INFORMAL LOGIC XVI.3, Fall 1994 The Deductive/Inductive Distinction* GEORGE BOWLES Key Words: Argument, deductive, inductive, relevance. Abstract: In this paper I examine five distinc- tions between deductive and inductive arguments, concluding that the best of the five defines a de- ductive argument as one in which conclusive favorable relevance to its conclusion is attrib- uted to its premises, and an inductive argument as any argument that is not deductive. This dis- tinction, unlike its rivals, is both exclusive and exhaustive; permits both good and bad arguments of each kind; and is both useful and needed in evaluating at least some arguments. In past years a debate flourished con- cerning the content and value of the dis- tinction between deductive and inductive arguments. Five ways of distinguishing deductive from inductive arguments fig- ured in that debate. In this paper I shall examine critically each of them, conclud- ing that one is, with minor revision, defen- sible against the many criticisms leveled at it. My procedure with regard to each of the five distinctions will be first to state the distinction, along with any available clarification and defense, and then to present objections to the distinction, along with any appropriate replies. For the sake of fairness and thoroughness, I shall include an objection or reply either if it has appeared in the literature or if it seems reasonable. In my discussion I shall assume the fol- lowing: where 'p' and 'q' stand for proposi- tions, 'p' is either relevant or irrelevant to 'q' . For example, 'Most papers are too long' is relevant to 'This paper is too long', whereas 'At least one cat has kittens' is not. If 'p' is relevant to 'q', then it is so either favorably or unfavorably. For example, 'Most papers are too long' is favorably, whereas 'Few pa- pers are too long' is unfavorably, relevant to 'This paper is too long'. If 'p' is favorably or unfavorably relevant to 'q', then it is so ei- ther conclusively or inconclusively. For in- stance, 'All papers are too long' is conclu- sively favorably relevant to 'This paper is too long', because the first of these propositions entails the second; whereas 'Most papers are too long' is inconclusively favorably relevant to 'This paper is too long', since the first of these propositions makes the second prob- able. And finally, irrelevance as well as any degree of favorable or unfavorable relevance may be actual, attributed, or both. For exam- ple, in The fact that most papers are too long makes it certain that this paper is too long. conclusive favorable relevance to the con- clusion, 'This paper is too long', is attrib- uted, but does not actually belong, to the premise, 'Most papers are too long'. But In The fact that most papers are too long makes it probable that this paper is too long. inconclusive favorable relevance to the same conclusion not only is attributed, but actually belongs. to the same premise. I shall also assume that the terms 'de- ductive' and 'inductive' are alike in that if one were evaluative, so would be the other. Now, the expressions 'deductive and good', 'deductive and bad', 'inductive and good', 160 George Bowles and 'inductive and bad' are neither pleo- nasms nor oxymorons, as they would be if 'deductive' or 'inductive' were evaluative. Therefore, neither term is evaluative. This means that deductive and inductive argu- ments can both be either good or bad. I. Distinction 1 The first distinction I shall discuss is the almost universally discarded traditional one saying that deductive arguments are those whose premises are general and whose con- clusion is particular, whereas inductive ar- guments are those whose premises are par- ticular and whose conclusion is general. 1 To this distinction I shall consider two objections. Objection 1. This distinction is unaccept- able because there are arguments that are deductive or inductive without conforming to its definitions of 'deductive argument' or 'inductive argument'. For instance, the argument Premise 1. All animals are mortals. Premise 2. All humans are animals. Conclusion. All humans are mortals. is deductive although, contrary to this dis- tinction, its conclusion is general rather than particular.2 And the argument Premise I. Amy is taller than Beatrice. Premise 2. Beatrice is taller than Carlene. Conclusion. Amy is taller than Carlene. is also deductive although, again contrary to this distinction, its premises are particular. 3 Reply. In logic, 'deductive argument' and 'inductive argument' are technical terms, used differently by different authors holding different theories about what dis- tinguishes deductive from inductive argu- ments. Therefore, since there are no pretheoretical intuitions about the correct use of technical terms, there are no pretheoretical intuitions regarding the proper use of 'deductive argument' and 'in- ductive argument' . So, whatever intuitions govern the use of the terms 'deductive ar- gument' and 'inductive argument' are in- formed by one or more theories. To appeal, then, to such intuitions in arguing against some distinction between deductive and inductive arguments would be to beg the question against that distinction by assum- ing the truth of one or more rival theories (namely, those shaping the intuitions ap- pealed to). For example, the present objec- tion begs the question against Distinction I by assuming a rival distinction (probably Distinction 2 or 3). So, in reply to this ob- jection, an advocate of Distinction I might simply decline to assent that the alleged counterexamples are deductive, and that would be that. To avoid such fruitless col- lisions of theories, I shall henceforth avoid appealing to intuitive judgments that given arguments are deductive or inductive. Objection 2. The deductive/inductive distinction is so related to the evaluation of arguments that at least three criteria4 of the adequacy of any particular way of dis- tinguishing deductive from inductive argu- ment are concerned, directly or in directly, with argument evaluation.s The first crite- rion is that the distinction should, if possi- ble, be both exclusive and exhaustive (i.e., it should be so drawn that [a] no argument can be both deductive and inductive, and [b] every argument is either deductive or inductive).6 On the plausible assumption that the deductive/inductive distinction is so tied to the evaluation of arguments that a deductive argument is evaluated by ref- erence to a different standard than an in- ductive argument, we can explain why the distinction should, if possible, be both ex- clusive and exhaustive. It should be exclu- sive in order to prevent the same argument's being evaluated by reference to more than one standard at once. And it should be ex- haustive in the interests of economy: it is undesirable needlessly to multiply the standards by reference to which arguments are evaluated. 7 The second criterion is that any adequate distinction between deductive and induc- tive arguments must at least permit both good and bad arguments of each kind. 8 For if the deductive/inductive distinction did not at least permit both good and bad argu- ments of each kind, we would be commit- ted to one or more of the following unac- ceptable consequences: (a) No deductive argument can be good. (b) No deductive argument can be bad. (c) No inductive argument can be good. (d) No inductive argument can be bad. The third criterion is that the distinction should, if possible, facilitate, or even be necessary to, the evaluation of arguments. 9 Distinction I fails to satisfy the first cri- terion, since it is not exhaustive (see the arguments used as counterexamples in the preceding objection).l0 It also fails to sat- isfy the third criterion, since the fact that an argument's premises are general and its con- clusion particular, or that its premises are particular and its conclusion general, is nei- ther necessary nor helpful in ascertaining whether it is good or bad. For these two rea- sons, Distinction I should be discarded. ll. Distinction 2 The second distinction that I shall con- sider divides the class of arguments into deductive and inductive arguments accord- ing to their families. Categorical syllogisms (the good ones as well as the bad) consti- tute one family of arguments, analogical arguments (again, the good as well as the bad) constitute another, and so on. An ar- gument is deductive if it belongs to one group of families and inductive if it belongs to the other. Categorical syllogisms, hypo- thetical syllogisms, disjunctive syllogisms, quantificational arguments, mathematical arguments, etc. belong to one group; if an argument belongs to a family in this first group, it is deductive. Generalizations from particular instances, analogical arguments, causal arguments, good-reasons arguments, etc. belong to the other group; if an argu- ment belongs to a family in this second group, it is inductive. II If it be asked why the families should be grouped as they are (with, for example, Deductive/Inductive Distinction 161 categorical syllogisms grouped with math- ematical arguments rather than with causal arguments), and why belonging to a fam- ily in the first group would make an argu- ment deductive whereas belonging to a family in the second would make an argu- ment inductive, the answer is as follows. If a family of arguments is such that formal features reveal whether its members' premises are conclusively favorably rel- evant to their conclusions, then that family belongs to the first group, and its members are deductive; if it is such that formal fea- tures reveal whether its members' premises are inconclusively favorably relevant to their conclusions, then it belongs to the sec- ond group, and its members are inductive. l2 This distinction is subject to the follow- ing two objections. Objection 1. The distinction will not be exhaustive unless every argument belongs to at least one family and every family be- longs to at least one of the two groups. Now, it is not obvious that every argument be- longs to at least one such family, and no proof or evidence is given. Moreover, given the rationale for assigning a family of ar- guments to one rather than the other of the two groups, it seems that not every family can be assigned to one group or the other, so that the distinction cannot be exhaus- tive. For that rationale depends on there being formal features of a family's argu- ments that reveal whether its members' premises are conclusively or inconclusively favorably relevant to their conclusions. But formal features alone do not always deter- mine whether the premises of arguments of a given family are inconclusively favorably relevant to their conclusions. 13 For instance, whether the premises of ana- logical arguments are inconclusively favorably relevant to their conclusions de- pends, in part, on the number of relevant disanalogies among the things mentioned only in the premises and those mentioned only in the conclusion and also on whether the analogy reasoned from is relevant to the analogy reasoned to; 14 but these features 162 George Bowles seem not to be fonnal. Therefore, the ra- tionale given for assigning a family of ar- guments to one group rather than the other seems to be inconsistent with the distinc- tion's being exhaustive. Objection 2. The distinction is not ex- clusive, because the rationale for assign- ing a family of arguments to one rather than the other of the two groups will sometimes require that the same family be assigned to both groups. Consider, for example, the family of arguments having the fonn of hypothetical syllogism. Formal features show that their premises are conclusively favorably relevant to their conclusions. According to the rationale, a family of ar- guments belongs to the first (the deductive) group iffonnal features of those arguments show whether their premises are conclu- sively favorably relevant to their conclu- sions. So, this family belongs to the first group, making arguments in the hypotheti- cal syllogism family deductive. Now, the premises of no argument can be at once both conclusively and inconclusively favorably relevant to its conclusion. By the same reasoning. then, the same fonnal fea- tures of arguments in the hypothetical syl- logism family that show that their premises are conclusively favorably relevant to their conclusions also show (a) that their premises are not inconclusively favorably relevant to their conclusion, and a fortiori (b) whether their premises are inconclu- sively favorably relevant to their conclu- sions. So, since the rationale says that a family of arguments belongs to the induc- tive group if fonnal features of those argu- ments reveal whether their premises are inconclusively favorably relevant to their conclusions, this family belongs to the sec- ond group too, making arguments having the fonn of hypothetical syllogism induc- tive as well as deductive. Similarly, if there are formal features of other arguments that show that their premises are inconclusively favorably relevant to their conclusions, the same features show whether their premises are conclusively favorably relevant to their conclusions; and so, again. the families to which the arguments belong would be in both groups at once, and hence those argu- ments would be both deductive and induc- tive. Suppose we attempt to avoid this objec- tion by revising the rationale to say that a fam- ily of arguments belongs to one of the two groups if formal features of those arguments show (not whether but) that their premises are conclusively or inconclusively favorably relevant to their conclusions. In that case, the distinction between deductive and inductive arguments would not be exhaustive. There are two reasons for this. (I) As observed above concerning analogical arguments, for- mal features alone sometimes do not deter- mine whether the premises of arguments of a given family are inconclusively favorably relevant to their conclusions; in such cases, then, formal features do not determine that the premises are inconclusively favorably rel- evant to their conclusions. So, according to the revised rationale, arguments belonging to such families would not be inductive. Nor would they be deductive if (as in the case of at least some analogical arguments) formal features alone did not detennine that their premises are conclusively favorably relevant to their conclusions. Consequently, such ar- guments would be neither inductive nor de- ductive. (2) Fonnal features of some argu- ments show that their premises are neither conclusively nor inconclusively favorably relevant to their conclusions, and so they would be neither deductive nor inductive. For instance, the argument ex pressed in the text Descriptive metaphysics has had a long and complicated history, and consequently there are no new truths to be discovered in it. IS has the form Premise. x has had a long and com- plicated history. Conclusion. There are no new truths to be discovered in x. which determines that the argument's premise is neither conclusively nor incon- clusively favorably relevant to its conclu- sion, so that the argument would be nei- ther deductive nor inductive. 16 Moreover, the revised rationale would not permit the existence of both good and bad deductive arguments. For according to it, a family of arguments would belong to the deductive group only if formal features of those arguments showed that their premises are conclusively favorably rel- evant to their conclusions. Every argument that belonged to a family in the deductive group, then, would be such that formal fea- tures of that argument show that its premises are conclusively favorably rel- evant to its conclusion. That means that every deductive argument would be good. 17 III. Distinction 3 Having rejected the first two distinctions between deductive and inductive argu- ments, we proceed to the third, which dis- tinguishes the two kinds of arguments ac- cording to the degree of favorable rel- evance that the premises actually have to the conclusion. It says that an argument is deductive if and only if its premises actu- ally are conclusively favorably relevant to its conclusion, and it is inductive if and only if its premises actually are inconclusively favorably relevant to its conclusion. IS To this distinction there are three objec- tions. Objection 1. The distinction is not ex- haustive. For there are some arguments whose premises are neither conclusively nor inconclusively favorably relevant to their conclusions (viz. arguments whose premises are either irrelevant or unfavorably relevant to their conclusions), so that, ac- cording to this distinction, they would be neither deductive nor inductive. For in- stance, the premise of the argument Premise. Descriptive metaphysics has had a long and compli- cated history. Conclusion. There are no new truths to be discovered in descriptive metaphysics. Deductive/Inductive Distinction 163 is neither conclusively nor inconclusively favorably relevant to its conclusion. Hence, according to this distinction, this argument would be neither deductive nor inductive. 19 Objection 2. The distinction does not permit both good and bad arguments of both kinds. For if a deductive argument is one in which the premises are conclusively favorably relevant to the conclusion, then there can be no distinction between good and bad (valid and invalid) deductive ar- guments. And if an inductive argument is one in which the premIses are inconclu- sively favorably relevant to the conclusion, then although there can be better and worse inductive arguments (distinguished by the degree of inconclusi ve favorable relevance of the premises to the conclusion), there can be no distinction between good and bad inductive arguments. 20 Objection 3. In the case of arguments with unexpressed premises, ascertaining whether the argument is deductive or in- ductive will at least sometimes be a circu- lar procedure. Suppose someone says, "The fact that Fido is a dog proves that he has fleas", so that his argument's conclusion is 'Fido has fleas' and its one explicit premise is 'Fido is a dog'. Is this argument deduc- tive or inductive? Under the present dis- tinction, we cannot say until we have com- pleted the argument by supplying its unexpressed premise; for until all of its premises have been made explicit, we can- not ascertain whether its premises jointly are conclusively or inconclusively favorably relevant to its conclusion. So, what is that unexpressed premise? It might be 'All dogs have fleas' or 'Almost all dogs have fleas' or 'Most dogs have fleas', and so on. How do we decide which to supply? Although several considerations (e.g., background beliefs, the Principle of Char- ity) might bear on this question, this much surely is relevant we should supply' All dogs have fleas' if the argument is deductive, but one of the other candidates if it is inductive. This brings us back to our initial question: is the argument deductive, or is it inductive?21 164 George Bowles Iv. Distinction 4 Having rejected the first three distinc- tions, we proceed to one that has been of- ten criticized but that, with some revision, is better than its competitors. It deals with attributed, rather than actual, favorable rel- evance and says that an argument is deduc- tive if and only if conclusive favorable rel- evance to the conclusion is attributed to the premises, and an argument is inductive if and only if inconclusive favorable rel- evance to the conclusion is attributed to the premises.22 It is deliberate that the agent of attribu- tion (i.e., who or what attributes to the premises one degree or another of favorable relevance to the conclusion) is here unspecified. Usually, the most con- spicuous such agent would be the arguer (i.e., the person or other rational being who offers the argument). But it need not always be so. Someone may have an argument in mind without accepting or offering it him- self, in which case he would not himself attribute to the premises any degree of favorable relevance to the conclusion. Moreover, an argument might be expressed independently of any rational agency - e.g., an improbable natural arrangement of colored pebbles on a beach might read The facts that all dogs have fleas and that Fido is a dog make it certain that Fido has fleas. thereby expressing the argument Premise 1. All dogs have fleas. Premise 2. Fido is a dog. Conclusion. Fido has fleas. so that there would be no arguer to attribute any degree of favorable relevance to the premises. In such admittedly atypical but nevertheless possible cases as these, al- though no arguer attributes to the premises any degree of favorable relevance to the conclusion, something else does. In the first case - the case of someone having an ar- gument in mind without endorsing it - the attribution is done not by the argument's uncommitted contemplator but by a propo- sition he has in mind (but does not accept) that sums up the whole argument (viz., a proposition that says that the premises are, to some degree, favorably relevant to the conclusion). For instance, if someone is considering, but not yet accepting or offer- ing, an argument whose premises are 'All dogs have fleas' and 'Fido is a dog' and whose conclusion is 'Fido has fleas', he does not attribute to the premises any de- gree of favorable relevance to the conclu- sion, since he does not accept the argument; but in order to contemplate the argument, he must at least have in mind a proposition that so relates some of its constituents (its premises) to another (its conclusion) that it attributes to the former some degree of favorable relevance to the latter; otherwise, what he contemplates would not be an ar- gument.23 Similarly, in the second case the case of an argument's being expressed independently of any rational agency the attribution is done not by any rational be- ing but again by the proposition that sums up the whole argument and is expressed by the entire text. For instance, the previously mentioned improbably but naturally ar- ranged colored pebbles on a beach express not only the propositions that are the argu- ment's premises and conclusion but also a further proposition saying that 'All dogs have fleas' and 'Fido is a dog' are jointly conclusively favorably relevant to 'Fido has fleas'. Even in cases where an arguer is involved, he attributes to the premises some degree of favorable relevance to the conclusion only insofar as he believes or expresses a proposition that does so. In all arguments, then, one agent of attribution is such a proposition; in some, another agent is the arguer. In favor of Distinction 4 it has been urged that it permits good and bad argu- ments of both kinds. 24 For among deduc- tive arguments (that is, those whose premises are said to be conclusi vely favorably relevant to their conclusions), the good ones are those whose premises really are conclusively favorably relevant to their conclusions, whereas the bad are the re- mainder. Similarly, among inductive argu- ments (that is, those whose premises are said to be inconclusively favorably relevant to their conclusions), the good ones are those whose premises really are inconclu- sively favorably relevant to their conclu- sions, whereas the bad are the remainder. 25 To this distinction have been raised four- teen objections of unequal merit. I wish to show that Distinction 4 (when suitably amended) is defensible against all of these objections, so that there is no impediment to its acceptance. Consequently, I shall ad- dress all of them. Objection 1. Arguers often claim that their premises are favorably relevant to their conclusion without explicitly attrib- uting to the premises one degree rather than another of favorable relevance to the con- clusion. In such cases, under Distinction 4, their arguments would be neither deduc- tive nor inductive. 26 Reply. The objection presupposes that an arguer who attributes to his premise some degree, or range of degrees, of favorable rel- evance to his conclusion must do so explic- itly (e.g., by means of expressions like 'proves', 'suggests', 'certainly', and 'prob- ably'). But this presupposition is false: an arguer may not make explicit all that he thinks concerning the relation between his premise and his conclusion: he may attribute to his premise some degree, or range of degrees, of favorable relevance to his conclusion with- out communicating that attribution.27 There- fore, contrary to the objection, even if an ar- guer does not explicitly claim that his premises are to some degree, or range of de- grees, favorably relevant to his conclusion, it does not follow that, on Distinction 4, his ar- gument is neither deductive nor inductive. Nor does it follow that we can have no knowledge or justified belief about the con- tent of such a tacit attribution. 28 For we may have pertinent knowledge about the argu- er's reasoning habits, about the reasoning habits of a class of reasoners to which he belongs, or about the reasoning habits of Deductive/Inductive Distinction 165 people generally.29 Consequently, even though an arguer says nothing about the degree, or range of degrees, of favorable relevance of his premise to his conclusion, Distinction 4 can still help to provide good reasons for thinking his argument deduc- tive or inductive. 30 The next two objections have a common reply. Objection 2. Distinction 4 is not exhaus- tive, because there may be occasions when an arguer attributes to his premises only favorable relevance, rather than any par- ticular degree of favorable relevance, to his conclusion. 31 This is possible because, al- though it is true that if his premises are favorably relevant to his conclusion, they must be so either conclusively or inconclu- sively, it is not true that if the arguer at- tributes favorable relevance to his premise, he must attribute either conclusive or in- conclusive favorable relevance to them. Objection 3. The distinction is not ex- haustive, because, as already noted, an ar- gument might be expressed independently of any rational agency. In such an instance, there might be neither explicit nor implicit attribution to the premises of conclusive or inconclusive favorable relevance to the conclusion. For example, suppose that an improbable natural arrangement of colored pebbles produces on a beach the following text: The facts that all dogs have fleas and that Fido is a dog make it at least prob- able that Fido has fleas. This would express an argument and at- tribute to the premises favorable rel- evance to the conclusion; but it would not, in any manner, attribute either con- clusive or inconclusive favorable rel- evance. Hence, according to this distinc- tion, the argument would be neither de- ductive nor inductive. Reply. To avoid the difficulties raised by Objections 2 and 3, Distinction 4's defini- tion of 'inductive argument' might be re- vised to this: "an argument is inductive if and only if it is not deductive".J2 An argu- 166 George Bowles ment, accordingly, is deductive if conclu- sive favorable relevance to the conclusion is attributed to the premises; otherwise, it is inductive. The argument mentioned in Objection 3, then, would be inductive, since in that instance conclusive favorable rel- evance to the conclusion is not attributed to the premises. 33 Distinction 4 would thus be exhaustive. 34 In accordance with this revision of Dis- tinction 4, the difference between good and bad arguments of both kinds might be re- vised thus: an argument is good if and only if the attributed and actual degrees of rel- evance of its premises to its conclusion agree. So, for instance, since in a deduc- tive argument conclusive favorable rel- evance to the conclusion is attributed to the premises, such an argument will be good if and only if its premises actually are con- clusively favorably relevant to its conclu- sion. Similarly, an inductive argument in which some degree of inconclusive favorable relevance to the conclusion is attributed to the premises will be good if and only if its premises actually are to that same degree inconclusively favorably rel- evant to its conclusion. Likewise, an induc- tive argument in which favorable relevance alone (without regard to whether it is con- clusive or inconclusive) to the conclusion is attributed to the premises will be good if and only if the premises actually are favorably relevant to the conclusion. 35 It might be objected that, aside from its ability to answer Objections 2 and 3, this amendment of Distinction 4 is arbitrary. For consider the following three classes of ar- guments. Class 1 contains arguments in which conclusive favorable relevance to the conclusion is attributed to the premises. According to Distinction 4, arguments in Class I are deductive. Class 2 contains ar- guments in which inconclusive favorable relevance to the conclusion is attributed to the premises. According to Distinction 4, arguments in Class 2 are inductive. Class 3 contains arguments in which neither con- clusive nor inconclusive but only an un- qualified favorable relevance to the con- clusion is attributed to the premises. Ac- cording to the amendment of Distinction 4 proposed by the present reply, arguments in Class 3 are inductive because they are not deductive. But why should we not say instead that they are deductive because they are not inductive? Why should we group arguments in Class 3 together with argu- ments in Class 2 rather than with arguments in Class I? The reasons why it is not arbitrary for the amendment to Distinction 4 to classify the arguments in both Class 2 and Class 3 as inductive are these. An assumption com- mon to all or most of those who distinguish deductive from inductive arguments is that the conditions necessary for an argument to be deductive, or for a deductive argu- ment to be good, are (in some sense) more stringent and less easily satisfied than those for an argument to be inductive, or for an inductive argument to be good. Now, in order for the deductive arguments in Class I to be good, their premises must be con- clusively favorably relevant to their con- clusions; in order for the inductive argu- ments in Class 2 to be good, their premises must be inconclusively favorably relevant to their conclusions; and in order for the arguments in Class 3 to be good, their premises must be (conclusively or incon- clusively - it doesn't matter) favorably relevant to their conclusions. So, since there are many, more easily attained degrees of favorable relevance short of the highest, the conditions necessary for an argument in Class 2 to be good are more easily satis- fied than are those necessary for an argu- ment in Class 1 to be good; and those nec- essary for an argument in Class 3 to be good are still more easily satisfied. The amend- ment to Distinction 4, then, accords with this common assumption in classifying the arguments in Class 3 with those in Class 2 rather than with those in Class I. It is there- fore not arbitrary. Objection 4. Distinction 4 is not exclu- sive. For different arguers can give the same argument,36 one attributing conclu- sive, and the other inconclusive, favorable relevance to the premises; so that the same argument can be both deductive and induc- tive. 37 Reply. This objection presupposes that arguments are individuated only by their premises and conclusions (Le., that argu- ments are different if and only if they have different premises, different conclusions, or both). For instance, these two texts The fact that most papers are too long makes it certain that this paper is too long. The fact that most papers are too long makes it probable that this paper is too long. would express a single argument, since the premises are the same and the conclusions are the same. But if this way of individuating arguments were correct, then, since the same argument would be ex- pressed in both of the above texts, if the argument expressed in the first text were bad with respect to the relation between its premise and its conclusion, so would be the argument expressed in the second; and if the argument expressed in the second text were good in the same respect, so would be the argument expressed in the first. Nei- ther of these consequents seems true: the first text seems to express an argument that is bad with respect to the relation between its premise and its conclusion, whereas the second seems to express an argument that is good in the same respect. Because the same argument cannot be both good and bad in the same respect, then, arguments must be individuated by more thanjust their premises and conclusions. 38 Objection 5. Distinction 4 depends on our ascertaining what degree of favorable relevance to a conclusion is attributed to premises. But that is alien to logic, which is concerned only with the degree of favorable relevance that actually obtains be tween premises and conclusion. Hence, logic cannot properly accommodate any reference to attributed favorable relevance. Deductive/Inductive Distinction 167 And so, in logic neither 'deductive argu- ment' nor 'inductive argument' can be properly defined even partly in terms of attributed favorable relevance. 39 Reply. Although the claim that logic is not concerned with attributed favorable relevance might be true of formal logic, it is not true of logic in general. For one of the things that such logic does is to describe how, through illatives like 'therefore' and 'because', we indicate our own, or detect others' , arguments; and illatives are expres- sions of attributed favorable relevance. If logic can be properly concerned with attributed favorable relevance in order to indicate or detect arguments, it can also properly be concerned with attributed favorable relevance in order to ascertain whether arguments are deductive or induc- tive. Objection 6. The concept of conclusive favorable relevance presupposes two dis- tinctions: (a) between logical and empiri- cal connections and (b) between premises' being true and their being properly related to a conclusion. If, then, Distinction 4 is right in saying that deductive arguments are those in which conclusive favorable relevance to the conclusion is attributed to the premises, whereas inductive argu- ments are those in which inconclusive, as distinguished from conclusive, favorable relevance to the conclusion is attributed to the premises, an argument will not be deductive or inductive unless the arguer grasps both of the above dis- tinctions. Now, these distinctions, having been drawn by philosophers and logi- cians, are foreign to most arguers. (Al- though everyday arguers sometimes em- ploy expressions like 'must' and 'shows conclusively', that does not prove that they understand the above distinctions.) Therefore, many arguments are neither deductive nor inductive. 40 Reply. The objection errs when it asserts that an argument will not be deductive or inductive unless the arguer grasps the dis- tinctions between logical and empirical 168 George Bowles connections and between premises' being true and their being properly related to a conclusion. For the purposes of Distinction 4, it is sufficient that an arguer understand the difference between conclusive and in- conclusive favorable relevance to the con- clusion being attributed to the premises; it is not necessary that he also understand that conclusive favorable relevance is a kind of logical rather than empirical connection, or that he understand that the premises need not be true in order to be conclusively favorably relevant to the conclusion. Simi- larly, even though a cold sore may be a vi- ral, rather than a bacterial, infection, I may say that someone has a cold sore although I do not understand the difference between viral and bacterial infections. Moreover, suppose the objection were strengthened to say that many arguers do not understand even what conclusive favorable relevance is, or what inconclu- sive favorable relevance is, or how the two differ from each other. In light of the amendment to Distinction 4 suggested in the reply to Objections 2 and 3, we could respond that all that is necessary to make someone's argument either deductive or inductive is that he grasp the concept of favorable relevance: if he attributes to his premises nothing more than favorable rel- evance to his conclusion, that suffices to make his argument inductive. If someone did not understand what favorable rel- evance is, he could not be an arguer in the first place.41 Therefore, even if the objec- tion were strengthened in the way de- scribed, the distinction between deductive and inductive arguments would still be ex- haustive. Objection 7. Even if Distinction 4 satis- fies the criterion that it permit both good and bad deductive and inductive arguments, it is neither needed nor helpful in evaluat- ing an argument. For the logical evaluation of an argument involves only actual and not also attributed favorable relevance: it involves ascertaining only whether certain logical relations really hold between the argument's premises and its conclusion. For instance, to say that an argument is valid is not to say that there is any special relation (like coincidence or inclusion) between the degrees of attributed and actual favorable relevance of its premises to its conclusion but only that its premises entail its conclu- sion; and to say that an argument is invalid is not to say that any such special relation fails to obtain but only that its premises do not entail its conclusion.42 Consequently, the logical evaluation of an argument is independent of the degree of attributed favorable relevance of its premises to its conclusion. Such favorable relevance may tell us something about the arguer's state of mind, and it may therefore pertain to the evaluation of the arguer; but it has nothing to do with the evaluation of his argument. 43 Therefore, this way of distinguishing de- ductive from inductive arguments is use- less in evaluating an argument. Reply. I have argued elsewhere that, con- trary to the objection, an argument is good with respect to the relation between its premises and conclusion if and only if the actual degree of relevance of its premises to its conclusion either coincides with or falls entirely within the limits of the attrib- uted. 44 I concede that if we know the ac- tual degree of relevance of its premises to its conclusion, knowing in addition whether an argument is deductive or inductive in the sense defined by the present distinc- tion is not always necessary for evaluating the argument. For even without knowing whether it is deductive or inductive, we can know that an argument whose premises are either irrelevant or unfavorably relevant to its conclusion is not good. Despite this con- cession, however, the objection's conclu- sion is false. For in other cases we cannot correctly evaluate an argument without knowing whether it is deductive or induc- tive. For example, knowing that an argu- ment's premises are inconclusively favorably relevant to its conclusion does not tell us whether the argument is good or bad with respect to the relation between its premises and conclusion unless we know at least whether or not conclusive favorable relevance to the conclusion is attributed to its premises, and that is the same thing as knowing whether the argument is deduc- tive or inductive, according to Distinction 4. So, it is not true that this way of distin- guishing deductive from inductive argu- ments is useless in evaluating arguments. Objection 8. If deductive arguments were distinguished from inductive ones on the basis of the degree of attributed favorable relevance of premises to conclu- sion, it would follow both that a timid reasoner offering a valid categorical syllo- gism but claiming that his premises make his conclusion only probable would be of- fering an argument that is inductive and so ought to be evaluated by the inductive standards that would be appropriate for such arguments as generalizations or ana- logical arguments, and also that a bold reasoner offering a strong generalization from particular instances but claiming that his premises make his conclusion certain would be offering an argument that is de- ductive and so ought to be evaluated by the deductive standards that would be appro- priate for such arguments as categorical syllogisms and mathematical arguments. The absurdity of these consequences shows that Distinction 4 is faulty.45 Reply 1. The objection begs the question by assuming a rival position - namely, that an argument is deductive or inductive (or- what would be the same thing for the objec- tors - ought to be evaluated by deductive or inductive standards) according either to the family to which it belongs (e.g., a categori- cal syllogism belongs to a family that makes it deductive, whereas a generalization from particular instances belongs to a family that makes it inductive) or to whether the premises are actually conclusively or inconclusively favorably relevant to the conclusion. Reply 2. Distinction 4 does not have the absurd consequences described by the ob- jection. For having ascertained, on the ba- sis of the degree of favorable relevance to Deductive/Inductive Distinction 169 the conclusion attributed to the premises, that an argument is deductive or inductive, one could proceed to evaluate the argument by ascertaining whether the premises ac- tually have the degree of favorable rel- evance to the conclusion that has been at- tributed to them. One would ascertain that by means of whatever formal or informal criteria pertain to the argument at hand. For instance, if it is a categorical syllogism, one would inquire whether its middle term is distributed, and so on; if it is an analogical argument, one would inquire whether the attributes mentioned only in the premises are relevant to those mentioned in the con- clusion, and so on. Having by one means or another ascertained the premises' actual degree of favorable relevance to the con- clusion, one could then compare the actual with the attributed degree of favorable rel- evance of the premises to the conclusion and evaluate the argument accordingly.46 Reply 3. Assuming (in accordance with the reply to Objection 4 above) that arguments are individuated not only by their premises and conclusions but also by the degree of favorable relevance to their conclusions at- tributed to their premises, to comply with the objection's implicit recommendation to evaluate someone's argument as if he attrib- uted to his premises a different degree of favorable relevance than he did would be to evaluate a different argument.47 Objection 9. It is wrong to distinguish deductive from inductive arguments ac- cording to degrees of attributed favorable relevance, as does Distinction 4, because doing so improperly permits the conversion of at least some inductive arguments into deductive ones. Such conversion can be accomplished in the following manner. Begin with a text possessing a form like x. Hence, it is likely that y. which would express what Distinction 4 would call an inductive argument, because the expression 'Hence, it is likely that' shows that inconclusive favorable rel- evance to the conclusion, y, is attributed to the premise, x. Now, let 'it is likely that' 170 George Bowles become part of the conclusion, so that the argument's conclusion is no longer simply 'y' but 'it is likely that y'. Finally, let 'z' stand for the new conclusion, so that the text now has the form x. Hence, z. This would express a deductive argument,48 and so the conversion of an inductive into a deductive argument is complete.49 Reply. It is sometimes correct, as the ob- jection assumes, to incorporate into an argu- ment's conclusion an expression like 'it is likely that' in the text above; for some con- clusions are propositions that attribute some degree of probability to something. But it is incorrect when the expression is meant to express, or to help to express, the degree of favorable relevance to the conclusion attrib- uted to the premises; for in that case the ex- pression does not constitute part of the con- clusion but instead helps to indicate how the premise is supposed to be related to the con- clusion. Whether an expression like 'it is likely that' constitutes part of the conclusion or attributes to the premises some degree of favorable relevance to the conclusion, and hence whether its incorporation into the con- clusion is correct or incorrect, depends on the argument at hand. When such incorporation is correct, the expression does not initially make the argument inductive according to Distinction 4, so that no conversion from inductive to deductive is possible; and when it is incorrect, it is not proper to im- port the expression into the conclusion, so that no such conversion is permissible. Therefore, the objection errs when it claims that Distinction 4 permits the conversion of at least some inductive arguments into deductive ones. 50 Objection 10. This distinction has the counterintuitive consequence that there will be inductive arguments that are deductively valid and deductive arguments that are in- ductively strong. 51 Reply. The counterintuitive character of the consequence arises only from the ver- bal affinity between 'deductively valid' and 'deductive' and between 'inductively strong' and 'inductive'. For if we remove these words but retain their meanings, the counterintuitive character of the conse- quence vanishes, since it now says that there are arguments to whose premises is not attributed conclusive favorable rel- evance to their conclusions but whose premises are actually conclusively favorably relevant to their conclusions, and that there are other arguments to whose premises is attributed conclusive favorable relevance to their conclusions but whose premises are actually inconclusively favorably relevant to their conclusions;52 and there is nothing counterintuitive about that. Objection 11. Degrees of attributed favorable relevance of premises to conclu- sion cannot distinguish deductive from in- ductive arguments. For (a) degrees of at- tributed favorable relevance do not divide arguments into two classes, one of which is subject to evaluation only by means of deductive logic, and the other of which is subject to evaluation only by means of in- ductive logic. For instance, deductive logic is required to evaluate the categorical syl- logism Premise 1. All animals are mortals. Premise 2. All humans are animals. Conclusion. All humans are mortals. regardless of the degree of favorable rel- evance to the conclusion attributed to the premises. And (b) an argument is deduc- tive if and only if it can be ascertained to be good or bad only by means of deduc- tive logic, and an argument is inductive if and only if it can be ascertained to be good or bad only by means of inductive logic. s3 Reply. The objection's second premise, (b), begs the question by assuming a rival distinction between deductive and induc- tive arguments. Moreover, (a) falsely as- sumes that it is possible to evaluate any argument by means of either deductive or inductive logic. For, as I have argued above and elsewhere, not every argument can properly be ascertained to be good or bad with respect to the relation between its premises and conclusion merely by ascer- taining to what degree, if any, its premises are actually relevant to its conclusion. 54 This is, however, the most that deductive or inductive logic can ever ascertain con- cerning any argument. Objection 12. Distinction 4 is wrong because it has the false consequences that the same argument can be both analogical and deductive and that the same argument can be both an instance of modus ponens and inductive. For someone arguing from analogy can attribute to his premises con- clusive favorable relevance to his conclu- sion; and that, according to Distinction 4, would suffice to make his analogical argu- ment deductive. And someone arguing by modus ponens can attribute to his premises inconclusive favorable relevance to his conclusion; and that, according to Distinc- tion 4, would make his argument induc- tive. 55 Reply. This objection begs the question by presupposing a rival distinction (prob- ably Distinction 2 or 3) when it assumes that if an argument is analogical, it cannot be deductive, and that if an argument has the form modus ponens, it cannot be induc- tive. Objection 13. On Distinction 4, some arguments are both deductive and induc- tive. For instance, the argument Premise 1. If the plane landed on time, he will be at the meeting at 9 a.m. Premise 2. He is at the meeting at 9 a.m. Conclusion. The plane landed on time. would be deductive. Yet because the premises, while not entailing the conclu- sion, do provide some evidence for it, the argument seems also to be inductive.56 Reply 1. Distinction 4 does not have ei- ther of the consequences that this objec- tion attributes to it. Since it distinguishes deductive from inductive arguments on the basis of the favorable relevance to the con- clusion that is attributed to the premises, it cannot justify claiming either that the ar- gument above is deductive until the degree Deductive/Inductive Distinction 171 of attributed favorable relevance of the premises to the conclusion is known or that an argument is inductive because its premises are actually inconclusively favorably relevant to its conclusion. Reply 2. The objection gives no reason why the argument would be deductive. If the reason were that the argument has the form of affirming the consequent, which would mean that it belongs to a family of arguments that are deductive, then the ob- jection would beg the question by assum- ing Distinction 2. Reply 3. The objection begs the ques- tion by assuming Distinction 3 when it rea- sons that because the argument's premises make its conclusion probable, the argument is (or appears) inductive. Objection 14. For application to every- day arguments, Distinction 4 requires that the vocabulary of our language be rich enough to permit arguers to express the degree of favorable relevance to the con- clusion that they attribute to their premises. But our vocabulary is not that rich. 57 Reply. On the contrary, the vocabulary of our language is rich enough not only to enable us to distinguish conclusive from inconclusive favorable relevance (e.g., by means of 'proves' as opposed to 'suggests' , and 'makes certain' as opposed to 'makes probable') but also to distinguish degrees of inconclusive favorable relevance from each other (e.g., by means of 'makes very probable', 'makes somewhat probable', 'makes the probability 0.7'). To summarize: Although many objections have been leveled at Distinction 4, they are all answerable except for Objections 2 and 3, which require that the distinction be re- vised to say that deductive arguments are those in which conclusive favorable rel- evance to the conclusion is attributed to the premises, whereas inductive arguments are the remainder. So revised, the distinction is both exclusive and exhaustive; it permits both good and bad arguments of both kinds; and it is both useful and needed in evaluat- ing at least some arguments. 172 George Bowles v. Distinction 5 Distinction 5 is a modification of the preceding one, since it defines deductive and inductive arguments in terms of the degree of favorable relevance to the con- clusion attributed to the premises; but it also incorporates features of Distinctions 2 and 3. According to this distinction, there is a deductive claim and an inductive claim, each of which may be made either by an argument or by an arguer. The deductive claim is that the argument's premises are conclusively favorably relevant to its con- clusion; the inductive claim is that the ar- gument's premises are inconclusively favorably relevant to its conclusion. An argument is deductive when and only when the deductive claim is made concerning it; it is inductive when and only when the in- ductive claim is made concerning it. There are both prima jacie indicators (called 'de- ductive indicators') that the deductive claim is made and prima jacie indicators (called 'inductive indicators') that the in- ductive claim is made. The significance of calling these indicators 'primajacie' is that, although none of them is either necessary or sufficient for an argument's being de- ductive or inductive, if a deductive indica- tor is present then, other things being equal, the argument is deductive; and if an induc- tive indicator is present then, other things being equal, the argument is inductive. If other things are not equal (i.e., if at least one deductive and at least one inductive indicator are present), then whether the ar- gument is deductive or inductive depends on the relative weights of the conflicting indicators: "If the balance of inductive in- dicators outweighs the balance of deduc- tive indicators, then the argument should be judged inductive. If the reverse, then deductive."58 Some deductive and inductive indicators are explicit; others are implicit.59 Explicit indicators are expressions that the arguer uses to show what degree of favorable rel- evance to the conclusion he attributes to the premises. For example, 'must' and 'shows conclusively' are explicit deduc- tive, whereas 'likely' and 'suggests' are explicit inductive, indicators. Implicit indicators are oftwo kinds. One is the family to which the argument be- longs. Categorical syllogisms (the good as well as the bad) constitute one family of arguments, analogical arguments (again, the good as well as the bad) constitute an- other, and so on. Families of arguments belong to one or the other of two groups, with, for example, categorical syllogisms, truth-functional propositional arguments, and mathematical arguments belonging to one group and analogical arguments and causal- arguments to the other. Whether a family belongs to one group or the other is determined by the following criterion: if a family of arguments is such that formal fea- tures reveal whether its members satisfy the deductive claim, then that family belongs to the first group; and if a family of argu- ments is such that formal features reveal whether its members satisfy the inductive claim, then that family belongs to the sec- ond group. An argument's belonging to a family in the first group is an implicit de- ductive indicator; its belonging to a family in the second group is an implicit induc- tive indicator.60 The second kind of implicit indicator is whether it is intuitive and obvious that the premises actually are conclusively or inconclusively favorably relevant to the conclusion. Its being intuitive and obvi- ous that the premises actually are con- clusively favorably relevant to the con- clusion is an implicit deductive indica- tor; its being intuitive and obvious that the premises actually are inconclusively favorably relevant to the conclusion is an implicit inductive indicator.61 Perhaps the absence of any explicit in- ductive indicator is itself an implicit deduc- tive indicator. In other words, each argu- ment may carry the presumption (which can be either defeated by explicit induc- tive indicators or reinforced by other de- ductive indicators) that the deductive claim is made concerning it. 62 When deductive and inductive indica- tors conflict, we might resolve the conflict by means of the rule that "(eJxplicit prima facie indicators always take precedence over implicit indicators."63 It is reasonable to accept this rule, because it leads to in- tuitively acceptable consequences. For in- stance, if we assume the rule, we are led to judge that the argument Washington was rationaL Lincoln was rational. Kennedy was rationaL So, all U. S. presidents must be ra- tional. is a bad argument, in that it is both deduc- tive and invalid; and this judgment is intui- tively correct64 Three considerations are advanced in favor of Distinction 5. First, it preserves the traditional distinction between deduc- tive and inductive arguments, endorsing the traditional classification of certain argu- ments as deductive and of others as induc- tive. Second, it preserves and harmonizes the preceding three distinctions according to families. actual favorable relevance, and attributed favorable relevance. And third, it accommodates disagreements between people about whether an argument is de- ductive or inductive; for people may disa- gree about the weight that should be as- signed to conflicting indicators. 65 To this distinction I shall raise four ob- jections which, jointly at least, warrant not accepting it. Objection i. There are gaps in Distinc- tion 5's account of the implicit indicators. How can either of the two implicit indica- tors (namely, the family to which the argu- ment belongs and the intuitiveness and ob- viousness of the premises' actually being conclusively or inconclusively favorably relevant to the conclusion) be prima facie indicators that the deductive or inductive claim is being made concerning that argu- ment? For suppose an argument belongs, say, to the family of categorical syllogisms. Deductive/inductive Distinction 173 That family belongs to the first group of families, because formal features of cat- egorical syllogisms show whether they sat- isfy the deductive claim. But how does that indicate, other things being equal, that a claim is being made (whether by the ar- guer or by the argument itself) that the premises are conclusively favorably rel- evant to the conclusion? Although there are some categorical syllogisms that would satisfy the deductive claim if it were made about them (because their premises are conclusively favorably relevant to their conclusions), there are others that would not Given only that this argument is a cat- egorical syllogism, then, why is it more likely than not, other things being equal, that the claim is made that this argument is one whose premises are conclusively favorably relevant to its conclusion, rather than one whose premises are not conclu- sively favorably relevant to their conclu- sions? Must it be assumed that all or most categorical syllogisms have premises that are claimed to be conclusively favorably relevant to their conclusions? If so, is there sufficient evidence for this assumption? The connection between an argument's belonging to a given family and its being the subject of either the deductive or the inductive claim has not been provided.66 And second, suppose an argument's premises are intuitively and obviously in- conclusively favorably relevant to its con- clusion. How does that indicate, other things being equal, that a claim is made that the premises are inconclusively favorably relevant to the conclusion? Although some arguments are such that there is agreement between the attributed and actual degrees of favorable relevance of their premises to their conclusions, others are not For in- stance, given only that it is intuitive and obvious that some argument's premises are inconclusively favorably relevant to its conclusion, why is it more likely than not, other things being equal, that the claim is made that this argument is one whose premises are inconclusively favorably rel- 174 George Bowles evant to its conclusion, rather than one whose premises are conclusively favorably relevant to its conclusion? Why is it not at least equally likely that, other things being equal, the claim is made that the premises of this argument are conclusively favorably relevant to its conclusion? Must it be as- sumed that in the case of all or most argu- ments in which the premises are intuitively and obviously inconclusively favorably relevant to the conclusion there is agree- ment between the attributed and actual de- grees of favorable relevance of the premises to the conclusion? If so, is there sufficient evidence for this assumption? Again, the connection has not been made out. Objection 2. Distinction 5 is subject to two objections already raised against other distinctions. First, it is subject to Objection 2 raised against Distinction 2. For Distinc- tion 5's rationale for assigning a family of arguments to the deductive group or to the inductive group will sometimes warrant assigning it to the other group as well. This would yield mutually inconsistent implicit prima facie indicators, neither of which would take precedence over the other. As long as no other, prevailing indicator is present, the result apparently would be that the argument is both deductive and induc- tive. Second, Distinction 5 is subject to Ob- jection 2 raised against Distinction 4. For there may be occasions when an arguer explicitly attributes to his premises only favorable relevance, rather than any par- ticular degree thereof. On such an occa- sion the arguer makes concerning his ar- gument a claim neither identical with nor implicative of either the deductive or the inductive claim. Moreover, on the suppo- sition that an argument is capable of mak- ing the deductive or inductive claim about itself, there is no reason why it might not instead make only the claim that its premises are favorably relevant to its con- clusion. In either case, according to Dis- tinction 5 the argument would be neither deductive nor inductive. Objection 3. Both the rule that explicit indicators should take precedence over implicit ones and its defense are problem- atic. (a) The rule would reduce the poten- tial for disagreement about whether some given argument is deductive or inductive; yet one part of the defense of Distinction 5 is that it accommodates such disagreement. Is such disagreement undesirable (in which case, it is good that the rule reduces it) or not (in which case it may be bad that the rule interferes with its accommodation)? (b) The defense of the rule (namely, that it is acceptable because if accepted it would yield judgments that are intuitively correct) has the logical form of affirming the con- sequent, and it is controversial whether the premises of arguments of that form con- firm (to say nothing of prove) their con- clusions. 67 Objection 4. The three points in the defense of this distinction present some difficulties. The first point is that Distinc- tion 5 preserves not only the traditional dis- tinction between deductive and inductive arguments but also the traditional assign- ment of certain arguments to one class or the other. But what is the "traditional" dis- tinction referred to, and what are the "tra- ditional" assignments? There seem to be three conspicuous possibilities. First, the "traditional" distinction might be Distinc- tion I, which is based on the generality or particularity of premises or conclusions. Because that distinction has been rejected by most logicians, however, it is doubtful that its conservation would seem desirable. Besides, Distinction 5 does not preserve that distinction's assignment of individual arguments to one class or the other. For an argument that would be, say, deductive ac- cording to Distinction 1 because its premises are general while its conclusion is particular might be inductive according to Distinction 5 because the arguer used an expression like 'probably' to express his claim that the premises are inconclusively favorably relevant to the conclusion. And an argument that would be inductive ac- cording to Distinction 1 because its premises are particular while its conclusion is general might be deductive according to Distinction 5 because the arguer used an expression like 'must' to express his claim that the premises are conclusively favorably relevant to the conclusion. So, it does not seem that the preserved "tradi- tional" distinction is Distinction 1. The two remaining possibilities are that the "tradi- tional" distinction referred to is that accord- ing to families (Distinction 2) or that ac- cording to actual favorable relevance (Dis- tinction 3). But in neither case would Dis- tinction 5 necessarily preserve "traditional" assignments of arguments to one class or the other. For, since Distinction 5 permits the ranking of explicit over implicit indi- cators, and since it says that both member- ship in a family and actual favorable rel- evance are implicit indicators, in some cases an argument may be deductive or in- ductive because of its explicit indicators and in spite of any implicit indicators of family membership or actual favorable rel- evance. The only earlier distinction that Distinction 5 preserves is Distinction 4, which was introduced too recently to be traditional. The second point in the defense of Dis- tinction 5 is that it harmonizes competing Distinctions 2, 3, and 4. But Distinction 5 harmonizes these rival distinctions only by transforming them all from definitions of deductive and inductive arguments to primafacie indicators that the deductive or inductive claim is being made: Distinction 2 becomes the implicit indicator of family membership, Distinction 3 becomes the implicit indicator of intuitive and obvious actual favorable relevance, and Distinction 4 becomes the explicit indicator of expres- sions employed by the arguer. In Objection I above I indicated the difficulties in inter- preting either of the two implicit indica- tors as prima facie indicators that the de- ductive or inductive claim is made. Deductive/Inductive Distinction 175 The final point in the defense of Dis- tinction 5 is that it accommodates disagree- ment about whether some argument is de- ductive or inductive. There are two reasons why this seems an odd thing to adduce in favor of the distinction. First, as already noted, it is incongruous with the inclusion of the rule assigning precedence to explicit indicators, since that rule reduces, rather than accommodates, disagreement about whether some argument is deductive or in- ductive. And second, other things being equal, the utility of any distinction is pro- portional to the extent to which it mini- mizes, rather than accommodates, disagree- ment about the things it distinguishes. So, if a distinction between deductive and in- ductive arguments accommodates disa- greements about whether a given argument is deductive or inductive, then, other things being equal, that is not a virtue but a defect of the distinction. For the reasons set forth in Objections 1-4, I conclude that Distinction 5 should not be accepted. VI. Conclusion Of the five distinctions between deduc- tive and inductive arguments examined here, I conclude that the best is the fourth - an argument is deductive if and only if conclusive favorable relevance to its con- clusion is attributed to its premises, and an argument is inductive if and only if incon- clusive favorable relevance to its conclu- sion is attributed to its premises. Although many objections have been raised to it, they are all answerable except for Objections 2 and 3, which require that the distinction be revised to say that an inductive argument is any argument that is not deductive. So revised, this distinction satisfies the crite- ria set forth earlier in the paper: it is exclu- sive and exhaustive; it permits both good and bad arguments of each kind; and it is both useful and needed in evaluating at least some arguments. 176 George Bowles Notes * My thanks to Thomas E. Gilbert, to my fonner colleagues in the Philosophy Department at George Washington University, and especially to Mark Vorobej for their many helpful com- ments on earlier drafts of this paper. This distinction is sometimes attributed to William Whewell (e.g., by Copi and Cohen [1990:46] and by Weddle [1979:2]). But Whewell (1840:11, xi, 5; Vol. 2, p. 214) not only refers to it as already familiar but also rejects its account of induction as incomplete ([1840:214-2151; see also [1860:XXII, pp. 240, 253-254, 255-256]). Copi and Cohen (1990:47-48). See also Yanal ( 1988:86-87). 3 If it be objected that the premises of this al- leged counterexample are singular, rather than particular, propositions, the following argu- ment might be substituted: Premise. Some women are runners. Conclusion. Some runners are women. These are not the only criteria. For instance, the distinction should also be clear rather than obscure and distinct rather than fuzzy. Not every classification of arguments is re- lated to argument evaluation in this way. The distinction between verificatory and explana- tory arguments and the distinction between elegantly and inelegantly expressed arguments are examples. Barker (1975:59) and Freeman (1984:38). It might seem that Nicholas Rescher's work on plausible reasoning supports "the idea that there is a third type of reasoning distinctive from deductive and inductive reasoning called plausible reasoning ... " (Walton [1992:33]). Rescher's approach, however, assimilates plausible reasoning to deductive reasoning, at least for the purposes of evaluation: "The pres- ently envisaged approach to plausible infer- ence thus proposes to assess the plausibility of a 'merely plausible' piece of reasoning in terms of the plausibility of the added enthymematic premisses needed to transform it into a valid deductive argument." (Rescher [1976:60-61 n. The variety and obscurity of C. S. Peirce's distinctions among deduction, induction, and abduction (or hypothesis, or retroduction) probably explains their neglect in the recent debate with which this paper deals. See Peirce (1958, 1960: 1.66-68, 2.267, 2.269, 2.270, 2.515,2.620,2.624,5.145,6.526,7.206, 8.209, 8.236). 8 Fohr (1980a:5), Freeman (1984:38), Machina (1985:572). Arguments may be evaluated logi- cally according to at least the following three criteria: whether their asserted, nonredundant premises are true or acceptable; whether they beg the question; and whether, apart from beg- ging the question, their premises are properly related to their conclusions. When, in this paper, I speak of an argument's being good or bad, it is with respect to this third criterion only. Objection. It is requiring too much of a dis- tinction between deductive and inductive ar- guments that it pennit both good and bad ar- guments of each kind. It would suffice if it permitted one argument to be better or worse than another. Reply. The main purpose of evaluating an ar- gument is to ascertain whether it is worthy of acceptance. This purpose is advanced by evaluating that argument as good or bad, not as better or worse than another. For given only that one argument is better or worse than an- other, it remains undetermined whether either ought to be accepted or rejected. It would, therefore, be pointless to require that a dis- tinction between deductive and inductive ar- guments permit the comparative evaluation of arguments as better or worse yet not re- quire also that it pennit the evaluation of ar- guments as good or bad. 9 Hitchcock (1981:7): "As practitioners of in- fonnal logic, we are oriented towards the ap- praisal of arguments which people actually advance in an attempt to convince others (or themselves) to believe or do something. The question at issue, then, is whether any ver- sion of the distinction between deduction and induction is helpful in appraising arguments. If so, which one?" 10 Objection. Distinction I could be made ex- haustive either by first defining a deductive argument as one whose premises are general and whose conclusion is particular and then defining an inductive argument as any that is not deductive, or by first defining an induc- tive argument as one whose premises are par- ticular and whose conclusion is general and then defining a deductive argument as any that is not inductive. Reply. Either amendment, while yielding an exhaustive distinction, would sacrifice the general- to- particular Iparti c u lar-to-general symmetry that seems essential to Distinction I. For example, the first amendment would entail that Premise I. All animals are mortals. Premise 2. All humans are animals. Conclusion. All humans are mortals. must be an inductive argument, since it does not fit the definition of a deductive argument. Likewise, the second amendment would en- tail that Premise. Some females are not mothers. Conclusion. Some mothers are not females. is deductive, since it does not fit the defini- tion of an inductive argument. An advocate of Distinction I would probably reject both of these consequences and conclude that the amendments did not so much preserve as de- stroy the distinction. II Weddle (1979:4): "It is tempting to say that what distinguishes deductive from inductive arguments is the sections of logic books in which they happen to be found." The clearest presentation of this distinction comes, by ad- aptation, from Freeman (1983:8-9) and espe- cially (1984:37-38). Although Professor Free- man would reject the distinction as stated above, he incorporates a modification of it in Distinction 5. 12 Adapted from Freeman (1983:8-9) and (1984:37 -38). I am indebted to the author for clarification of his arguments in private cor- respondence (September 12, 1988). For a dif- ferent account of what distinguishes the two groups from each other, see F. Johnson ( 1980:5). 13 This is not to say that formal features never determine whether the premises of arguments of a given family are inconclusively favorably relevant to their conclusions. 14 See, for example, Freeman (1988:322, 324). 15 Adapted from Strawson (1959:xiv). This ex- Deductive/Inductive Distinction 177 ample was found by a student, Fred Sandal. 16 For additional criticisms, see F. Johnson ( 1980:5). If my example is objected to, the same point can be made by other means. Let the argu- ment be Premise. Either descriptive metaphysics has had a long and complicated history or not. Conclusion. There are and are not new truths to be discovered in descriptive metaphysics. It has the form Premise. Either p or not-po Conclusion. q and not-q. which determines that the argument's premise is neither conclusively nor inconclusively favorably relevant to the conclusion, so that this argument too would be neither deductive nor inductive. 17 The argument in this paragraph could also be used against a further revision meant to in- sure exhaustiveness namely, • An argument is deductive if it belongs to a family such that formal features reveal that its members' premises are conclusively favorably relevant to their conclusions; otherwise, it is induc- tive'. 18 Baum (1981 :96), Damer (1980:2), Guttenplan and Tamny (1971:4), Machina (1985:572), Manicas and Kruger (1968:22), Nolt (1984:5- 7), M. Salmon (1984:32, 39, 48), W. Salmon (1963:14). See also Brody (1967:62, 66) and Fritz. (1960: 127). who distinguish deductive from inductive inferences in terms of actual favorable relevance. A similar way of distinguishing deductive from inductive arguments says that an argu- ment is deductive if its premises are conclu- sively favorably relevant to its conclusion; otherwise, it is inductive. (Fritz. [1960: 128], Manicas and Kruger [1968:24), and Manicas and Kruger [1976:52].) This would make in- ductive all arguments whose premises are ei- ther inconclUsively favorably relevant, irrel- evant, or unfavorably relevant to their con- clusions, so that there would be no distinc- tion between good and bad (valid and invalid) deductive arguments. (See Objection 2 to Dis- tinction 3.) 19 Freeman (1984:39). 178 George Bowles A similar conclusion follows concerning the following distinction: a deductive argument is one in which conclusive favorable relevance to the conclusion either actually belongs, or at least is attributed, to the premises, whereas an inductive argument is one whose conclu- sion is empirical. (Barker [1989:258,260] and Govier [1988:260)). This distinction is nei- ther exclusive nor exhaustive. It is not exclu- sive, because it implies that the argument ex- pressed in the text The fact that all papers are written by inhab- itants of Earth makes it certain that this paper is written by an inhabitant of Earth. would be both deductive (since conclusive favorable relevance to the conclusion 'This paper is written by an inhabitant of Earth' both actually belongs, and is attributed to, the premise 'All papers are written by inhabit- ants of Earth') and inductive (because the conclusion is empirical). The distinction also is not exhaustive, because according to it the argument expressed in the text The fact that John says that the square of 13 is 169 makes it probable that it is, since John is usually right when it comes to mathemat- ics. would be neither deductive (since conclusive favorable relevance to the conclusion 'The square of 13 is 169' neither actually belongs, nor is attributed, to the premises 'John says that the square of 13 is 169' and 'John is usu- ally correct when it comes to mathematics') nor inductive (since the conclusion is not empirical). 20 Barker (1975:62). Fohr (I 980a:5-6), Freeman (1984:39), Hitchcock (1980:9). Machina (1985:572), Noh (1987:419), and Yanal (1988:87). It would be unreasonable to suggest that, say, those inducti ve arguments whose premises are, at least to some specified degree, favorably relevant to their conclusions are good, whereas the remainder are bad. For that would mean that inductive arguments whose premises are, to less than that specified de- gree, favorably relevant to their conclusions are, despite that favorable relevance, to be evaluated as bad, just as they would be if their premises had been irrelevant or unfavorably relevant to their conclusions. For the theory of argument evaluation as- sumed here, see Bowles (1991 :9·11). And for a probabilistic explication of the notion of relevance employed, see Bowles (I 990:65- 67). 21 Fritz (1960: 129). The circularity described in this objection is avoided by at least some other distinctions between deductive and inductive arguments. For instance, on Distinction 4 the argument would be deductive because the expression 'proves that' shows that conclusive favorable relevance to the conclusion is attributed to the premises: the argument's classification would not be contingent on its unexpressed premise's being supplied. 22 Beardsley (1950:201), Copi and Cohen (1990:49), Fohr (1980a:7), Halverson (1984:5-6), Hurley (1982:21), Moore and Parker (1986: 189), Rescher (1964:60), Simco and James (1983: I, 310, 313), and Terrell (1967: 10). These writers speak not of attribu- tion in general, as I do here, but of some spe- cific kind of attribution. Beardsley, Copi and Cohen, Rescher, and Simco and James speak of a claim; Terrell speaks of a supposition; Fohr, Hurley, and Moore and Parker speak of an intention; and Rescher speaks of an at- tempt. 23 Bowles (1989). 24 Fohr (1980a:8). 25 Copi (1986:547), Halverson (1984:5-6), Hur- ley (1982:29-30), R. Johnson (1987:26), and Simco and James (1983:184). 26 Govier (I 980a: II) and (I 987b:30), Hitchcock (1983:109), and NoIt and Rohatyn (1988:21, note 3). 27 Fohr (l980b:6): " ... a person doesn't have to state an intention explicitly, or even be think- ing of something, in order to have an inten- tion." 28 Contra Hitchcock (1980:9): " ... the only way of detecting such intentions is to notice what the arguer claims." 29 Patrick J. Hurley (1982:21-22) suggests that if the premises are actually conclusively favorably relevant to the conclusion, "one may usually assume" that the arguer meant them to be. Both he and Barrie Wilson (1980:262) suggest that, in the absence of verbal clues, we can infer the arguer's intentions from the family to which his argument belongs. For ex- ample, if an arguer gives a categorical syllo- gism, we may assume that he means his premises to be conclusively favorably relevant to his conclusion. But if he gives an analogi· cal argument, we may infer that he means his premises to be inconclusively favorably rel- evant to his conclusion. (But see Objection I to Distinction 5, below.) 30 Fohr (1980b:6) and Hurley (1982:25) offer dif- ferent replies to this objection. Fohr says that, in the absence of verbal clues from the ar- guer, we should, if possible, ask him for clari- fication. If that is not possible, we should observe the Principle of Charity and classify the argument as deductive if it would be a better deductive than inductive argument, or as inductive if it would be a better inductive than deductive argument. Hurley implicitly admits that the deductive-inductive distinc- tion is not exhaustive when he says that it is an analytical tool and that " ... no analytical tool can ever be expected to fit every possi- ble set of circumstances." 31 Hitchcock (1980: 10), NoH (1987:419-420), and Yanal (1988:87). 32 See Carney and Scheer (1980: 10-11). Objection 1. Such a definition of an induc- tive argument would be entirely negative. AI· though it would tell us that an inductive argu- ment is not deductive, it would not tell us anything positive about what all inductive arguments have in common. (Govier [1987a:57] and [l987b:50, 51-2]. See also Govier [1980a: II], [1980b: 8], and [1988:260].) Reply 1. It is not necessary for a definition to specify something positive, rather than nega- tive, common to the members of the class being defined. The complement of a class, for instance, can properly be defined negatively. Reply 2. Under the revised definition, induc- tive arguments would have in common at least what all arguments have in common namely, that favorable relevance to their con- clusions is attributed to their premises. (See Bowles [1989].) Reply 3. Under the revised Distinction 4, com- mon to all inductive arguments would be the Deductive/Inductive Distinction 179 ability to be good with respect to the relation between their premises and their conclusions even if their premises are not actually con- clusively favorably relevant to their conclu- sions. Objection 2. 'Therefore' (like 'so', 'for', 'hence', 'since', and 'because') is a neutral illative, in the sense that when someone says something of the form 'x; therefore, y' he tells us that he believes that 'x' is favorably rel- evant to 'y' but not that 'x' is conclusively, nor that it is inconclusively, favorably relevant to 'y'. (Allen [1988:60]) According to the pro- posed revision of Distinction 4, then, if some- one were to say, "x = 7; therefore, xl = 49", although he would attribute to his premise favorable relevance to his conclusion, he would not attribute either conclusive or in- conclusive favorable relevance, so that his argument would be inductive. But it seems at least odd to call this argument inductive, since it is a mathematical argument, in which the premise actually is conclusively favorably relevant to the conclusion; and no arguer of normal competence would attribute to the premise 'x = 7' anything less than conclusive favorable relevance to the conclusion 'Xl = 49'. Reply. The objection assumes that when some- one uses only a neutral illative like 'there- fore', he attributes to his premises only favorable relevance - not conclusive and not inconclusive favorable relevance - to his conclusion. But two kinds of attribution are distinguishable in such a case: (I) what the person attributes explicitly by means of words; and (2) what he attributes tacitly, without ex- pressing his attribution in words. Because 'therefore' is a neutral illati ve, it is quite true that when someone says something of the form 'x; therefore, y', he explicitly attributes to his premises only favorable relevance - not con- clusive and not inconclusive favorable rel- evance - to his conclusion. Still, he may, and in the case of 'x = 7; therefore. Xl = 49' surely WOUld, have in mind (but not say) that the premise is conclusively favorably relevant to the conclusion. (See Bowles and Gilbert [1993:256. Reply to Objection 4].) And the attribution that matters for Distinction 4 is that which the arguer has in mind, whether he ex- presses it (correctly) in words or not There- fore. on the plausible assumption that some- one who said "x = 7; therefore, X Z = 49" would tacitly attribute to his premise conclusive 180 George Bowles favorable relevance to his conclusion, it fo[- lows that for Distinction 4 such an argument would be deductive, not inductive. 33 The argument referred to may be good, in the sense that there is agreement between the at- tributed and actual degrees of relevance of its premises to its conclusion (see the next para- graph in the body of the paper), although it is inductive and its premises are conclusively favorably relevant to its conclusion. For it is inductive because the attributed degree of favorable relevance of the premises to the conclusion is "at [east probable", which ac- commodates certainty, the actual degree of favorable relevance of the premises to the conclusion. 34 Fohr ([ 980b:6) responds differently, saying, "When a person utters something which could be construed as an argument but has no intention about the relationship of the premisses to the conclusion then that per- son has not really expressed a unique argu- ment." This reply seems inapt, since the objection says not that the arguer "has no intention about the relationship of the premisses to the conclusion" but that a[- though he attributes to the premises favorable relevance, he does not attribute to them any specific degree of favorable re[- evance, to the conclusion. 35 Bow[es (199 [:9- II). 36 F. Johnson (1980:5). 37 Hitchcock (198 [:8). 38 Fohr (1980b: [0). In a similar vein, Mark Yorobej (1992: [06) notes that "the very iden- tity of the argument being presented will of- ten rest onjust this issue [namely, the strength of the logical link between the premises and the conclusion from the author's perspec- tive]", and A[ec Fisher (1990: I) suggests that two arguments are the same "if they make the same commitments". 39 Machina (1985:573-574, 577, 578). See also No[t (1987:420): "Intentional definitions may have some use in the psychology of argumen- tation. But I shall lay them aside, since I am concerned with logic." 4ll Adapted from Govier (I980b:7-8). 41 See Bowles (1989). 42 Machina (1985:577): "Properly logical evalu- ation merely considers whether certain evidentiary relations hold between the prem- iss set and the conclusion." See also Carter (1977:14), Nolt and Rohatyn ([988:2[, note 3), and Weddle (1980:12). 43 Machina ([ 985:573-574, 577). See also Hitchcock (198 [:8-9). 44 Bowles (1991:9-11 and 7, Objection [ and its Reply). 45 Weddle (1979:2). See also Hitchcock ( 1980: [0) and ([ 983: [09). 46 Objection. Evaluating an argument by com- paring the actual with the attributed degrees of favorable relevance of the premises to the conclusion amounts to employing a different standard of evaluation for each argument, which makes the evaluation of an argument independent of whether it is deductive or in- ductive - contrary to the third criterion. Reply. It is not true that evaluating an argu- ment by comparing the actual with the attrib- uted degrees of favorable relevance of the premises to the conclusion amounts to em- p[oying a different standard of evaluation for each argument. For in some arguments (namely, those that would be deductive ac- cording to Distinction 4) conclusive favorable relevance to the conclusion is attributed to the premises, so that these arguments would all be evaluated according to the same standard (viz., whether their premises actually were conclusively favorably relevant to their con- clusions); and in other arguments (namely, some of those that would be inductive accord- ing to Distinction 4) inconclusive favorable relevance to the conclusion is attributed to the premises, so that these arguments would all be evaluated according to the same standard (viz., whether their premises actually were in- conclusively favorably relevant to their con- clusions.) 47 Fohr (1980b:7). 48 See Bowles and Gilbert (1993:256, Reply to Objection 4). 49 Weddle (1979:3). 50 Freeman (1983:3-8). See also Hempel (1965:58-59) and W. Salmon (1963 :61). 51 Govier (I980b:7-8). See also Machina ( 1985:576). 52 See Fohr (1980b:7). 53 Machina (1985:575-6). 54 Bowles (1991 :2-6). 55 Nolt (1987:420). See also DeWitt (1992:9). 56 Yanal (1988:87). 57 Govier (l987b:30). 58 Freeman (1984:34). In (1988:225-229) Professor Freeman offers a variant of Distinction 5. But inasmuch as it is not clear to me how much of Distinction 5 is to be retained in the variant, and the vari- ant is not as carefully worked out as Distinc- tion 5 (probably because it appears in an el- ementary textbook), I confine my attention in this paper to the original. 59 It seems permissible to surmise that explicit indicators are prima facie indicators that the arguer is making the deductive or the induc- tive claim, whereas implicit indicators are prima facie indicators that either the arguer or the argument itself is making that claim. 60 For a detailed explanation why membership in some families is a deductive indicator, whereas membership in others is an induc- tive indicator, see Freeman (1984:37-38). 61 The preceding three paragraphs are based on Freeman (1983:8-10) and (1984:36-38). The crucial definitions of deductive and inductive arguments are never stated there but seem to be assumed, especially in the latter essay. Anticipations of the implicit indicators may be found in Hurley (1982:21-22) and Wilson (1980:262). Hurley suggests that if the premises are actually conclusively favorable to the conclusion, "one may usually assume" that the arguer meant them to be. Both he and Wilson suggest that. in the absence of verbal clues, we can infer the arguer's intentions from the family to which his argument be- longs. For example. if an arguer gives a cat- egorical syllogism, we may assume that he means his premises to be conclusively favorably relevant to his conclusion. But if he gives an analogical argument, we may in- fer that he means his premises to be incon- clusively favorably relevant to his conclusion. The status of Distinction 5's second kind of implicit indicator is unclear. In (1983:9) Pro- fessor Freeman says, "The fact that an argu- ment is intuitively and obviously deductively Deductive/Inductive Distinction 181 valid or that its premises clearly give good inductive support to the conclusion [n.b.:] could be a prima facie mark that the argument is deductive or inductive." But later on the same page, without explanation, he drops this indicator from his formulation of Distinction 5. 62 Freeman (1983:9), (1988:228). 63 Freeman (1983:9). Although Professor Free- man does not insist on this rule, his position, especially as elaborated in (1984), seems to commit him to it. For it would be implausible to say that an implicit prima facie indicator is a better indicator than an explicit prima facie indicator of whether the deductive or the in- ductive claim is made, because the explicit indicator helps to make that claim overtly. On the same page the author seems to sug- gest that the first of the two implicit prima facie indicators should take precedence over the second. In a case where the premises of a categorical syllogism are actually only incon- clusively favorably relevant to the conclusion, he asks rhetorically, "wouldn't the mark that the argument belonged to a traditional deduc- tive family override the mark of the premises supporting the conclusion?" 64 Freeman (1983:9) and (1984:36). 65 Freeman (1983:9-10). 66 Reply. The reason why an argument's being a categorical syllogism is a prima facie indica- tor that the claim is made that the premises of that argument are conclusively favorably rel- evant to its conclusion is that the premises of a categorical syllogism may be conclusively favorably relevant to its conclusion (if the syllogism is valid), or they may fail to be at all favorably relevant to it (if the syllogism is invalid), but they cannot be inconclusively favorably relevant to it. Some categorical syl- logisms, in short, are capable of satisfying the deductive claim, but none is capable of satis- fying the inductive claim. That is why the fact that an argument is a categorical syllogism creates a presumption that the deductive claim is made concerning it: it could satisfy that claim, but it could not satisfy the inductive claim. Similarly, the reason why the fact that an argument is, say, an analogical argument is prima facie evidence that the inductive claim is made concerning it: such an argu- ment could satisfy the inductive claim, but it could not satisfy the deductive claim. 182 George Bowles Rejoinder. It appears that, on the contrary, some categorical syllogisms can satisfy the inductive claim, and some analogical argu- ments can satisfy the deductive claim. For example, the categorical syllogism Premise I. All murderers of Enoch J. Drebber are men who are more than six feet tall; are in the prime of life; have small feet for their height; wore coarse, square-toed boots; smoked Trichinopoly ci- gars; came with the victim to the site of the murder in a four- wheeled cab pulled by a horse with a new shoe on its off fore- leg; and had a florid face and unusually long nails on their right hands. Premise 2. All suspects named 'Jefferson Hope' are men who are more than six feet tall; are in the prime of life; have small feet for their height; wore coarse, square-toed boots; smoked Trichinopoly ci- gars; came with the victim to the site of the murder in a four- wheeled cab pulled by a horse with a new shoe on its off fore- leg; and had a florid face and unusually long nails on their right hands. Conclusion. All suspects named 'Jefferson Hope' are murderers of Enoch J. Drebber. or a variant of it, appears to be such that its premises are inconclusively favorably relevant to its conclusion. (See Weddle [1979:3 J and Free- man [1983:9].) And the analogical argument Premise I. Arnold, Beth, Carl, and Dora are fast runners. Premise 2. Arnold, Beth, and Carl are run- ners. Conclusion. Dora is a runner. seems to be such that its premises are conclu- sively favorably relevant to its conclusion. Therefore, it seems, some categorical syllo- gisms are capable of satisfying the inductive claim, and some analogical arguments are capable of satisfying the deductive claim. 61 See, for example, W. Salmon (1963:81-84) and (1975:34-35). References Allen, Derek, "Inferential Soundness", Informal Logic, Vol. X, No.2 (Spring 1988), 57-65. Barker, S. E, "Must Every Inference Be Either Deductive or Inductive?", in Philosophy in America, ed. Max Black (Ithaca, New York: Cornell University Press, 1975),58-73. Barker, Stephen E, The Elements of Logic, Fifth Edition (New York: McGraw-Hili Book Com- pany, 1989). Baum, Robert, Logic, Second Edition (New York: Holt, Rinehart and Winston. 1981). Beardsley, Monroe c., Practical Logic (Englewood Cliffs, N. 1.: Prentice-Hall, Inc., 1950). Bowles, George and Gilbert, Thomas E., "The Probabilistic Import of Illatives", Argumen- tation. Vol. 7, No.3 (1993), 247-262, Bowles, George, "Evaluating Arguments: The Premise-Conclusion Relation", Informal Logic, Vol. XIII, No. I (Winter 1991), 1-20. Bowles, George, "Favorable Relevance and Ar- guments", Informal Logic, Vol. XI, No. I (Winter 1989), 11-17. Bowles, George, "Propositional Relevance", In- formal Logic, Vol. XII, No.2 (Spring 1990), 65-77. Brody, Boruch [sic] A., "Glossary of Logical Terms", in The Encyclopedia of Philosophy, ed. Paul Edwards (New York: Macmillan Pub- lishing Co., Inc., & The Free Press, 1967), Vol. 5, 57-77. Carney, James D. and Scheer, Richard K., Fun- damentals of Logic, Third Edition (New York: Macmillan Publishing Co., Inc., 1980). Carter, K. Codell, A Contemporary Introduction to Logic (Beverly Hills: Glencoe Press, 1977). Copi, Irving M. and Cohen, Carl, Introduction to Logic, Eighth Edition (New York: Macmillan Publishing Company, 1990). Copi, Irving M., Introduction to Logic, Seventh Edition (New York: Macmillan Publishing Company, 1986). Damer, T. Edward, Attacking Faulty Reasoning (Belmont, California: Wadsworth Publishing Company, 1980). DeWitt, Richard, "Critical Thinking and Sexing Chickens", Inquiry: Critical Thinking Across the Disciplines, Vol. 10, No. I (September 1992), 8-11. Fisher, Alec, private communication, May, 1990. Fohr, Samuel D., "Deductive-Inductive: Reply to Criticisms", Informal Logic Newsletter, Vol. III, No. I (October 1980b), 5-10. Fohr, Samuel D., "The Deductive-Inductive Dis- tinction", Informal Logic Newsletter, Vol. II, No.2 (April 1980a),5-8. Freeman, James B., "Logical Form, Probability Interpretations, and the InductivelDeductive Distinction", Informal Logic Newsletter, Vol. V, No.2 (June 1983), 2-10. Freeman, James B., "Reply to Englebretsen", Informal Logic, Vol. VI, No.3 (December 1984),34-40. Freeman, James B., Thinking Logically: Basic Concepts for Reasoning (Englewood Cliffs, New Jersey: Prentice Hall, 1988). Fritz, Charles A., Jr., "What is Induction?", The Journal of Philosophy, Vol. 57 (February 18, 1960), 126-138. Govier, Trudy, "Beyond Induction and Deduc- tion", in Argumentation: Across the Lines of Discipline, ed. Frans H. van Eemeren, Rob Grootendorst, J. Anthony Blair, and Charles A. Willard (Dordrecht-Holland: Foris Publi- cations, 1987a), 57-64. Govier, Trudy, "Critical Review of Carl Wellman's Challenge and Response", Infor- mal Logic Newsletter, Vol. II, No. 2 (April I 980a), 10-15. Govier, Trudy, "More on Deductive and Induc- tive Arguments", Informal Logic Newsletter, Vol. II, No.3 (June 1980b), 7-8. Govier, Trudy, A Practical Study of Argument, Second Edition (Belmont, California: Wadsworth Publishing Company, 1988). Govier, Trudy, Problems in Argument Analysis and Evaluation (Dordrecht, Holland: Foris Deductive/Inductive Distinction 183 Publications, 1987b). Guttenplan, Samuel D. and Tamny, Martin, Logic: A Comprehensive Introduction (New York: Basic Books, Inc., 1971). Halverson, William H., A Concise Logic (New York: Random House, 1984). Hempel, Carl G., "Inductive Inconsistencies", in Carl G. Hempel, Aspects of Scientific Expla- nation (New York: The Free Press, 1965), 53- 79. Hitchcock, David, "Deduction, Induction and Conduction", Informal Logic Newsletter, Vol. III, No.2 (March 1981),7-15. Hitchcock, David, "Deductive and Inductive: Types of Validity, Not Types of Argument", Informal Logic Newsletter, Vol. II, No.3 (June 1980), 9-11. Hitchcock, David, Critical Thinking: A Guide to Evaluating Information (Toronto: Methuen, 1983). Hurley, Patrick J., A Concise Introduction to Logic, First Edition (Belmont, California: Wadsworth Publishing Company, 1982). Johnson, Fred, "Deductively-Inductively", Infor- mal Logic Newsletter, Vol. Ill, No. I (Octo- ber 1980),4-5. Johnson, Robert M., A Logic Book (Belmont, California: Wadsworth Publishing Company, 1987). Machina, Kenton F., "Induction and Deduction Revisited", Nous, Vol. XIX, No.4 (Decem- ber 1985),571-578. Manicas, Peter T. and Kruger. Arthur N., Essen- tials of Logic (New York: American Book Co., 1968). Manicas, Peter T. and Kruger, Arthur N., Logic: The Essentials (New York: McGraw-Hill Book Company, 1976). Moore, Brooke Noel and Parker, Richard, Criti- cal Thinking: Evaluating Claims and Argu- ments in Everyday Life (Mountain View, Cali- fornia: Mayfield Publishing Company, 1986). Nolt. John and Rohatyn. Dennis, Schaum's Out~ line of Theory and Problems of Logic (New York: McGraw-Hili Book Company, 1988). Nolt, John E., "Dilemmas of the Inductive/De- ductive Distinction", in Argumentation: Per- l84 George Bowles spectives and Approaches, ed. Frans H. van Eemeren, Rob Grootendorst, J. Anthony Blair, and Charles A. Willard (Dordrecht-Holland: Foris Publications, 1987), 418-425. Nolt, John Eric, Informal Logic: Possible Worlds and Imagination (New York: McGraw-Hili Book Company, 1984). Peirce, Charles Sanders, Collected Papers of Charles Sanders Peirce, ed. Charles Hartshorne, Paul Weiss, Arthur W. Burks (Cambridge, Massachusetts: Harvard Univer- sity Press, 1958, 1960), Vol. I-VIII. Rescher, Nicholas, Introduction to Logic (New York: St. Martin's Press, 1964). Rescher, Nicholas, Plausible Reasoning (Assen/ Amsterdam: Van Gorcum, 1976). Salmon, Merrilee H., Introduction to Logic and Critical Thinking (San Diego: Harcourt Brace Jovanovich, 1984). Salmon, Wesley c., "Confirmation and Rel- evance", in Minnesota Studies in the Philoso- phy of Science, Volume VI: Induction, Prob- ability, and Confirmation, ed. Grover Maxwell and Robert M. Anderson, Jr. (Minneapolis: University of Minnesota Press, 1975), 5-36. Salmon, Wesley C., Logic (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1963). Simco, Nancy D., and James, Gene G., Elemen- tary Logic, Second Edition (Belmont, Cali- fornia: Wadsworth Publishing Company, 1983). Strawson, Peter, Individuals: An Essay in De- scriptive Metaphysics (n. p.: Methuen & Co. Ltd., 1959; Garden City, New York: Doubleday & Company, Inc., 1963). Terrell, D. 8., Logic: A Modern Introduction to Deductive Reasoning (New York: Holt, Rinehart and Winston, Inc., 1967). Vorobej, Mark, "Defining Deduction", Informal Logic, Vol. XIV, Nos. 2 & 3 (Spring and Fall 1992), 105-118. Walton, Douglas, "Rules for Plausible Reason- ing", Informal Logic, Vol. XIV, No. I (Winter 1992), 33-51. Weddle, Perry, "Good Grief! More on Deduction! Induction", Informal Logic Newsletter, Vol. III, No. I (October 1980), 10-13. Weddle, Perry, "Inductive, Deductive", Informal Logic Newsletter, Vol. II, No. I (November 1979), 1-5. Whewell, William, On the Philosophy of Discov- ery (1860; reprinted New York: Burt Franklin, 1971). Whewell, William, The Philosophy of the Induc- tive Sciences, Founded upon their History (London: John W. Parker, 1840). Wilson, Barrie, The Anatomy of Argument (Wash- ington, D.C.: University Press of America, 1980). Yanal, Robert J., Basic Logic (St. Paul: West Publishing Co., 1988). GEORGE BOWLES 4466 ARLINGTON BLVD. ARLINGTON, VA 22204-/34066 a