(此文原載于《Argumentation》網絡版發表于2016年4月)

Must a Successful Argument Convert an Ideal Audience?

Xingming Hu

Nanjing University

Forthcoming in Argumentation.

Online version: http://link.springer.com/article/10.1007%2Fs10503-016-9402-x

Abstract: Peter van Inwagen defines a successful argument in philosophy as one that can be used to convert an audience of ideal agnostics in an ideal debate. Sarah McGrath and Thomas Kelly recently argue that van Inwagen’s definition cannot be correct since the idea of ideal agnostics is incoherent with regard to an absolute paradigm of a successful philosophical argument. This paper defends van Inwagen’s definition against McGrath and Kelly’s objection.

Key words: argument; philosophy; ideal agnostics; the epistemological approach; Peter van Inwagen

1. Introduction

In the Republic, Socrates instructs his interlocutor Adeimantus that “whatever direction the argument blows us, that’s where we must go” (Plato 1997, 394d). Surely, Socrates does not mean that we should follow any argument where it leads. After all, some arguments are bad. They are not worth following. What Socrates claims is that we should always follow good arguments.

But what is a good argument? Many philosophers (e.g. Feldman 1994 Goldman 1999, and Lumer 1991) take an epistemological approach to this question, claiming that an argument is good or successful if it achieves the aim of argument: to enable the addressee to obtain knowledge of the conclusion or rational (or justified) belief in the conclusion. Since the aim of argument entails belief in the conclusion, the epistemological approach takes (rational) persuasion as a necessary condition for good argument. If an argument fails to persuade the addressee to believe its conclusion, it is not good.

In this paper, I focus on a specific version of the epistemological approach: Peter van Inwagen’s account of good argument in philosophy. Van Inwagen (2006) proposes that a philosophical argument for p is good or successful iff it can be used, in an ideal debate, to convert an audience (who has no initial opinions about p) to belief in p. This is a version of the epistemological approach because the audience in an ideal debate is perfectly rational so that they would not believe anything irrationally. If an argument for p is capable of converting a perfectly rational audience (who has no initial opinions about p) to belief in p, then it can help the audience acquire the new rational belief p.

Van Inwagen’s account is significant for at least two reasons. First, it is an original contribution to the philosophy of argumentation. To be sure, the notion of an ideal audience or observer is by no means new and has received considerable attention in ethics. Many philosophers (e.g. Firth 1952, Brandt 1995, Hare 1981, and Carson 1984) hold that an action is morally right if and only if an ideal audience or observer would approve of the action (in some particular way). But it is van Inwagen who first develops an interesting theory of argumentation in terms of the notion of an ideal audience. All other well-developed versions of the epistemological approach to argumentation such as Feldman (1994), Goldman (1999), and Lumer (2005) define a successful or good argument in terms of real people. Second, if van Inwagen’s account is correct, then those other well-developed versions of the epistemological approach are wrong. For example, suppose our argument fails to convince the real people who we argue with. Then the addressees of the argument do not acquire knowledge of, or justified belief in, the conclusion. In light of van Inwagen’s account, our argument might still be good. However, those other well-developed versions of the epistemological approach entail that our argument is bad.

Van Inwagen’s account of successful argument in philosophy has been discussed by several philosophers, such as Fischer and Tognazzini (2007), Ballantyne (2014a), and Hanna (2015). McGrath and Kelly recently raise a new objection, which states that van Inwagen’s idea of a perfectly rational audience is incoherent with regard to “an absolute paradigm of a successful philosophical argument” (forthcoming, p. 14). This objection is worth close examination because if it were correct, not only would van Inwagen’s account of successful argument be seriously flawed and beyond repair, but also any attempt to define successful argument in terms of a perfectly rational audience would be doomed to failure.

In this paper, I argue that McGrath and Kelly’s objection is untenable. I shall proceed as follows. First, I will briefly explain van Inwagen’s criterion of successful argument and how it is different from alternative versions of the epistemological approach. Next I will introduce McGrath and Kelly’s objection, identify what I take to be a flaw in their approach, and consider some possible replies from them. I will close with a summary of the discussion and a brief note on a different worry about van Inwagen’s criterion of successful argument, that is, it is not helpful in practice because we—real people — cannot know how an ideal audience would react to an ideal debate.

2. Van Inwagen’s criterion of successful argument

We make arguments when we debate with others. Van Inwagen thinks that an argument should be evaluated only in relation to a debate. His idea of debate is based on the forensic model: two representatives of opposed positions argue with each other before an audience. And the aim of the two debaters is not to convert each other but rather to convert the audience.  

Van Inwagen’s basic account is teleological, that is, an argument is successful iff it achieves the aim of either debater—to convert the audience. But in real life, the audience might be emotional or irrational or unable to think logically. They might already have taken sides before hearing the arguments. And the debaters might appeal to the emotions rather than the reason of the audience. In addition, the time of a real debate is always limited: it usually lasts for less than two hours, and often much less. Finally, the room where the debate happens might be ill-equipped such that the audience sometimes cannot hear what the debaters say. If an argument successfully converts the audience because the audience didn’t hear clearly what the counterarguments are, or because there was no time for the audience to hear objections, or because the audience’s minds were too clouded to think logically, etc., then it seems that such argument cannot be deemed a success. To handle this problem, van Inwagen defines successful argument in philosophy in terms of ideal debate. Specifically,

An argument for p is a success just in case it can be used, under ideal circumstances, to convert an audience of ideal agnostics (agnostics with respect to p) to belief in p— in the presence of an ideal opponent of p. (van Inwagen 2006, p. 47)

Call this criterion “van Inwagen’s Criterion” or “VIC” for short. An ideal circumstance is one where the debaters —proponents and opponents of p—and the audience are given “a quiet, comfortable room with a blackboard, and enough chalk and time” (van Inwagen 2006, p. 43).  And ideal debaters satisfy the following two conditions:

1. [They] are of the highest possible intelligence and of the highest degree of philosophical and logical acumen, and they are intellectually honest in this sense: when they are considering an argument for some thesis, they do their best to understand the argument and to evaluate it dispassionately.

2. [They] have unlimited time at their disposal and are patient to a preternatural degree; they are, like General Grant, prepared to fight it out on this line if it takes all summer, and if their opponents think it necessary to undertake some lengthy digression into an area whose relevance to the debate is not immediately evident, they will cooperate. (van Inwagen 2006, pp. 42-43)

An audience of ideal agonistics is attributed the same unlimited leisure and superhuman patience, as well as the same high intelligence and high degree of logical and philosophical acumen and intellectual honesty. What differentiates the audience from the debaters is that each member of the audience bears no initial allegiance or predilection to either position though they are interested in finding out which one is correct.

The audience of ideal agnostics is not agnostic about everything, however. They may have some beliefs that are logically relevant to the subject matter under debate. This is because, on van Inwagen’s view, the audience of ideal agnostics is drawn from our time and our culture. So they share the beliefs in our time and culture, that is, the beliefs they have are widely accepted as true in our society. Since some of the widely accepted beliefs in our society might be false, an argument that convinces ideal agnostics might still have false premises and a false conclusion. So the conclusion of a successful argument, according to van Inwagen, might be false.

As we have seen, a salient feature of VIC is its appeal to the idea of ideal audience. This feature also distinguishes it from other well-developed versions of the epistemological approach, such as Feldman (1994), Goldman (1999), and Lumer (2005), which define successful argument in terms of a real audience (people in the real world). On this view, an argument is successful iff it can achieve the aim of argument, which is to help a real audience obtain knowledge of, or rational (or justified) belief in, its conclusion. Unlike an ideal audience, the members of a real audience often differ from each other in background knowledge and reasoning abilities. Some are more knowledgeable in a certain area than others; some can figure out that a complex argument is deductively valid while others cannot. Those who define successful argument in terms of a real audience take these differences into account and make the status of an argument relative to its audience: whether an argument is good always depends on to whom it is addressed. If an audience is not convinced (because they don’t follow the argument or they don’t think some premises are true), then the argument is not good for them. It makes no sense to say that an argument is successful independent of any context. Put differently, it is “successful argument for real audience S” rather than “successful argument” that is the proper object for philosophers of argumentation to define.  Suppose AR is a non-circular and deductively valid argument whose premises are all true. It has been addressed to two different groups of real audience. Both groups didn’t believe the conclusion before hearing the argument. After hearing the argument, Group 1 is convinced of the conclusion because they not only believe that all the premises are true and that the conjunction of all the premises entails the conclusion but also are justified in believing so. And the argument is not defeated for them. In contrast, Group 2 is not convinced of the conclusion by the argument because they think some of its premises are false. Then AR is a successful argument for Group 1 but not successful for Group 2.

However, a relativism of this sort (call it “Ahistorical Relativism”) is counter-intuitive. For instance, it is counterintuitive to say that Andrew Wiles’ magnificent argument for Fermat's Last Theorem is successful for a group of mathematicians but unsuccessful for a group of politicians. Moreover, Ahistorical Relativism implies that an argument is unsuccessful for those who cannot follow it simply because they cannot follow it. This implication is not only counter-intuitive but also has undesirable consequences. It would make people reject a complex argument too easily without making efforts to understand it (“It is not a good argument for me because I don’t believe that the premises lend strong support to the conclusion”).

Of course, it is entirely possible for the philosophers who endorse Ahistorical Relativism to come up with a plausible response to this problem. But VIC can avoid this problem to a certain extent since it makes the status of an argument completely depend on an ideal audience. In fact, VIC implies that if an argument can be used to convince the ideal audience in an ideal debate, then it is a successful argument even for those who are not convinced by it. To be sure, VIC still has an element of relativism, for on van Inwagen’s view, the ideal audience is drawn from our time and our culture. And he explicitly admits, “It is certainly possible that an argument that would have succeeded in, say, convincing an eighteenth-century audience that space was infinite would not succeed with an audience of our contemporaries” (van Inwagen 2006, p. 47). So he seems to endorse the following form of relativism about successful argument: an argument is successful in a certain culture and time iff it can be used in an ideal debate to convince the ideal audience who is drawn from that culture and time. Call this form of relativism “Historical Relativism,” in contrast to Ahistorical Relativism. Historical Relativism might also have problems, but it does not lead to the counter-intuitive consequence that an argument is unsuccessful for those who cannot follow it simply because they cannot follow it.

3. McGrath and Kelly’s Objection

So far, I have explained what VIC is and how it is different from other versions of the epistemological approach. In what follows, I will examine an interesting objection to VIC recently raised by Sarah McGrath and Thomas Kelly. It runs as follows:

1. Ideally rational agents possess a kind of logical omniscience such that they would not fail to believe propositions that are logically entailed by the things that they know.

2. “A transparently valid argument for a philosophically substantive conclusion from generally known premises …would be an absolute paradigm of a successful philosophical argument” (McGrath and Kelly forthcoming, p. 14). Call this argument “Successful Philosophical Argument” or “SPA” for short.

3. If a proposition is generally known, then ideally rational agents must know it.

4. Therefore, ideally rational agents would already believe the conclusion of SPA before SPA was ever presented to them.

5. An ideal agnostic with respect to p is an ideal agent who has no initial opinion about whether p is true, and moreover, “no predilection, emotional or otherwise” to accept either p or not-p.

6. So an ideally rational agent cannot also be an agnostic with respect to the conclusion of SPA. That is, there is no ideal agnostic with respect the conclusion of SPA.

7. If VIC is correct, then if there is no ideal agnostic with respect to the conclusion of an argument, then the argument cannot be successful.

8. Therefore, VIC is incorrect. (Call this objection “McGrath-Kelly” or “M-K” for short.)

McGrath and Kelly seem to think Premise (1) is true by definition. On their view, van Inwagen’s ideal agnostics are logically omniscient agents because van Inwagen claims ideal agnostics possess the highest possible intelligence and the highest degree of logical acumen.

Premise (2), according to McGrath and Kelly, is not negotiable: “it is a condition of adequacy on any proposed criterion of philosophical success that it returns the verdict that an argument with these characteristics would be a successful philosophical argument” (McGrath and Kelly forthcoming, p. 14).

McGrath and Kelly think Premises (3) and (5) are actually endorsed by van Inwagen. (4) necessarily follows from (1), (2), and (3). And (6) necessarily follows from (4) and (5). Premise (7), on McGrath and Kelly’s view, is a logical truth. And the conclusion (8) necessarily follows from (2), (6), and (7).

 Some might argue that van Inwagen does not have to accept (3). He can release the tension by taking ideal agnostics to be thinkers with the ability to reason impeccably, and hence to see what follows from their beliefs. Basically, he might think the kind of rationality the idealized thinkers enjoy is dispositional, and so they have the relevant rationality in advance of having any beliefs or knowledge, just because they've got the relevant abilities. So far, it seems as though we can imagine how these thinkers could be both agnostic and ideal: they don't share the common knowledge that makes up the premises.

But McGrath and Kelly do not think that van Inwagen can make this move, for “Once knowledge of the propositions employed as premises is removed or bracketed, the presentation of the argument might very well fail to persuade otherwise ideal individuals of the conclusion, but that hardly shows (or even provides reasons to think) that the argument is a failure” (forthcoming, pp. 15-16). Here they seem to reason as follows: if ideal agnostics do not know the generally known premises, then they will not be persuaded by SPA, and consequently, SPA is not successful given VIC. But SPA is successful—there is no good reason to deny this. So if ideally rational agents do not know the generally known premises, van Inwagen has to reject VIC.

4. A reply to M-K

M-K might appear plausible at first glance, but it cannot stand up to close examination.  Specifically, Premise (1) of M-K is too strong and should be replaced by the following claim:

LA: Ideally rational agents possess the highest possible degree of logical acumen such that they would not fail to believe p if p is logically entailed by a set of propositions that they know and they consider these propositions together.

If one never considers p and q together, that is, one never associates p with q, one would have no idea what the conjunction of p and q would imply even if one knows p and q separately. For example, suppose I know that flowers are the plant's reproductive structures. I also know that some people send flowers to graves. But I never consider these two things together. So I never believe that some people send the plant's reproductive structures to graves. But had I considered the two things together, I would immediately “see” that some people send the plant's reproductive structures to graves because I have sufficient logical acumen to figure out the logical relationship between them. It seems that an ideally rational agent might know each premise of an argument but never associate them with each other before being presented with the argument. So they might fail to believe the conclusion even if it is actually entailed by the premises they know.

In light of LA, an argument might qualify as successful if it converts the ideal agnostics by making manifest the previously unrecognized connections between its premises.

5. Some objections and replies

McGrath and Kelly actually consider this reply but reject it as unpromising. On their view, the only good reason why ideally rational agents fail to consider together the premises of an argument they know is that the number of the premises is quite large and these premises are seemingly disparate and unrelated, for “full rationality does not require one to believe logical consequences of what one knows that are extremely difficult to recognize” (forthcoming, p. 17. Also cf. Cherniak 1986 and McGrath 2010). But as McGrath and Kelly point out, we may imagine a transparently valid philosophical argument that employs a relatively small number of generally known premises. (It is not a sketchy argument. All its premises are spelled out.) Call this argument “Small Argument” or “SA” for short. McGrath and Kelly claim that SA would surely qualify as a success. But they think ideally rational agents must be able to see the connections between the premises of SA and thereby would not fail to believe the conclusion before SA is presented to them. Although McGrath and Kelly do not explain why ideally rational agents must have this ability, they might make the following reasoning: an agent who fails to see the connections cannot be called “ideal” in the sense van Inwagen uses the term. This is because, on van Inwagen’s view, ideally rational agents are not only logically omniscient and highly intelligent but also have unlimited leisure and superhuman patience. So they would consider carefully all things they know before hearing any argument in an ideal debate. As a result, they would be able to “see” the connections between the premises of SA that only employs a relatively small number of generally known premises (even though they might still fail to consider together a large number of seemingly disparate and unrelated propositions).

However, van Inwagen may respond to this objection by making a distinction between logical acumen and psychological acumen. An agent of logical acumen is able to recognize that (p and q) entails r (suppose the entailment holds), but she might still fail to associate p with q on her own. To associate p with q on one’s own, one needs to have psychological acumen. Now van Inwagen just claims that ideal agnostics have the highest degree of logical acumen; he does not say that they also have a very high degree of psychological acumen. This reading of van Inwagen’s view is at least acceptable to McGrath and Kelly since they tend to agree that full rationality does not require the ability to recognize the connections between a quite large number of propositions that are seemingly disparate and unrelated, an ability which is impossible without a very high degree of psychological acumen. Now if ideal agnostics do not have a high degree of psychological acumen, they might still fail to consider together the premises of SA that only employs a relatively small number of generally known premises. Note that here we don’t have to assume that ideal agnostics have a very low degree of psychological acumen, for an agent of normal degree of psychological acumen might be unable to consider together a relatively small number of generally known propositions and consequently fail to see what follows from the conjunction of them. For instance, it is generally known that everything is made of atoms. It is also generally known that some human beings hate each other. But even very smart people in the real world might be unable to associate these two generally known facts (and thus fail to figure out what the conjunction of them entails) on their own. Van Inwagen can stipulate that ideal agnostics, though having the highest degree of logical acumen, only have a normal degree of psychological acumen.

But McGrath and Kelly might find this reply inadequate. They might insist that even if ideal agnostics only have a normal degree of psychological acumen, they must be still able to see the connections between the premises of SA (and thereby would not fail to believe the conclusion) before SA is presented to them. Here is the reason. Ideal agnostics can consider together a relatively small number of the generally known propositions in a systematic way. For example, suppose the generally known propositions are P1, P2, and P3 …, and Pn. Ideal agnostics can first consider together P1 and P2 and figure out what the conjunction entails. Then they can consider (P1 and P3), (P1, P2, and P3),(P1 and P4), etc.in turn.This way of thinking does not seem to require a high degree of psychologicalacumen. It is kind of machinelike. Of course, it’s extremely boring and time-consuming. But recall that ideal agnostics, according to van Inwagen, have unlimited leisure and superhuman patience. So this would not be a problem for them. And they would be able to see the connections between a relatively small number of the generally known propositions whatever the propositions are.

To address this objection, van Inwagen may add a further condition on ideal agnostics: they will be given unlimited leisure and superhuman patience only when they serve as audience in an ideal debate.  Before hearing any argument in an ideal debate, they don’t have unlimited time to consider together all things they know separately in a machinelike way described above. This constraint surely diminishes the ideality of ideal agnostics. But it does not make them less rational or less able in logic. The key insight of VIC is preserved.

Nevertheless, McGrath and Kelly might not think the maneuver of making ideal agnostics less ideal (i.e. they don’t have the highest degree of psychological acumen nor do they have sufficient time to consider together everything they know in a machinelike way before hearing any argument) could save VIC. They might point to the fact that van Inwagen claims that ideal agnostics are drawn from our time and our culture. Since normal people of our time and our culture are able to independently consider together at least some things that they know, ideal agnostics must also be able to do this before entering into an ideal debate. Now we can imagine a philosophical argument that satisfies all the following three conditions:

(a) it is transparently valid;

(b) it employs a relatively small number of generally known premises;

(c) normal people of our time and our culture have sufficient psychological acumen to consider these premises together before being presented with the argument (though they lack sufficient logical acumen to see what these premises entail).

Call this argument “Normal Small Argument” or “NSA” for short. Given (c), ideal agnostics must also be able to consider together the premises of NSA before being presented with NSA. Given their extraordinary logical acumen, they would already believe its conclusion before hearing NSA. Hence, according to VIC, NSA is not a successful argument. But McGrath and Kelly would say VIC is wrong since NSA surely qualifies as a success.

There are two problems with this reply, however. First, it is impossible for us to imagine NSA in an ideal debate. According to van Inwagen, arguments that we can properly evaluate in light of VIC can only arise in a debate between ideally rational agents who, like ideal agnostics, possess the highest degree of logical acumen. If ideal agnostics can recognize the connections between the premises of NSA and “see” the conclusion necessarily follows from these premises and thereby already believe the conclusion before being presented with NSA, then ideal debaters must also be able to do so. Hence, the latter would not debate whether the conclusion is true since they already agree that it is true. That is, NSA would not arise in the first place. Second, it is doubtful that NSA could arise in a real debate. Note that NSA is supposed to be transparently valid. So normal real debaters (of average or higher degree of psychological acumen and logical acumen) must also be able to “see” its conclusion necessarily follows from the premises before hearing NSA if the ideal debaters can do it. Accordingly, normal real debaters would not debate whether the conclusion is true since they already agree that it is true. Put differently, it is unlikely that (a), (b), (c) can be all true in reality.

But McGrath and Kelly might further object that the very fact that NSA cannot arise in an ideal debate counts against van Inwagen’s criterion of successful argument, for van Inwagen’s criterion implies that NSA cannot be a successful argument. However, NSA is a successful argument, since it can be used, in a real debate, to convince a real audience who does not believe the conclusion of NSA before being presented with NSA. Of course, people of average or higher psychological acumen and logical acumen can figure out the conclusion before hearing NSA since it is transparently valid. But there are psychologically normal people who do not have sufficient logical acumen and thus cannot “see” what the premises of NSA entail before NSA is presented to them. So they might be converted to belief in its conclusion after hearing NSA. If NSA can be used to convert the audience (who is initially agnostic with respect to its conclusion) to belief in its conclusion in a real debate, it is a successful argument.

To this objection, van Inwagen may respond that NSA cannot be an absolute paradigm of a successful philosophical argument. First of all, NSA cannot be used in a debate to convince normal real people (i.e., the people who possess average or higher psychological acumen and logical acumen), simply because they already believe the conclusion before NSA is addressed to them. If an argument cannot be used to convince normal real people, it cannot be an absolute paradigm of a successful philosophical argument.

Moreover, NSA cannot convince by its logical force the people who are psychologically normal but possess such a low degree of logical acumen that they cannot figure out the conclusion before hearing NSA. Specifically, an argument might convince its audience in two ways: (a) by its logical or rational force and (b) by its non-logical or rhetoric force. Consider the following argument:

1. Human beings only possess a right to life in virtue of their dignity.

2. If a person commits murder, they lose their dignity.

3. Therefore, if a person commits murder, they lose their right to life. (cf. Pojman 2005, p. 108)

The argument is valid. Suppose the two premises are true and generally known. Some people are convinced by this argument because they “see” its logical force. But other people might be convinced by this argument merely because it is touching—it reminds them that someone they deeply love was murdered by a drug dealer, not because they “see” the logical force of the argument.

Now imagine that someone presents NSA to the people who are psychologically normal but possess such a low degree of logical acumen that they cannot figure out the conclusion before hearing NSA. Also imagine they are convinced of its conclusion after hearing NSA. How does NSA convert them to its conclusion? Obviously, it is not because they “see” the logical force of NSA, for despite the fact that NSA is transparently valid, they don’t have sufficient logical acumen that would enable them to “see” the conclusion necessarily follows from the conjunction of the premises. So NSA can merely convince them by its non-logical force. For example, the conclusion of NSA touches their heart, or NSA is presented to them by a person they trust. If an argument can convince people of its conclusion merely by its non-logical or rhetoric force, then it is not epistemically interesting at all.

6. Concluding Remarks

Summing up, I have shown that McGrath and Kelly fail to make a conclusive objection to van Inwagen’s criterion of successful argument. Van Inwagen defines a successful argument as one that can be used to convince an audience of ideal agnostics in an ideal debate. McGrath and Kelly argue that van Inwagen’s idea of ideal audience is incoherent with regard to an absolute paradigm of a successful philosophical argument, i.e., a transparently valid argument for a philosophically substantive conclusion from generally known premises. This is because an ideal audience cannot be agnostic with regard to the conclusion because they, being perfectly rational, are able to “see” the conclusion necessarily follows from the premises they know before hearing the argument. However, this objection does not succeed, for the fact that we know a few propositions separately does not entail that we are able to consider them together and figure out what the conjunction of them implies.

I consider three possible responses from McGrath and Kelly: (1) the ideal audience possesses an extraordinarily high degree of psychological acumen so that they are able to consider together any two or more propositions they know; (2) the ideal audience merely possesses an average degree of psychological acumen but is given unlimited time and leisure before hearing any argument so that they can consider together any two or more propositions they know in a machinelike way; (3) the premises of the absolute paradigm of a successful philosophical argument are such that people of an average degree of psychological acumen can easily consider together before hearing the argument.

I argue that van Inwagen may reject (1) and (2) without compromising his key idea that an argument should be evaluated in terms of a perfectly rational audience who is agnostic about the conclusion before hearing the argument. The ideal audience can still be perfectly rational even if (1) and (2) are false. In addition, it is doubtful that there is an absolute paradigm of a successful philosophical argument as required by (3), for the paradigm argument is supposed to be transparently valid. If the ideal audience can “see” its conclusion necessarily follows from the premises before hearing the argument, then real people, as long as they possess an average or higher degree of psychological acumen and logical acumen, can also “see” as much. Put differently, if the ideal audience cannot be agnostic with regard to the conclusion, nor can real people.

If the idea of an ideal audience as perfectly rational agnostics is not incoherent, then van Inwagen’s account seems to be a more promising version of the epistemological approach to argumentation. All other well-developed versions of the epistemic approach hold that an argument is good only if it can rationally convince real people. But real people are sometimes irrational and intellectually lazy. Intuitively, we cannot regard an argument as bad simply because it fails to rationally convince the people who are irrational or too lazy to figure out the argument is valid. Van Inwagen’s account nicely captures this intuition by holding that an argument should be evaluated in terms of an ideal audience.

One might worry that the idea of an ideal audience is useless in helping us assess arguments because people in the real world are not ideal debaters or ideal audience members and that we— real people – cannot know how an ideal audience would react to an ideal debate. Van Inwagen actually provides a reply to this worry: if an argument has been examined by normally well-informed and rational people in reality for a long time, then it has been tested under circumstances that sufficiently approximate the circumstances of an ideal debate. If the conclusion of this argument is still highly controversial, then it is likely that the argument is not good. However, if the conclusion becomes much less controversial after being examined by normally well-informed and rational people for a long time, then it is likely that the argument is good. Some people will no doubt find this view problematic (e.g. they might think that even members of an ideal audience could disagree with each other on whether some arguments are good), but it is an interesting view that is worth further research.

Acknowledgments: I'd like to thank Nathan Ballantyne, Stephen Grimm, David Kovacs, Emily Sullivan, and two anonymous referees for their helpful comments on earlier versions of this paper.

References

Ballantyne, N. (2014a). Knockdown arguments. Erkenntnis, 79, 525–543

Ballantyne, N. (2014b). Counterfactual philosophers. Philosophy and Phenomenological Research, 88(2), 368-387.

Brandt, R. B. (1955). The definition of an ideal observer theory in ethics. Philosophy and Phenomenological Research, 407-413.

Carson, T. (1984). The Status of Morality. Dordrecht: D. Reidel.

Cherniak, C. (1986). Minimal Rationality. MIT Press.

Feldman, Richard (1994). Good Arguments. In: Frederick F. Schmitt (ed.), Socializing Epistemology: The Social Dimensions of Knowledge. Lanham, MD: Rowman & . Littlefield. Pp. 159-188.

Firth, R. (1952). Ethical absolutism and the ideal observer. Philosophy and Phenomenological Research, 317-345.

Fischer, J. M., & Tognazzini, N. A. (2007). Exploring evil and philosophical failure. Faith and Philosophy, 24(4), 458-474.

Goldman, Alvin I. (I999). Knowledge in a Social World. Oxford: Clarendon.

Hanna, H. (2015). Philosophical success, Philosophical Studies, 172 (8),2109-2121.

Hare, R.M. (1981). Moral Thinking. New York: Oxford University Press.

Lumer, Christoph (1991). Structure and Function of Argumentations. An Epistemological Approach to Determining Criteria for the Validity and Adequacy of Argumentations. In Frans H. van Eemeren [et al.] (eds.), Proceedings of the Second International Conference on Argumentation. Amsterdam: Sicsat, 98-107.

Lumer, C.(2005a).The Epistemological Approach to Argumentation-A Map. Informal Logic 25(3), 189-212.

Lumer, C.(2005b).The Epistemological Theory of Argument-How and Why? Informal Logic 25(3), 214-232

McGrath, S.(2010).Moral Realism Without Convergence. Philosophical Topics 38(2): 59-90.

McGrath, S. and Kelly, T. (forthcoming) Are there any successful philosophical arguments? In Being, Freedom, and Method: Themes from van Inwagen, ed. John Keller, Oxford University Press.

Plato (1997). Complete Works. Eds. Cooper, J. M., & Hutchinson, D. S. Indianapolis: Hackett.

Pojman, L. (2005). A Defense of the Death Penalty, in Contemporary Debates in Applied Ethics, eds. Andrew Cohen and Christopher Wellman, Blackwell Publishing, pp. 107–23.

Siegel, H.(1997). Rationality Redeemed?: Further Dialogues on an Educational Ideal. Routedge.

van Inwagen, P. (2006). The Problem of Evil. Oxford University Press.