INFORMAL LOGIC XI.3, Fall 1989 Beardsley's Theory of Analogy EVELYN M. BARKER The University of Maryland 1. Is the Argument from Analogy a Fallacy? In the various editions of Practical Logic and Thinking Straight Monroe Beardsley pioneered informal logic, the assessment of reasoning techniques employed in actual human situations. 1 Although acknowledg- ing the prominence of analogies in reason- ing, he contended that analogy has some heuristic uses, but an argument based on one is inherently fallacious: Analogies illustrate, and they lead to hypotheses, but thinking in terms of analogy becomes fallacious when the analogy is us- ed as a reason for a principle. This fallacy is called the "argument from analogy." (PL 107) Beardsley's view implies that no serious reasoner will use analogies between things in drawing conclusions, or be persuaded by another's introduction of them. According- ly, an important source of human knowledge is proscribed: historical analogy, reasoning based on direct comparison of past situations with the present. In practical reasoning, however, one must often reach a judgment about the crucial features of some particular situation in order to determine an appropriate course of action with respect to it. Analogizing its features to another whose favorable or un- favorable outcome is already known, far from being fallacious, is at least common sense and at most wisdom. For example, in a recent newspaper commentary head- lined "A Stable World: Lessons of World War II", Henry Kissinger compares the current international situation with that leading to World War II, warning of the dire consequences of an "abdication of statesmanship."2 Kissinger cites historical analogy as the only basis for sound judg- ment in international diplomacy: History is the only experience on which statesmen can draw. But it does not teach its lessons automatically. It demonstrates the consequences of comparable situations, but each generation has to determine what situa- tions are in fact comparable. Such analogical arguments, as Kissinger points out, are directed to finding suitable premises for wise and prudent action. Beardsley's intellectualist approach limits rational argumentation to asserting premises as evidence for the truth of a conclusion. Consequently he overlooks or misconceives the character of some forms of analogical argument prevalent in practical reasoning. It is misleading to speak of "the argument from analogy," as Beardsley does, as if all are of the same kind. 3 Historical analogy, or "learning from experience," most nearly resembles arguments of inductive analogy: From observed facts about previous instances one draws an inference about what is likely to be true of a new instance sharing salient characteristics with them. Beardsley denies, however, that inductive analogy is "a suc- cesful kind of inductive argument" (TS2 66). Beardsley's theory of analogical reason- ing exemplifies overformalization in reduc- ing diverse kinds of analogy arguments to a single form, then assessing them all fallacious because of the logical characteristics of that form. This results in a "denaturalized epistemology" of analogical arguments, one with inappro- 186 Evelyn M. Barker priate concepts of knowledge, unrelated to reasoning techniques successfully employed in a range of actual situations. In order to track Beardsley's claim that an analogical argument is inherently fallacious, we must have in view his general theory of argument. Like many, he divides arguments into two exclusive kinds: deduc- tive and inductive. A deductive argument makes the claim (correctly or incorrectly) that the truth of its conclusion follows necessarily from the truth of the premises, because the argument is formally valid. He calls all nondeductive arguments inductive: Such arguments make a lesser claim that "the evidence is sufficient to make the con- clusion, at the very least, more likely to be true than false." (PL 197-203, TS228-29, TS423-27) He insists that "any compelling argu- ment must be one of these kinds, and moreover, it must conform to the standards of its kind." (PL 197). As Beardsley describes the argument from analogy, it is neither formally valid, nor does it fit his two models of inductive reasoning, explanatory hypothesis and inductive generalization. Not surprisingly, an analogical argument fulfills Beardsley'S conception of a fallacy: An argument is a good one only if it pro- ceeds in accordance with some rule of in- ference and thus can be justified by appeal to that rule. If it purports to conform to a rule of inference, and thus acquires some plausibility, though in fact it violates that rule, then it is said to be fallacious, or to contain a fallacy. (TS4 26) As we shall see, the analogical argument turns out not to have a justifying "rule of inference" of its very own, or more exact- ly, its "rule of inference" contradicts a principle Beardsley enunciates. In defining a fallacy relative to its peculiar' 'rule of in- ference" Beardsley betrays a formalistic view of what constitutes "good argument" as well as what constitutes a "fallacy". A broader conception of both need make no reference to the existence of a peculiar' 'rule of inference." For example, one could define a fallacy as an argument that seduces one into thinking that the evidence provid- ed by its premises suffices for the truth of its conclusion. 2. Beardsley's Formalization of Analogical Reasoning. Beardsley first formulated the principle of analogical argument in this way: "If X and Y have a number of characteristics in common, it is likely that any further characteristics found in Y will also be found in X." (PL 108) Few if any analogical arguments make such a sweeeping claim. He subsequently toned down the conclusion, schematizing the argument as follows: X and Y both have certain characteristics a, b, c ... And it is known that Y also has another characteristic q. Therefore: X has the characteristic q. (TS2 65) His reason for calling the analogical argument fallacious remains constant, a thesis we may call the Principle of Infinite or Indefinite Variability, italicized by me below: .. .if two things have a good deal in com- mon ... the likeness may justify a further in- vestigation to see whether they actually do have more in common. But it does not justify our believing that they have more in com- mon without the further investigation. No matter how many characteristics a pair of things have in common, there may be any number of other ways in which they are dif- ferent. (PL 108, TS2 66) With a little logic-chopping it is quite true that any two things, even identical twins, can readily be shown to have an in- definitely large number of different characteristics. But this logical point is ir- relevant to whether there are significant similarities between the things analogized on which a reasonable judgment affirming an additional similarity may be based. In- ductive reasoning is useful because observ- ed similarities enable us to to make predic- tions about the future, which eventually are confirmed or disconfirmed. In holding that analogical reasoning is fallacious because a conclusion may tum out in fact to be false, Beardsley applies a stricter standard to it than to ordinary in- ductive reasoning-one appropriate to deductive reasoning. A "good" deductive argument has both 1) a valid logical form and 2) true premises, so that the truth of the conclusion follows necessarily from the truth of the premises. In effect, he accuses the analogical argument form of being in- valid, so that the truth of its premises does not guarantee the truth of its conclusion. But an analogical argument need not, and usual- ly does not, make that deductive claim. Beardsley's formalization of the analogical argument tempts one to suppose the assumed resemblances a, b, c and the inferred resemblance q are independent variables. Yet Beardsley's own definition of an analogy points out that they are all interconnected: An analogy is an "extend- ed simile" in which two things are com- pared in terms of the structural relationships holding between their characteristics or part-whole relationships (PL 105-6). This interrelatedness among the resemblances mentioned in the premises tends to justify an inference about a further resemblance. 4 Like inductive reasoning generally, an inductive analogy is not based merely on its stated premises but on all our knowledge about the world. The grounds for the argu- ment include our independent knowledge about the things analogized and the characteristics attributed to them. Conse- quently the following argument, cited by Beardsley, would hardly be taken seriously: John's parents read Greek, and so do Jim's; John likes horseradish; therefore, Jim must like horseradish. (TS2 66) We are aware that a person's eating tastes are quite independent of parental language skills, and would not credit any such argument. On the other hand, the following argument is supported by our knowledge of the genetic correlation be- Beardsley on Analogy 187 tween blue-eyed parents and their offspring: John's parents both have blue eyes, and so do Jim's; John has blue eyes; therefore, Jim must have blue eyes. (TS2 66) Beardsley's comparison of these two arguments shows that they have the same logical form, but that form is of such a kind that the premises may be true without the conclusion also being true, that is, the argu- ment is formally invalid. In the first, we would not be surprised by the falsity of the conclusion; in the second, we would be. If we insist on interpreting the argument from analogy as making a deductive claim, which requires a valid argument form as well as true premises, we would indeed have to call it fallacious. But the issue is, should we view' 'the analogical argument" as making a deduc- tive claim that the conclusion follows necessarily from the premises? Since analogical arguments are based on the characteristics of the referents of the two terms analogized, the semantic interpreta- tion of the terms drives the argument rather than the logical form of the argument statements. The "must be" can reasonably be understood as the less insistent "most likely" of an inductive claim. An invalid deductive argument readily converts into a respectable inductive argument, as many in- troductory logic books say. If this is the case, an analogical argument, rather than being inherently fallacious, has more or less weight depending on what we know about its actual terms and their known or infer- red resemblances. Another misleading feature of Beard- sley's formalization of the argument from analogy is the specification of its premises as a comparison between a pair of things. In standard formulations of inductive analogy arguments, a number of previous- ly observed instances, YI .... Yn are com- pared with a new instance X.5 So stated, the truly inductive character of an argument is visible. The inductive canon applies to an argument reaching the singular conclu- 188 Evelyn M. Barker sion that a new like instance has an addi- tional characteristic q as well as to one in- ferring a generalization that a class of things with defining characteristic a has the fur- ther characteristic q. The canon includes, for example, the rule that the greater the number of previously observed instances, the more likely the conclusion is to be true; that a known connection between the pro- perty q and characteristics a, b, c increases the probability of the conclusion. Once the inductive canon is seen to be relevant to an argument from analogy, it is difficult to hold that it is by nature fallacious. Like other forms of inductive argument, its conclusion may turn out to be false, even though the inductive reasoning contains no inductive fallacy. For example, suppose you observe that my husband and I both have brown eyes. By sound induc- tive principles, you would reasonably con- clude that our two sons also have brown eyes. In fact, the younger is blue-eyed. But the falsity of the conclusion does not show any fallacy in the inductive reasoning. It is much more likely that his eyes would be brown than blue. The inductive canon contains a special rule for arguments of inductive analogy. They reach a singular conclusion (not a generalization) that an individual Y, (not a class of things with characteristic y) shares a further characteristic q with the individuals XI .... X n. Now, the greater the resem- blances between this individual Y and the other individuals XI .... Xn described in the premises, the stronger the inductive argu- ment. For the more characteristics known to be common to individuals Y and XI .... Xn, the more likely the inference that individual Y will have the further characteristic q that individuals XI .... Xn are already known to have. For example, if you observe that Mrs. Smith, her two daughters and her mother are all blue-eyed and blond, you may well expect that Mrs. Smith's sister is also blue- eyed and blond. If told that Mrs. Smith's sister is her twin, the resemblance between them is increased, and the likelihood of the conclusion that she is also a blue-eyed blond is more probable. If told that Mrs. Smith's sister has a different mother, however, then the degree of resemblance between sister and Mrs. Smith, her mother and two daughters is decreased. The likelihood of the conclusion that the sister is blue-eyed and blond correspondingly decreases. Beardsley denies this: "You can't even say that the more known resemblances there are between X and Y, the more likely it is that X will have any further characteristic found in Y." (P L 108). His skepticism is due to the Principle of Indefinite Variability of individuals on which he grounds the fallaciousness of analogy. This diffidence goes counter to common sense and the in- ductive canon even with respect to a pair of things. In summary, Beardsley's rejection ofthe argument from analogy is grounded on an abstract formalization of the argument and an abstract logical principle. Since the argu- ment scheme is formally invalid, it falls out- side "good" deductive reasoning. Because of the abstract logical principle about the diversity of individuals he artificially limits induction from like instances to a conclu- sion about a class, throwing the argument from analogy outside inductive reasoning (TS 4 Ch. 2).6 Since deduction and induc- tion are, in Beardsley's view, the only two "good" forms of reasoning, an argument from analogy has no place among legitimate reasoning strategies (TS2 26-31). In order to be respectable, it must find rental accom- modation with a generalization. 3. The "Hidden Generalization" Thesis. Beardsley claims that a plausible argu- ment from analogy is a confused version of some other argument form. He explains it is often convertible into a deductive argu- ment, whose major premise is a true generalization. Both terms of the analogy are instances of the same generalization. The analogical argument itself is logically dispensable, once the unstated generaliza- tion is recognized. Suppose I say, 'John's parents both have blue eyes, and so do Jim's; John has blue eyes; therefore, Jim must have blue eyes.' Now, taken as it stands, this is an argument from analogy, and evidently a very feeble one. (TS2 66, my italics) The reason for its being "feeble" is that the conclusion, 'Jim must have blue eyes,' is "wildly jumped to" from the premises about the eye color of John, his parents and Jim's parents. Beardsley defends this by pointing out that it has the same form as the other analogy argument citing parental and child characteristics known to be unrelated, that is, the parents reading Greek and the son liking horseradish. According to Beard- sley's analysis both arguments from analogy are fallacious, because of their common form. But if an argument from analogy is not a deductive argument, their common form is neither grounds for nor against the sound- ness of the argument. In the case of a nondeductive argument, the form only serves to identify the kind of argument. Whether an argument is sound depends on substantive matters, not merely on the form of the argument. A reasonable person takes account of the content of an argument, not just its form; she would judge the arguments differently because of the varying contents of their premises and conclusion. Beardsley attributes the deceptive plausibility of an analogical argument to a "hidden generalization": "When we make that generalization explicit, we can throw away the rest of the analogy." (TS4 113).7 The original fallacious analogical argument is reduced to a "good one", a deductive argument in which a previously tested generalization is applied to a new instance, as follows: John's parents both have blue eyes, and John has blue eyes. All people whose parents both have blue eyes are people with blue eyes. Beardsley on Analogy 189 But Beardsley's account is mistaken: Here the generalization is not an implicit assump- tion of the reasoner, simply made explicit. If a reasoner has independent grounds for supposing the generalization true, there is no point in offering weaker premises about particular instances. The generalization can- not be merely "hidden." Only when a covering generalization is not known to be true does it make sense to offer statements about particular instances as premises in lieu of the generalization. When its singular con- clusion is derivable from a "previously tested generalization, " the argument from analogy rather than being reducible to a "good" deductive argument, is superfluous. Beardsley's ' 'hidden generalization" thesis underlies a recent text's presentation of "a new way to cast arguments by analogy in schematic form." 8 The argument is divided into two steps, the first an "induc- tive step", in which premises about one term of the analogy are evidence for an in- ductive generalization. In the second "deductive step", this generalization is ap- plied deductively to the other term of the analogy as a new instance falling under the generalization. According to this schematization, the "hidden generalization," instead of being "previously tested" (as Beardsley holds), is itself proved in the course of the argu- ment, then is used to prove the conclusion together with the premise about the other term of the analogy. Beardsley did not so misread the argument claim: He correctly fixed the distinctive feature of the argument by analogy as the direct comparison of the structural resemblances between two wholes. Both views err in introducing a generalization as an essential component of the analogical argument. The argument from analogy is useful in the absence of a "previously tested generalization" justify- ing the conclusion. The success of a series of arguments by analogy about particular instances could suggest some generalization 190 Evelyn M. Barker about a class of things with some characteristic they share. Beardsley's exam- ple, 'All people whose parents have blue eyes are people with blue eyes,' achieves the status of "previously tested generaliza- tion" partly because of verified predictions about the offspring of blue-eyed persons, as well as Mendelian genetic theory of recessive and dominant heritable characteristics. Few successful arguments from analogy are clumsy attempts to say what a good deductive argument contain- ing a previously tested generalization does better. Beardsley, I suspect, was influenced by contemporary philosophy of science in the theory of explanation. In order to justify belief in the truth of a singular conclusion, he supposes there must be a true inductive generalization (known or unknown) in- tegrating singular premise statements about particular instances with the singular state- ment asserted in the conclusion. He appears to be appealing to a Hempelian "covering- law" model of explanation for singular con- clusions, in judging an inductive analogy argument in need of a true "hidden generalization. " In his own view a true singular statment (not proved deductively) is inductively justified only as the best explanation for a diverse set of facts (PL Ch. 2, #10). Only a generalization is to be inferred from premises about like instances. This theoretical bias generates a cramped view of a proven useful form of reasoning. In practical life, we often make singular predictions without the help of "covering laws" or generalizations. Let me give a homely example familiar to us academics. At some point in a course a stu- dent will ask an instructor her prospects of passing, (or receiving A or B, and so forth). The instructor has a plentiful supply of generalizations for this purpose. But the problem is that one and the same student instantiates a number of different generalizations leading to conflicting con- clusions: She is diligent, but her previous work has shown little aptitUde in the sub- ject; she is highly motivated, but is emo- tionally unstable; the future assignments are not difficult but she is undertaking longer work hours, and so on. An individual (as Beardsley reminds us in his Principle of Indefinite Variability) may show diverse characteristics whose in- terplay in special circumstances must be assessed because they point different ways. This is not achieved simply by adding generalizations, for the applicable generalizations would warrant different judgments, inconsistent with one another. The experienced instructor will often make an analogy between the student before her and past students whose academic success and failure the instructor has observed. The analogy is the basis for a holistic judgment about the present student's potentialities in her situation. Analogical reasoning is effec- tive for dealing with variable combinations of traits found in individual cases, which exhibit "family resemblances," rather than some constant characteristic. The com- parison to other individuals cannot be read- ily replaced by some very complex generalization, for the aim is to reach a con- clusion fitting this particular individual in these special circumstances. The judgments of Art concerning in- dividuals, who sometimes show an anomalous set of traits, or whose characteristics may vary from a norm, will often make good use of inductive analogy, in conjunction with applicable generaliza- tions. Medical diagnosis and treatment often involves such analogical reasoning. A pa- tient presents conflicting symptoms, or symptoms suggesting different diagnoses. Or her medical condition includes several problems; for one, a type of therapy is in- dicated, for another, that same therapy is counterindicated. In distinguishing signifi- cant from less important factors, a physi- cian's accumulation of experiences with analogous cases is often a patient's best security for accurate diagnosis and beneficial treatment. 3. The Phenomenology of Noninductive Analogy. Many arguments present an analogy be- tween two types of thing, sometimes from different categories: the ancient analogy be- tween the "body politic" and the human organism, and Judith Jarvis Thomson's con- temporary analogy between a woman be- ing pregnant and being plugged into an ail- ing musician as his kidney dialysis machine are well known. Beardsley's thesis that the argument from analogy is a fallacy is less paradox- ical with such analogies, for the difference in type of things analogized means that it is naive to infer a factual conclusion about one type merely from its analogy with the other. His view that the argument from analogy is a fallacy parallels his assertion that a metaphor is always literally false (PL 101). The former is, of course, formally in- valid, but only the most unsophisticated reasoner would credit an argument based on analogy as formally valid, just as one who thought a metaphor literally true would miss its point. S. F. Barker calls this kind of analogy "noninductive," for reasoning based on it does not issue in a conclusion verifiable in principle by subsequent experience. 9 Its role is phenomenological, as Beardsley su~­ gested in his first treatment of analogy III a chapter on figurative language (PL Ch. 2), where he observed that such analogies "put an object in a new light." (TS4 112). Beardsley approved the nonargumen- tative uses of such analogy, noting its capacity to explain abstractions in terms of concretely perceptible things, as the dependence of the individual upon the com- mon good appears in the analogy between body parts and divisions of the state. He elo- quently sketches the phenomenology of their use in science, illustrating or clarifying a general principle, and even suggesting a testable hypothesis: It helps us to understand the nature and fimc- tion of the blood stream to think of it as 'the Beardsley on Analogy 191 internal environment,' the bit of ocean ... that the early creatures carried with them when they first undertook to live on land ... When Nils Bohr constructed his model of the atom as a miniature solar system, with the elec- trons revolving around the nucleus, this pic- ture led scientists to think of many ideas they could test experimentally. (TS2 64) Noninductive analogy as "the favorite in debate and polemic" plays a role different both from such heuristic uses and inductive analogy. It is usually directed to locating or relocating a thing in our world-view. 10 Beardsley's common formal characteriza- tion of all kinds of analogy arguments leads him to misconceive the phenomenological goal, as we can see in his analysis of an argument making an analogy between por- nography and an infectious disease. (TS4 112-15) Here the writer is not merely presenting an analogy between pornography and disease: that each is harmful, that neither has positive value, that one can be 'infected' by either, that both 'infections' can be spread, etc. He is presenting an analogy to support a con- clusion about pornography .... In the present case the property p is perhaps somewhat unusual: It is the property of being something that ought to be eliminated. Beardsley's formalist approach makes him interpret the argument as a straightfor- ward argument reaching a conclusion whose validity depends on the plausibility of the hidden generalization: "Whatever does harm but no good should be eliminated." With this assumption made explicit, the argument becomes a deductive argument in which the analogy of pornography with an infectious disease makes "no logical con- tribution to supporting the conclusion." Pornography does harm but no good. Whatever does harm but no good should be eliminated. Pornography should be eliminated. In trying to capture the argument's "logic," Beardsley has overlooked that the analogy serves a phenomenological func- tion, directed to making us see pornography 192 Evelyn M. Barker in a way that will affect our moral attitude toward it: not as an innocent and harmless private pleasure, but as an insidious threat to self, spreading to others. The "somewhat unusual property" disguises this practical phenomenological aim in the predicate of the argument's conclusion, in Beardsley's awkward reduction. The "hidden generalization" is a vapid ineffective truism, that does not replace the provocative analogy of the original argument either logically or phenomenologically. Beardsley's condemnation of analogy arguments incites him to offer three helpful strategies for revealing its fallaciousness: 1) Weaken the analogy by questioning the assumed resemblances and pointing out differences between the two types of things. 2) Expose the invalidity of the' 'hidden generalization" of the reduced argument that justifies the inference. 3) Extend the analogy "to the point where it boomerangs," leading to an ab- surd conclusion (PL 110-112). But these methods often fail to blunt the phenomenological point of the analogy. For example, Thomson's analogy provides a powerful defense of abortion because it por- trays pregnancy as an involuntary offensive intrusion in a woman's life, a violation of her own dignity that makes abortion look justified in a moral scheme stressing in- dividual rights and voluntary personal rela- tionships. II It is easy to attack Thomson's analogy as Beardsley instructs, by pointing out that the natural function of a woman's womb is to nourish the fetus, and that it does so without impairing a woman's life or dignity, while no human body is designed to perform kidney dialysis for another adult human. Beardsley attributes the strength of an analogy to the degree of isomorphism be- tween the things analogized, as a map represents an area's geographic features (PL 106). If so, Beardsley's three strategies would always work. But the power of an analogy often lies in its depressing some features of a thing, while simultaneously highlighting others, to make it fit smoothly into a specific world-view, eliminating trouble-making bits. Thus the fetus, in the Thomson analogy, is no longer an innocent dependent being biologically linked to the woman, but an unwelcome exploitative alien. It is sometimes a whole world view, rather than a "hidden generalization, " that empowers the analogy. A counteranalogy is sometimes more ef- fective in dealing with the phenomenolog- ical use of analogy. In a recent legal case involving the fate of frozen embryos belong- ing to a couple engaged in divorce pro- ceedings, competing analogies with joint property and human lives dominated the reasoning: A doctor supporting the implan- tation of the embryos in the biological mother exploits the analogy with the' 'tiny human": "Putting tiny human beings in a very cold space, deprived of liberty, depriv- ed even of time, they are suspended as it were in a concentration camp. It is not as hospitable a place as the secret temple of a woman's womb. "12 4. Reprise: Is There a Fallacy of Denaturalized Epistemology? Despite his negative estimate of the argument, Beardsley took analogical reasoning seriously, more seriously than a host of textbook writers in this acme of in- formal logic. He advanced our understan- ding both of its powers and limitations. He appreciates the fruitfulness of analogies in scientific thinking, because they suggest working hypotheses capable of empirical proof. He denigrates appeals to analogy in polemical writing, where verification is not in order, advising the reasoner to attack the weakness of the premises or ridicule the conclusion by reducing it to an absurdity. He is silent, however, on the presence of analogies in reflective legal and philosophical reasoning where they are in- troduced within an extended course of argument, as in Thomson's paper. This gap reveals the deficiency of Beardsley's theory of argument: It looks to logic and empirical science for the two models of "good" deductive and inductive argument respectively. Logic and empirical science deal with subject matter abstractable from its setting in the concrete circumstances of human life. When the subject of the discourse is not so "clean," one often deploys a number of dif- ferent thinking strategies to bear on the mat- ter at hand, in a way that resists formaliza- tion. Arguments using analogy comprise one such strategy. In foregoing sections I point out how analogical reasoning func- tions successfully in a number of situations-discovering historical precedents for prudent action, predicting individual outcomes, evoking morally significant features of a thing, classifying unusual or novel phenomena. Beardsley's theory of argument leads him to concentrate on the form of an argu- ment in search of some valid principle of inference which negotiates the step from premises to conclusion. Since arguments us- ing analogy are diverse, Beardsley's uniform analysis and common abstract for- mulation, by confounding them, makes his condemnation of the argument type credi- ble. Since its "principle of inference" violates the abstract Principle of Indefinite Beardsley on Analogy 193 Variability, he rejects it from the canon of "good argument." Consequently, when a "critical thinker" meets an argument by analogy, he charitably looks for the' 'hid- den generalization, " or rigorously follows the rules to expose the fallacy. Like many of us, Beardsley's aim was to make principles of inference discovered by philosophers accessible to nonlogicians, so they can do "critical thinking." "Critical thinking" can always produce the "rule of inference" that legitimates each step of a piece of reasoning, after one has schematiz- ed it to reveal the form, deductive or in- ductive. Only such arguments are "convinc- ing," in opposition to the hodge-podge of reasoning strategies characterizing the "rhetorical point of view," which has the power but not the right to "persuade." (TS4 1-6) Is this "critical thinking" model of good reasoning applicable to all thinking we rightly admire? Does it express an elitist conception of "good argument" which favors abstract reasoning, overlooking reasoning that does not move in a linear lockstep or fit the twin Procrustean beds of deductive and inductive argument? When we persuade our students that only "critical thinking" is good reasoning, and immunize them with a list of fallacies to ward off arguments that merely "persuade," are we committing the fallacy of "denaturalized epistemology"? Notes I In this paper I discuss Monroe C. Beardsley's treatment of analogies in Practical Logic, 1950. New York: Prentice Hall; Thinking Straight. 2nd edition. 1956 and 4th edition. 1975. Englewood Cliffs: Prentice Hall. I refer to the first as PL; the other two as TS2 and TS4. 2 ••... our generation faces a problem similar to that which engaged statesmen 70 years ago- how to construct a stable international order. Most of the postwar period has been characterized by a relatively stable European equilibrium. Now that two-power world is disintegrating-more in the East than in the West.'· Article printed in The Baltimore Sun. Monday. August 28, 1989. 7 A. 3 For example, I distinguish arguments of induc- tive analogy, non inductive analogy. using an analogy to support a moral judgment or defend a classification, in Everyday Reasoning. Englewood Cliffs: Prentice-Hall, 1981. 4 By "tends to justify" I mean the premises about 194 Evelyn M. Barker observed similarities always or mostly confer some plausibility to the supposition of an addi- tional similarity. Much depends on whether the further resemblance is independently known to be causally or structurally related to the observ- ed similarities. 5 Note a contemporary formulation in Irving M. Copi, Introduction to Logic, 2nd edition, New York: The MacMillan Company, 1956 (also in later editions). 6 Beardsley makes inductive generalization and explanatory hypothesis into mutually exclusive types of inductive reasoning: But a generaliza- tion, as well as a singular judgment, may be regarded as an explanatory hypothesis for a set of data; and each may function predictively as a conclusion from like instances. 7 S. F. Barker points out Beardsley-type inter- pretations of analogy arguments as implying a generalization make it commit a non sequitur or circular reasoning in Elements of Logic, 5th edition, New York: McGraw-Hill, 1989, n. 46, p.282. 8 David Kelley, The Art of Reasoning with Sym- bolic Logic, New York: W. W. Norton & Company, 1990, ch. 16. 9 S. F. Barker, Elements, pp. 225-228. 10 Howard Kahane stresses the importance of "background beliefs" and "world-views" in more recent editions of Logic and Contem- porary Rhetoric, Belmont: Wadsworth Publishing Company. 11 1. J. Thomson, "A Defense of Abortion", Philosophy and Public Affairs I (1971). 12 Dr. Jerome Lejeune, quoted in various news reports of The Baltimore Sun, August 1989. EVEL YN M. BARKER PHILOSOPHY DEPARTMENT THE UNIVERSITY OF MARYLAND BALTIMORE COUNTY CAMPUS CATONSVILLE, MD 21228 o