—Hume – Reasoning to Cause or Effect
Any and all things are situated. Billiard balls are in situations. The situations are in situations too. The weather is in a situation no less than a billiard ball. The range of situations in which a thing might be is a concomitant of a thing’s identity. The range of situations in which a thing might be is learned from one’s total experience. The task of sorting out what is what and what can become what in what situations is undertaken not only by scientists and philosophers (Armstrong 1991, 137–39), but by everyone, child and adult (Keil 1989, 159–215).
Hume would want to emphasize that learning from experience is a matter of forming habits. He tried to squeeze far too much into that mode of learning. (On Hume’s attempt to explain generalization in terms of habit, see Stroud 1988, 38–41. Habituation accounts of conceptual knowledge have not fared well in recent years; see Rips 1989 and Jackendoff 1987, 143–48.) In his earlier work, the Treatise, Hume seems to be cognizant of complications. He allows that we sometimes reason about causes and their effects not only according to custom arising from similar past conjunctions, but from a principle of identity. He allows that in cases “more rare and unusual,” we may assist elementary custom by conscious reflection on past experience and arrive at “custom [belief] in an oblique and artificial manner.”
I explain myself. ’Tis certain, that not only in philosophy, but even in common life, we attain the knowledge of a particular cause merely by one experiment, provided it be made with judgment, and after a careful removal of all foreign and superfluous circumstances. Now as after one experiment of this kind, the mind, upon the appearance either of the cause or the effect, can draw an inference concerning the existence of its correlative; and as a habit can never be acquir’d merely by one instance; it may be thought, that belief cannot in this case be esteem’d the effect of custom. But this difficulty will vanish, if we consider, that tho’ we are suppos’d to have had only one experiment of a particular effect, yet we have many millions to convince us of this principle; that like objects, plac’d in like circumstances, will always produce like effects; and as this principle has establish’d itself by a sufficient custom, it bestows an evidence and firmness on any opinion, to which it can be apply’d. (T I.3.8)
We might want to qualify and refine the general principle to which Hume appeals, the principle that “like objects, placed in like circumstances, will always produce like effects,” but it is at any rate clear that Hume is here squirming out of his official position that knowledge of fact is simply a matter of custom or habit.[19
] Also in the Treatise,
when writing on our knowledge of the continued and independent existence of bodies (T I.4.2) and when writing on our understanding of why things, e.g., clocks, may sometimes behave one way and sometimes another (T I.3.7), Hume senses the inadequacies of his habituation account of inductive inference and tries to make accommodations. In his later work, the Enquiries,
he deals with these complications either by not bringing them up or by offering only meager “hints “ of their solution (E 47).
Hume’s commonsense principle that same causes yield same effects was endorsed also by Aristotle: “It is a law of nature that the same cause, provided it remain in the same condition, always produces the same effect” (GC 336a27-28
). Ockham endorsed the principle in a form close to Hume’s: “Causes of the same kinds are effective of effects of the same kinds” (Weinberg 1965, 142
). Ockham took this principle to be necessary and self-evident. As the principle is formulated by Ockham or Hume, it is subject to two interpretations. One, a broad
one, I shall endorse in a moment. The other—and this is what both Ockham and Hume (E 64) most likely meant—is just the principle as stated without ambiguity by Aristotle. I think we should be wary of Aristotle’s principle. Hereafter, I shall refer to it as the narrow
mode of causality. Although it obtains throughout vast regions of our experience, throughout much of existence, it evidently does not obtain for physical processes in quantum regimes nor in classical chaotic regimes. I suggest we reformulate the principle more broadly, thus: “Identical existents, in given circumstances, will always produce results not wholly identical to results produced by different existents in those same circumstances.” Application of the law of identity to action or becoming would seem to require only this much (contrary to Peikoff 1991, 14–15
It is not always the case that identical things placed in the same circumstances yield a single (repeated) result. Some existents yield single distinctive results; others yield distributions of distinctive results. Only if one allows for the latter possibility in one’s construction of the principle that “same causes yield same effects” is it universally true. Within the realm of classical mechanics, which covers billiards, Hume is correct in saying that same causes yield same (single) effects.
Out of all the conditions that obtain in a situation, we typically take only one or a few as cause of some distinctive result, only a select portion of the ways in which the law of identity applies to an action or a becoming (Minsky 1988, 129; van Fraassen 1991, 318–27; Gasper 1991, 293–95
). We try to discover among antecedent conditions ones that will make a certain result under a wide range of variations in the remaining variable conditions. In a general commonsense way, everyone knows that characteristic of causes. Then they know, though perhaps only dimly, the principles of induction enshrined by John Stuart Mill: the principle of agreement, the principle of difference, and the principle of concomitant variation (Mill 1973, 3.8; Copi 1961, 363-407; Kelley 1988, 276–87
). Aristotle knew something of these principles (Phys. 199b15–20; Top. 146a2–12
); more so did Robert Grossette, Albert the Great, Duns Scotus, Ockham, and Francis Bacon (Weinberg 1965, 136–45, 153
). That these are effective techniques for arriving at causes is knowable from everyday experience and all the more from modern scientific practice.[21
] To rely on these techniques is to affirm the principle of identity operative. In reasoning to causes, we follow most conspicuously not habit but identity.
In a primary sense, causes make things happen.[22
] We witness one billiard ball setting another in motion. We would usually say that what happened was that one billiard ball struck the other and set it in motion. According to Hume, we have gotten the idea that the first ball made the second move by custom of having seen like conjunctions
of billiard balls in the past. Then why could we not as well say that the second billiard ball’s acceleration is what causes the first ball to strike it rather than what we usually say? (In some circumstances, young children do take effects to precede causes; Bullock, Gelman, Baillargéon 1982, 216
.) Perhaps we are more often interested in prediction than postdiction; animals are surely more oriented to the future than the past. Nevertheless, when we look only backwards, why can we not believe that the acceleration of the second caused the first to strike it? I suggest it is because we have direct experience of causal powers, contrary to the fabulous sayings of Hume (E 50), and we know where they lie in this situation (Boyd 1991, 355–66
). The idea of causality is more than the (tightly fastened) idea of regular conjunction in regular temporal order even for the simple billiards case.
Hume acknowledges that we think of causes as having some sort of necessary connection with their effects. This sense of necessity he supposes to be a psychological compulsion arising from our past observations of repeated conjunctions between causes and their effects. “We immediately feel a determination of the mind to pass from one object to its usual attendant. . . . Upon the whole, necessity is something, that exists in the mind, not the objects. . . . Necessity is nothing but that determination of the thought to pass from causes to effects and from effects to causes, according to their experienced union” (T I.3.14). As Hume would have it, even the necessities we find in the interactions of our bodies with other objects are merely projections from necessities we find in our mental operations (T I.4.4). This is absurd. (This error has more subtle relatives, e.g., the thesis of Aquinas that necessity in affirmation of existence is gotten from necessity in intelligibility; see Hoenen 1952, 165–67, 179–81
By about three years of age, we have grasped the general commonsense adult principles of causality; we understand that occurrences have causes and that these causes come down to specific mechanisms. The striking of one billiard ball by another is perceptually salient. It is a sharp event. We naturally would seek a trigger of such an event, and in this case, what we would seek is readily apparent. All are agreed that the second ball’s being set in motion had a cause, namely, the first ball’s striking it. This is a causal explanation at the level of common sense. Even at this level, though, not all causal relations are so obvious. That the snowman melted because it became too warm or that the living room became filled with smoke because the damper was not open are a little less obvious. Identification of these causal relations requires, in addition to specific experience, a little more causal reasoning. Such reasoning is a major mode of induction.
It is not from more repetitions of experience, but from a more extensive network of experience and from greater skill in reasoning on causal principles, that we become capable of composing commonsense causal explanations (Bullock, Gelman, Baillergéon 1982
] Contrary to the implications of Hume’s model, repetition of experience is not the main influence in our identification of causes. Hume recognized the importance, for humans, of causal reasoning, but he managed to not notice that this is at variance with his insistence on the preeminent role of habit (E 84n1).[24
When one comes to formulate ideas about inanimate motion more generally, things become much
less obvious, and customary experience can put one in the wrong frame of mind. It was very difficult for man to get straight which motions needed to be explained, which motions have causes in the primary sense. Today, students of physics learn the answer when they are taught the law of inertia, the law that a body will continue at constant speed and in a straight line (or will remain at rest) unless acted upon by a force. At a more advanced level, students learn that and how the inertia principle has been recast in more general forms: in the Lagrangian mechanics, as an extremal principle which tells how a body will move when or when not subjected to external forces; and in general relativity, as the principle that free bodies travel along geodesics of spacetime whether curved or flat.
It is only the elementary form of the law of inertia that concerns us here. Aristotle and his followers held to principles contrary the inertia principle. Terrestrial bodies, when moving, naturally tend to move towards certain places of repose (Phys. 199b14–19, 208b9–15, 215a1–21, 230a19–231a17
). The heavenly spheres, upon which ride the heavenly bodies, naturally and always move in circular ways. Here let us confine attention to terrestrial bodies whose natural direction of motion is towards the earth: the earthly bodies, not water, air, nor fire. Aristotle’s camp took the free fall of earthly bodies as natural and as standing in no need of special explanation; no external force is being applied to keep such bodies falling to the earth. Any other motion of an earthly body, any motion that is not free fall, needs special external explanation. Moreover, any motion at all requires some explanation. “Everything that is in motion must be moved by something. For if it has not the source of its motion in itself it is evident that it is moved by something other than itself, for there must be something else which moves it” (Phys. 241b34–36
What could be more sensible? In our life experiences here on the surface of the earth, we have countless confirmations of Aristotle’s thesis every day. To get an object moving requires effort, to keep it moving requires effort, and the object will sooner or later return to rest. The strings of the harp will return to silence. For Galileo and his followers to propose that motion, provided it be uniform, required no explanation, no efficient cause, but that non-uniform motion, including coming to rest, did, they had to put on new thinking caps. (Although it does not affect my argument, I should note for historical accuracy that it was not until Descartes that uniform motion was surely taken to be along a straight line; Galileo took it to be along a circle of constant radius, with center at the center of the earth.) These men could not leave the answer to habitual experience (nor to tradition). Formulating physical principles simply according to the most usual observations would not have led men to the law of inertia. Until this law and its conceptual vantage were discovered, the scientific revolution could not happen (Butterfield 1965, 14–28, 67–85; McClosky 1983
In common sense and in most scientific reasoning, we make the tried and true presumption that occurrences have causes. Hume discussed this principle in the venerated form “whatever begins to exist, must have a cause of existence.” He contended that this principle is not a necessary truth. He may have actually doubted the truth of the principle (T 78–82). We should distinguish two interpretations of the principle. In one we take cause as material cause, and in the other, we take cause as efficient cause. As to material cause, the principle seems to have held up perfectly in the two and a half centuries of science since Hume. It holds for all elementary particles; every particle gets made from some others. As to efficient cause, the principle holds always for cause in the broad mode; each type of elementary particle has its distinctive ways of coming about. (It seems to me that Kant’s defense of the principle as pertaining to efficient causality, in his celebrated Second Analogy, succeeds for the broad mode, but not for the narrow; Kant 1965 A190-211 B233–56; Brittan 1978, 170–71, 181–82
.) Again as to efficient cause, the principle evidently does not
hold in the narrow
mode for elementary particles. There is no narrow cause of a particle decay, so there can be no narrow cause of the decay products. Remember, too, the proverb of particle physicists: “Seek not reasons for decay, but seek the barriers to decay.” At the level of elementary particles, we seek reasons for stability (Frauenfelder and Henley 1974, 83–87; Sachs 1987, 100–103, 175-77; Weinberg 1981
We have observed that Mill’s methods of induction—agreement, difference, and concomitant variation—are essential to the growth of scientific knowledge and that they are clearly wedded to a fundamental law of all existents, the law of identity. We have observed also that Hume’s proposed basis for induction—habit—makes no suggestion as to why those techniques should be effective, why they should expose new regularities, orders, and unities of nature. Hume’s account also leaves utterly opaque that great engine of scientific discovery, the hypothetico-deductive method (Copi 1961, 433–51; Hempel 1966, 10–28; Kelley 1988, 344–65
From the statement of a hypothesis (say, Newton’s law of gravitation), together with some established truths (e.g., the orbit of Uranus) and plausible presumptions, we deduce consequences (a planet beyond Uranus). On the character of confirmation, see Mackie 1981; Newton-Smith 1981, 183-97, 226–32; and Armstrong 1991, 41–46
. Both the hypothesis and its consequences are claims about the world. Hypotheses are arrived at by some mix of induction and imagination. We really do not know all that much about how hypotheses are formed (Reilly 1970, 36–38; Hempel 1966, 15–18; Drake 1980; Cohen 1981
). We really do not know that much either about how one selects relevant established truths. I assume that not all are selected ahead of the grasping of consequences, but my argument does not depend on this. We do know how the consequences are constructed, or at any rate, how they can be re
constructed (Minsky 1988, 186–89
). The consequences are deductions from the hypothesis in combination with select established truths. Now consider the case of a hypothesis that has been subsequently well confirmed by observation of predicted, deduced consequences. On Hume’s model of induction and causality, is this not a minor miracle? Why, on Hume’s view, should deduction yield consequences about what is in the world?
On the view that identity is a deep and general law of reality, the success of the hypothetico-deductive method is intelligible. Validity of our deductions assures us in some measure that we are not in opposition to the universal law of identity. Presumably, that is why validity is desirable. From the vantage of Rand’s principle of identity, we have some idea of why the hypothetico-deductive method works. It is marvelous but not miraculous.
In Hume’s view, as in Ockham’s, the existence of one object cannot be inferred from the existence of another (E 132). Hume contends that, excepting mathematical objects, nothing can be demonstrated of any objects of reasoning (E 131). “All belief of matter of fact or real existence is derived merely from some object, present to the memory or senses, and a customary conjunction between that and some other object” (E 38). Yet in reasoning from a hypothesis to an observable, but as yet unobserved consequence, we can hardly be relying merely on custom. Apparently, Hume never squarely confronted the hypothetico-deductive method. He speaks of “hypothetical arguments, or reasoning upon supposition,” but in these he says there is no “belief of a real existence” (T I.3.4; E 37). Insofar as Hume does begin to consider scientific reasoning, he turns from custom to identity. He may boldly proclaim that “all inferences from experience, therefore, are effects of custom, not of reasoning” (E 36), but in a note, he remarks that scientific knowledge cannot be established purely by experience, but requires “some process of thought, and some reflection on what we have observed, in order to distinguish its circumstances and trace its consequences” (E 36n1; see also E 84n1).Continued below—
—Hume – Necessity
—Hume – Uniformity
—Existence is IdentityReferences
Aristotle 1984 [c. 348–322 B.C.
]. The Complete Works of Aristotle
. J. Barnes, editor. Princeton. Princeton University Press.
Armstrong, D. M. 1991 . What Is a Law of Nature?
Cambridge: Cambridge University Press.
Brittan, G. G. 1978. Kant’s Theory of Science
. Princeton: Princeton University Press.
Boyd, R. 1991 . Observations, Explanatory Power, and Simplicity: Toward a Non-Humean Account. In Boyd, Gasper, and Trout 1991.
Boyd, R., Gasper, P., and J. D. Trout, editors, 1991. The Philosophy of Science
. Cambridge, MA: MIT Press.
Bullock, M., Gelman, R., and R. Baillargéon 1982. The Development of Causal Reasoning. In The Developmental Psychology of Time
. W. J. Friedman, editor. New York: Academic Press.
Butterfield, H. 1965 . The Origins of Modern Science
. Revised ed. New York: Free Press.
Cohen, I. B. 1981. Newton’s Discovery of Gravity. Sci. Amer
Copi, I. M. 1961 . Introduction to Logic
. 2nd. Ed. New York: Macmillan.
Drake, S. 1980. Newton’s Apple and Galileo’s Dialogue. Sci. Amer
van Fraassen, B. C. 1991 . The Pragmatics of Explanation. In Boyd, Gasper, and Trout 1991.
Frauenfelder, H., and E. M. Henley 1974. Subatomic Physics
. Englewood Cliffs, NJ: Printice-Hall.
Gasper, P. 1991. Causation and Explanation. In Boyd, Gasper, and Trout 1991.
Hempel, C. G. 1966. Philosophy of Natural Science
. Englewood Cliffs, NJ: Prentice-Hall.
Hoenen, P. 1952. Reality and Judgment according to St. Thomas.
H. F. Tiblier, translator. Chicago: Henry Regnery.
Hume, D. 1975 [1893, 1748]. Enquiries Concerning Human Understanding
. 3rd ed., L A. Selby-Bigge, editor. Oxford: Clarendon Press.
——. 1978 [1888, 1740]. A Treatise of Human Nature
. 2nd ed., L. .Selby-Bigge, editor. Oxford: Clarendon Press.
Jackendoff, R. 1987. Consciousness and the Computational Mind
. Cambridge, MA: MIT Press.
Kant, I. 1965 [A-1781 B-1787]. Critique of Pure Reason
. N. Kemp Smith, translator. New York: St. Martin’s Press.
Keil, F. C. 1989. Concepts, Kinds, and Cognitive Development
. Cambridge, MA: MIT Press.
Kelley, D. 1988. The Art of Reasoning
. New York: W. W. Norton.
Mackie, J. L. 1981 . The Paradox of Confirmation. Reprinted in The Philosophy of Science
. P. H. Nidditch, editor. New York: Oxford University Press.
McClosky, M. 1983. Intuitive Physics. Sci. Amer
Mill, J. S. 1973 . A System of Logic Ratiocinative and Inductive
. Toronto: University of Toronto Press, Routledge & Kegan Paul.
Minsky, M. 1988 . The Society of Mind
. New York: Simon & Schuster.
Newton-Smith, W. H. 1981. The Rationality of Science
. Boston: Routledge & Kegan Paul.
Peikoff, L. 1991. Objectivism: The Philosophy of Ayn Rand
. New York: Dutton.
Reilly, F. E. 1970. Charles Peirce’s Theory of Scientific Method.
New York: Fordham University Press.
Rips, L. J. 1989. Similarity, Typicality, and Categorization. In Similarity and Analogical Reasoning
. S. Vosniadour and A. Ortony, editors. Cambridge: Cambridge University Press.
Sachs, R. G. 1987. The Physics of Time Reversal
. Chicago: University of Chicago Press.
Stroud, B. 1988 . Hume
. London: Routledge.
Weinberg, J. R. 1965. Abstraction, Relation, and Induction
. Madison: University of Wisconsin Press.
Weinberg, S. 1981. The Decay of the Proton. Sci. Amer
~~~~~~~~~~~~~~~~This essay of 1991 had no notes. I will now add a few endnotes to indicate changes or emendations to the positions I took in this essay nineteen years ago. I will also add some hyperlinks within the text.
here means established in one’s conceptual framework, as in Harriman 2010
, 31–34. Compare those pages with Critique of Pure Reason,
Bxiii–xiv. On Kant’s connection of induction to taxonomic systematization of nature in the organization of experience, see Allison 2008
, 140–51. On the connection of induction to conceptualization, in Francis Bacon and in William Whewell, see McCaskey 2004
. On abstractive induction, see here
. Allison points out that in warranting induction from experience of a single case, Hume is not only relying on his general principle that “like objects, placed in like circumstances, will always produce like effects.” Hume is also relying on his principle that every event has some cause (Allison 2008, 156
). On this latter principle, see third paragraph from last in the present section (and Allison 2008, 93–111, 137–38
Allison observes also that those two conditions are required, in Hume’s account, not only for induction based on a single case, but for virtually all inference from something observed to something unobserved (e.g., T I.3.13). Hume’s text on inductive inference from the single case (T I.3.8), which I quoted in the third paragraph, shows acutely the power of judgment as acting independently of custom concerning the case at hand. Making the right judgment concerning causally relevant factors in this single-case base for inference in subsequent cases is not helped by appeal to the general rule that like objects placed in like circumstances produce like effects. As a matter of fact, Hume’s two general rules are also not helpful in making the required particular judgment where the observed cases are multiple or numerous. Sorting objects and actions into classes according to usual, manifest similarities does not help either (Allison 2008, 159–60; see also here and Harriman 2010, 9, 31–34
Hume helps himself, in the quoted passage and in others, to the human power of judgment. That is a power Hume is not entitled to invoke, given his model of human cognition. Barry Stroud writes that Hume’s theory of ideas
obstructs proper understanding of the role, or function, or point of various ideas in our thought about the world because in representing “having” an idea as a matter of a certain object’s simply being “in” the mind, it leaves out, or places in a secondary position, the notion of judgment, the putting forth of something that is true or false. For Hume, ideas exist in the mind and have their identity completely independently of any contribution they might make to judgments or statements that have a truth-value. He sees judging as just a special case of an object’s being present to the mind. . . . He does not see that without an account of how ideas combine to make a judgment or a complete thought he can never explain the different roles or functions various fundamental ideas perform in the multifarious judgments we make, or in what might be called the “propositional” thoughts we have. Consequently, he does not arrive at even the beginnings of a realistic description of what “having” the idea of causality actually consists in. (1988, 232)20
. There were two errors in this paragraph. Firstly, unlike quantum regimes, classical chaotic regimes hold no exceptions to Aristotle’s principle (a
). Secondly, my broader formula intended to be the minimum implied by the law of identity did not quite reach the minimum, which would be: “For some given circumstance or other, identical existents will produce results not wholly identical to results produced by different existents in those same circumstances.”*
As is seen in the link, that picayune revision to the broad formula is occasioned by a consideration that applies to all physical regimes, including the classical regular regime.
The gravamen of the broad formula was captured perfectly well in my 1991 statement: “Identical existents, in given circumstances, will always produce results not wholly identical to results produced by different existents in those same circumstances.” In contrast Leonard Peikoff had maintained earlier that year that Rand’s law of identity entails the following: “In any given set of circumstances, there is only one action possible to an entity, the action expressive of its identity” (1991, 14
). Dr. Peikoff’s formula can be read as not in contradiction with mine if his phrase only one action possible
is taken to mean only one kind and range of action possible
. But that is not the plain reading of his text. In his 1976 lectures The Philosophy of Objectivism
), also, he had maintained that Rand’s law of identity applied to action entailed that only a single action was physically possible to a thing in a given circumstance. Rand gave notice that those lectures were an accurate representation of her views, so I expect she shared the erroneous view expressed by Peikoff concerning uniquely determined outcome. (That there is a unique outcome in all cases is not in dispute; the issue is whether in all cases only that unique outcome was physically possible; see my 1997 reply to Rafael Eilon, 159–62
So I expect Rand meant “uniquely determined” in her 1973 formula for the law of physical causality: “All the countless forms, motions, combinations, and dissolutions of elements within the universe—from a floating speck of dust to the formation of a galaxy to the emergence of life—are caused and determined by the identities of the elements involved” (MvMM, 25
). In any case, the error is easily corrected without major revision to her metaphysics or to its counters to Hume’s account of causation.21
. Cf. Harriman 2010, 67–71, on the first two methods, the third glossed only as a species of the second. Kelley 1988 discusses all three methods, the third on pages 283–85.22
. An ambitious attempt at developing this elementary thesis into a full-blown account of causation and explanation in science is made in Woodward 2003
. On the origins and elaboration of causal understanding in development, see also Chapter 6 of Carey 2009
along with Gopnik and Schulz 2007
. It was misleading for me to say that Hume “managed to not notice” that the importance of causal reasoning is at variance with his insistence on the preeminent role of habit in causal attributions. What he managed to “not notice” was the ineffectiveness of his gestures at reconciling that variance. That is, the variance remains.25
. Cf. Harriman 2010, 14–15, 44–46, 49–53. See further, Miller 2006
. More on the stability of matter: Lipkin 1995
and Lieb 2009
. Contrast my representation of the hypothetico-deductive method in science with its representation by Harriman (2010, 145–46
). I have not supposed that the method entails that hypotheses are mere guesswork, which is not the way the method has been employed by any research in physical science with which I am familiar. However much later philosophers of science took hypotheses to be guesswork, that was not the view of William Whewell (Snyder 2006
Rand’s theoretical philosophy and my own understanding of the methods of science are consonant with the following superbly informed view of Ernan McMullin 1992
. (See also
Let us restrict the term abduction to the process whereby initially plausible and testable causal hypotheses are formulated. This is inference only in the loosest sense, but the extensive discussions of the logic of discovery in the 1970’s showed how far, indeed, it differs from mere guessing. The testing of such hypotheses is of the most varied sort. It does, of course, involve deduction in a central way, as consequences are drawn and tried out. Some of these may be singular, others may be lawlike and hence involve induction. But we shall not restrict induction to the testing of causal hypotheses, as Peirce came to do. (89–90)
[Our concern] is with the process of theoretical explanation generally, the process by which our world has been so vastly expanded. This is the kind of inference that makes science into the powerful instrument of discovery it has become. . . . As a process of inference, it is not rule-governed as deduction is, nor regulated by technique as induction is. Its criteria, like coherence, empirical adequacy, fertility, are of a more oblique sort. They leave room for disagreement, sometimes long-lasting disagreement. Yet they also allow controversies to be adjudicated and eventually resolved.
It is a complex, continuing, sort of inference, involving deduction, induction, and abduction. Abduction is generally prompted by an earlier induction (here we disagree with Peirce). The regularity revealed by the induction may or may not be surprising. Deductions are made in order that consequences may be tested, novel results obtained, consistency affirmed. The process as a whole is the inference by means of which we transcend the limits of the observed, even the instrumentally observed.
Let us agree to call the entire process retroduction. We are “led backwards” from effect to cause, and arrive at an affirmation, not simply a conjecture. Retroduction in this sense is more than abduction. It is not simply the initial plausible guess. It is a continuing process that begins with the first regularity to be explained or anomaly to be explained away. It includes the initial abduction and the implicit estimate of plausibility this requires. It includes the drawing of consequences, and the evaluation of the match between those and the observed data, old or acquired in light of the hypothesis. Tentative in the first abduction, gradually strengthening if consequences are verified, if anomalies are successfully overcome, if hitherto disparate domains are unified, retroduction is the inference that in the strongest sense “makes science.” (92–93)