This past Summer I had the great good fortune to participate in the Summer Institute in Argumentation hosted by CRRAR. The Summer Institute preceded the OSSA conference, so the whole experience turned out to be about two and half weeks of really great discussions on all kinds of topics in argumentation theory and rhetoric.
One of the topics that’s been bouncing around in the back of my thoughts since then has been the question of whether or not an argument must have true premises in order to be good. The question was raised in a fantastic session on Chapter 7 of Hamblin’s Fallacies that was led by David Hitchcock during the Summer Institute. Hamblin, of course, answers this question in the negative, and I think it fair to say that the consensus of most of those attending agreed with him in that. For my part, I’ve been mulling it over since then and a few thoughts are beginning to emerge.
To start, I observe that if true premises were required for an argument to be good, it could only be after the fashion of necessary conditions. Just because the premises of an argument were true would not guarantee that they offered sufficient (or even any) support to the conclusion. So we have to begin with the admission that having true premises, if it is a criterion for a good argument, is only one of many potential criteria that belong in the full suite of those used to evaluate arguments.
Then there’s the question of whether, if true premises are required, it would be enough for the premises to be true apart from anyone’s actually knowing that they are true, or whether it would have to be known, at least to the hearers of the argument, that the premises are true. One wants to say that obviously it must be the latter, but recall that the question is “must an argument have true premises in order to be good”? Here it makes a difference whether we are thinking of argument as process or product, and whether by ‘good’ we mean ‘successful’ or something else. David Hitchock thinks the former–that a good argument is a successful one. I respectfully disagree with Hitchcock in that I think it important to separate criteria for an argument’s being good and an argument’s being successful. I don’t think the two notions are profitably collapsed. It seems to me that they do different work. I think of success as a criterion that applies primarily to argument as process. A successful argument in that sense, then, is one that does what it is intended to do given the dialogue in which occurs. An argument’s being successful in this way involves a happy synchronicity of sorts between the triad of arguer, argument, and audience such that the argument persuades its hearers of its conclusion. But of course, owing to qualities in either the arguer or the audience or both, logically bad, disingenuous, or fallacious arguments can be successful in some cases. So we should hesitate to define argumentative goodness in terms of success (I think). Of course one hopes that argumentative goodness and argumentative success will be found together, and certainly virtuous arguers try to achieve both aims, but it is all too obvious that it is not always so.
If that is what it is for an argument to be successful, then what is it for an argument to be good? To arrive at the notion of what it is for an argument to be good, I think it will be helpful to isolate the notion of argument in order to treat it as product. We can do this by idealizing the other two points in the triad: arguer and audience. Let us say, then that the ideal arguer would exhibit certain more-or-less epistemic and communicative virtues, e.g. honesty, clarity and economy of communication, etc.. Similarly the idealized audience (not unlike Perelman and Olbrechts-Tyetca’s universal audience, on Tindale’s reading) would exhibit particular epistemic and communicative virtues, for example: being free from clouding prejudices, being endowed with patience, having a desire to have true beliefs wherever possible, or at least to avoid being taken in by specious arguments that would lead them into false beliefs, having the ability to recognize a good argument when they hear one, and so on. A good argument then would be one that would convince under these conditions–where the force of the argument would be all that mattered. Tentatively, suppose the suggestion was that we think of a good argument as one that ought to convince our ideal audience of its conclusion. Would true premises be necessary under such a definition of goodness?
Certainly having true premises would count in favor of an argument’s overall goodness on such a definition, but it still seems too strong to say that an argument must have true premises in order to be good. For very often it will be difficult (or perhaps even impossible) for even an ideal audience of less-than-omniscient agents to know that the premises of a particular argument are true. If knowledge that the premises of an argument are true is going to be out of reach of what most agents can accomplish, then it seems as though we gain little by saying that an argument, to be good, must have true premises. The criterion would often be inapplicable owing to the epistemic limitations of the audience.
It seems far more realistic to say that, except in cases where their background knowledge is such that it allows them to make judgments with more precision, both arguers and agents presented with arguments will have to make “folk” judgments about the relative likelihoods of the premises. What I mean by this is not that agents will subjectively assign numeric probabilities to the premises, but that they will make judgments as best they can as regards the seeming-truth or seeming-likely-truth of the premises as given. Does this mean that having true premises is not required for an argument to be good? In my view, no. It means that it is not required premises be such that agents know that they are true before they endorse an argument as providing support for the conclusion. Instead, the requirement I would propose is this: for an argument to be good the premises must be such that a reasonable, informed, and (more-or-less) cognitively normal agent can incur no charge of epistemic vice for thinking that they are (at least) more likely than not to be true, given the information available at the time of judgment.
For an argument to be good, then, it isn’t necessary for its premises to be true, but it is necessary for the premises to be likely to be true, on balance, given the information available. One wonders: is this the same thing as premise acceptability?
The criterion of acceptability for premises goes back at least to Logical Self Defense, by Johnson & Blair. As is no doubt well-known by now, Johnson modifies his view in Manifest Rationality (p. 189 or thereabouts) to hold that acceptability for premises is not a replacement for the criterion of premise truth, but a different standard altogether, and that both acceptability and truth are required for an argument to be good. On the other hand, Trudy Govier, in A Practical Study of Argumentation, seems to subsume truth under acceptability as one way, among others, that a premise might be acceptable. At present I’m undecided as to which way, Johnson’s or Govier’s is best. Johnson’s view has the virtue of being able to explain different evaluations of the same argument by different agents. An evangelical Christian, for example, will find acceptable premises that are drawn from the Bible whereas a hardline atheist will not–even though both might agree that the premises, if true, would support the conclusion. In favor of Govier’s view, however, what is their disagreement with respect to the Biblical premises really about, if not their likely truth or falsehood? It’s something I’m still thinking through.
Whether Johnson’s view or Govier’s view (or another view altogether) of the relationship between acceptability and truth is the correct one, I do think it is important to maintain the likelihood of a premise’s being true as a criterion for a good argument in the manner I sketched above. In addition to providing a kind of “regulative ideal” for argumentative discourse, I think retaining a role for truth helps with other tasks too. In particular, retaining truth will help us explain one common sort of change that occurs over time in the evaluation of some arguments. It’s not as if once an argument is declared sound that judgment holds for all time. We can and do frequently change our minds about how to evaluate the same arguments as time progresses and our knowledge base improves. This is especially true in philosophy and in the sciences. (Philosophy of mind in particular is rife with such cases.) In many of these cases we one day discover that the premises on which we have, up to the present point, believed a conclusion have on further investigation turned out to be false. The revealed falsehood of those premises then counts against continuing to maintain the conclusion of that argument. As a result we come to reject arguments which we once accepted–even if we find no fault in the logical movement from premises to conclusion. We cannot explain that sort of rejection using premise acceptability alone. If acceptability (of the sort unconnected to truth) were all we had, it seems as though all we could do is hold the argument before us and say that they, our counterparts in the past, accepted the same premises we reject. The difference then becomes a matter for psychology or history or cultural anthropology. Particularly when it comes to arguments for conclusions like the role of neurochemistry in mental illness, or the Earth’s position in relation to the Sun, one wants to say that the salient difference isn’t a psychological or an anthropological one at all–it’s that we have more information than they did. Moreover, we want to say that if they had understood then what we understand now, they would have judged matters in a similar fashion. Retaining a place for truth in the evaluation of arguments lets us say this. I’m not sure a non-alethic conception of acceptability can do the same work, or if it can, as well.
So, my present thinking is that while it isn’t necessary for the premises of an argument to be true in order for that argument to be good, it does seem to me that it is necessary that the premises of that argument are such that they can be taken to be true, on balance, by epistemically virtuous agents operating with the best information available to them given the situation in which they have to evaluate the argument.
Now if that makes any sense, all I have to do is figure out what truth is. No problem. 😀