by Scott F. Aikin and Robert B. Talisse
We’re currently finishing work on the manuscript for our forthcoming book, Why We Argue (And How We Should), so we’ve been thinking a lot recently about argumentation. We’ve been especially concerned with how arguments can go wrong. When evaluating an argument, one of the central questions to ask is whether the stated premises support the proposed conclusion. When the premises fail to provide the right kind of support for the conclusion, we often call the argument (and its form) fallacious. Fallacies are so pervasive precisely because they are cases in which it looks as if the stated premises provide propose support for a proposed conclusion, but in fact they don’t. Take, for example, a simple textbook fallacy, that of asserting the consequent:
If Bill’s a bachelor, Bill is male.
Bill is male, therefore Bill is a bachelor.
The trouble with an argument of this form is that it presents an invalid inference — the premises, if true, don’t guarantee the truth of the conclusion. So even were the premises and the conclusion true, the proposed argument fails. Note that the failure is a matter of the proposed argument’s form rather than its content. The objective of fallacy detection in the formal mode is to reveal cases in which the truth of the stated premises fail to provide the proper kind of support for the conclusion.
In the formal mode, we also can identify different degrees in which premises provide support for a conclusion. The highest degree of support that premises can provide for a conclusion is the guarantee of its truth, given the truth of the premises. Arguments that manifest that feature are called deductively valid. But note that deductive validity does not depend on the stated premises actually being true. That is, with a valid argument, the conclusion is guaranteed to be true, if the premises are true. Accordingly, an argument can be deductively valid even if every one of its stated premises is false.
Thus we require an additional metric of formal success. It would seem that an argument that is both deductively valid and has premises that in fact are all true would be bombproof. Such arguments are called deductively sound. Notice that deductive soundness encompasses deductive validity in that every sound argument is valid. A deductively sound argument is a deductively valid argument that has true premises. Since a deductively valid argument is one that guarantees the truth of its conclusion provided that its premises are in fact true, it should be no surprise that deductive soundness is often considered the gold standard for argumentative success. Every deductively sound argument actually establishes the truth of its conclusion. Who could ask for more than that?
Yet there are problems with taking deductive soundness as the benchmark of argument quality. For one thing, we don’t regularly have access to the kinds of high-quality premises that deductive soundness demands. There are no guarantees in life, and the world’s a messy place; we’re often stuck with rough-and-ready pieces of reasoning that are just good enough for our given purposes. We don’t want to make the perfect the enemy of the good in logic, so we must look for more forgiving standards of argumentative success. And so we look to the fallible, but reliable, inductive arguments — arguments where true premises yield conclusions that are more likely true than false.
But there is another problem. As it turns out, deductive soundness isn’t all it’s cracked up to be. Not even deductive soundness guarantees that an argument does what it needs to do. To see this, imagine a dispute over the question of who was the only President of the United States that served two non-consecutive terms. One disputant says: Chester A. Arthur. Another says: Grover Cleveland. An argument is needed to resolve the dispute. Now imagine our second disputant making the following argument:
Grover Cleveland was the only President of the United States to serve non-consecutive terms. Therefore, Grover Cleveland was the only President of the United States to serve non-consecutive terms.
That’s obviously an impotent argument. But notice that it’s deductively sound. The truth of its premise indeed guarantees the truth of its conclusion. And, moreover, the premise is in fact true! Yet, again, the argument is patently lacking. It looks as if the supposed benchmark for argumentative success just failed.
So there must be more to argumentative success than meeting the requirements of deductive soundness. When arguing, we are looking to address questions, resolve disagreements, further inquiry, and flesh things out. The trouble with the Grover Cleveland argument above is that it begs the question. The conclusion is identical to the premise, so the argument attempts to resolve a disputed question by merely asserting one disputant’s answer. But disputes arise between people who disagree, and arguments must be designed to address disputes in ways that could provide a resolution to disagreements. In order to resolve a disagreement, the arguments provided must actually address the disputants and attempt to provide them with reasons to come to agree. Accordingly, even a formally unimpeachable argument can fail if it is unable to fulfill the social role of argumentation. To put the matter in a nutshell: in order to be successful, arguments must be sufficiently dialectical.
What is it for an argument to be sufficiently dialectical? Here are two rough desiderata. First, an argument must be composed not merely of reasons that support its conclusion, but of reasons that its target audience can recognize as reasons. Accordingly, a flat-footed appeal to the authority of the Pope in a dispute among Catholics and non-Catholics about the permissibility of stem-cell research is a dialectical failure. Second, an argument must address the most pressing concerns and doubts that prevail among the target audience. That is, in order to attempt resolve a disagreement, we must not only assess the reasons for one of the sides, we must assess the reasons for both sides. So a dialectically proper argument presents not merely a case for one’s preferred view; it must also take into account the going criticisms and objections to one’s conclusion. Those arguments that fail to satisfy these desiderata beg the question.
Now, we can see that this dialectical requirement for success in argument brings with it additional responsibilities for arguers. In order to argue well, one must be in a good position to know or have compelling reasons to believe one’s conclusion true. But one also must know something about those with whom one disagrees. One needs to know something about their reasons, and why they might (reasonably, perhaps) reject what may seem so clearly true. Winning at argument, then, isn’t what many people think it is. To win at argument is not to silence one’s opposition or prove them silly or foolish. Such ends are served better by rhetoric than by reason. Winning at argument rather requires something on the order of coming to see, and perhaps even in some ways appreciate, the rationale of one’s opponents. That, of course, is just the beginning to a well-run argument. But notice that this means that argument does not put an end to discussion, but rather helps to continue one.