Kind concepts as idealised models

01/07/2025

The world is populated with things belonging to various natural kinds - electrons, mammals, stars and so on. On one everyday view, what scientists do is go out into the world, discover these natural kinds and give names to them. In this blog post Dr Ellen Clarke suggests that scientific activity is a much more active process of conceptual engineering, in which our observations are heavily idealised and bent into models that help us navigate reality.

What happens when scientists discover new things in the natural world? According to an influential account of reference by philosophers Saul Kripke and Hilary Putnam, these are baptism events. Instead of a vicar dipping a baby in a font, and then announcing its name, a scientist points at the new thing they’ve observed and communicates the name to others in journals. But this ‘causal theory of reference’ left much about the nature of scientific baptism ceremonies vague, because Kripke and Putnam were more concerned about how names come to stick to things, than about how things themselves are found or pointed at in the first place. But I think many folk who aren’t involved in coal-face science – the science that deals with hard stuff, that aims to excavate new truths, rather than the stuff in textbooks – assume that discoveries are largely a matter of looking at things really carefully, using fancy technologies to make tiny things visible or to bring long-gone things back into view. We write those names on museum specimens to educate everyone about the interesting things the world contains - here is a dinosaur bone, there a slab of quartz. We refine those labels over time into more fine-grained names until we have arrived at categories whose members are all essentially the same. All water molecules have the same bonds between their component hydrogen and oxygen molecules. All quartz samples share the same sort of crystalline structure. The concept of such a natural kind just faithfully reports what all the members have in common. All phoenixes are born out of fire – science can search around and tell us whether there are any such things.

This naïve view overlooks the fact that scientific discoveries are actually acts of conceptual engineering. I argue that scientific conceptualisation is much more active than this naïve view assumes - more a process of moulding the soft clay of reality than of excavating its parts. In fact, scientific concepts are best understood in much the same way as we understand scientific models: As representations in which the goal is not to faithfully replicate our target feature of the world, but rather to create an idealisation of it, an interpretation, which stretches and distorts the truth in illuminating ways. These models make our world navigable. Just as our perceptual apparatus converts a field of spinning sub-atomic particles into a landscape populated by discrete objects with apparent edges, so scientific concepts convert the complexities of the scientific realm into theorisable maps of cause and effect.

Kinds go beyond facts, in the same way as a line on a graph goes beyond the dots it connects. We can track discontinuities in the causal properties of things, but those properties tend not to cluster neatly. For example, when Socrates died, we expect that many of his important properties ceased. He stopped breathing, talking, thinking. But if we forget about our conventions, and think about it more carefully, we’ll notice that there probably isn’t usually a particular moment when all of a person’s defining properties change all at once. Socrates likely stopped talking before he stopped thinking. His breathing likely slowed down before it stopped altogether. Likely he stopped breathing before his heart stopped beating. His hair and nails will have continued growing long afterwards. Many of his cells will have continued their vital processes for hours, days even. All this leaves a fairly wide temporal window within which we can, if we choose, designate some moment the conventional moment at which Socrates ended.

My main study concept is the organism. We think we can see these, that no theoretical work is needed to make them visible. But when you dig into details, like where exactly the edges of a dog are, and which of the things inside its body are its parts, then you start leaning on idealisation. We have collectively, unconsciously, constructed the dog idea so that we converge on shared conventions about where and when dogs begin and end. We are so good at developing sensible joints for our concepts that it’s as if they fell from the sky, were found rather than imposed. The creative distortion becomes invisible. But dogs could have had different edges. There is no law of nature that forbids us from conceiving dog life cycles as beginning at the haploid stage, instead of with zygotes. We could view them as mobiles, sentient colonies of cells instead of as individuals, with no change of empirical facts at all.

Such contingency is everywhere, and we forget it at our peril. It is not possible for observable phenomena to determine the nature of our concepts. No matter how closely we look or how fancy our instruments, data alone will not decide where to put the edges of an idea. No quantity of facts about hormones or chromosomes or stable equilibria will settle who qualifies as a woman, just as no experiment will determine whether some substance is lithium, because the boundaries of a kind concept always have to be chosen. We can do this in more and less silly ways, given the empirical constraints and the facts about what we want to achieve. But no categorisation can occur until somebody has invented a category. We must make active choices about how to idealise the tangle of mashed properties that we observe in the world.

So forget about scientists as unearthing shy specimens from the conceptual shale and holding them aloft for all to see. The concepts that stick are the end products of long trial and error processes where the mashed felt of reality is teased and twisted into different designs, with some forms just holding their shape better than others, better suiting our practices, goals and perceptual habits. It’s not that organisms aren’t real, don’t exist. It’s just that real living things are far messier, far vaguer, far more imbricated with everything else than the neat shiny boxes we shoehorn them into.

Dr Ellen Clarke is an Associate Professor at the University of Leeds’ School of Philosophy, Religion and History of Science, and Director for the Centre for History and Philosophy of Science. She works in philosophy of biology, blogs as Philosomama and her book ‘The Units of Life’ is out now.