it is curious which of these theories of perception is true, commitment to either theory is unnecessary for the purposes of the present discussion. It is sufficient to say that perception is a rich source of nondoxastic justification.
6 Now, suppose S forms the belief that q, where q = I have pain. Then, S’s experience of having pain justifies her in believing q. In this case, it is not perception, but (2) introspection that justifies S’s belief. Again, suppose S forms the belief that r, where r = a black marker just flew across the classroom. What justifies S in believing r is her (3) memory of seeing a black marker fly across the classroom. In this instance, it is not perception or introspection that justifies S’s belief, but memory. Finally, Suppose S forms the basic belief that m, where m = all red triangles are triangles. S’s justification for believing m is nondoxastic, but is not grounded in perception, introspection, or memory; rather, S’s justification is (4) a priori. These are the four nondoxastic sources of justification of basic beliefs. They are perception, introspection, memory, and the a priori. Now that we have a grasp of the sources of nondoxastic justification, we shall investigate an argument against the possibility of nondoxastic states conferring justification on beliefs.
4. Arguments Against Nondoxastic Justification 4.1 The E.T. problem In “The Raft and the Pyramid”, Ernest Sosa (1980) gives a fantastic argument against nondoxastic justification. Suppose an extraterrestrial—E.T.—has a radically different observational system than humans, yet most of his beliefs about his environment are justified.2 Also assume that E.T.'s belief-forming processes are somewhat analogous to ours. To grasp the analogy, let us make the rather simplified assumption that our belief-forming process works something like this: when the human, Smith, has a visual
Sosa formulates the argument in terms of knowledge. I am substituting justification for knowledge because it is nondoxastic justification that I am concerned with in this paper.
7 experience of F, she forms the belief that F. Analogously, when E.T. has the experience of Ψ, he forms the belief that Ψ. One reason for thinking Smith and E.T. are justified in believing F and Ψ, respectively, is that both beliefs are derived from or have their origins in the experience of F and Ψ, respectively. But there is a problem, which Sosa presents in the form of a dilemma: ...regarding the epistemic principle that underlies our justification for believing that something here is [F] on the basis of our visual experience of something [F], is it proposed as a fundamental principle or as a derived generalization? (146). If the principle is a derived generalization, then we have no reason for accepting it unless we discover the principle from which it is derived. Sosa quickly abandons this project in favor of the second horn of the dilemma. If the principle is in fact a fundamental principle, then the basis of our belief that something is F must include not only visual experiences, but also auditory experiences, tactile experiences, olfactory experiences, and so on. What these have in common is that they are all sensory experiences that involve sensible characteristics. Now, what would such a fundamental principle look like? Here is the principle Sosa offers to account for the justification of our beliefs by way of sensory experience. I call it the Perceptual Epistemic Principle (PEP): If Φ is a sensible characteristic, then the belief that there is something with Φ before one is (prima facie) justified if it is based on a visual experience of something with Φ in conditions that are normal with respect to Φ (146). Sosa points out that there are two problems with PEP. The first is that we lack a viable conception of sensible characteristics, because these characteristics are so diverse. They include colors, shapes, tones, odors, and so on. I suspect that if Sosa would have kept listing sensible characteristics, then he would have eventually created a complete list of sensible characteristics. A complete list would give us a notion of sensible
8 characteristics that we could work with: very simply, a sensible characteristic could be any one of the items on the list or any bundle of combinations of items on the list. Although I like this idea, part of the problem with such a list is that it would be a list only for us. Perhaps we would need a radically different or partially overlapping list for E.T. Sosa’s second problem with the PEP is that even if we had a good notion of sensible characteristics, it is not obvious that the principle would be fundamental enough to encompass beings like E.T. Sosa thinks we may imagine E.T. not having any sensory experiences, yet E.T. succeeds in forming true beliefs about the world. If such a being could exist, then we have to wonder what makes sense experience so special, since it would play no role in providing justification for E.T.'s beliefs about the state of the world. Now, if all of this speculation about E.T. were the case, then the Perceptual Epistemic Principle would no longer be a fundamental principle; rather, it would be a derived generalization only applicable in instances of human belief. We would need a more abstract and fundamental principle to cover both human and E.T. beliefs about their respective worlds. As it is, we are in the dark as to what that fundamental principle would look like. In response to Sosa’s argument, I doubt the importance of formulating a principle that encompasses both E.T. and humans. In defending my doubt, it is important to remember that we have a good idea of the role of contemporary epistemology. According to Alvin Goldman (1992), our job as epistemologists is to describe how human beings acquire and employ epistemic concepts, and then prescribe improvements for the practice of believing and knowing. Nowhere does he mention that part of our
9 project involves understanding and describing the epistemological practices of extraterrestrials. Nor does Goldman—or any other epistemologist—claim that normative epistemology should take into account the practices of any rational species other than human beings. On his view, epistemology is part cognitive science and part conceptual analysis. As epistemologists looking from the standpoint of cognitive science, we should not care about the belief system of E.T. because (1) E.T. probably does not exist, and (2) even if he did exist, we would not have to change our views about our cognitive processes to understand his cognitive processes. Furthermore, it is doubtful that we would be able to understand how E.T.’s belief-forming system works, let alone formulate abstract principles for both him and us. Maybe his system would be so radically different from ours that we could not say anything about it at all. In light of this, I propose we figure out humans first and then worry about E.T. when he pays us a visit. It may be objected that we could say the same thing about all the other characters that appear in epistemological thought experiments, because these characters do not exist either. I fail to see the force of this objection, because I think epistemology should be naturalized. By ‘naturalized’, I mean ‘explained in non-epistemic terms’. The nonepistemic terms that are the best candidates to explain epistemic terms belong to the realm of science and, more specifically, to the realm of cognitive science and its cognate disciplines. If you think epistemologists should wait for all the verdicts to come in from the cognitive scientists before we start analyzing epistemological concepts, then you will not have a worry about E.T., but you will have to worry about finding a job as an epistemologist. On the other hand, if you think naturalizing epistemology to that degree is a dangerous thing to do, then E.T. will present big problems for you. I prefer a
10 moderate approach. Cognitive science has produced a huge corpus of data, analysis, and theory. In light of this we, qua epistemologists, have more than enough concepts on our plate to analyze. You might object to my approach by saying, “Look, we have strong reasons to believe that our perceptual beliefs are justified. They also happen to be a) reliably formed, b) beliefs about which we have metabeliefs attributing a high degree of reliability, and c) accompanied by sensory states. Which of a)-c) is why they’re justified? We have to look to counterfactual cases (E.T., et al) where not all our criteria for justification obtain, and then see whether beliefs are justified or not.” We can circumvent this objection by appealing to the usefulness of rigorous scientific research. We do not have to invent counterfactual cases that at best provide us with intuitions about the sources of justification. Instead, we can construct scientific experiments that are designed to eliminate certain possibilities and that use actual people as data points. Surely we can find a person who has perceptual beliefs accompanied by sensory states and metabeliefs, but whose beliefs are not reliably formed. It is not that big of a stretch to suppose that that person could provide us with the kind of information we would need to show that reliability is a necessary component of justification. But even if it were too big a stretch for some, it certainly isn’t as big a stretch as basing one’s epistemology on a sci-fi character like E.T
4.2 Sellars’ Dilemma [by way of (the old) Bonjour] Perceptual experience is like Aristotle's unmoved mover: it can give justification, even though it does not have justification. The Sellarsian dilemma (Sellars 1956) is
11 designed to show that this is impossible: nothing can confer justification that does not have justification to begin with. Suppose S has the experience of being appeared to redly and roundly. (1) This kind of experience is either propositional (like a belief, it has a truth-value) or nonpropositional. (2) If S's experience is propositional, then it is capable of conferring justification, but it is in need of justification itself. Because the experience does not have justification, it cannot terminate a regress. (3) If S's experience is nonpropositional, then it does not need justification, but it is incapable of conferring justification. (4) Therefore, either way, nondoxastic experience could not be responsible for justification. That is, it is impossible for there to be something that is not in need of justification, but can confer justification. One may respond to the Sellarsian dilemma in one of three ways: (i) nondoxastic experience is in some sense propositional. (ii) nondoxastic experience is nonpropositional. Finally, (iii) the propositional status of nondoxastic experience is irrelevant because epistemic properties supervene on nonepistemic properties. There is an important distinction between (i) and (ii), and (iii).3 (i) and (ii) tell us that the justificatory ground for basic beliefs is sense experience, which may be either propositional or nonpropositional. In other words, sensory experience gives justification to a basic belief if and only if the basic belief is based on the sensory experience. In this instance the basing relation may or may not be a causal relation. On the other hand, the notion of supervenience expressed by (iii) is what makes a basic belief justified. This relationship is supposed to be non-causal as there is, in effect, a state of affairs that is metaphysically sufficient for the basic belief’s being justified. One possible supervenience base is the instance of a reliable cognitive process. If justification 3
We owe this distinction to Jack Lyons’ “Nondoxastic Justification and the Brown Argument”.
12 supervenes on a reliable cognitive process and the basic belief was formed by said reliable cognitive process, then the belief will be metaphysically justified. Nondoxastic foundationalism is a popular theory because it confirms our intuition that there is something special about sense experience: sense experience can serve as the grounds for a belief while also providing the supervenience base for justification. Because sense experience plays such a crucial role in the justification of basic beliefs, we must investigate it more fully.
5. Steup’s Unrestricted Foundationalism In “Unrestricted Foundationalism and the Sellarsian Dilemma”, Matthias Steup (2000) defends the claim that appearance states justify basic beliefs. He also claims that physical object beliefs can be basic and they are basic when they are justified by appearance states. In making these claims, Steup commits himself to nondoxastic sources of justification. Steup raises several good reasons to prefer a theory that allows physical object beliefs into the corpus of basic beliefs rather than beliefs about appearance states. The best reason he gives for this preference is supervenience. If beliefs about appearances terminate the regress of justification, then we must admit that justification supervenes on other epistemic properties. This position does not keep with the spirit of explanation in epistemology, in which the goal is to explain epistemic properties in non-epistemic terms. Because appearance states themselves terminate the regress and appearance states are non-epistemic, Steup can explain justification in non-epistemic terms. Thus, his view is
13 compatible with supervenience. How is it, then, that appearance states terminate the regress of justification by justifying basic beliefs? Appearance states—“A-states” for short—justify basic beliefs, because A-states have the properties necessary to justify basic beliefs. A-states are propositional attitudes that involve: 1) concept application, 2) discriminatory competence, and 3) propositional content. For an object to appear to you as an F, it must be presented to you as having the property of being an F. Steup uses a tomato as an example. In order for a tomato to appear to you as a tomato (and not an orange, let’s say), it must be presented to you as having the property of being a tomato. For it to appear to you as having the property of being a tomato, you must a) have the concept of a tomato, and b) apply your tomato concept to your experience. This is concept application, which is necessary for being in an A-state. Discriminatory competence requires the possession of the concept of F-ness, and the capacity to associate instances of F-ness with certain sensory qualities. This capacity enables you to tell one A-state from another. To use the tomato example again, you can tell you are in the state in which it appears to you that an object is a tomato because you can match your tomato concept to the various sensations you are having. Without this capacity, you would not be able to tell tomatoes from oranges. From the notion of discriminatory competence, it follows that you cannot be appeared to in certain ways because you cannot discriminate one appearance from another. You cannot be in the state in which it appears to you that there are 5999 leaves on a tree, because it is outside the realm of human ability to tell a tree with 5999 leaves from a tree with 6000 leaves.
14 Finally, and most importantly, A-states have propositional content. If you are in a state in which it appears to you that an object is an F, then the propositional content of the A-state is this object is an F. With respect to the tomato example, the propositional content of that A-state is this object is a tomato. A-states are a type of intentional state, where an intentional state is understood as a mental attitude directed toward a proposition. Beliefs, fears, hopes, wishes, suspicions, and desires are also intentional states, as are perceptual states, which Steup defines as veridical A-states. In addition to perceptual states, introspective, intuitional, and memorial states are also A-states. These are modes of presentation that, together with propositional content, identify and individuate particular A-states. One of the most interesting aspects of Steup’s account is his explanation of how A-states justify beliefs. Suppose you are in the state in which it appears to you that something is red. Let’s use A:Fr as shorthand for this state. Steup thinks A:Fr justifies you in believing that there is something red, because you have evidence for believing that A:Fr is a reliable indicator that something is red. Steup calls this view presumptive reliabilism. Presumptive reliabilism is limited, since works only for prima facie justification, which is a kind of justification that admits of defeasibility. Basic beliefs require ultima facie justification—justification that is indefeasible relative to a person’s total body of evidence. Steup tweaks presumptive reliabilism to get ultima facie justification by requiring that A-states be reliable indicators in one’s current circumstances. Here is Steup’s (2000) definition of ultima facie justification: S’s being in the state A:Fa ultima facie justifies S in non-inferentially believing that a is F if and only if (i) S non-inferentially believes that a is F; (ii) S is in the state A:Fa (iii) S has undefeated evidence for believing that, under the present
15 circumstances, his being in the state A:Fa is a reliable indicator of a’s being F (87-88). In a footnote, Steup acknowledges that his definition appeals to the epistemic notion of evidence. Thus, his account of ultima facie justification is problematic if he wants to adhere to the principle that epistemic properties supervene on non-epistemic properties. The second point he makes in the footnote is that his account does not depend on beliefs about the reliability of A-states; rather, his account requires that the agent have evidence for believing that A-states are reliable indicators of physical objects. Still, it makes no sense not to infer that a person who has the requisite evidence will also have the concomitant belief. Moreover, because evidence of the reliability of A-states together with accompanying beliefs about the reliability of one’s A-states is the defining feature of Steup’s definition of ultima facie justification, we can reasonably infer that Steup’s account is internalist. We now have a robust enough description of A-states to understand how Steup uses them to side-step the Sellarsian Dilemma. Remember that Sellars’ Dilemma is designed to prove that A-states cannot stop an infinite regress of justification. The first horn of the dilemma is that if A-states are nonpropositional, then they cannot have justification, and therefore cannot confer justification on beliefs.4 The second horn is that if A-states are propositional, then they can have justification and therefore give justification, but they must be in need of justification themselves. If A-states need justification, then they cannot serve as justifiers for basic beliefs, which means our putative basic beliefs are not basic after all. Because Steup claims that A-states are propositional, he must do battle with the second horn of the dilemma. 4
Steup concedes this point in his short online paper, “Foundationalism, Sense-Experiential Content, and Sellar’s Dilemma”
16 The argument for the second horn of the dilemma works like this: Assume A-states have propositional content. Then, 1) 2) 3) 4) 5)
If A-states have propositional content, then they must be able to transmit justification to basic beliefs. If A-states are capable of transmitting justification to basic beliefs, then they must already possess justification. If they already possess justification, then they are in need of justification themselves. If they are in need of justification themselves, then they cannot terminate the regress of justification. Therefore, A-states cannot termination the regress of justification.
Borrowing from James van Cleve’s work (1979), Steup invokes the distinction between transmission and generation of justification to dismantle the first premise. The idea is that since justification can be transmitted from one belief to another ad infinitum, if we are to avoid skepticism and end the regress, then there must be a point at which justification is generated. A-states provide that point of generation because a) they have propositional content, and b) we have evidence for believing that they are reliable indicators of the presence of actual physical objects. Before I go further into Steup’s rebuttal, I would like to point out that invoking the distinction between transmission and generation of justification begs the question. The main question for foundationalism is whether or not there is a foundation of justified basic beliefs or instead of a foundation, an infinite regress of justified beliefs. To answer this question with the transmission/generation distinction is to answer that there must be a foundation where justification is generated. But this is exactly what we want to know! Since this is question-begging, this line of response to the second horn of the Sellarsian Dilemma is fallacious and therefore unacceptable.
17 Steup’s second tactic is to show that premise two is false and that A-states can justify basic beliefs even though A-states are not the kinds of things that have justification. His argument goes like this: 1) 2) 3) 4) 5) 6)
A-states have propositional content. If A-states have propositional content, then they are a species of intentional states, i.e., mental states directed toward a proposition. Some intentional states do not have truth-vales (e.g., fears, hopes, desires, and wishes). If intentional states do not have truth-values, then they are incapable of possessing epistemic justification. A-states are intentional states that do not have truth-values. Therefore, A-states are incapable of possessing epistemic justification.
This argument hinges on the following question: Are A-states more like beliefs or more like hopes, fears, and desires? Steup claims that A-states are more like hopes, fears and desires. Together these make up the class of intentional states that do not have truthvalues; thus, these intentional states cannot possess justification. Beliefs, on the other hand, have truth-values and thus may possess justification. Since beliefs and A-states belong to different classes of intentional states, having propositional content does not turn A-states into belief-like states any more than having propositional content turns desires into belief-like states. For this reason, A-states are not the type of mental states that can be justified. Is this a good argument? In terms of validity, it adheres to the rules of logical inference, but in terms of soundness, several premises are suspect. Let’s consider the truth of each of the premises. 1) A-states have propositional content. It may very well be the case that Steup is grappling with the wrong horn of the Sellarsian Dilemma. It is by no means clear that appearance states have propositional content.
18 Moreover, Steup has done nothing to forward an argument for this claim. The best he gives us is this: A-states are sense-experiential states in which it appears to one that so-and-so is the case. Consequently, they have propositional content…If you look at an object that appears to you as a tomato, then you are in a sense-experiential state the propositional content of which is: this object is a tomato (82). Steup has not given a reason—let alone a good enough reason—to ascribe propositional content to A-states. We could just as easily conclude that A-states have qualia as their content. So, when you look at an object that appears to you as a tomato, then you are in a sense-experiential state the content of which is: tomato quaila. We could give further description of our tomato qualia as red, roundish, juicy, seedy, firm, acidic, and so on. In saying this, I am not claiming that A-states have no content, because I think they do. My contention is that their content is not necessarily propositional—it could be phenomenal. Furthermore, nothing Steup writes in his paper recommends the view that A-states are propositional over the view that they are phenomenal. Positing A-states with phenomenal content might be an even better move for Steup, since there is no debating the claim that phenomenal mental states—like pain, for example—do not admit of epistemic justification. If Steup were to take this route, he could potentially have a better explanation of the content of A-states, but he would have to take up the issues with the first horn of the Sellarsian dilemma. 2) If A-states have propositional content, then they are a species of intentional states, i.e., mental states directed toward a proposition. I assert that our best epistemological theories and our best defenses against dilemmainduced skepticism are born out of cross-pollination with cognitive science. In order to determine whether or not A-states are a species of intentional states, we must first find
19 out what it is for a cognitive state to be an intentional state. We must find out if our theories of what it is for a cognitive sate to be about something, or to refer to something, or to be true of something—we must find out if those theories and our epistemological theories have any kind of congruency or compatibility. I have the same leanings when I consider Steup’s third premise: 3) Some intentional states do not have truth-vales (e.g., fears, hopes, desires, wishes, etc). To ascertain the truth of this premise, we have to find out what is it for a cognitive state to have a truth condition. Fears, hopes, desires, wishes and the like do not have truth conditions, but they do have satisfaction conditions. The satisfaction conditions of an intentional state justify us in claiming that some fears are irrational and that some desires are fulfillable, and that some hopes are realistic. When we consider the mental state expressed by “I fear that a million poisonous snakes will crawl out of the walls and kill me”, we feel that that fear is irrational because its content a million poisonous snakes will crawl out of the walls and kill me picks out a state of affairs that is unlikely to ever obtain. The improbability of this content being satisfied justifies us in asserting that the person has an irrational fear. We may construct similar examples for fears, desires and hopes. 4) If intentional states do not have truth-values, then they are incapable of possessing epistemic justification. 5) A-states are intentional states that do not have truth-values. Whether or not A-states are intentional states or their content is propositional, one might have the intuition that A-states do not have truth-values. On the other hand, we are certain that beliefs do have truth-values. We can see that the strength of 4) and 5)
20 depends on the strength of the analogy between A-states and mental states that do no not have truth-values and the strength of the disanalogy between A-states and beliefs, which do have truth-values. Steup invokes a tenuous distinction to show the difference between beliefs and Astates: He says, What, then, is the difference between these two types of intentional states? The difference is that we can be related to a proposition in two different ways: assertively and non-assertively. When you are related to a proposition p because you are convinced of p, believe that p, or suspect that p, then you are necessarily related to p in an assertive way: you take p to be true…However, when we hope, fear, or desire that p, wonder whether p, contemplate p, or understand p, we are not related to p in an assertive way. This, I suggest, explains why it would be odd to ascribe truth-values to A-states (95). We are to infer from this passage that we are related to A-states in a non-assertive way; hence, we do not take A-states to be true. Rather, we take A-states to be either veridical or non-veridical. Because we do not relate to beliefs in this way, beliefs and A-states lack the kind of similarity that allows us to claim that A-states can be epistemically justified. I admit that this turn in Steup’s argument puzzles me. For one, it does not strike me that whether or not an intentional state has a truth-value depends on me taking its proposition to be true. That opens the door to an unwholesome kind of relativism about truth. Aside from that, if the propositional content of a belief and an A-state is the same (e.g., that the tomato is red), then I claim that the belief is true if the proposition that the tomato is red is true; likewise for the A-state. I would not, however, assert that the belief is veridical, although I would assert that the A-state is. The propositional content of beliefs and A-states are either true or false. They both have truth-values. The difference between these two states is their mode of evaluation: beliefs are justified or unjustified
21 whereas appearance states are either veridical or non-veridical. The difference between beliefs and A-states does not depend on content or on an assertiveness relation. Their difference depends on how we evaluate them Michael Heumer’s apprehensions are much like Steup’s A-states. Heumer (2001) gives a convincing account of A-states that involves the claim that they are assertive mental representations. Apprehensions are mental states that are involved in awareness. They have representational content, which means that they are the kinds of mental states that are always about something. Beliefs, fears, wishes, hopes, desires, imaginings, perceivings, and so on, are a few examples of mental states that have aboutness. Contrast these with tickles, itches, and pains—mental states of which it makes no sense to speak of what they are about, because they are not representational. Apprehensions are a kind of representation. They are (1) mental representations that (2) represent their contents as actualized. This second feature Heumer calls “assertiveness”. Assertive mental states always purport to represent the world as it actually is, although it is oftentimes the case that our apprehensions are wrong and the world gets misrepresented. Where does Heumer’s account place A-states in relation to beliefs? The best we can tell is that they are placed side-by-side: Beliefs are a kind of apprehension, as are the experiences we have when perceiving things…However, desires, emotions, and exercises of the imagination are not apprehensions, because they do not represent their contents assertively (54). If Heumer is right, then states in which we are appeared to are assertive just as our beliefs are assertive, which is to say that the representational content of beliefs and A-states is presented as true or actualized. In Steup’s terminology, we take A-states to be true.
22 According to Heumer’s view, Steup gets it right that fears, desires, and hopes are not assertive; his error is in the assertiveness-based classification of A-states with fears and desires, and not with beliefs.
Conclusion In order to defeat the second horn of the Sellarsian Dilemma, Steup must show that the first two premises of the argument in favor of the second horn are false. Remember the first two premises are: 1) If A-states have propositional content, then they must be able to transmit justification to basic beliefs. 2) If A-states are capable of transmitting justification to basic beliefs, then they must already possess justification. Against the first premise, Steup invokes the distinction between transmitting and generating justification to show that A-states are generators and not transmitters of justification. I found his appeal to this distinction to be question-begging. Thus, the first premise of the second horn of the Sellarsian dilemma stands. Against the second premise, Steup argues for a disanalogy between beliefs and Astates. The difference, he claims, is that because we are related to beliefs assertively, we take their propositions to be true. Our relationship to the propositional content of Astates is non-assertive; hence, we think A-states lack truth-values. If we accept that in order to possess epistemic justification, intentional states must have truth-values, then Astates cannot possess justification. Aside from the fact that Steup’s explanation of this distinction is horribly deficient, we may reject the distinction on the grounds that we are willing to assent to the truth or falsity of the propositional content of an A-state, although we evaluate the state
23 itself as being either veridical or non-veridical. Since Steup’s disanalogy is weak, he has failed to prove that A-states are dissimilar enough from beliefs to render A-states incapable of possessing epistemic justification. Finally, because Steup neither sufficiently explains nor argues for the assumption that A-states are propositional, it is possible that his argument against the second horn of the dilemma is a non-starter. If A-states are nonpropositional, then nondoxastic foundationalist of Steup’s ilk must focus their energy on winning that battle.
References Bonjour, Laurence (1978). "Can Empirical Knowledge Have a Foundation?" In Epistemology: An Anthology. ed. Ernest Sosa & Jaegwon Kim. Massachusetts: Blackwell Publishers, 2000, 261-273. Chisholm, Roderick (1964). "The Myth of the Given" In Epistemology: An Anthology. ed. Ernest Sosa & Jaegwon Kim. Mass: Blackwell Publishers, 2000. Goldman, Alvin (1992). "Epistemic Folkways and Scientific Epistemology." In Epistemology: An Anthology. ed. Ernest Sosa & Jaegwon Kim. Mass: Blackwell Publishers, 2000, 438-443. Heumer, Michael (2001). Skepticism and the Veil of Perception. Lanham, MD: Rowman & Littlefield. Lyons, Jack. “Externalism and the Sellarsian Dilemma”. Draft. Sellars, W. (1956). “Empiricism and the philosophy of mind” In H. Feigl & M. Scriven, eds. Minnesota studies in the philosophy of science, vol 1. Minneapolis: University of Minnesota Press.
24 Sosa, Ernest (1980). "The Raft and The Pyramid." In Epistemology: An Anthology. ed. Ernest Sosa & Jaegwon Kim. Mass: Blackwell Publishers, 2000, 134-153. Steup, Matthias (1996). Introduction to Contemporary Epistemology. New Jersey: Prentice-Hall. _____ (2000) “Unrestricted Foundationalism and the Sellarsian Dilemma”, Grazer Philosophische Studien, 60: 75-98, 2000. _____ (2001) “Foundationalism, Sense-Experiential Content, and Sellars’s Dilemma”, Colloquium Paper, Pacific Division Meeting of the American Philosophical Association, San Francisco. URL =