Scientists and philosophers down the ages have argued that there should be no overlap between science and metaphysics. Others have claimed, to the contrary, that science needs a foundation that only metaphysics can provide. I would challenge both these lines of thought, but the first is the main target of this talk.
Consider the following claims, which are widely accepted:
The Universe was created in a great explosion whose effects remain visible;
Biological species are the product of a chancy process involving mutation and natural selection;
What happens in our minds is the outcome of neural processes.
Not only are these all metaphysical (in a broad sense): scientific research is also relevant to them, even if they aren’t simple to decide.
If we understand metaphysics broadly to be concerned with general and abstract questions of existence and reality, science is clearly concerned with metaphysics. But how good is science at answering metaphysical questions like these, and under what conditions? I address that question historically, by looking at two well-known failures (causal determinism and Newtonian mechanics; a claimed refutation of vitalism); and some important but neglected successes (the achievements of 19th-century chemistry).
I conclude by arguing that metaphysical claims can be supported by science if they have had an active, intimate and indispensable heuristic and conceptual role in the development of science. Otherwise, it’s too soon to tell.
Models are epistemic artefacts, constructed and put to many uses, including understanding, prediction, data mining, and design. In many cases, the key to their epistemic value is erotetic: they are built to address pending theoretical or empirical questions, and that is where their epistemic value lies, not in a model’s correspondence to a determinable real-world target system. I defend this thesis and dispel a double bind in the account’s recent reception. It has been claimed, on the one hand, that the artefactual account is covertly representational and, on the other hand, that as a non-representational account, it cannot supply the content that property attribution and model-world comparison require. These charges pull in opposite directions and cannot both land. The artefactual account does attribute content to models, located in the interpreted representational modes and media used in their construction, but does not assume that the epistemic value of models is due to the representational relationship between model and target. Artefactualism stays neutral between different theories of representation, though closer to deflationary, inferentialist accounts. No longer organised around the model-target relation as its unit of analysis, the account clarifies the use of cross-domain model templates, the modal reach of modelling, and models without determinable targets, such as large language models and synthetic models.
Many theories of quantum gravity appear to imply that the spacetime of general relativity will be effective or emergent. This talk explores the consequences of such a view, and argues that, on any plausible account on which general relativistic spacetime emerges, other relativistic and non-relativistic spacetime structures will emerge in relevant contexts and at relevant scales. This has surprising and under-appreciated consequences – for example, that, at certain scales, binary black holes live in Newtonian spacetime.