Epicureanism

7. Scientific method

Starting from the direct data of the senses, broader and less accessible truths can be established. Epicurus apparently divided the accessing process into two classes, ‘attestation’ (or ‘witnessing’, ‘confirmation’, and so on: epimarturēsis) and ‘non-contestation’ (or ‘lack of counterevidence’, ‘non-confirmation’, and so on: ouk antimarturēsis). Attestation plays a relatively minor part in scientific inference, since it seems confined, if not to truths directly supplied by the senses, at most to inductive generalizations derived from these. The main concern of Epicurean science is the extension of such knowledge beyond the realm of direct experience, into inaccessibly remote regions of the universe such as the heaven or other worlds, and into the microscopic realm of atoms and void. Here theories cannot be directly confirmed, but are tested instead by their consistency with empirical data: hence ‘non-contestation’ by phenomena.

Not that mere consistency with experience is enough to establish truth. A theory is only entertained in the first place if it has some explanatory power or recommends itself in some comparable way. But there may be several alternative theories of equal merit in this regard, and if so each must be tested for ‘non-contestation’ by phenomena, to see which survives. There are many intrinsically credible general theories of matter, for example, including the four-element theory and the monistic fire theory of Heraclitus, and likewise many intrinsically plausible theories of vision, and so on. But all except the Epicurean theory fail somehow to be consistent with the entire range of phenomena against which they are tested.

In some other cases, however, especially astronomical ones, Epicurus admits that two or more rival theories survive the test. This might be considered an indication that the consistency test can at best prove a theory possible, not true. But Epicurus insists that in such cases all the successful theories are indeed true. Sometimes this means that several different explanations of the same phenomenon – for example, thunder, or the generation of perceptual ‘images’ (see §6) – operate concurrently in our world. In other cases it means no more than that, given the intrinsic possibility of the hypothesized causal process, it must operate somewhere in the infinite universe, even if not here.

All such ‘sign-inferences’ (sēmeiōseis) are from something evident to something non-evident. Often this is from the macroscopic to the microscopic: either causal inference from a macroscopic effect to a microscopic cause (for example, from the observed regularities of nature to the existence of unchangeable elements; see §2), or analogical inference (for example, from the mobility of observed spheres to that of spherical atoms, see §5). Others are from the macroscopic to the macroscopic: either from accessible to inaccessible entities or events (for example, from the mechanism of the water-wheel to the rotation of the stars), or from the present to the past or future (for example, from the current structure of a social institution like law or language to its historical origin, or from the impermanence of familiar compounds to the future destruction of the world).

The nature and validity of such sign-inferences is debated between the Epicureans and the Stoics in Philodemus, On Signs, including an important controversy about the justification of induction.