El Conocimiento

Throughout his career, C.S. Peirce argued for the existence of a third kind of inference or reasoning, in addition to deduction and induction, which he called abduction.16 However, during his 50 years of writing about the topic, his account of what distinguishes abduction from the traditional forms of inference changed repeatedly. Commentators usually distinguish two main phases in Peirce’s thinking about abduction: the first comprising his writings between 1860 and 1890, and the second, mature period emerging post-1890.17

In his early view, Peirce regarded abduction and induction as two separate kinds of probable inferences, differing in the kinds of premises they rely on, and the kinds of conclusions they support. Whereas induction infers general laws from observed regularities, abduction infers causes or explanatory hypotheses from their observed effects. However, the two inferences are similar in that both provide some degree of non- demonstrative, probable support for their conclusions.

15 _{Parts of this section is based on the Peirce exegesis which I contributed to Stanley and Nyrup}

(forthcoming).

16 _{Peirce’s terminology varies. He also sometimes calls this third form of inference retroduction,}

hypothesis¸ hypothetic inference, or presumption. I follow commentators in simply using the term abduction (cf. Fann 1970: 5, note 19).

17 _{This division was introduced by Burks (1946). Burks regards the period between 1890 and 1900 as a}

transitionary period between Peirce’s early and mature views. See also Fann (1970), Niiniluoto (1999) and Psillos (2011a) for overviews of the development of Peirce’s thought on abduction.

Post-1900, Peirce came to regard his previous discussion as badly confused. For instance, in a 1902 manuscript, he concedes that his earlier conception of abduction “had necessarily confused two kinds of reasoning”, since “probability proper [i.e. empirically grounded probabilities]18 had nothing to do with the validity of Abduction, unless in a doubly indirect manner” (CP2.102).19 _{Here, and in other writings around the same time,}

Peirce redraws the distinction between abduction and induction in terms of the role they play in scientific inquiry. He now classifies as inductive all inferences which provide empirically based probable support for a hypothesis. These include both arguments from random samples to general statistical regularities, which he sometimes calls ‘quantitative induction’, and inferences which support a hypothesis by confirming its observable consequences, which he calls ‘qualitative induction’, the latter resembling what he had earlier called abduction.

In his mature view, abduction instead becomes an inference, or line of reasoning, by which a new hypothesis is in some sense introduced into scientific inquiry. This does not (in itself) provide any kind of probable support for the hypothesis, except insofar as a hypothesis introduced by abduction can subsequently be supported through successful inductive testing. Thus, in a 1903 lecture series at Harvard, Peirce writes that abduction:

is the only logical operation which introduces any new idea. … Its only justification is that from its suggestion deduction can draw a prediction which can be tested by induction, and that, if we are ever to learn anything or to understand phenomena at all, it must be by abduction that this is brought about. No reason whatsoever can be given for it, as far as I can discover; and it needs no reason, since it merely offers suggestions. (CP5.171).

18 _{Peirce around this time tends to reserve the term ‘probability’ for empirically grounded judgements about}

how likely a hypothesis is to be true. He distinguishes this from ‘likelihood’, by which he means a priori judgements about likeliness. Peirce thinks the latter just expresses our prejudices and so should generally not be trusted.

19 _{Following standard conventions, Peirce (1932-58) is cited in the format ‘CP[volume].[paragraph]’. As}

these volumes are not chronologically organised, I will sometimes add the approximate year of writing to these citations.

Despite this merely suggestive role, Peirce emphasised that abduction is a form of reasoning, that it involves giving reasons and that there is a difference between good and bad abductions.

In some places, Peirce still seems to regard abduction as providing some kind of epistemic support for a theory. For instance, in the same lectures, he characterises abduction as “the operation of adopting an explanatory hypothesis” (CP5.189), arguing that it follows the following inference schema:

The surprising fact, C¸ is observed;

But if A were true, C would be a matter of course, Hence, there is reason to suspect that A is true.

Here, abduction is supposed to provide some “reason to suspect” that the explanatory hypothesis is true, suggesting that abduction provides at least a weak form of epistemic justification. In his Lowell lectures also given in 1903, Peirce similarly characterises abduction as “any mode or degree of acceptance of a proposition as a truth, because a fact or facts have been ascertained whose occurrence would necessarily or probably result in case that proposition were true” (CP5.603). Here, it seems that abduction can even provide enough justification to accept an explanatory hypothesis as true. Yet, in the preceding paragraph, Peirce insisted that “abduction commits us to nothing. It merely causes a hypothesis to be set down upon our docket of cases to be tried” (CP5.602). Here, “adopting” a hypothesis through abduction merely consists in giving it priority for testing. Furthermore, Peirce stresses in a number of places that it can be reasonable to test a hypothesis because it is easily falsifiable, rather than because it is likely to be true: “The

best hypothesis, in the sense of the one most recommending itself to the inquirer, is the one which can be the most readily refuted if it is false. This far outweighs the trifling merit of being likely”. (CP 1.120, c. 1896).

As McKaughan (2008) has persuasively argued, the support abduction provides for a hypothesis is first and foremost to justify giving a theory “a high place in the list of theories of those phenomena which call for further examination” (CP2.776, 1902), i.e. to provide reasons for pursuing it.20 The normative standard guiding abduction is, according to Peirce, “economy”, which in the context of research he takes to be “how, with a given expenditure of money, time, and energy, to obtain the most valuable addition to our knowledge” (CP7.140, 1879). Considerations of economy are crucial to Peirce because:

Proposals for hypotheses inundate us in an overwhelming flood, while the process of verification to which each one must be subjected before it can count as at all an item, even of likely knowledge, is so very costly in time, energy, and money—and consequently in ideas which might have been had for that time, energy, and money, that Economy would override every other consideration even if there were any other serious considerations. In fact there are no others. (CP5.602, 1903).

Since there are only a limited amount of resources available for scientific research at any given point, scientists ought to prioritise them such that they would contribute the most to our knowledge. While considerations of how likely a hypothesis is to be true can still play some role in evaluating this, it is only in an indirect way: “the likelihood would not weigh with me directly, as such, but because it would become a factor in what really is in all cases the leading consideration in Abduction, which is the question of Economy—

20 _{Thus, abduction cannot be assimilated to the modern notion of “inference to the best explanation”, as}

pointed out by many Peirce scholars (Hintikka 1998, Minnameier 2004, Paavola 2004, Campos 2011 and Pietarien and Belucci 2014).

Economy of money, time, thought, and energy” (CP5.600). Furthermore, how considerations of likeliness affect the decision to pursue a hypothesis varies. In some cases, a hypothesis may be so unlikely that it is not worth spending energy on: “if a man came to me and pretended to be able to turn lead into gold, I should say to him, ‘My dear sir, I haven’t time to make gold” (ibid.). Peirce allows that if a hypothesis has a “marked probability of the nature of an objective fact, it may in the long run promote economy to give it an early trial” (CP6.534, 1901). However, the opposite may also be the case: if a hypothesis can be easily tested and “promises not to detain us for long, unless it be true”, then “Sometimes the very fact that a hypothesis is improbable recommends it for provisional acceptance on probation” (CP6.533). Finding out which hypotheses are false can in itself be a valuable contribution, partly because it “leaves the field free” for further investigations (CP1.120-21, 1896). Furthermore, Peirce argues that the resources invested in pursuing a hypothesis will gradually yield diminishing returns of new knowledge so that, at some point, it will no longer be worthwhile to pursue it further (CP1.122). As I will argue in the next chapter, Peirce’s account can plausibly be reconstructed in decision- theoretic terms.

A remaining tension in Peirce’s characterisation of abduction concerns in what sense abduction is supposed to “offer suggestions” or “introduce” hypotheses into inquiry. Sometimes he characterises abduction as the process of generating or

formulating a new explanatory hypothesis. In the Harvard lectures, Peirce claims that

“Abduction consists in studying facts and devising a theory to explain them” (CP5.145, emphasis added) and that “Abduction is the process of forming an explanatory hypothesis.” (CP5.171, emphasis added). At other times, however, he seems to characterise abduction as an inference where a hypothesis is (in some sense) adopted because it could potentially explain an otherwise puzzling set of phenomena. When

abduction is written as an inference schema, such as the one above, the inferred hypothesis (‘A’) is explicitly mentioned in the premises, and so it seems to have been formulated before the abductive inference was made (Frankfurt 1958, Kapitan 1997). Furthermore, Peirce usually characterises reasoning as controlled and voluntary thinking, remarking in the 1903 Harvard lectures that “To criticize as logically sound or unsound an operation of thought that cannot be controlled is no less ridiculous than it would be to pronounce the growth of your hair to be morally good or bad” (CP5.109). But since we do not control which ideas occur to us, the question is how abduction could “suggest” ideas to us and still be considered an inference subject to any kind of normative criticism. There are different proposals for how to reconcile the tensions between these two characterisations of abduction. McKaughan (2008) argues that the interpretation of abduction as generative reasoning should be rejected. However, some commentators (Fann 1970; Curd 1980; Psillos 2011a) have however argued that the tension can be resolved by construing abduction as a “dual process” (Psillos 2011a: 133), one which encompasses both the generation of a hypothesis and its adoption for pursuit. As Fann (1970: 42) notes, simply coming up with a new hypothesis is easy; the challenge in generating or formulating a new hypothesis is not merely to think of any hypothesis whatsoever. Peirce illustrates this point with the following example:

Consider the multitude of theories that might have been suggested. A physicist comes across some new phenomenon in the laboratory. How does he know but the conjunctions of the planets have something to do with it or that it is not perhaps because the dowager empress of China has at that same time a year ago chanced to pronounce some word of mystical power or some invisible jinnee may be present” (CP5.172)

The problem for the physicist here is not to generate new hypotheses. Rather, since he cannot examine every conceivable hypothesis, the problem is to come up with a good hypothesis, one that it is worth considering further. The normative criteria for generating good hypotheses in this context are the same as the criteria for adopting a hypothesis for pursuit. Although the concrete thought processes in generating and adopting hypotheses of course differ, it makes sense to classify them together under the label ‘abduction’ because they aim to satisfy the same normative standards, distinguishing them from induction and deduction.

This interpretation also suggests an answer to the objection that generative reasoning is not subject to control. What we can control are decisions about how to attempt to formulate new ideas, and such choices can be subject to normative criticism on the basis of how effectively they lead to theories that are worth pursuing. A physicist who simply starts freely associating all sorts of possible causes of a puzzling phenomenon, in the manner parodied by Peirce above, would rightly be criticised by her colleagues for wasting their time. Even if she cannot strictly speaking control which ideas occur to her, she can still exert some control on the direction of her thoughts, choosing for instance to focus on known physical causes which it would be possible to test. In some contexts, more systematic heuristic strategies may be available or, if the desiderata on a satisfying problem solution are sufficiently constrained, it may be possible to deductively derive all hypotheses that could be of interest. The choice to adopt such methods can also be evaluated in terms of how effectively and efficiently they will generate an adequate range of pursuit worthy hypotheses.

Part of the choice here also concerns when or whether to generate new hypotheses and when to stop. Scientists at any given stage of inquiry will only be able to effectively consider a limited range of hypotheses and so it will often be reasonable to stop generating

hypotheses once a few good candidates for pursuit have been found. Allan Franklin (1986: ch. 1) describes one such an example. In the 1950s, particle physicists were faced with a puzzling phenomenon: for certain observed decay patterns, the principle that all particles have a unique mass indicated that the decay products stemmed from the same particle, whereas the principle of parity conservation ruled this out (8-10). At a conference in 1956 where the problem was discussed, physicists proposed several possible explanations (34-5). After a range of possibilities had been discussed the chair “felt that the moment had come to close our minds” (35), i.e. to start thinking about how to test the salient proposals rather than generate new ones. The following year, experiments designed to test whether parity conservation is violated in these interactions confirmed (much to the surprise of most physicists at the time) that this was indeed the case. In this case, at least, the decision to stop generating new hypotheses seems reasonable for the same reason that it was reasonable to start pursuing the already generated hypotheses: the latter represented a more cost-effective use of their time and resources. Here, again, the normative criteria for generation overlaps with those for pursuit.