Lat. abductio, “action of taking”, by an outwardly directed movement (see infra, meaning 2).

1. Abduction as inference from facts to hypothesis

The concept of abduction was introduced in modern philosophy by the philosopher Charles Sanders Peirce. According to Peirce, there are two kinds of inferences: deductive inference and abductive inference or abduction. Abduction starts from the observation of a fact “contrary to what we should expect” Peirce ([1958], § 202), that is to say, a fact that does not fit into an available explanatory system. Abduction is a kind of inference by which one proposes a hypothesis accounting for this fact.
This hypothesis is not the product of the application of a “discovery algorithm”, but the fruit of a creative process, “abduction is, after all, nothing but guessing” (Peirce [1958], § 219).

Abduction is not an issue in logic, but rather a scientific method (id., Chap. 6). Scientific work consists in proposing, on the basis of facts, plausible hypotheses “suggested” by these facts. Abduction is the first step in this process.
The practice of abduction is not guided by logical rules but by general principles, such as the principle of exclusion of so-called metaphysical hypotheses, that is to say, hypotheses which would have no experimental consequences, or the principle according to which every fact has an explanation: an abducted hypothesis is interesting “if it seems to make the world reasonable” (id., §202).

Unlike abduction, which starts from facts in search of theory, the Peircian deduction starts from a theory in search of facts; that is, it seeks to identify the crucial experimental consequences of a hypothesis.

Much more than a form of deduction or induction, argumentation should be seen as a form of abduction: because the light is on, “I abduct”, I make the hypothesis, that there is someone in the room; but this hypothesis still needs to be checked, S. Probable, Plausible, True.

Woods redefines abductions as “responses to ignorance-problems. An agent has an ignorance-problem in relation to an epistemic target that cannot be hit by the cognitive resources presently at his command, or within easy and timely reach of it” (Woods, 2009; Gabbay & Woods, 2005). The study of argument as an abductive process has proved especially fruitful in the fields of medicine, science and law (Walton 2004).

2. Abduction as reduction of uncertainty

In its Peircian sense, abduction is a kind of inference by which one arrives at a hypothesis accounting for this fact. Aristotle defines abduction as a kind of dialectical syllogism (Aristotle, PA, II, 25), whose major premise is true, the minor just probable, and, consequently, the conclusion also probable. The conclusion alone, without the minor, is more improbable than the minor. The minor therefore strengthens the relative acceptability of the conclusion. This situation recalls the Ciceronian definition of argumentation, S. Argumentation (I).

For example, if the question is: “can virtue be taught?” we can reason as follows:

A true premise: it is clear that science can be taught.
A doubtful premise: virtue is a science.
Conclusion: virtue can be taught.

Though uncertain, the veracity of the second premise is still less in doubt than the conclusion “virtue can be taught”. This second premise may therefore serve as an argument for the conclusion. We find this montage in speeches such as:

Citizenship can be taught.
Citizenship is essentially a set of social knowledge and practices.
Knowledge is being taught and all practical skills can be improved by teaching. So, citizenship can be taught.

Argument functions “for want of better”. Reduction of uncertainty serves to modify relevantly the epistemic status of a belief. This is a logic not of elimination but of reduction of doubt and uncertainty, S. Default reasoning.