analysis:
the strength of
arguD1ents
798
Introduction
The relationships we examined in Part II have to do, in one way or another, with the question "How do the parts of this argument hang together?" Whether or not an argument is sound depends on whether or not the required connections between the parts of that argument are or are not present at all. A conclusion which is groundless, or an inference which is unwarranted, or a warrant which is baseless, is no conclusion, inference or warrant. The arguments embodying it will be wholly unsound.
Once the presence of the required connections has been demonstrated, how ever, a further set of questions can then be raised. These further questions have to do with the strength of the connections on which the argument depends. Granted that we have constructed an argument that is sound enough, so far as it goes, how much weight will it bear? From what has been said so far, one might get the impression that all rationally based arguments are either totally and unconditionally sound or shaky. It may seem as though the supporting material presented in the arguments must either be totally and unconditionally sufficient to support the claim, or else must entirely fail to support it.
Yet separating questions of soundness from questions of strength, as we have done here, is never more than a first simplification. Clearly, in real life, the "warrants" in our arguments do not authorize the steps from particular "grounds" to particular "claims" absolutely. In many situations we have to deal with claims, arguments, and trains of reasoning in which reliability is somewhat less than abso lute. Only within the abstract arguments of pure mathematics can our statements be linked together by relations of "absolute necessity": in all practical realms, the connections that we have to deal with are more or less qualified, and more or less
conditional. If we always waited until absolutely rigorous arguments could be con structed before we acted with reasoned confidence, we would be overtaken by events before we had occasion to act. In practice, therefore, it is often reasonable to base our conclusions on something less than absolutely perfect evidence. We then put forward our claims, not as beingformally irrefutable, but rather as being
practically strong or reliable.
Even where the connections embodied in our arguments are sound in them selves, we may often be justified in relying on those arguments only generally, not necessarily or invariably, or else on certain conditions. rather than absolutely and unconditionally.
How, then, do we think and talk about the strength of arguments, as con trasted with their soundness? In Part III, we shall be looking at four groups of issues. We shall begin by considering the qualifying phrases that are commonly employed to mark the degree and kind of certainty that attaches to different claims. Some warrants lead us to the required conclusion invariably; others do so frequently, but not with 100 percent reliability; others again do so more often than not; and so on. We accordingly present some conclusions as certainly the case, others as probably the case, others again as very possibly so. There is a whole string of adverbs and adverbial phrases that have characteristic functions in dif ferent types of practical arguments.
Secondly, we shall consider how conditions and exceptions are allowed for, in the critical presentation and discussion of claims or arguments. Some warrants lead us to the required conclusions unconditionally, others do so in all normal cases, others again do so only in exceptional circumstances. Correspondingly, we present our conclusions as presumably the case, as normally the case, or as occa sionally so. Once again, there is a whole string of terms and phrases that have characteristic uses; these indicate how the strength of our arguments depends on the circumstances and conditions of their presentation. In particular, we shall have to pay special attention to the notion of rebuttal. Very often, we present arguments which we have reason to believe are strong arguments, but we do not state explicitly all of the conditions and assumptions on which that confidence rests. Just because we have reason to believe that the conditions hold, we are enti tled to presume that the conclusion of our argument holds true. If someone is now able to demonstrate that one of our assumptions is, in fact, incorrect, she may then be able to upset our conclusions simply by pointing out that fact. (As we may say, her objection rebuts our original presumption.)
Thirdly, we shall consider the connected ideas of burden of proof and quan daries. In many practical situations, we need to take decisions in a rationally defensible way, either in the absence of sufficient information, or even in the pres ence of information that points us in conflicting directions. Where the evidence is insufficient, it may yet be irrational to suspend judgment, because urgent practical priorities may demand prompt and timely action. It is often reasonable to start by assuming one particular conclusion in the absence of evidence to the contrary. Similarly, where the evidence is conflicting, it is often reasonable to start by assuming one of two possible positions until these conflicts have been resolved.
Finally, we shall have to consider, in more general terms, how practical arguments fit into the contexts of their use, and what is meant by talking of their "relevance." The critical analysis of particular arguments (we shall find) leads us
INTRODUCTION
83
back, in the end, to an examination of the different human enterprises whose com mon purposes those arguments serve. In the last resort, we shall best understand what gives practical argumentation its rational force and power only by investi gating how it serves the purposes both of specialized professional activities like law and science and of our everyday nonprofessional lives and "common-sense" needs.
AN EXAMPLE
A patient turns up at the doctor's office with a sore throat, a headache, and a running nose that just will not clear up. The physician recognizes the symptoms as frequently associated with a bacterial infection of the upper respiratory tract, and her first inclination is to prescribe penicillin. Why? Because penicillin has frequently been found to be effective against the general run of upper respiratory infections. There is implicit in the physician's prescription the following argument:
Backing
Warrant
This patient has an upper respiratory
infection. Grounds
As repeated
clinical experience shows
Penicillin is effective against most upper respiratory infections
So, most likely
Qualifier
FIGURE 8-1
Penicillin will cure this patient's
condition. Conclusion
However, as is well known, a small proportion of patients are penicillin-sensitive; that is, if they take penicillin they are liable to suffer a severe or even fatal reaction to the drug. Even if penicillin kills the bacteria responsible for the infection, it would in that case do serious damage to the patient. The physician therefore has to ask her patient whether he has any past history of reactions to antibiotics. With this inquiry, the physician's argument takes on a further aspect:
In Figure 8-1 , the phrase most likely illustrates the kind of qualifying phrase that we use to register the strength of our conclusions. In Figure 8-2, the word presumably illustrates the conditional character of the conclusion, while the further reference to penicillin-sensitivity indicates the circumstances under which the conclusion would be rebutted.
Backing
Warrant
Penicillin will most likely cure this patient's respiratory infection Grounds
The general goals of medicine being what they are,
The most effective medicine is normally the one to prescribe
so, presumably Unless the patient is sensitive to penicillin Rebuttal FIGURE 8-2 Penicillin is the right medicine to prescribe.
9
Qualified
claiD1s
and
tentative
discoveries
"So certainly-if anything in sports can ever be certain-the 49'rs should make it to the Super Bowl this year."
"So evidently enough-granted the triviality of the overall plot-the new King Kong makes more psychological sense than the original."
"So presumably-within the limits of our bacteriological test procedures the food service equipment was the source of infections."
"So on the face of it-the market being what it is at present-municipal bonds are our best bet."
Here the strengths and limitations of the initial claims are indicated by the addition of qualifiers. These are phrases that show what kind and degree of reli ance is to be placed on the conclusions, given the arguments available to support them.
At this point, the interrogator, I, can direct fresh critical questions at the assertor, A, to bring to light the precise strength of A's supporting argument. Even after A has placed on the table all the grounds, warrants, and backing on which
she is relying, two further sets of questions can still be raised. These have to do (1) with the strengths, and (2) with the conditions of relevance of the proposed argument. They represent the strengths, and conditions of relevance, of the link between the original claim (C) on the one hand and the entire body of supporting material (G. W. and
B)
on the other.THE NATURE OF QUALIFIERS
We may begin, once again, by noting some colloquial phrases. Suppose that the questioner, I. asks the assertor, A. "How strong is your claim?" I may expand the question as follows:
"I mean-are you making this claim unconditionally and without qualifi cation? Are you saying that it's certainly and necessarily so, or only that it's prob ably, very likely, or quite possibly the case?"
In a word, every argument has a certain kind of strength. Any claim is presented with a certain strength or weakness, conditions, and/or limitations. We possess a familiar set of colloquial adverbs and adverbial phrases that are customarily used to mark these qualifications. Their function is to indicate the kind of rational strength to be attributed to C on the basis of its relationship to G. W. and
B.
Such adverbs and adverbial phrases include the following:Necessarily Certainly Presumably In all probability
So far as the evidence goes For all that we can tell Very likely Very possibly Maybe Apparently Plausibly Or so it seems
Grammatically all these phrases have one feature in common. They can all be inserted into the statement "G. so C" immediately after the word so. In this way, they yield such modally qualified statements as the following:
G. so in all probability C. G. so certainly C. G. so apparently C.
QUALIFIED CLAIMS AND TENTATIVE DISCOVERIES
87
In each case, this addition of the adverb or the adverbial phrase has the effect of indicating what sort of reliance the supporting material entitles us to place on the claim, C.
At one extreme, consider a situation in which ( 1 ) we have all the grounds we could reasonably need, (2) our warrant is unambiguous and clearly relevant, and (3) the solidity of its backing is unchallenged. In that event, it may be legit imate to advance our claim emphatically and unconditionally:
"G, so certainly c."
Alternatively we may be in a weaker position. Either the available grounds may point toward C strongly but not conclusively, or else the backing for our warrant may indicate a strong rather than a 100 percent correlation between the relevant facts and the present claim. In that event, we can appropriately state the argument in a less emphatic, more qualified manner:
"G, so probably c."
Again the warrant may be one that applies in cases like the present one only
in certain conditions. That is to say, there may be certain exceptions or disquali fications that would invalidate the application of the warrant-even though, in the present situation, they are assumed not to do so. This kind of case too can be indicated by the use of the appropriate qualification:
"G, so presumably c."
Thus, in general, we may extend our basic pattern of analysis to include "modal qualifiers" as in Figure 9-1 .
B Given our general experience in the field concerned,
W rules or principles, In accordance with the resulting
G Q c
These grounds support, in a qualified way, the claim.
The letter Q will be used to indicate the particular modal qualifier used in any argument.
THE USES OF QUALIFIERS In law
The degree of strength and certainty that any legal claim is entitled to depends on two things: the quality of the evidence available and the precise force of the relevant legal statutes or precedents. In legal arguments, modal qualifiers can be used to do either of two things: to reflect how good the evidence is or to register limitations on the relevant legal provisions.
Take an example: George has died without leaving a known will, and the question arises, "What will happen to his widow, Mary?" (See Figure 9-2.)
w governing i ntestacy Given the normal rules
So presumably
G
Q
FIGURE 9·2
The bulk of the estate will in due course pass to Mary as widow
c
Here, the modal term presumably marks the fact that the claim, C. is less than absolutely certain and makes that claim conditional.
Why should this qualifier be necessary? It can be for either of two reasons.
Either a more thorough search may yet bring a will to light. In that event, the existence of new evidence (G) will inevitably oblige us to reconsider the claim ( C). Or else, a more careful scrutiny of the relevant law may affect the reliability of the warrant ( W). Although when a husband or wife dies without leaving a will most American jurisdictions deal with the rights of the widow or widower in the same general way, there is some variation in the details. Until these finer points of difference have been looked into, the extent of Mary's right to inherit in the absence of a proven will must remain unclear. Still, as an initial position, we can reasonably assume that the available facts support our qualified conclusion, failing
either the discovery of new documents or else some quirks in the local inheritance laws, either of which may upset it. Those are the kinds of possibilities we guard against by using the modal term presumably instead of certainly.
In science
QUALIFIED CLAIMS AND TENTATIVE DISCOVERIES
89
In fields such as medicine and natural science, also, we may find ourselves similarly placed, with something less powerful by way of an argument than we would prefer. This may happen either through lack of a theoretical understanding of the events in question or through an insufficiency of factual evidence. In each case, we can register this fact by presenting our claims as provisional and there fore uncertain. On the one hand, there may be no doubt what theoretical argu ments are relevant to a particular case, and so what warrants are applicable. But the quantity of relevant factual information at our disposal may be inadequate. On the other hand, we may have a substantial body of factual data in our posses sion. But this evidence may not indicate unambiguously just what theoretical con siderations are relevant in the present case.
As an illustration of the first situation:
A physician is treating a hitherto fatal disease with a new drug. This drug has been strikingly effective in preliminary clinical trials, but the statistics from this testing program are still fragmentary. In these circumstances, the physician cannot say, "This treatment will work." Rather, all that he can conclude is "The drug has shown some signs of effectiveness, so there is a real possibility that it will work in this case."
As an illustration of the second situation:
Faced with a new phenomenon, a physicist is in some initial uncertainty. Is this a matter for the theory of gravitation, for electromagnetic theory, or what? If the factual evidence gathered so far suggests that the phenomenon is gravitational but does so only tentatively, the physicist will be limited to a conditional conclusion: "On the supposition that this is a purely gravitational effect, G, so presumably c." For just so long as the gravitational hypothesis remains acceptable, the rel evant theoretical statement will warrant the inference "G, so presumably c." But, if that initial presumption falls, the inference will-correspondingly-be undercut entirely. Once we are sure that the phenomenon is not after all a gravitational one, conclusions drawn from gravitation theory cease even to be "very possible," far less "likely."
And in everyday life
The use of such qualifying phrases is not, of course, limited to technical fields such as law, medicine, and science. On the contrary, we can hedge our claims in similar ways in all kinds of situations and all types of argumentation. Whatever other differences there are between the modes of argumentation appro priate to our different activities and enterprises, we frequently have occasion
1 . To present our claims tentatively, without staking our whole credit on them.
2. To put them into debate in an uncommitted way, merely for purposes of discussion.
3. To treat them as serious but conditional conclusions. 4. To offer them simply as a good bet.
As a result, the relevant modal qualifiers-probably, possibly, presumably, very
likely, and so on-have a part to play in all kinds of arguments.
Indeed the technical uses of these qualifiers are best understood as special cases of the more general uses. The meaning of adverbs like probably is entirely familiar to us from everyday life and provides a basis for moving on to more abstract technical terms, such as the noun probability itself. For instance, when the weather service reports that "the probability of precipitation is 30 percent," this means the same kind of thing as "The odds that it will rain or snow are 100 to 30 against." Even within the exact theoretical discussions of such sophisticated sciences as quantum mechanics, the link between probabilities and reasonable expectations-as contrasted with scientific certainties and solidly founded predic tions-remains the same as it is in ordinary life. No doubt, the scientific statement "The probability of such an event's occurring by chance is 0.08 1 " can serve as the warrant in much more exact inferences than the everyday statement "That kind of thing is very unlikely to happen," but the meaning of the two statements is very similar. This is just one of those cases in which we should try to avoid being flab bergasted by figures.
Not all forms of argumentation are as tightly organized as the proceedings in a law court or the transactions of a scientific academy. Where less hangs on the outcome of our arguments, we may speculate more freely and speak more loosely. So in casual conversation, we often put forward claims that are extremely tentative-or even only half serious-in order to see how far the available argu ments will take us:
"It sometimes seems to me as though . . . " "I sometimes think that . . . "
In doing so, of course, we run the risk of being misunderstood. So we put our more