Anti-Skeptical Arguments∗

Daniel Bonevac

“... the word of truth is sent from God.”

—Justin Martyr [1885], viii, viii, i.

“But the light of the human mind is God.... Knowledge of the truth isDivine.”

—Lactantius [1885], iii, iii, i.

Skeptical arguments attack the possibility of knowledge—but also of other things, in-cluding truth, communication, and content. As David Hume [1739] famously noted,“they admit of no answer and produce no conviction. Their only effect is to cause thatmomentary amazement and irresolution and confusion, which is the result of scepti-cism.” With a bit of philosophical judo, however, they can be turned into arguments forthe existence of God. I will present six such arguments here: arguments from truth,knowledge, intersubjectivity, communication, interpretation, and content.1 They areclosely related, and might even be seen as versions of the same core argument.

The strategy of converting skeptical arguments into arguments for something ob-jective and independent of human minds originates with Plato. I begin, therefore, byconsidering both epistemological and metaphysical arguments for skepticism. I thenconsider some Platonic moves turning them into arguments for the forms. Philo ofAlexandria and early Church Fathers such as Justin Martyr, Lactantius, and Augustineadd premises that allow us to extend those arguments to arguments for the existenceof God. That effort culminates in Augustine’s argument from truth. The argumentsfrom truth, knowledge, communication, intersubjectivity, interpretation, and contentthat I develop are Augustinian in spirit but independent of many aspects of Platonic orneo-Platonic metaphysics.∗I developed these ideas in a seminar on Natural Theology East and West at the University of Texas at

Austin in 2015. I am grateful to my co-instructor, Stephen Phillips, and to our students for their insightfulcomments and criticisms as well as their enthusiasm and support.

1The argument from knowledge is, in essence, what Alvin Plantinga [2007] calls the Putnamian argumentor the argument from global skepticism. The argument from content is what he calls the Kripke-Wittgensteinargument from plus and quus. Plantinga’s paper, under that latter heading, simply says, “See SupplementaryHandout,” and I have not been able to track down a copy of that handout. So, I make no guarantee thatthe argument from content as I develop it corresponds to the argument Plantinga had in mind. Some ofthe arguments I present are relatives of Plantinga’s arguments from intentionality, positive epistemic status,reference, intuition, and the confluence of proper function and reliability, for all could be seen as anti-skeptical arguments, turning a skeptical challenge into an argument for God’s existence.

1

I shall develop a general picture of skeptical challenges, focusing especially onthat posed by Saul Kripke’s [1982] interpretation of Wittgenstein [1953]. It raises askeptical puzzle—one variously described as concerning meaning, content, and rule-following—and then offers a skeptical solution. Paul Boghossian [1989, 1990] usesthe skeptical puzzle to argue for a realist, non-reductive, and judgment-independentaccount of content. My goal is to extend Boghossian’s argument to arguments for theexistence of God.

The key idea is straightforward. Content is, among other things, infinitary andnormative. These features of content, Kripke argues, make it impossible to account fora speaker’s content in terms of facts about that speaker’s past usage, mental history, oreven dispositions, since a finite being’s dispositions are finite. I shall go further: Thenormative character of content transcends any naturalistic relation or set of facts. Itsinfinitary character transcends any relation to any finite set of finite minds. Contentthus requires a non-naturalistic relation to an infinite set of finite minds or to an infinitemind. The only live options for accounting for content are thus pragmatism and theism.If pragmatism fails, as I shall argue it does, then theism is the only remaining option.

1 Skeptical ArgumentsSkepticism, as generally understood (e.g., by Striker [1980]), holds that knowledge isimpossible (in its Academic form) or at any rate refrains from asserting that knowl-edge is possible (in its Pyrrhonian form). That understanding locates skepticism firmlywithin epistemology. But skeptical challenges can extend beyond the theory of knowl-edge to theories of meaning and reference and thus to metaphysics in general. W. V.Quine’s [1960] thesis of the indeterminacy of translation is a skeptical challenge, as isthe Kripke-Wittgenstein puzzle concerning plus and quus [Kripke 1982].

Say that A has a belief or perception s that portrays the world as being a certainway—that portrays an object or circumstance o as having a property P (or of an indi-vidualized property Po, or a trope P/o; the differences are immaterial for our purposes).We can ask whether A’s mental state portrays the world as it actually is, at least in thisrespect. That is, in this case, we can ask whether o really does have P.

To make this slightly less abstract, imagine that A sees a triangle and thinks, “That’sa triangle.” On what I will call a traditional understanding of meaning, running, roughly,from Plato through Locke, the word ’triangle’ stands for an idea or concept, which wemight designate as *triangle*. Suppose that someone else, B, looks at the same sceneand also thinks, “That’s a triangle.”

We can now raise a number of skeptical challenges. The skeptic challenges theconnection between two classes of items:

• between

– what A means or refers to and

– what B means or refers to,

• between

2

– what A means or refers to and

– objects and properties in the world, and

• between

– what A means and

– any content or interpretation of what A means, even that of A himself.

The skeptic sees the possibility of a mismatch between mental or linguistic entitiesand something else—something those entities are meant to portray or represent, suchas states of the world or mental or linguistic contents, or something to which theyare meant to be similar or identical, such as other mental or linguistic entities. Thechallenge moreover may be epistemic or constitutive. The skeptic may raise the meta-physical or epistemic possibility of mismatch [Boghossian 1989, Greco 2012]. Fromthis point of view, skeptical challenges fall into six categories (Table 1).

Table 1: Kinds of Skeptical Challenge

Constitutive EpistemicMental state – mental state intersubjectivity communication

Mental state – world truth knowledgeMental state – content content interpretation

Let’s examine these in turn.

1.1 Intersubjectivity and CommunicationFirst, consider challenges involving the relation between one person’s mental states andthose of another: the intersubjectivity and communication challenges. What guaranteesthat A and B are in similar mental states, that they share the same concept, refer tothe same thing, and, so, in some sense think the same thought? We might neutrallyrepresent the situation this way, recognizing that A and B might not have the sameconcept of a triangle and so might not be using ‘triangle’ in exactly the same way:

A: That’s a triangleA (standing for the concept *triangleA*)B: That’s a triangleB (standing for the concept *triangleB*)

We might observe the same point about the object these two have in mind. Perhaps oneintends to refer to a physical object, while the other intends to refer to its shape. So, itmight be better to represent the situation as

A: ThatA (standing for objectA) is a triangleA (standing for the concept*triangleA*)B: ThatB (standing for objectB) is a triangleB (standing for the concept*triangleB*)

3

We can now ask how A, B, or some third party could know whether ‘triangleA’ meansthe same as ‘triangleB’, whether *triangleA* = *triangleB*, whether ‘ThatA’ and ‘ThatB’corefer, or whether objectA = objectB. That is the epistemic question. We can also askthe constitutive question: could ‘triangleA’ mean the same as ‘triangleB’, etc.? Theworry could be either that A and B could not possibly be thinking the same thoughtor that there might be no fact of the matter about whether they are thinking the samethought.

We might go further, inspired by Quine [1960], and realize that A and B might noteven share concepts of predication, thus taking this even further:

A: ThatA (standing for objectA) (is a)A triangleA (standing for the concept*triangleA*)B: ThatB (standing for objectB) (is a)B triangleB (standing for the concept*triangleB*)

Perhaps A, a nominalist, thinks that ‘triangle’ applies to that thing, while B, a Platonist,thinks it exemplifies triangularity.

We can, as we have seen, interpret these challenges constitutively or epistemically.That is, we can think that the possibility that A and B are thinking different thoughts isa metaphysical possibility—maybe even a metaphysical inevitability—or an epistemicpossibility. We may wonder whether A and B are or could be thinking the same thought,or whether A, B or a third party could know whether they are thinking the same thought.

The epistemic challenge assails communication. If I cannot know what you meanwhen you speak, we cannot be said to communicate, at least in any sense that couldgenerate common knowledge. Thus an ancient skeptic, Cratylus, gave up on the pos-sibility of communication and refused to speak [Popkin 2002, 50]. In a similar spirit,Clitomachus complained that he could never figure out what the skeptic Carneadesactually thought [Striker 1980].

The constitutive, metaphysical challenge assails the very possibility of intersubjec-tivity. Can two people be in the same mental state—not the same token state, of course,but the same type of state? A relativist answers no, going on to say that meaning andtherefore truth are relative to the set of concepts, the perspective, or the conceptualframework of each individual. A different sort of skeptic might deny that there is anyfact of the matter about whether two people are in the same mental state.

1.2 Knowledge and TruthSecond, consider the challenge concerning the relation of mental states to the world. Inthe situation we have been envisioning, A and B seem to agree in describing somethingas a triangle. But the skeptic’s usual challenge isn’t to interpersonal agreement andcommunication, but to an individual’s knowledge of the world. Think of A’s mentalstate as portraying an object o as having a property P. The question whether A’s con-cepts, terms, and referents match B’s has the same structure as the question whetherthey match the world, that is, in this case, the object and its property. We might repre-sent the situation as

4

A: ThatA (standing for objectA) (is a)A triangleA (standing for the concept*triangleA*)The world: o has property P

We can now ask the epistemic question, how A or anyone else could know whether‘triangleA’ or *triangleA* stands in the appropriate relation to P, whether ‘ThatA’ refersto o, whether ‘is aA’ stands in the appropriate relation to the relation we call predication(indicated by ‘has’ above), and whether A’s thought stands in the appropriate relationto the world. We can also ask the constitutive question of whether A’s words, concepts,and thoughts could stand in the appropriate relations to the world. The epistemic ques-tion represents the skeptic’s familiar challenge to knowledge. The constitutive questionrepresents a metaphysical challenge to the possibility of truth.

1.3 Interpretation and ContentThird, consider the relation between mental states and their interpretations or contents.We can use the situation involving two people, A and B above, to raise skeptical worriesabout the idea of content. Most simply, imagine that A and B are not different peoplebut different temporal stages of the same person or counterparts of the same personin different possible worlds. The questions of how we might know that two peoplemean the same thing by what they say or whether they could even mean the samething thus turns into the question of whether I could mean or know that I mean thesame thing I meant five minutes ago, or might have meant if today were not Saturday,if I had not just blinked my eyes, or if my cat were not about to jump off the table.But the puzzle is not just about modal or temporal parts. Think of B as someoneattributing content to A’s utterance and concepts. Think, in other words, of ‘ThatB’as an articulation of an interpretation of ‘ThatA,’ ‘triangleB’ as an articulation of aninterpretation of ‘triangleA,’ etc. We can then ask, just as we did above, how anyonecould know whether ‘triangleA’ means the same as ‘triangleB’, whether *triangleA* =

*triangleB*, whether ‘ThatA’ and ‘ThatB’ corefer, whether ‘is a’A means the same as‘is a’B, or whether objectA = objectB. We could also ask whether it is even possiblefor any of these things to hold. The epistemic question: How could anyone know whatA is referring to or what A means? The constitutive question: Is there any fact of thematter about what A is referring to or what A means? Are meaning and reference evenpossible?

This reasoning applies even to A himself. *TriangleB* in that case is not sim-ply *triangleA*, which is of course self-identical, but an account or interpretation of*triangleA*, perhaps A’s own account or interpretation. And A’s interpretation of hisown concepts is not necessarily correct. The same holds of linguistic items, as Plato’searly dialogues dramatically illustrate. Socrates shows his interlocutors that their inter-pretations of their own terms and concepts are inadequate. Content is opaque.

2 Skeptical Arguments in EpistemologyLet’s begin with epistemological skepticism. Classic skeptical arguments typically startfrom a mental state—typically, a perception or belief—and invoke a skeptical scenario

5

designed to undermine our faith in its veridicality or truth. The mental states in questionare intentional; they represent, stand for, or are of or about things, events, properties,or states of affairs. They present them as being a certain way. If they are that way, themental state is veridical (if it is a perception) or true (if it is a belief). If not, the mentalstate is illusory or not true.

Let’s start with some preliminary definitions, soon to be generalized. Call a situa-tion in which an agent has a veridical or true mental state directed at a state of affairs amatch for that mental state. Hilary Putnam [1981] is dismissive of this locution: “Butthe notion of a transcendental match between our representation and the world in itselfis nonsense” (134). In his view, the realism implied by ‘match’ and skeptical scenariosstand or fall together, and he is happy to dismiss the skeptical arguments to be advancedin this section as nonsense. But a match in the sense I am outlining does not have to be“transcendental,” whatever that means. You think that Concord is the capital of NewHampshire, and it is; that’s a match. I go to the eye doctor, look at the astigmatismchart, and see some of the radial lines as much darker than others; that’s not a match.A variety of authors, including Wittgenstein [1953] and Sellars [1956], find this kindof talk acceptable case-by-case or for a limited portion of discourse, but not globally;they think the skeptic can challenge any given putative item of knowledge, but not allat once. For now, let’s table these considerations, which will return in the section onpragmatism below.

A skeptical scenario for a mental state is a situation, indiscernible from a match forthat state from the agent’s point of view, in which the same mental state is illusory ornot true.2 In a skeptical scenario, the situation is indiscernible from a match not justgiven that mental state itself but given the totality of all the agent’s possible mentalstates. Someone seeing what appears to be a puddle on the road ahead may not be ableto tell whether this is a mirage or a veridical perception of a puddle, but traveling a bitfurther will reveal it to be one or the other. People under the influence of Descartes’sevil genius, however, cannot tell whether any of their experiences are veridical or anyof their beliefs are true, no matter how much experience they might accumulate or howmuch reasoning they do. The same holds of other skeptical scenarios, for example, thatI am a brain in a vat or someone trapped in the Matrix.

There is an interesting question whether all interesting skeptical scenarios are thusglobal. Consider Holliday’s [2012] definition: “The skeptic describes a scenario v inwhich all such beliefs [i.e., all beliefs the agent holds in v] are false, but the agentis systematically deceived into holding them anyway.” It seems possible that a demonmight deceive me on a proper subset of my beliefs while leaving others intact, however;consider a demon who deceives me solely with respect to my perceptual beliefs, forexample, or with respect to my beliefs about conscious beings. Indeed, the brain-in-a-vat scenario is of this kind; it is not clear how that scenario threatens my logical or

2Some writers (e.g., Williamson [2007] and Kung [2011]) include under the heading of skeptical sce-narios cases in which a belief is true but unwarranted, so that it does not count as knowledge. On thatunderstanding, Gettier cases, fake barn cases, and the like count as skeptical scenarios. I refrain from doingthat here for two reasons. First, these are not cases that are truly indiscernible from matches from the agent’spoint of view: The agent could come to learn that the other person also has ten coins in his pocket, thatthe key is not to Jones’s car, that there are many fake barns in the vicinity, etc. Second, the question ofwarrant pertains to knowledge, and so has bearing only within epistemology, having no obvious correlate inmetaphysical skepticism.

6

mathematical knowledge, much less my knowledge that I exist. Interesting skepticalscenarios can be and generally are more limited than challenges to all of everyone’sbeliefs. Hence, skepticism is worth taking seriously even if Wittgenstein and Sellarsare right that a truly global skepticism would be unintelligible.3

2.1 The Knowledge ChallengeGiven a perception or belief, we can distinguish situations that are matches, in whichthe perception is veridical or the belief is true, from situations that are not matches,in which the perception is illusory or the belief is not true, and specifically from thesubset of those that are skeptical scenarios, in which the agent cannot discover that theperception is illusory or the belief is not true. We begin with a skeptical premise:

(1) Given a state of mind s, among our epistemic possibilities are skeptical scenar-ios for s.

To go further, we need to think about what it would take to defeat the skeptic’sstrategy. There are many options concerning the needed relationship between skeptical

3Some ancient skeptical tropes do not generate skeptical scenarios in the sense with which I am con-cerned. For a superb survey, see Annas and Barnes [1985]. The argument from illusion, for example, appearsto be more limited, pointing to cases in which the senses have been deceived and urging the possibility ofdeception in any given case. It has an ∀∃ logical structure: in any perceptual situation, there is a possibilitythat the perceiver is suffering an illusion. Plato’s version of the argument from dreaming, however, is moregeneral:

Socrates. A question which I think that you must often have heard persons ask: How canyou determine whether at this moment we are sleeping, and all our thoughts are a dream; orwhether we are awake, and talking to one another in the waking state?Theaetetus. Indeed, Socrates, I do not know how to prove the one any more than the other,for in both cases the facts precisely correspond;—and there is no difficulty in supposingthat during all this discussion we have been talking to one another in a dream; and when in adream we seem to be narrating dreams, the resemblance of the two states is quite astonishing.Socrates. You see, then, that a doubt about the reality of sense is easily raised, since theremay even be a doubt whether we are awake or in a dream. And as our time is equally dividedbetween sleeping and waking, in either sphere of existence the soul contends that the thoughtswhich are present to our minds at the time are true; and during one half of our lives we affirmthe truth of the one, and, during the other half, of the other; and are equally confident of both.(158c, d)

Arguments from dreaming and from variability, especially species variability can generate global skepticalscenarios, which have an ∃∀ structure: there is a possibility that every situation is illusory. (I might be ofa species that systematically misperceives or misconceives the world, for example.) The question, ‘Whatif I am dreaming right now?’, does not describe a skeptical scenario, but ‘What if it’s all a dream?’ does.Zhuangzi too puts the argument from dreaming in a global form:

Once upon a time, I, Zhuangzi, dreamt I was a butterfly, fluttering hither and thither, to allintents and purposes a butterfly. I was conscious only of following my fancies as a butterfly,and was unconscious of my individuality as a man. Suddenly, I awaked, and there I lay,myself again. Now I do not know whether I was then a man dreaming I was a butterfly, orwhether I am now a butterfly, dreaming I am a man. [1889, 1923]

Plato and Zhuangzi in effect argue that we perceive the world in two radically different ways, in waking anddreaming states, but cannot tell which way is which or which way reflects the way the world is. Similarissues arise in cases in which a psychiatrist, say, tells someone that everything they have been perceivingand believing is delusional. (This is a popular dilemma in recent fiction. See, for example, “Normal Again,”episode seventeen of season six of Buffy the Vampire Slayer, or “Frame of Mind,” episode twenty-one ofseason six of Star Trek: The Next Generation.)

7

scenarios and matches:4

• We can rule skeptical scenarios out. [See DeRose and Warfield 1999.]

• Skeptical scenarios must be sufficiently rare. (We might give this a frequencyinterpretation—almost no scenarios are skeptical scenarios, or, stronger, almostall scenarios are matches—but we need not. This might be for example theinterpretation of an indicative or subjunctive conditional as understood by Adams[1970]. Or it might be straightforwardly statistical.)

• The closest scenarios are not skeptical scenarios, or, stronger, are matches. (Thisis the Lewis [1973] interpretation of a counterfactual of the form The agent is inmental state s �→ not Skeptical Scenario s or The agent is in mental state s �→Match for s, given the limit assumption.)

• The close scenarios are not skeptical scenarios, or, stronger, are matches. (Thisis the interpretation of a counterfactual of the form The agent is in mental state s�→ not Skeptical scenario for s or The agent is in mental state s �→Match fors as understood by Nute [1975] or Nozick [1991].)

• Scenarios are normally not skeptical scenarios, or, stronger, are matches. (Thisis the interpretation of a normality conditional of the form The agent is in mentalstate s→ not Skeptical scenario for s or The agent is in mental state s→Matchfor s as understood in a non-monotonic logic, as in Reiter [1980], McCarthy[1986], Asher and Morreau [1991, 1995], Brewka [1991], or Morreau [1997].)

• Skeptical scenarios must be sufficiently rare among some subset of the scenarios—the closest ones, sufficiently close ones, normal ones, etc.—or some other desig-nated set of scenarios. (This would be the interpretation of a similar conditionalon a neighborhood conception, as in Bonevac, Dever, and Sosa [forthcoming].)

• Skeptical scenarios are “somehow inferior” to matches [Brueckner 1992].

This list is by no means exhaustive. I will use “matches are closer than skeptical sce-narios,” or, turning it around, “skeptical scenarios are more remote than matches,” asplaceholders for whatever the appropriate condition might be. But I do not intend theselocutions to decide among the options I have outlined; they are to be understood asshorthand for an adequate characterization.

We can now formulate a second premise:

(2) We have no grounds for ranking skeptical scenarios for s more remote thanmatches for s.

A third premise:

4I am simplifying somewhat in thinking solely about possible relationships between skeptical scenarios;one might instead choose to focus on relationships between skeptical scenarios and other scenarios, includingcorrigible non-veridical states as well as matches. Or, one might choose to focus on relationships betweenmatches and non-matches, which might fit skeptical arguments such as the argument from illusion morefaithfully than what appears here in the text.

8

(3) We can know that a state of mind s is veridical only if we have grounds forranking skeptical scenarios for s more remote than matches for s.

The conclusion, of course:

(4) We cannot know that our states of mind are veridical.

This is one way to think of skeptical arguments; there are many others in the lit-erature. (See, for example, some of the approaches in Annas and Barnes [1985] andDeRose and Warfield [1999].) But this will be useful for bringing out structural simi-larities between various kinds of skeptical arguments.

2.2 The Communication ChallengeSo far, I have focused on the knowledge challenge. But we can generalize to obtainthe communication and interpretation challenges. In place of the mind-world or word-world connection we have so far discussed, consider the mind-mind connection. Howcould anyone know whether two people are thinking the same thought or making thesame assertion? Even if they use the same words, they may be using them differentlyand attach different meanings to them. In this context, a match for an agent’s mentalstate is a situation in which that state is sufficiently similar, perhaps type-identical toa designated target mental state. A skeptical scenario for a mental state is a situation,indiscernible from a match for that state from the agent’s point of view, in which thesame mental state is not sufficiently similar to some target state.

(5) a. Given an agent A’s state of mind s, among our epistemic possibilities areskeptical scenarios for s—situations indiscernible from a match from theagent’s point of view in which, for example, s is not sufficiently similar toany mental state t of B.

b. We have no grounds for ranking skeptical scenarios for s more remote thanmatches for s.

c. We can know that a state of mind s is sufficiently similar to t only if we havegrounds for ranking skeptical scenarios for s more remote than matches fors.

d. So, we cannot know that our states of mind are sufficiently similar to thoseof anyone else.

It follows that we cannot know whether anyone else thinks what we think, feels whatwe feel, or means what we mean. We cannot know whether we are communicatingwith anyone else or whether they are communicating with us. Of course, this appliesjust as well to our own temporal and modal stages. So, we cannot know whether weare thinking what we thought a moment ago or mean today what we meant yesterday.

2.3 The Interpretation ChallengeWe can generalize in a different way to encompass the interpretation challenge. In placeof the mind-world or mind-mind connection, think of a mind-content connection. In

9

this context, a match for an agent’s mental state is a situation in which that mentalstate has a designated content. A skeptical scenario for a mental state is a situation,indiscernible from a match for that state from the agent’s point of view, in which thesame mental state does not have that content.

(6) a. Given a state of mind s, among our epistemic possibilities are skeptical sce-narios for s—situations indiscernible from a match for s from the agent’spoint of view in which that state of mind does not have its designated con-tent.

b. We have no grounds for ranking skeptical scenarios for s more remote thanmatches for s.

c. We can know that a state of mind s has its designated content only ifwe have grounds for ranking skeptical scenarios for s more remote thanmatches for s.

d. So, we cannot know that our states of mind have their designated contents.

Some will think this is obviously absurd, since they will see an essential or intrinsicconnection between a mental state and its content. Perhaps they will think that a mentalstate is a content—of a linguistic entity, for example, or a quasi-linguistic entity suchas a proposition. So, it may be valuable to frame this argument separately for linguisticitems. Throughout this discussion, I have been thinking of the connections in questionas mind-world, mind-mind, or mind-content. But we might just as readily focus onlinguistic entities and think of the relevant connections as word-world, word-word, andword-content. The last will include the word-mind connection, if we identify contentswith mental entities. In any case it would be possible to add that connection to thepicture. I will not do that explicitly, for its tracks the mind-content and word-contentconnections closely. Throughout the remainder of this paper I will use ‘match’ and‘skeptical scenario’ broadly enough to include all of these interpretations.

(7) a. Given an expression s, among our epistemic possibilities are skeptical sce-narios for s—situations indiscernible from a match for s from the agent’spoint of view in which that expression does not have its designated content.

b. We have no grounds for ranking skeptical scenarios for s more remote thanmatches for s.

c. We can know that an expression s has its designated content only if we havegrounds for ranking skeptical scenarios for s more remote than matches fors.

d. So, we cannot know that an expression has its designated content.

In short, we cannot know what our words and utterances mean.One advantage of thinking about mind-mind and mind-content connections as well

as mind-world connections is that it shows that Putnam’s [1981] attempt to turn skepti-cal arguments into weapons against realism must sweep broadly if it is to succeed at all.Putnam’s idea that metaphysical realism inevitably leads to skepticism applies most di-rectly to the mind-world connection and indeed attempts to undermine it. If we dividemind from world, holding that reality and truth are radically non-epistemic [Putnam

10

1978], then, he contends, the skeptical argument against the possibility of knowledgeis insurmountable. If metaphysical realism is true—if the world is truly something in-dependent of the mind—then I have no way of discounting skeptical scenarios (e.g.,that I am a brain in a vat) and thus cannot know anything (e.g., that I am not a brain ina vat). But I do know that I am not a brain in a vat; thus, metaphysical realism is false.Or so Putnam argues.

Before we consider ways of defusing the argument, notice that it would apply justas readily to mind-mind and mind-content connections. There is nothing distinctiveabout the mind-world form of the argument. Your ideas and thoughts are or could bejust as external to me as my table. The same is true of the contents of my thoughts.Putnam’s thought experiment has all the brains connected together, so that we suffer a“collective hallucination” (6) and skeptical scenarios involving communication cannotarise. But we can set up a parallel scenario lacking this feature: maybe I alone am abrain in vat. Putnam’s collective scenario challenges the mind-content connection, ashis argument about reference shows. If we were all brains in vats, then my use of ‘brain’would no longer refer to brains, Putnam contends, and I could not think anything withthe content of ‘I am a brain in a vat.’ The skeptical scenario thus threatens the contentof ‘brain’ as well as my knowledge that I am not a brain in a vat. So, an anti-realismmotivated by Putnam’s arguments must apply to communication and content. Realityand truth emerge as radically epistemic. So does content. Your thoughts cannot bemetaphysically independent of my thoughts, if we are to be able to communicate.

None of the language I have been using, note, commits me to anything like a “copy”theory of content. Nor does it force me to reify contents. Talking of relations betweenlinguistic or mental entities and their contents is natural and convenient, but the skepti-cal problem arises just as well if we think of having a content as being of a certain kindor playing a certain role. Counterfeit kinds and roles are just as epistemically possibleas counterfeit contents of other kinds.

3 Skeptical Arguments in MetaphysicsMy reason for choosing this particular formulation of the skeptical argument is thatit generalizes to what I have called metaphysical skepticism, which includes skepti-cism about content, intersubjectivity, and truth. As an intermediate step, think aboutconcepts, and, in particular, think about them as pointing to associated properties. Amatch, in this setting, is a situation in which something falls under the concept if andonly if it has the associated property—in which, I will say, the concept fits the property.A skeptical scenario is indiscernible from a match from the agent’s point of view eventhough it is not one. We can now construct a skeptical argument for concepts:

(8) a. Given a concept c, among our epistemic possibilities are skeptical scenar-ios for c.

b. We have no grounds for ranking skeptical scenarios for c more remote thanmatches for c.

c. We can know that a concept c fits its associated property only if we havegrounds for ranking skeptical scenarios for c more remote than matches for

11

c.d. So, we cannot know that our concepts fit their associated properties.

So far, this is just the epistemological skeptical argument applied to concepts. It is asmall step, however, from this argument to a metaphysical skeptical argument that hasnothing directly to do with the theory of knowledge. We can replace talk of knowingthat our concepts fit associated properties with talk of the fitting itself. We can speakof metaphysical possibilities rather than epistemic possibilities. And we can talk aboutthere being grounds rather than having grounds. The result is a metaphysical argumentthat our concepts cannot fit their associated properties: that all our concepts are, inLocke’s [1690] sense, inadequate.

(9) a. Given a concept c, among our possibilities are skeptical scenarios for c.b. There are no grounds for ranking skeptical scenarios for c more remote

than matches for c.c. A concept c can fit its associated property only by virtue of some fact.d. If there were a fact by virtue of which c fit its associated property, there

would be grounds for ranking skeptical scenarios for c more remote thanmatches for c.

e. So, our concepts cannot fit their associated properties.

This may not seem a very compelling argument; the second, third, and fourth premisesare not particularly well-motivated. We frequently do have empirical ways of dis-covering the adequacy or inadequacy of our concepts. The argument becomes muchstronger, however, if we think not in terms of concepts and their associated propertiesbut in terms of concepts and their contents.

(10) a. Given a concept c, among our possibilities are skeptical scenarios for c.b. There are no grounds for ranking skeptical scenarios for c more remote

than matches for c.c. A concept c can have a specific content only by virtue of some fact.d. If there were a fact by virtue of which c had a specific content, there would

be grounds for ranking skeptical scenarios for c more remote than matchesfor c.

e. So, our concepts cannot have any specific content.

3.1 The Content ChallengeWe are now in the region of the Kripke-Wittgenstein puzzle. It is easy to extend thisto a general argument against the possibility of content. Let b be anything thought tohave content—a word, a phrase, a sentence, a proposition, a thought, a perception, aconcept, etc. Call it a content-bearer. We can think of b as ‘+,’ for example, and askwhether it is possible for ‘+’ to have any specific content, for example, addition ratherthan quaddition. This is in part a puzzle about meaning: how is it possible for ‘+’ tomean plus rather than quus?

12

(11) a. Among our possibilities for ‘+’ are skeptical scenarios for ‘+,’ e.g., inwhich it means quus instead of plus.

b. There are no grounds for ranking scenarios in which ‘+’ means quus asmore remote than scenarios in which ‘+’ means plus.

c. ‘+’ can mean plus and not quus only by virtue of some fact.d. If there were a fact by virtue of which ‘+’ mean plus and not quus, there

would be grounds for ranking scenarios in which ‘+’ means quus as moreremote than scenarios in which ‘+’ means plus.

e. It is not true that ‘+’ means plus and not quus.f. So, ‘+’ has no specific content.

More generally, how is it possible for a content-bearer to have one content rather thana counterfeit? Here is an argument that it cannot:

(12) a. Given a content-bearer b, among our possibilities are skeptical scenariosfor b.

b. There are no grounds for ranking skeptical scenarios for b more remotethan matches for b.

c. A content-bearer b can have a specific content only by virtue of some fact.d. If there were a fact by virtue of which b had a specific content, there would

be grounds for ranking skeptical scenarios for b more remote than matchesfor b.

e. So, content-bearers cannot have any specific content.

But that seems to imply that there are no content-bearers. It at least implies that con-tents are vague; no symbol or thought can have any specific content, but can at besthave a certain kind of content.

3.2 The Intersubjectivity ChallengeSo far, we have focused on the mind-content connection. Focusing instead on themind-mind connection, we obtain a challenge to intersubjectivity:

(13) a. Given a content-bearer b (e.g., of A’s), among our possibilities are skep-tical scenarios for b—situations indiscernible from a match for b fromthe agent’s point of view in which b is not sufficiently similar to a targetcontent-bearer c (e.g., of B’s).

b. There are no grounds for ranking skeptical scenarios for b more remotethan matches for b.

c. A content-bearer b can be sufficiently similar to a target content-bearer conly by virtue of some fact.

d. If there were a fact by virtue of which b were sufficiently similar to c, therewould be grounds for ranking skeptical scenarios for b more remote thanmatches for b.

e. So, content-bearers cannot be sufficiently similar to other content-bearers.

13

Thus, no two people have have the same concept or the same thought. No two peoplecan mean the same thing by what they say. This poses a severe challenge not only tointersubjectivity but to the continuity of thought, for A and B might be temporal stagesof the same person. The argument thus purports to establish the hyper-Heracliteanconclusion that no one can think the same thought twice. A similar twist would showthat thought is radically contingent; no one could think the same thought in anotherpossible world.

3.3 The Truth ChallengeThe challenge to the possibility of truth stems from focusing instead on the mind-worldconnection.

(14) a. Given a belief or perception b, among our possibilities are skeptical sce-narios for b—situations indiscernible from a match for b from the agent’spoint of view in which b is not true or veridical.

b. There are no grounds for ranking skeptical scenarios for b more remotethan matches for b.

c. A belief or perception b can be true or veridical only by virtue of somefact.

d. If there were a fact by virtue of which b were true or veridical, there wouldbe grounds for ranking skeptical scenarios for b more remote than matchesfor b.

e. So, a belief or perception cannot be true or veridical.

We might speak just as easily of propositions or utterances.Note that this is not an argument against the possibility of knowing that a belief

is true or that a perception is veridical. It is an argument against truth or veridicalityitself. The worry is that our beliefs and perceptions might be radically out of line withthe world, and that there are no grounds for discounting this possibility. It is not just thatwe have no grounds; there are no grounds. From the point of view of a correspondencetheory of truth, this challenge may appear absurd. The fact by virtue of which we candiscount skeptical scenarios is just the fact that makes the proposition, utterance, orbelief true. Other conceptions of truth, however, cannot wave the challenge aside soeasily. Indeed, on some idealist assumptions it may appear unanswerable.

4 Judo, Academy-StyleThe problem of distinguishing contents from counterfeits is the heart of the Kripke-Wittgenstein puzzle. It is also the heart of Plato’s early dialogues. A question arisesabout courage (Laches), piety (Euthyphro), temperance (Charmides), friendship (Ly-sis), or beauty (Hippias Major)—in effect, about the content of the associated terms.Interlocutors propose definitions. Socrates asks questions to undermine them, that is,to identify them as signifying counterfeits rather than accurate contents. In the end, the

14

group gives up the attempt. It would be tempting to draw the skeptical conclusion thatthere is no way to tell contents from counterfeits.

In the middle dialogues, however, Socrates proposes and defends definitions, andintroduces the theory of forms to solve the problem of counterfeits. Plato offers severalarguments for the existence of forms in the middle dialogues.5 I do not propose tosurvey those arguments here. Instead I want to consider Plato’s general strategy for de-fusing the skeptical arguments sketched above—arguments that motivate the Sophists’relativism as well as epistemological and metaphysical skepticism.

4.1 The Epistemic FlipThe central strategy, as I see it, is to deny the conclusion of these arguments and therebyargue for the denial of one of the premises. Consider the epistemological argumentsketched above:

(15) a. Given a state of mind s, among our epistemic possibilities are skepticalscenarios for s.

b. We have no grounds for ranking skeptical scenarios for s more remote thanmatches for s.

c. We can know that a state of mind s is veridical only if we have grounds forranking skeptical scenarios for s more remote than matches for s.

d. So, we cannot know that our states of mind are veridical.

Plato turns the argument on its head:

(16) a. Given a state of mind s, among our epistemic possibilities are skepticalscenarios for s.

b. We can know that some of our states of mind are veridical.c. We can know that a state of mind s is veridical only if we have grounds for

ranking skeptical scenarios for s more remote than matches for s.d. So, we have grounds for ranking skeptical scenarios for s more remote than

matches for s.

The Platonic strategy admits that among our epistemic possibilities are skeptical sce-narios. I might be a victim of a Cartesian deceiver. I might be a brain in a vat. I mightbe in the Matrix. My entire species might be perceptually and cognitively misalignedwith reality as a result of evolutionary oddities. I cannot rule out the possibility of thesescenarios. But I can know that some of my states of mind are veridical. I can knowthat I have hands. I can know that I am in pain. I can know that this mango tastessweet to me. I can know that 7 + 3 = 10 and that I exist. (These last examples areAugustine’s [1955].) So, I must have ways of discounting skeptical scenarios. I mustbe able to rank them as more remote than matches, recognizing their possibility whilealso realizing that they do not prevent me from having knowledge.6

5These have been dubbed the One Over Many Argument, the Accurate One Over Many Argument, theArgument from Imperfection, the Object of Thought Argument, the Argument from Relatives or ConflictingAppearances, and the Arguments from the Sciences. See, e.g., [Annas 1974], [Fine 1980, 1995, 2003], and[Waterlow 1982].

6From a contemporary point of view, this seems to commit Plato to denying closure. The argument:

15

It may be worth reflecting on how Plato’s judo differs from Putnam’s. Putnamargues that if metaphysical realism is true I could not know that I am not a brain ina vat. But I do know that; therefore, metaphysical realism is false. From Putnam’spoint of view, Plato omits a crucial presupposition embodied in talk of matching andthus implicit in the idea of a skeptical scenario itself. From a Platonic point of view,however, Putnam’s argument too omits some crucial steps. We might see more clearlyhow they relate by making things more explicit:

(17) a. If metaphysical realism is true, then, given a state of mind s, among ourepistemic possibilities are skeptical scenarios for s.

b. We can know that some of our states of mind are veridical.c. We can know that a state of mind s is veridical only if we have grounds for

ranking skeptical scenarios for s more remote than matches for s.d. So, either metaphysical realism is not true or we have grounds for ranking

skeptical scenarios for s more remote than matches for s.

From this combined perspective, both arguments are too quick. Plato ignores his realistpresuppositions. And Putnam ignores the possibility of finding grounds for discountingskeptical scenarios. In putting it this way, I do not mean to endorse Putnam’s allegationthat the very idea of a skeptical scenario presupposes metaphysical realism; it seemsto me that an inverted spectrum scenario offers grounds for a skeptical argument with-out any commitment to a metaphysically loaded thesis. But for now I will frame theargument to remain as neutral as possible on the question.

4.2 The Constitutive FlipPlato similarly flips the metaphysical skeptical argument:

(18) a. Given a content-bearer b, among our possibilities are skeptical scenariosfor b.

b. Content-bearers have specific content.c. A content-bearer b can have a specific content only by virtue of some fact.

1. I know that 7 + 3 = 10.

2. I do not know that I am not under the spell of Descartes’s demon. (Among my epistemic possibilitiesis such a skeptical scenario.)

3. I know that if I am under the spell of Descartes’s demon then 7 + 3 , 10.

4. By contraposition, I know that if 7 + 3 = 10 then I am not under the spell of Descartes’s demon.

5. By closure, if I know that 7 + 3 = 10, then I know that I am not under the spell of Descartes’s demon.

6. I know that I am not under the spell of Descartes’s demon; contradiction.

But there is another plausible option: denying contraposition. Aristotle and Alexander of Aphrodisias appearat some points to rely on contraposition, but it creates problems for Aristotle’s view that true conditionalshave possible antecedents, since A→ B does not guarantee the possibility of ¬B, which is a prerequisite, onhis view, for the truth of ¬B→¬A. Boethius appears to be the first person to state contraposition explicitly[Bonevac and Dever 2012.] So, Plato might indeed have objected to the argument at just that stage. The firststatement of a closure principle is in Paul of Venice (1369–1429), so Plato may have objected to it as well.

16

d. If there were a fact by virtue of which b had a specific content, there wouldbe grounds for ranking skeptical scenarios for b more remote than matchesfor b.

e. So, there are grounds for ranking skeptical scenarios more remote thanmatches.

Once again, Plato admits the possibility of skeptical scenarios, but insists that theremust be grounds for discounting them. Notice Kripke’s own reaction to the skeptic hedescribes; he calls the skeptic’s claim that ‘+’ means quus “obviously insane” (8). Ofcourse ‘+’ means plus. If so, there must be grounds for ranking skeptical scenarios asmore remote than matches—for declining to allow the possibility that ‘+’ means quusto interfere with its possession of a specific content, namely, meaning plus.

Putnam once again would object that distinguishing contents and content-bearerspresupposes realism about content. But Plato would counter by pointing to the pos-sibility of finding grounds for discounting skeptical scenarios. We might expand theargument to incorporate both:

(19) a. If semantic realism is true, then, given a content-bearer b, among our pos-sibilities are skeptical scenarios for b.

b. Content-bearers have specific content.c. A content-bearer b can have a specific content only by virtue of some fact.d. If there were a fact by virtue of which b had a specific content, there would

be grounds for ranking skeptical scenarios for b more remote than matchesfor b.

e. So, either semantic realism is not true or there are grounds for rankingskeptical scenarios more remote than matches.

Once again we seem to have two options: embrace anti-realism or find grounds fordiscounting skeptical scenarios.

4.3 Transcendent GroundsFor the moment, I will set the anti-realist option aside, returning to it in a later section.For now, let’s seek grounds for ranking skeptical scenarios more remote than matches.The question in both epistemological and metaphysical contexts is, How? What kindsof grounds are there for discounting skeptical scenarios? What fact underlies a content-bearer’s having its content? How could we have epistemic access to these grounds?

Plato’s answer is the theory of forms. The keys are the following premises:

(20) For a content-bearer b, there could be grounds for ranking skeptical scenar-ios for b more remote than matches for b only if b is anchored to somethingtranscendent, the relation to which gives b its content.7

7I use ‘anchored’ advisedly, generalizing the idea in Kamp and Reyle (1990): anchoring is a relation byvirtue of which something has the content that it has. It is typically a causal connection, such as the linkbetween Aristotle and the name ‘Aristotle,’ or the connection between H2O and the word ‘water.’ Thinkingof anchoring in this way explains why virtually every basic term designates rigidly.

17

(21) For a content-bearer b, we could have grounds for ranking skeptical scenariosfor b more remote than matches for b only if b is anchored to something tran-scendent and epistemically accessible, the relation to which is also accessibleand gives b its content.

This is, of course, the crux of the argument. What does ‘transcendent’ mean?

x is transcendent if and only if x is

• independent of individual, finite minds

• temporally and modally stable

• infinitary

• normative

• objective

Let’s take these in turn to see why Plato and his followers find each aspect of transcen-dence important to solving skeptical puzzles. These arguments are implicit in Plato,but appear explicitly in certain early Church Fathers, and especially Augustine.

• Independence from individual, finite minds. The danger here is relativism, thethought that truth is relative not only to a conceptual framework as that is oftenunderstood in contemporary philosophy but even to a conceptual idiolect. ThePlatonic worry can be expressed as a thesis: If a content-bearer b is anchored tosomething dependent on individual, finite minds, and that anchoring gives b itscontent, then there would be no grounds for ranking skeptical scenarios for b asmore remote than matches for b. Kripke’s skeptic insists that ‘+’ means quus;perhaps in his idiolect it does. If so, we may find it difficult to know whether weare communicating. Every term I use might mean something different comingfrom someone else. And we presumably cannot assess the truth of ‘68 + 57 =

125’ in absolute terms, for it will be true in my conceptual idiolect and false inhis.8

• Temporal and modal stability. These problems arise even apart from interper-sonal communication. How do I know what I meant by ‘+’ yesterday? How doI know what I will mean tomorrow? Is there any determinate answer to what Imeant yesterday or will mean tomorrow, quite apart from epistemological con-siderations? The dangers posed by temporal variation also apply to modal vari-ation. Would I have meant what I do by ‘+’ if it had rained today, or if I were

8This argument is explicit in Augustine [1955], who argues that what anchors content must be “commonto all”: “Hence, is not the one truth common to both of us, which we both see with our individual minds?...Can we deny that this is true, and one, and common to the sight of all who know it, although each sees it withhis own mind, and not with yours or mine or anyone else’s? For that which is seen is present in common toall who see it.... Will you not also agree that the following propositions are absolutely true, and are presentin common to you and me and all who see them: we ought to live justly; the better should be preferred tothe worse; like should be compared with like; every man should be given his due?” (II, 10.28, 107–108;emphasis added) “So you would certainly not say that what you and I perceive in common, each with hisown mind, shares the nature of the mind of either of us. You could not say that what the eyes of two peoplesee at the same time is the eyes of either of them; it is something else to which both of them direct theirsight.” (II, 12.33, 114; emphasis added)

18

currently standing rather than sitting? If the danger of dependence on individ-ual finite minds is a relativism born of a personal kind of parochialism, then thedanger of dependence on particular times and worlds is a structurally similar rel-ativism born of temporal or modal parochialism. The relativistic dangers here arein fact worse, for it is not just that I cannot know whether I am communicatingwith someone else; I cannot know whether I am communicating with myself.I cannot know, in the moment it takes to think a thought, whether my thoughthas the same content at the end of that moment that it had at the beginning. Icannot know whether the cat’s presence on the table is changing the meaningsof my words and thoughts. Modal and temporal relativism will apply across theboard, even to ‘7 + 3 = 10’ and ‘The better should be preferred to the worse.’ Ifthere are such eternal truths, then what anchors content will have to be maximallytemporally and modally stable: eternal and necessary.9

• Infinitary character. Contents are infinitary. This is obvious in the case of ‘+’or ‘number,’ since there are infinitely many numbers and infinitely many triplesx,y,z such that x + y = z. But it is true of other contents as well. ‘Just,’ ‘unjust,’‘courageous,’ ‘beautiful,’ ‘red,’ ‘hungry,’ and other predicates might apply or failto apply to infinitely many possible objects or situations. The content of a term orconcept somehow determines its applicability or lack of applicability in infinitelymany possible circumstances. It will not do to point to recursive definitions, fortheir ability to characterize something infinite in finite terms, while important,presupposes the contents of the terms appearing in the definitions, which arethemselves infinitary.10

9These examples are Augustine’s. Augustine [1955] again, stressing the unchangeability as well as thepublic character of truth (in all cases, emphasis added): “It is enough that you see, as I do, and admit to bequite certain that those principles and illuminations, so to speak, in which the virtues appear, are true andunchangeable and, whether separately or all together, are present in common to the sight of those who cansee them, each with his own reason and mind.” (II, 11.29, 108–109) “Therefore the principles of number aretrue and unchangeable; their law and truth are, as you said, present unchangeably and in common to all whosee them. In the same way the principles of wisdom are true and unchangeable.” (II, 11.29, 110) “But, whenwe begin, as it were, to mount upward, we find that numbers pass beyond our minds and abide unchangeablyin truth itself.” (II, 11.31, 112) “Therefore you would by no means deny that there exists unchangeable truth,containing all those things which are unchangeably true. You could not call this yours or mine or any man’s,but it is present and offers itself in common to all who behold unchangeable truths, like a light which in awonderful fashion is both secret and public. No one could say that anything which is present in commonto all who have reason and understanding belongs to the nature of one individual.” (II, 12.33, 113) “Youcannot grasp with bodily sense or attention of the soul any changeable thing you see which is not possessedby some form of number: take this away, and it falls back to nothing. Therefore have no doubt that thereis some eternal and unchangeable form, in order that changeable things may not cease, but, with measuredmovement and distinct and varied forms, may pass through their temporal course. This eternal form isneither contained, nor, as it were, spread in space, nor prolonged nor altered in time; it enables those otherthings to receive their forms, and according to their nature to realise and use the numbers proper to placeand time.” (II, 16.44, 125–126) “So we conclude that body and soul are given their forms by a form which isunchangeable and everlasting.” (II, 17.45, 126)

10Augustine: “How, then, do we recognise that this fact, which we recognise throughout all numbers,is unchangeable, sure, and certain? No one is aware of all numbers with any bodily sense, for they areinnumerable. How, then, do we know that this holds good throughout them all? By what idea or image dowe see so sure a truth so confidently throughout innumerable instances, unless we do it by an inner light,unknown to the bodily sense?” (II, 2.23, 101–102).

19

• Normativity. Contents determine correctness conditions. They specify not merelyhow a term is used but how it ought to be used. There are, of course, normativeterms themselves, which wear their normative character proudly, but every termis normative in the sense that it has accompanying correctness conditions. Skep-tics who say that 57 + 68 = 5, given the ordinary meaning of ‘+,’ are wrong;their assertion is incorrect. They ought to get the answer ‘125.’ Similarly, a childwho calls a baby bear a cat and a beginner in Spanish who calls a dog pero aredoing it wrong.11

• Objectivity. Whatever anchors our concepts must be objective, not only in thesense of being independent of individual, finite minds, but also in the sense ofmatching the nature of objects. It must solve the problem of truth, providinggrounds for our utterances and our thoughts correctly describing the way thingsare. That means that whatever anchors our contents match the structure of theworld in appropriate ways [Burnyeat 1983]. Our concepts must be able to matchthe properties they are meant to capture—not always, of course, but sometimes,and perhaps even normally. The same is true of our thoughts and utterances,which must be able to match the facts. What anchors content must connect tothe world in ways that make these relationships possible.12

4.4 Plato’s Theory of FormsPlato’s forms are supposed to have precisely these characteristics. Recall Plato’s intro-duction of them in the Republic:

And do you not know also that although they [students of geometry, arith-metic, and the kindred sciences] make use of the visible forms and reasonabout them, they are thinking not of these, but of the ideals which they re-semble; not of the figures which they draw, but of the absolute square andthe absolute diameter, and so on—the forms which they draw or make, andwhich have shadows and reflections in water of their own, are convertedby them into images, but they are really seeking to behold the things them-selves, which can only be seen with the eye of the mind? (510 d–e)

It is crucial that the forms can be seen—we can have epistemic access to them—butnot in the usual, naturalistic way, via the senses, but instead in some other way, “withthe eye of the mind.” (“The many, as we say, are seen but not known, and the ideas areknown but not seen” (507c). Compare the Phaedo: “the unchanging things you canonly perceive with the mind—they are invisible and are not seen” (79a).)

11See Boghossian [1989, 513]:

Suppose the expression ‘green’ means green. It follows immediately that the expression‘green’ applies correctly only to these things (the green ones) and not to those (the non-greens).The fact that the expression means something implies, that is, a whole set of normative truthsabout my behaviour with that expression: namely, that my use of it is correct in application tocertain objects and not in application to others.

Plato sees this point clearly, but Augustine’s chief focus is on normative terms and concepts themselves.12Augustine: “Every changeable thing must necessarily be able to realise its form.” (II, 17.45, 126)

20

The forms ground truth (“that which imparts truth to the known and the power ofknowing to the knower” (508e)) and are the essences of things. They are thus objec-tive. (“Not only is the Form itself always entitled to its own name, but also what isnot the Form, but always has, when it exists, its immanent character” (Phaedo 103e;cf. 103bff.). Compare Aristotle: “Sensible things, [Plato] said, were all named after[Ideas], and in virtue of a relation to them; for the many existed by participation in theIdeas that have the same name as they.” (Metaphysics A 987b 3ff., translated by Ross).)They give content to our words and thoughts. The forms must be outside any individualmind, to defeat the Sophist, and they must be knowable, to defeat the skeptic.

They must be infinitary—“The sound which passes through the lips whether of anindividual or of all men is one and yet infinite” (Philebus 17b). Compare, also fromthe Philebus: “all figures are comprehended under one class; and yet particular figuresmay be absolutely opposed to one another, and there is an infinite diversity of them”(12e–13a).

They are also modally and temporally stable. “Is that idea or essence, which in thedialectical process we define as essence of true existence—whether essence of equality,beauty, or anything else: are these essences, I say, liable at times to some degree ofchange? or are they each of them always what they are, having the same simple, self-existent and unchanging forms, and not admitting of variation at all, or in any way, orat any time? They must be always the same, Socrates, replied Cebes.” (Phaedo 78d)

The forms are also normative, as Plato recognizes in the Gorgias :

Chaerephon. My question is this: If Gorgias had the skill of his brotherHerodicus, what ought we to call him? Ought he not to have the namewhich is given to his brother?

Polus. Certainly.

Chaerephon. Then we should be right in calling him a physician?

Polus. Yes.

Chaerephon. And if he had the skill of Aristophon the son of Aglaophon,or of his brother Polygnotus, what ought we to call him?

Polus. Clearly, a painter. (448bff.; emphasis added)

Plato’s forms are thus paradigms of transcendent anchors. They are also meant tobe epistemically accessible. They are essentially intelligibles, graspable by the mind[Sellars 1974].

Let’s return to the broader argument. For something to have content, it must be an-chored to something transcendent. For us to know its content, we must be able to haveepistemic access to the anchor of its content and to its anchoring. Our thoughts andwords must not only be able to have content; we must be able to know what those con-tents are—again, not always, for it can take work to uncover the contents of our wordsand thoughts—but sometimes, and maybe even normally. Contents must be intelligi-ble. On the anchoring picture, this means that the transcendent entities, the anchoringrelation, and the entities’ relation to the world must be epistemically accessible. Wemust have access to what gives content to a given expression or thought and also to thefact that the expression or thought has that content. We must, in short, have access not

21

only to a realm of transcendent entities but also to the connections between that realmand our expressions and thoughts. Augustine thus speaks of an “inner light” revealingcontents to us ([1955], II, 2.23, 101–102). Moreover, if what we say and think is to becapable of objectivity and of truth, we must be able to have access to the links betweenthe anchors and the world—what Plato would call the relation of participation.

But viewing the argument above as simply an argument for the theory of forms givesit too little credit. In a way, the argument does not engage the realist–conceptualist–nominalist debate at all. It argues for transcendent anchors. Are they abstract universalsexisting independently of the mental altogether? Are they mental constructions, not ofindividual, finite minds but perhaps of collective mental activity? Are they linguisticrules or roles? Conceptual or functional rules or roles? Semantic values? Nothing saidso far addresses these questions. One cannot therefore dismiss the argument as internalto Platonism. It remains independent of the metaphysical status of the transcendentanchors it champions.

5 Platonic Arguments for the Existence of GodSo far, we have an argument for something like Plato’s theory of forms, but withoutany commitment to the metaphysical status of the entities to which it points. Contentsmust be anchored to something transcendent—independent of individual, finite minds,modally and temporally stable, infinitary, normative, objective—and epistemically ac-cessible. It did not take long, however, for Plato’s theory to disintegrate in the face ofskeptical challenges. Its obvious weak spot is epistemic accessibility. Why should wethink that we can have epistemic access to anything independent of individual, finiteminds? Why should we think that we can have epistemic access to anything objective?Modally and temporally stable? Infinitary? Normative? Why moreover think that onekind of entity can fulfill all those roles at once? Why think we could gain access to itif it did?

Plato’s talk of “the eye of the mind” (Republic 510e), his metaphor of recollection(Meno 98a), and his almost mystical attribution of causal power to the form of the Good(Republic 508e–509a, 517b,c) seem to fail even on their own terms. If we cannot seehow we have epistemic access to the forms now, how can it help to refer to our accessto them at some earlier time? If we cannot see how forms could have causal powerof a kind needed to anchor content and to enable our knowledge of it, how can it helpto refer to the causal power of a specific form, the form of the Good? If the point isthat forms are quite unlike objects of sense or other naturalistically accessible entities,how can an analogy with vision help? No wonder that ‘Academic’ quickly came todesignate a form of skepticism.

It is possible, nevertheless, to transform skeptical arguments, even those directedat the theory of forms itself, into arguments for God’s existence. In this section Iwill focus on questions of content and outline an argument that follows the thinkingof such early Christian writers as Justin Martyr, Lactantius, and Augustine. In latersections I will develop streamlined versions of arguments from knowledge, content,interpretation, truth, intersubjectivity, and communication.

22

5.1 Epistemic AccessEarly Church Fathers revived Plato’s theory, but with some emendations. First, anadditional premise concerning epistemic access:

(22) Something independent of individual, finite minds can be epistemically acces-sible only if there is something with causal power, independent of individualfinite minds, that makes such access possible.

The idea is that, under its own power, the mind can have access only to its own cre-ations, things that are dependent on the mind itself. If a table, for example, is indepen-dent of the mind, then something with causal power, independent of the mind, mustmake our access to the table possible if it is possible at all. That power might be thetable itself, light, a light source, sound waves, or some combination of these. The thesisdoes not commit to a view of the mind as passive; it may be able to generate objectsand gain knowledge of its own constructions. What it cannot do is generate knowledgeof what is independent of it without some additional input from outside. What is trueof individual minds is also true of a collection of minds. They may be able to generateobjects and gain knowledge of their own constructions. But they too cannot generateknowledge of what is independent of them without input from outside.

Now, there are two paths forward. The simplest argues that only a transcendentcausal power could make it possible for us to have epistemic access to something tran-scendent.

(23) Only a transcendent causal power could make possible epistemic access to thetranscendent.

How could something finite make it possible for us to know the infinite? How couldsomething changeable make it possible for us to know the eternal? How could some-thing contingent enable us to know the necessary? How could something purely factualenable us to know the normative?

At first glance, these questions seem to have easy answers. A finite recursive def-inition can specify and thus enable us to know an infinite function. A changeable,contingent teacher can convey eternal and necessary truths. Purely factual occurrencescan reveal normative truths. But all these cases concern the transfer of content orknowledge, not its generation. The recursive definition characterizes an infinite func-tion given the infinitary character of the primitive terms it contains. We have to knowthe meanings of the primitives in order to know the meaning of the function. Teach-ing transfers content and knowledge from teacher to student. The occurrence revealsa normative truth because of the normative character of its nature or effects—e.g., theevent causes widespread damage, and we infer the badness of events of that kind fromthe badness of the damage. But how do we know the infinitary meanings of the prim-itives? Was there a first teacher, and how did that person come to know eternal andnecessary truths? How do we know the ultimate normative truths from which we drawother normative conclusions?

I have phrased these questions in ways that suggest a sort of atomism, but a co-herentist picture is no better, for the issue is what enables us to connect the finite and

23

infinite, the changeable and unchangeable, the contingent and necessary, and the de-scriptive and normative, whatever the nature of those connections might be.

The other path concerns itself with naturalism. We need the premise

(24) Nothing natural is transcendent.

We then link this to causation.

(25) Anything natural can cause or anchor only what is natural.

It might seem as if this, too, has an easy counterexample. The natural can cause oranchor conventions of a kind that underlie, for example, analytic truth. Analytic truthscan be necessary, eternal, independent of individual finite minds, infinitary, and evennormative: we ought to apply ‘bachelor’ only to unmarried men. But this, of course,is just what the Kripke-Wittgenstein puzzle challenges. Conventions can work thisway only on the assumption that we have a clear notion of rule-following and thus ofcontent. Conventions cannot solve the problem of content; from one point of view, theyare the problem of content.

So far we have an argument that content must be anchored in something supernat-ural, and that the anchoring takes place by virtue of the causal power of somethingtranscendent and supernatural.

5.2 Mental AnchorsAs various early Church Fathers develop it, the argument now turns to what couldanchor something like content. We have two models of such anchoring. First, in theirview, our words have content by virtue of their connection to ideas or concepts. Signs,symbols, language, and other non-mental entities with meanings get their content byway of a connection with something mental. Second, concepts may be able to betransferred from one person to another through teaching or other forms of training; asPutnam notes, I may anchor my concept to yours even if I don’t understand its content.This too requires a connection to ideas and concepts. So, in either case, our models ofanchoring employ mental entities as anchors.

(26) The best explanation for the anchoring of content employs mental entities asanchors.

Let’s call these mental anchors Ideas. The Ideas must be transcendent and therebysupernatural. They are clearly not to be found in any individual finite mind.

Nor can they be located in any finite collection of finite minds. This in itself seemsto defeat the skeptical strategy which anchors content in use. It also defeats some otherforms of anti-realism. With respect to the key properties involved in transcendence, afinite collection of finite minds is no better than a single finite mind [Blackburn 1984].The skeptical strategy Kripke explores from this perspective appears to change the sub-ject, replacing truth with assertibility and normativity with facts of communal agree-ment, censure, approval, etc. It is hard to assess the importance of the community inthis formulation, for communal acceptability seems to depend on individual accept-ability [Blackburn 1984, Boghossian 1989]. The skeptical strategy moreover seems to

24

give up on infinitude altogether. If no finite mind can account for the infinitary andnormative features of content, then no finite collection of finite minds can either, forwe still have finitely many occasions of use. Such a collection is still, in aggregate,finite; the uses, occurrences, or acts of such a collection still form a finite set. It thuscannot account for the infinitary character of content. Nor can it account for norma-tivity; those employing a skeptical strategy replace normativity with practice in such away that to be incorrect is just to deviate from the usage of the group. This makes ithard to distinguish a reform from a change in fashion. Sometimes, after all, the deviantusage is right, or at least better, than the common usage. A finite collection of finiteminds moreover cannot explain temporal and model stability; the group may changeits mind or change its usage, sometimes gradually, sometimes quickly. Finally, such acollection cannot explain objectivity; it replaces the notion with a shared subjectivity.

(27) A finite collection of finite minds does not suffice to explain the grounding ofcontent.

Two possibilities remain: that Ideas are to be found in an infinite collection of finiteminds, and that they are to be found in an infinite mind. The early Church Fathersdid not contemplate the former possibility, living well before Kant’s regulative ideals,Hegel’s vision of absolute knowledge, and Peirce’s transformation of it into an ideallimit of scientific inquiry. But the argument depends on ruling it out. So, we need athesis:

(28) An infinite collection of finite minds does not suffice to explain the groundingof content.

It may seem as if we have forgotten the possibility of anti-realism. But from a Platonicpoint of view we have already covered that option:

(29) Anti-realism anchors content to some feature of a collection of finite minds.

We may now draw the conclusion:

(30) The best explanation for the anchoring of content requires an infinite mind.

Since that infinite mind must have causal power sufficient for grounding content andour knowledge of it, moreover,

(31) The best explanation for the anchoring of content requires an infinite, transcen-dent, and supernatural mind.

And the best explanation for that is this: our words and thoughts have content by virtueof being anchored to Ideas, which are best explained as ideas in the mind of God.

(32) The best explanation of content requires the existence of God.

This argument must have the form of an inference to the best explanation, for there aregaps that it does not by itself address. The argument does not by itself establish that theIdeas inhabit a single infinite mind. The argument establishes God’s transcendence, butit does not by itself establish omnipotence, omniscience, benevolence, or other compo-nents of the classical conception of God. The argument by itself does not provide the

25

explanation of content; it points to a kind of explanation, the details of which must befilled in theologically.

Early Church Fathers undertook these tasks. Lactantius, for example, sees a de-sign argument and a theory of the general providence of God as key components ofthe explanation tying God’s mind to content, truth, and knowledge. Augustine seesthe explanation as part of an account of human free will as a part of the Divine plan.Descartes sees God’s goodness as central. Providing the explanation to which the ar-gument alludes requires the development of a comprehensive theology.

6 PragmatismPragmatism poses a more serious threat to the family of arguments I have been devel-oping than the skeptical solution does, for it holds much more promise for explainingtranscendence. If Ideas are characterizable in terms of an infinite collection of finiteminds—if they are for example limits of the activities of finite minds—they can be in-dependent of individual, finite minds. They can be infinitary, for the collection of finiteminds underlying them is infinite. If they are understood as limits, they do not have tobe reachable at any finite stage.

Thinking of Ideas as arising from an infinite collection of finite minds does not byitself account for temporal or modal stability. But if that collection includes all timesand all possible worlds, it can. It is hard to see, however, how to give that notionmuch plausibility. Ideas understood as limits can be temporally stable, for limits do notchange as we move along a path toward them. They cannot be objective or modallystable in the usual, realist sense, but they can be something close to that if we can showthat the limits themselves are independent of initial conditions and paths taken to reachthem. This is a tall order—Peirce [1878] gives us no reason to think that the eventualagreement of all who inquire, even to the extent that it could be achieved at all (andwhy think the series of stages of inquiry converges?), would be path-independent andinvariant across initial conditions—but it is hard to show that such an argument wouldbe impossible.

6.1 NormativityThe central question is normativity. How does an appeal to an infinite collection offinite minds explain normativity? The thought might be that the infinitary character ofcontent underlies the problem of normativity in the following sense. Why not identifycontent with use? Up to any given point in time, actually usage has been finite, andthat finite collection does not tell us how to go on to apply our term or concept topreviously unconsidered cases. If the usage to which we can appeal is infinite, coveringall possible cases, however, this problem goes away. Over an infinite set, one mightargue, we can identify content with use.

Matters are not so simple, however. How do we distinguish competence from per-formance [Chomsky 1965]? Among the finitely or infinitely many usages to whichwe might in principle have access are some, perhaps many, incorrect usages. Peoplehave to learn to use the term or concept. Circumstances are suboptimal. People make

26

mistakes. Expressions are ambiguous or vague. Usages change; the contents of termsand concepts sometimes changes and sometimes resists changes despite altered usage.How do we know which usages to take as canonical, helping to define content, andwhich to discard?

The skeptical solution provides a simple answer: See which usages meet with dis-approval or rejection and discard them. Whether such a strategy can succeed is in partan empirical question. As a practical matter, most incorrect usages go uncorrected.People tend to avoid giving negative feedback and so fail to correct mistakes. Evencommitted language scolds tire of explaining the correct uses of ‘hopefully’ and ‘begthe question.’ Incorrect uses may outnumber correct uses not only in these cases butin cases where terms have popular and more precise uses: think of ‘fish,’ ‘vegetable,’‘germ,’ and ‘flu.’ And suboptimal circumstances can lead to systematic errors [Boghos-sian 1989].

The most popular and most plausible strategy for distinguish competence from per-formance is to order the infinite collection of finite minds into a progression of partiallycorrected, better, broader, or at any rate later stages which tend toward a limit. Thisis the classic pragmatist move, and it offers some hope of being able to define compe-tence in terms of performance at the limit given certain idealizations and constraints.What idealizations and constraints are justifiable? Do they beg the question, sneakingin normativity at the outset? Do we have reason to expect convergence to a limit? Whatare we to say about belief or usage that appears at no finite stage but would appear atan ideal limit, if it were possible to survey all the finite stages at once? Will the contentat the limit be independent of initial conditions? Will it be path-independent? Whatare we to say about beliefs that achieve consensus arbitrarily at some finite stage andnever meet disapproval or counterevidence? Does the limit conception lead to a form ofrelativism? These are difficult questions, subject to intense debate. (See, e.g., Putnam[1981] and Rorty [1998].)

Even if the strategy can be refined to achieve empirical adequacy, however, it is notclear that it succeeds in explaining normativity any more than “You might get caughtand go to jail” explains what is wrong with burglary. If you call a tomato a vegetable,you might face disapproval—unlikely in most contexts, I would say—but that raises aEuthyphro question: Is your usage incorrect because you face disapproval, or do youface disapproval because your usage is incorrect? The skeptical solution implausiblydemands the former.13

This concern underlies Bertrand Russell’s broader objection to pragmatism as atheory of truth. Say that the ideal inquirers would eventually converge on p. Why?The realist would say, because p is true: truth explains agreement. For the pragmatist,

13The skeptical solution, one might say, offers a Mean Girls theory of normativity:

Gretchen: Regina, you’re wearing sweatpants. It’s Monday.Regina: So...?Karen: So that’s against the rules, and you can’t sit with us.Regina: Whatever. Those rules aren’t real.Karen: They were real that day I wore a vest!Regina: Because that vest was disgusting!Gretchen: You can’t sit with us!

27

this is tautological rather than explanatory: agreement explains truth. Russell arguesthat the pragmatic theory “seems to suggest that if I infer a world, there is a world.Yet I am not the Creator. Not all my inferences and explanations could prevent theworld from coming to an end to-night, if so it were to happen.... Whatever accusationspragmatists may bring, I shall continue to protest that it was not I who made the world”[Russell 1919, 26].

6.2 Further ArgumentsKripke’s objections to dispositional accounts appear to apply to accounts appealing toinfinite collections of finite minds, whether they follow a pragmatist strategy or not.The appeal of identifying competence or truth with usage or belief across some infiniteset or at an infinite limit inevitably goes beyond our finite evidence: it requires us toask what would be accepted in some idealized space. In short, this turns every questionof content or truth into a counterfactual question. And if the counterfactual holds, itmust be in virtue of some fact. But what sort of fact could this be? We are back in theheart of the Kripke-Wittgenstein puzzle.

There is a further infinite regress worry: if every question turns into a counterfac-tual question, then the counterfactual question raises a further counterfactual question,which raises yet another, and so on. This may or may not be vicious depending on thedetails:

p is true

if and only if

p would be accepted throughout a given space S ,

which is true if and only if

‘p would be accepted throughout S ’ would be accepted through S ,

which is true if and only if

“‘p would be accepted throughout S ’ would be accepted through S ” wouldbe accepted throughout S ,

and so on. The lengthier counterfactuals might follow trivially if acceptance throughoutS is transparent in S . But such transparency is not automatic; it has to be established.14

14It may also be destructive of this option, as Stillwell [1989] argues, elaborating Plantinga [1982]. Callan ideally rational scientific community an IRS. This might be a community at a Peircean limit; it might bean idealized community somehow capable of surveying an infinite space in some other fashion. Given thepragmatist’s understanding of truth,

p is true if and only if, is there were an IRS, it would accept p.

Plantinga and Stillwell argue that an IRS would be transparent to some extent:

If there were an IRS, it would accept that there is an IRS.

By the pragmatist’s own lights, that entails that there is an IRS. But that is plainly false.The Plantinga-Stillwell argument relies on the principle Bonevac, Dever, and Sosa [2006] call contraction:

(A→ (A→ B)) =⇒ (A→ B), which is in a Lewis-Stalnaker system equivalent to weak centering, that is, tomodus ponens. It remains open to the pragmatist to reject that principle.

28

Another worry: on some matters, such as unbounded universal assertions, accep-tance throughout an infinite collection could plausibly be identified with truth only byvirtue of including an infinite and in fact complete data set; all possible evidence willhave been gathered and all possible arguments will have been considered at some pointin the space. It is important to note, however, that if each mental act of a finite inquirertakes some finite time of at least length n, and if the space is itself countable, it cannotinclude all possible theories, for there will be uncountably many of those.15 So, on apragmatist conception, we cannot guarantee that ideal inquirers will settle on the besttheory, for even if they can settle on the best theory they have been able to consider,there may be better theories that they will not consider even in the infinite long run.More generally, we cannot guarantee that acceptance throughout the space will be ac-ceptance of the best possible theory unless the space is itself uncountable. That is, tomake an identification of truth or competence with acceptance over an infinite spaceplausible, that space would have to be uncountable. We would have to be able to tell astory making that plausible.

Finally, how will this strategy be able to discount skeptical scenarios with respectto content? What grounds could we have for treating such skeptical scenarios as moreremote than matches? In matters of content, access to an infinite space of usage mightseem to solve the problem of moving from the finite to the infinite. That assumes,however, (a) that usage has not changed throughout that infinite space and (b) thatusage reveals a single content rather than, say, multiple contents reflecting ambiguityor dialect differences. Surveying an infinite space offers no obvious way to discountthose possibilities. Since skeptical scenarios and matches are indiscernible from theagent’s point of view, even given unlimited evidence, ideal inquirers at the limit orhaving access to an infinite space are no better off than we are when it comes to thecentral problem of this paper. Indeed, from the perspective of skeptical challenges,there is no decisive difference between the ideal inquirers and us, their far-from-idealcounterparts.

7 Metaphysical Arguments: The Arguments from Con-tent, Intersubjectivity, and Truth

We are now in a position to articulate metaphysical anti-skeptical arguments for God’sexistence. These arguments follow the general pattern of the argument I have con-structed from the argumentative strategies of Plato, Justin Martyr, Lactantius, and Au-gustine. But they are streamlined by jettisoning two aspects of those arguments.

The early Church Fathers identify Platonic forms with ideas in the mind of God.Their model for anchoring content requires anchors to be mental. They have good rea-sons for this; Plato’s forms are abstract, not mental, and the question of how mindscan relate to them thus becomes central. This generates a problem of epistemic access,already discussed. It also generates a metaphysical mind-form problem analogous in

15Arguments are finite, and a countable set has countably many finite subsets, so the number of argumentsthat can be constructed in a given finitely specifiable language is countable. But theories can be infinite, anda countably infinite set has uncountably many subsets.

29

some ways to the mind-body problem. If forms and mental acts are of radically dif-ferent kinds, how can they relate to one another? How can something abstract anchormental or linguistic content? If the forms are themselves mental, in contrast, the meta-physical puzzle goes away, for then Ideas and mental acts are similar in kind.

From a contemporary point of view, however, this move may seem less attractive.Direct reference theorists anchor names, demonstratives, and referential definite de-scriptions to objects: Aristotle, Nixon, that dog, Smith’s murderer, the man with (whatI think is) a martini, etc. These entities are not mental. So, the thought that our bestmodel of the anchoring of content treats anchors as mental entities seems dubious.

There is nevertheless something to be said for it. On the direct reference picture,my content derives from the content of the person or people from which I learned theterm (and perhaps other factors); the causal chain goes back to an initial baptism. Thetheory has a recursive structure: there is a theory of content transfer and a theory ofinitial content acquisition. The theory of initial content acquisition remains somewhatvague—Aristotle’s parents give him a name, but how do they refer to what they arenaming so that it is clear who or what is being named?—and it may be that somethingmental, in addition to the object itself, serves as an anchor in that part of the story.

However this may be, the reliance on mental anchors may appear to be a weaknessof the argument. It is moreover unnecessary. Whatever the anchors are, and whatevermetaphysical status they might have, there must be a causal power by virtue of whichwe relate to them. If the anchors are transcendent, then that causal power must betranscendent as well.

We can abstract from the details of the Platonic argument in another way, too. Wedo not have to commit ourselves to a relational theory of meaning or content. Quine,Putnam, Sellars, and many others have criticized such views. Without endorsing thosecriticisms, we can frame the argument in a way that sidesteps them. We do not have totalk of anchoring as a relation or anchors as entities. We can speak instead of ground-ing. Meaning must be grounded in fact. Boghossian articulates “a criterion of adequacyfor theories of meaning: any proposed candidate for being the property in virtue ofwhich an expression has meaning must be such as to ground the normativity of mean-ing...” [1989, 509; emphasis added]. The primary issue is grounding, not mirroring,labeling, copying, or other metaphors that have been used to assail realism. If meaningis transcendent, moreover, then so is whatever grounds it. Transcendence of contentrequires transcendence of ground.

7.1 The Argument from ContentLet’s begin with the argument from content. How is it possible for our words andthoughts to have content? One simple way to put the argument is that there is no goodnaturalistic explanation for our ability to refer to things in the world and mean things bywhat we think and say. Any account of semantic capacities must at some point resortto magic.16 And the best explanation we have for that magic involves God.

16Though the terms comes from Putnam [1981], here I mean it to indicate a non-naturalistic element; Ido not mean to imply that all semantic relations are in any sense necessary. It is entirely contingent that thename ‘Winston Churchill’ refers to Winston Churchill.

30

(33) a. If realism is true, then, given a content-bearer b, among our possibilitiesare skeptical scenarios for b—situations indiscernible from matches for bfrom the agent’s point of view in which b does not have its real content butsome counterfeit.

b. Content-bearers have specific contents.c. A content-bearer b can have a specific content only by virtue of some fact.d. If there were a fact by virtue of which b had a specific content, there would

be grounds for ranking skeptical scenarios for b more remote than matchesfor b.

e. There could be grounds for ranking skeptical scenarios for b more remotethan matches for b only if b’s content is grounded in something transcen-dent.

f. Something independent of individual, finite minds can ground content onlyif there is something with causal power, independent of individual finiteminds, that makes such grounding possible.

g. Only a transcendent causal power could make possible grounding in some-thing transcendent.

h. Nothing natural is transcendent.i. Anti-realism grounds content in some feature of a collection of finite minds.j. A finite collection of finite minds does not suffice to explain the grounding

of content.k. An infinite collection of finite minds does not suffice to explain the ground-

ing of content.l. The best explanation for the existence of a supernatural, transcendent causal

power grounding content in the transcendent includes the existence of God.m. So, there is a God.

Recall the definition of transcendence. The argument establishes the existence of asupernatural, infinite, eternal, necessary, objective, normative, and independent causalpower capable of grounding content. That causal power is normative in the sense thatit defines standards for correctness and incorrectness, right and wrong, truth and false-hood, virtue and vice.

This argument is abstract, but it expresses a common religious intuition. God alonegives meaning to the world. In relation to God, this world, our words, our thoughts,and our actions have meaning. If there were no God, there would be no meaning.Existentialism notwithstanding, furthermore, there would be no way for us to assignanything meaning. This is true not only in the cosmic, meaning-of-life sense, but in themundane sense that ‘apple’ could not mean apple if there were no God.

7.2 The Argument from IntersubjectivityLet’s turn now to the argument from intersubjectivity. How is it possible for us to sharemeanings, to refer to the same things, to think the same thoughts? How can you andI think the same thing or mean the same thing? How can I even mean what I meant

31

when I began speaking or thinking? Again, there is no non-magical explanation. Andthe only magical explanation we have that makes any sense involves God.

(34) a. If realism is true, then, given a content-bearer b (e.g., of A’s), among ourpossibilities are skeptical scenarios for b—situations indiscernible frommatches for b from the agent’s point of view in which b is not sufficientlysimilar to any content-bearer c (e.g., of B’s).

b. Content-bearers can be sufficiently similar to other content-bearers.c. A content-bearer b can be sufficiently similar to a target content-bearer c

only by virtue of some fact.d. If there were a fact by virtue of which b were sufficiently similar to c, there

would be grounds for ranking skeptical scenarios for b more remote thanmatches for b.

e. There could be grounds for ranking skeptical scenarios for b more remotethan matches for b only if b and c are grounded in something transcendentthat gives b and c their content—in fact, the same content.

f. Something independent of individual, finite minds can ground some aspectof finite minds only if there is something with causal power, independentof individual finite minds, that makes such grounding possible.

g. Only a transcendent causal power could ground the transcendent.h. Nothing natural is transcendent.i. Anti-realism grounds content in some feature of a collection of finite minds.j. A finite collection of finite minds does not suffice to explain the grounding

of similarity of content.k. An infinite collection of finite minds does not suffice to explain the ground-

ing of similarity of content.l. The best explanation for the existence of a supernatural, transcendent causal

power grounding the similarity of content includes the existence of God.m. So, there is a God.

This argument too establishes the existence of a supernatural, transcendent causalpower. It also reflects a common religious intuition—that apart from God we are ut-terly alone. Only by way of relation to God are we able to relate to one another. Godunderlies community. It goes beyond the argument from content in explaining not justhow my words and thoughts can have content but how our words and thoughts can haveand share content.

7.3 The Argument from TruthLet’s turn finally to the argument from truth. How is it possible for us to relate tothe world? How do our thoughts and words manage to connect to a world that isindependent of our minds? Again, the idea is that there is no non-magical explanation.The only plausible account we have in the end relies on God.

32

(35) a. If realism is true, then, given a belief or perception b, among our possibil-ities are skeptical scenarios for b—situations indiscernible from matchesfor b from the agent’s point of view in which b is not true or veridical.

b. A belief or perception can be true or veridical.c. A belief or perception b can be true or veridical only by virtue of some

fact.d. If there were a fact by virtue of which b were true or veridical, there would

be grounds for ranking skeptical scenarios for b more remote than matchesfor b.

e. There could be grounds for ranking skeptical scenarios for b more remotethan matches for b only if b is grounded in something transcendent con-necting b to the world.

f. Something independent of individual, finite minds can ground a connectionto the world only if there is something with causal power, independent ofindividual finite minds, that makes such grounding possible.

g. Only a transcendent causal power could ground the transcendent.h. Nothing natural is transcendent.i. Anti-realism grounds content in some feature of a collection of finite minds.j. A finite collection of finite minds does not suffice to explain the grounding

of a connection to the world.k. An infinite collection of finite minds does not suffice to explain the ground-

ing of a connection to the world.l. The best explanation for the existence of a supernatural, transcendent causal

power grounding a connection between thought or language and the worldincludes the existence of God.

m. So, there is a God.

This argument goes further than the arguments from content and intersubjectivity, forits explains not only how our words and thoughts can have and share content but alsohow the content of our words and thoughts can relate to the world in such a way thatwe can perceive the world as it is and say and think true things. This too correspondsto a religious intuition: that apart from God we are blind; we can do nothing. Apartfrom God, we are radically separated from the world. As Justin Martyr declares, “Theword of truth is sent from God.”

Nothing in the arguments from content and intersubjectivity entail God’s omni-science or omnipotence, though of course they are part of the best explanation as welland in that way receive indirect support from those arguments. The argument fromtruth, however, seems to offer somewhat more direct support. To be capable of relatingour minds to the world, God must be master of the world as well as of the intellectualrealm. That provides some ground, if not itself decisive ground, for holding that Godis or has a mind and that God has complete knowledge and power over us as well asthe world.

33

8 Epistemological Arguments: The Arguments from In-terpretation, Communication, and Knowledge

Now let’s turn to epistemological arguments, streamlined to remove some controversialand unnecessary commitments from the original Platonic/Augustinian arguments.

8.1 The Argument from InterpretationFirst, consider the argument from interpretation, the epistemic correlate of the argu-ment from content. It asks how we can know the contents of our words and thoughts.The core idea is simple: there is no naturalist way to explain our own semantic knowl-edge. We at least in some cases know the meanings of what we say, what we think,and what we do. We know what we are referring to, what we intend, what we perceive,and what we remember. There is no way to account for this without at some pointappealing to something like magic. And the only explanation we have of that magic isreligious, involving God.

Since the problem of interpreting language might be thought to differ from theproblem of interpreting thought, let’s split this into two arguments, one focusing onour knowledge of the contents of language and the other focusing on our knowledgeof the contents of thoughts. It is important to remember that the problem is not onlyhow I can interpret your words and thoughts but how I can interpret my own. As Quine[1960, 1969] observes, the indeterminacy of interpretation begins at home.

(36) a. If realism is true, then, given a state of mind s, among our epistemic possi-bilities are skeptical scenarios for s—situations indiscernible from matchesfor s from the agent’s point of view in which that state of mind does nothave its designated content.

b. We can know that our states of mind have their designated contents. (Weare capable, that is, of understanding at least some of our thoughts.)

c. We can know that a state of mind s has its designated content only ifwe have grounds for ranking skeptical scenarios for s more remote thanmatches for s.

d. For a state of mind s, we could have grounds for ranking skeptical scenar-ios for s more remote than matches for s only if s is grounded in somethingtranscendent and epistemically accessible which gives s its content.

e. Something independent of individual, finite minds can be epistemicallyaccessible only if there is something with causal power, independent ofindividual finite minds, that makes such access possible.

f. Only a transcendent causal power could make possible epistemic access tothe transcendent.

g. Nothing natural is transcendent.h. Anti-realism grounds content in some feature of a collection of finite minds.i. A finite collection of finite minds does not suffice to explain our epistemic

access to content.

34

j. An infinite collection of finite minds does not suffice to explain our epis-temic access to content.

k. The best explanation for the existence of a supernatural, transcendent causalpower making it possible for us to have veridical perception and true be-liefs includes the existence of God.

l. So, there is a God.

The linguistic variant is similar:

(37) a. If realism is true, then, given an expression s, among our epistemic possi-bilities are skeptical scenarios for s—situations indiscernible from matchesfor s from the agent’s point of view in which that expression does not haveits designated content.

b. We can know that an expression has its designated content. (We can, inother words, sometimes know what a word means.)

c. We can know that an expression s has its designated content only if we havegrounds for ranking skeptical scenarios for s more remote than matches fors.

d. For a state of mind s, we could have grounds for ranking skeptical scenar-ios for s more remote than matches for s only if s is grounded in somethingtranscendent and epistemically accessible which gives s its content.

e. Something independent of individual, finite minds can be epistemicallyaccessible only if there is something with causal power, independent ofindividual finite minds, that makes such access possible.

f. Only a transcendent causal power could make possible epistemic access tothe transcendent.

g. Nothing natural is transcendent.h. Anti-realism grounds content in some feature of a collection of finite minds.i. A finite collection of finite minds does not suffice to explain our epistemic

access to content.j. An infinite collection of finite minds does not suffice to explain our epis-

temic access to content.k. The best explanation for the existence of a supernatural, transcendent causal

power making it possible for us to have veridical perception and true be-liefs includes the existence of God.

l. So, there is a God.

These arguments demonstrate the existence of a supernatural, infinite, eternal, neces-sary, and independent being, one who illumines our minds, sets standards of right andwrong, good and bad, correctness and incorrectness, and enables us to know what theyare. The arguments reflect a common religious intuition: that without God we couldassign no meaning to the world or to our own thoughts, words, and actions. The ar-gument from content reflects a similar intuition, that without God none of these wouldhave meaning. The argument from interpretation maintains that without God we could

35

not know that anything had any meaning. Apart from God, we could not help but seethe world as “sound and fury, signifying nothing.”

8.2 The Argument from CommunicationSecond, let’s turn to the argument from communication. How is it possible for usto understand one another? How can I know what you mean? Why should I have anyconfidence that can understand what you say or think or do? How could I know whetheryou and I agree? Think the same about something? Mean the same by our words anddeeds? How, in short, can we communicate, and know that we are communicating?There is no naturalistic way to explain that. Our best explanation invokes the power ofGod.

This problem, too, begins at home. How can I know what I meant yesterday, anhour ago, or even as I began this sentence? How can I know that my meaning hasn’tchanged? How can I know what I would have meant had the world been slightlydifferent? How then could I know the meaning of any speech or thought involving acounterfactual scenario, or, for that matter, a conditional of any kind? Apart from God,I have no explanation.

(38) a. If realism is true, then, given an agent A’s state of mind s, among our epis-temic possibilities are skeptical scenarios for s—situations indiscerniblefrom matches for s from the agent’s point of view in which, for example, sis not sufficiently similar to any mental state t of B.

b. We can know that some of our states of mind are sufficiently similar tothose of someone else (or of ourselves at another time, or in a counterfac-tual circumstance).

c. We can know that a state of mind s is sufficiently similar to t only if we havegrounds for ranking skeptical scenarios for s more remote than matches fors.

d. For a state of mind s, we could have grounds for ranking skeptical scenar-ios for s more remote than matches for s only if s is grounded in somethingtranscendent and epistemically accessible which gives s and t their content.

e. Something independent of individual, finite minds can be epistemicallyaccessible only if there is something with causal power, independent ofindividual finite minds, that makes such access possible.

f. Only a transcendent causal power could make possible epistemic access tothe transcendent.

g. Nothing natural is transcendent.h. Anti-realism grounds content in some feature of a collection of finite minds.i. A finite collection of finite minds does not suffice to explain our epistemic

access to similarity of content.j. An infinite collection of finite minds does not suffice to explain our epis-

temic access to similarity of content.

36

k. The best explanation for the existence of a supernatural, transcendent causalpower making it possible for us to communicate, knowing that contents aresimilar or dissimilar, includes the existence of God.

l. So, there is a God.

Once again we have an argument for a supernatural, transcendent being who makesit possible for us to understand one another and ourselves. It goes beyond the argu-ment from interpretation in pointing out that God makes possible not only my self-knowledge but my knowledge of others. This argument shows why a single transcen-dent ground is preferable to multiple grounds, for there must be something groundingthe meaning of both s and t, for arbitrary s and t and arbitrary agents.

This too reflects a common religious intuition: that apart from God I can understandnothing. I cannot understand you; I cannot understand myself. As with the argumentfrom intersubjectivity, God underlies the possibility of community. Communion withothers and even with ourselves presupposes communion with God.

8.3 The Argument from KnowledgeFinally, consider the ultimate anti-skeptical argument, the argument from knowledge.

(39) a. If realism is true, then, given a state of mind s, among our epistemic possi-bilities are skeptical scenarios for s—situations indiscernible from matchesfor s from the agent’s point of view in which s is not true or veridical.

b. We can know that some of our states of mind are true or veridical.c. We can know that a state of mind s is true or veridical only if we have

grounds for ranking skeptical scenarios for s more remote than matchesfor s.

d. For a state of mind s, we could have grounds for ranking skeptical scenar-ios for s more remote than matches for s only if s is grounded in somethingtranscendent and epistemically accessible that connects s appropriately tothe world, allowing s to be true or veridical.

e. Something independent of individual, finite minds can be epistemicallyaccessible only if there is something with causal power, independent ofindividual finite minds, that makes such access possible.

f. Only a transcendent causal power could make possible epistemic access tothe transcendent.

g. Nothing natural is transcendent.h. Anti-realism grounds content in some feature of a collection of finite minds.i. A finite collection of finite minds does not suffice to explain our access to

something transcendent.j. An infinite collection of finite minds does not suffice to explain our access

to something transcendent.k. The best explanation for the existence of a supernatural, transcendent causal

power making it possible for us to have veridical perception and true be-liefs includes the existence of God.

37

l. So, there is a God.

This argument goes beyond the arguments from interpretation and communication tosay that God is part of the best explanation of our knowledge of the world. Apart fromGod, I cannot understand the contents of my words and thoughts or their connectionto the world. I cannot know whether my perceptions are ever veridical. I cannot knowwhether my statements and beliefs are ever true. I certainly cannot know that any givenperception is veridical or any given statement or belief is true.

This too reflects a common religious intuition: Without God, I have no reason tothink of this world as anything but a hostile environment, a “field of death” [Nishitani1982], that is ultimately unintelliglble to me. God aligns my mind with the nature ofreality, making it possible for me at least sometimes to grasp the world as it is. God isthe light of the mind, as Lactantius declares, as well as the light of the world.

38

9 ReferencesAdams, E.W. [1970] “Subjunctive and Indicative Conditionals,” Foundations of Lan-guage, 89–94.

Annas, J. [1974] “Forms and First Principles,” Phronesis 19, 3, 257–283.

Annas, J., and Barnes, J. [1985] The Modes of Scepiticism. Cambridge: CambridgeUniversity Press.

Asher, N., and Morreau, M. [1991] “Commonsense Entailment: A Modal Theory ofNonmonotonic Reasoning,” In Logics in AI, Berlin, Heidelberg: Springer, 1–30.

Asher, N., and Morreau, M. [1995]. “What Some Generic Sentences Mean,” in Carl-son, G., and Pelletier, J. (ed.), The Generic Book. Chicago: University of ChicagoPress.

Augustine. [1955] The Problem of Free Choice (De Libero Arbitrio). Westminster,Maryland: The Newman Press and London: Longmans, Green and Company.

Boghossian, P. A. [1989] “The Rule-following Considerations,” Mind, 507–549.

Boghossian, P. A. [1990] “The Status of Content,” The Philosophical Review, 157–184.

Bonevac, D., and Dever, J. [2012] “A History of the Connectives,” in Dov Gabbay andJohn Woods (ed.), Handbook of the History of Logic, Volume 11. Amsterdam: ElsevierNorth Holland, 175–234.

Bonevac, D., Dever, J., and Sosa, D. [2006] “The Conditional Fallacy,” The Philosoph-ical Review 115, 3, 273–316.

Bonevac, D., Dever, J., and Sosa, D. [forthcoming] “There Goes the Neighborhood.”

Brewka, G. [1991] “Cumulative Default Logic: In Defense of Nonmonotonic InferenceRules,” Artificial Intelligence, 50, 2, 183–205.

Brueckner, A. [1992] “Semantic Answers to Skepticism,” Pacific Philosophical Quar-terly 73, 200–219.

Brueckner, A. [1994] “The Structure of the Skeptical Argument,” Philosophy and Phe-nomenological Research 54, 4, 827–835.

Burnyeat, M. [1983] “ Can the Sceptic Live His Scepticism?” in Burnyeat (ed.), TheSkeptical Tradition. Berkeley: University of California Press, 117–148.

39

Chomsky, N. [1965] Aspects of the Theory of Syntax. Cambridge: MIT Press.

Chuang Tzu. [1889, 1923] Translated from the Chinese by Herbert A. Giles. London:Bernard Quaritch.

DeRose, K., and Warfield, T. [1999] Skepticism: A Contemporary Reader. Oxford:Oxford University Press.

Fine, G. [1980] ‘The One Over Many,” Philosophical Review 89, 2, 197–240.

Fine, G. [1995] On Ideas: Aristotle’s Criticism of Plato’s Theory of Forms. Oxford:Oxford University Press.

Fine, G. [2003] Plato on Knowledge and Forms: Selected Essays. Oxford: ClarendonPress.

Greco, D. [2012] “The Impossibility of Skepticism,” Philosophical Review 121(3),317–358.

Holliday, W. H. [2013] “Epistemic Logic and Epistemology,” in Hansson, S. O., andHendricks, V. F. (ed.), Handbook of Formal Philosophy.

Hume, D. [1739] An Enquiry Concerning Human Understanding. London.

Justin Martyr. [1885] On the Resurrection, Fragments of the Lost Work of Justin onthe Resurrection, I; in Roberts, A., Donaldson, J., and Coxe, A. C. (ed.), Ante-NiceneFathers, volume 1: The Apostolic Fathers with Justin Martyr and Irenaeus. Edinburgh:T&T Clark; Grand Rapids: Wm. B. Eerdmans Publishing Company.

Kamp, H., and Reyle, U. [1993] From Discourse to Logic. Dordrecht: Kluwer.

Kripke, S. [1982] Wittgenstein on Rules and Private Language. Cambridge: HarvardUniversity Press.

Kung, P. [2011] “On the Possibility of Skeptical Scenarios,” European Journal of Phi-losophy, 19(3), 387–407.

Lactantius. [1885] A Treatise on the Anger of God Addressed to Donatus, in Roberts,A., Donaldson, J., and Coxe, A. C. (ed.), Ante-Nicene Fathers, volume 7: Fathers ofthe Third and Fourth Centuries: Lactantius, Venantius, Asterius, Victorinus, Dionysius,Apostolic Teaching and Constitutions, Homily. Edinburgh: T&T Clark; Grand Rapids:Wm. B. Eerdmans Publishing Company.

Lewis, D. [1973] Counterfactuals. Cambridge: Harvard University Press.

40

Locke, J. [1690] An Essay Concerning Human Understanding. London.

McCarthy, J. [1986] “Applications of Circumscription to Formalizing Common-senseKnowledge,” Artificial Intelligence, 28(1), 89–116.

Morreau, M. [1997] “Fainthearted Conditionals,” The Journal of Philosophy, 187–211.

Nishitani Keiji. [1982] “Science and Zen,” in Franck, F. (ed.), The Buddha Eye: AnAnthology of the Kyoto School. New York: Crossroad.

Nozick, R. [1981] “Knowledge and Skepticism,” Philosophical Explanations. Cam-bridge: Belknap Press, 172–85.

Nute, D. [1975] “Counterfactuals and the Similarity of Words,” The Journal of Philos-ophy, 773–778.

Peirce, C.S. [1878] “How to Make Our Ideas Clear,” Popular Science Monthly 12, 286–302.

Plantinga, A. [1982] “How To Be an Anti-Realist,” Proceedings of the American Philo-sophical Association 56, 57–70.

Plantinga, A. [2007] “Two Dozen (or so) Theistic Arguments,” in Baker, D. P. (ed.),Contemporary Philosophy in Focus. Cambridge: Cambridge University Press, 203–228.

Popkin, R. H. [2002] Skeptical Philosophy for Everyone. New York: Prometheus.

Putnam, H. [1978] Meaning and the Moral Sciences. London: Routledge and KeganPaul.

Putnam, H. [1981] Reason, Truth and History. New York: Cambridge University Press.

Quine, W. V. O. [1960] Word and Object. Cambridge: MIT Press.

Quine, W. V. O. [1960] Ontological Relativity. New York: Columbia University Press.

Reiter, R. [1980] “A Logic for Default Reasoning,” Artificial Intelligence, 13(1), 81–132.

Rorty, R. [1998] “Hilary Putnam and the Relativistic Menace,” in Truth and Progress.New York: Cambridge University Press.

Russell, B. [1919] “Professor Dewey’s Essays in Experimental Logic,” The Journal ofPhilosophy 16, 1, 5–26.

41

Schofield, M., Burnyeat, M., and Barnes, J. (ed.) [1980] Doubt and Dogmatism: Stud-ies in Hellenistic Epistemology. Oxford: Clarendon Press.

Sellars, W. [1956] “Empiricism and the Philosophy of Mind,” in Feigl, H., and Scriven,M. (eds.), Minnesota Studies in the Philosophy of Science, vol. I, Minneapolis: Uni-versity of Minnesota Press, 253–329.

Sellars, W. [1974] “Reason and the Art of Living in Plato,” in Essays in Philosophyand Its History. Dordrecht: D. Reidel, 3–26.

Stalnaker, R.C. [1981] “A Theory of Conditionals,” in Harper, W. L., Stalnaker, R., andPearce, G. (ed.), Ifs (pp. 41–55). Amsterdam: Springer, 41–56.

Stillwell, S. [1989] “Plantinga and Anti-Realism,” Synthese 78, 87–115.

Striker, G. [1980] “Sceptical Strategies,” in Schofield, Burnyeat, and Barnes (ed.), 54–83.

Waterlow, S. [1982] “The Third Man’s Contribution to Plato’s Paradigmatism,” Mind91, 339–357.

Williamson, T. [2007] “On Being Justified in One’s Head,” in Timmons, M., Greco,J., and Mele, A. R. (ed.), Rationality and the Good: Critical Essays on the Ethics andEpistemology of Robert Audi. Oxford: Oxford University Press, 106–22.

Wittgenstein, L. [1953]. Philosophical Investigations. London: Blackwell.

42