Contrast scientific realism with instrumentalism and anti-realism, and assess the no-miracles argument and the pessimistic meta-induction

What this dot point is asking

SEAB wants you to engage the central metaphysical debate in the philosophy of science: when a successful theory talks about unobservable things like atoms, fields or genes, should we believe those things really exist and the theory is true, or should we treat the theory merely as a useful instrument for predicting what we can observe? Your task is to contrast realism with instrumentalism, weigh the two master arguments, and consider the middle positions that try to keep the best of both.

The answer

Scientific realism

Scientific realism holds that our best, mature scientific theories are at least approximately true, and that the unobservable entities they posit genuinely exist. On this view, when physics talks about electrons, it is describing real things, not merely a convenient fiction; science aims at, and to a large extent achieves, true description of a mind-independent world, including its unobservable parts. Theories are to be believed, not just used.

Instrumentalism and anti-realism

Instrumentalism holds that theories are instruments or tools for systematising and predicting observations, and that their talk of unobservables need not be taken as literally true. What matters is whether a theory works, that is, whether it is empirically adequate. Anti-realists more broadly are sceptical that we can know the truth about unobservables, even if there is a fact of the matter. On these views, two empirically equivalent theories that posit different unobservables need not be ranked by truth at all.

The no-miracles argument for realism

The leading argument for realism is the no-miracles argument, an inference to the best explanation. Science is strikingly successful: it makes novel, precise predictions and underpins reliable technology. The best explanation of this success is that our theories are at least approximately true and their posited entities real; otherwise the success would be a miracle, an inexplicable cosmic coincidence. So we should believe our best theories are approximately true.

The pessimistic meta-induction against realism

The leading argument against realism is the pessimistic meta-induction. The history of science is full of theories that were predictively successful in their day yet are now regarded as false, positing entities (such as caloric or the luminiferous ether) we no longer believe exist. If past successful theories turned out false, then by induction our current successful theories may also be false. So success is not a reliable sign of truth, which undercuts the no-miracles argument.

Middle positions

Two influential positions try to keep the best of both sides. Constructive empiricism accepts theories only as empirically adequate, saving the observable phenomena, while remaining agnostic about the truth of claims about unobservables; it argues that empirical adequacy, not truth, is enough to explain success. Structural realism holds that what we can know, and what survives theory change, is the structure or pattern of relations the world has, even if our beliefs about the nature of the underlying entities change; this answers the meta-induction by pointing out that mathematical structure often carries over across revolutions even when ontology does not.

Examples in context

Example 1. Atoms from useful fiction to real entities. Atoms were long treated by some scientists as a useful calculating device rather than real objects, until converging independent measurements of atomic quantities made their reality hard to deny. The episode supports the realist: the convergence of many independent methods on the same value for an unobservable quantity is exactly what an instrumentalist struggles to explain away as mere convenience, and it is a model case for the no-miracles intuition.

Example 2. The ether's disappearance. Nineteenth-century physics posited a luminiferous ether as the medium for light waves, and theories employing it had genuine successes. The ether was later abandoned as non-existent. This is a stock example for the pessimistic meta-induction: a successful theory whose central unobservable entity we now reject, suggesting that present success is no guarantee that today's posited entities will survive future theory change.

Try this

Q1. Distinguish scientific realism from instrumentalism. [6 marks]

• Cue. Realism: best theories are approximately true and their unobservable entities really exist, so theories are to be believed. Instrumentalism: theories are predictive tools whose unobservable talk need not be literally true, only empirically adequate.

Q2. Explain the no-miracles argument for realism. [6 marks]

• Cue. An inference to the best explanation: science's striking predictive and technological success is best explained by its theories being approximately true and their entities real; otherwise the success would be an inexplicable miracle.

Q3. Explain how the pessimistic meta-induction challenges realism and how structural realism responds. [8 marks]

• Cue. Past successful theories were later judged false with abandoned entities, so success need not signal truth; structural realism replies that the structure (relations) typically survives theory change even when the entities do not, locating the durable truth in structure.