Why pain is not a natural kind

As our #BadFeelings week draws to a close, we turn out attention to pain, with Dr Jennifer Corns, Lecturer in Philosophy at the University of Glasgow. Jennifer works on the philosophy of mind, with a focus on pain. She has published a number of papers, including “Pain eliminativism: scientific and traditional” (2016) in Synthese, “The social pain posit” (2015) in Australasian Journal of Philosophy and “The inadequacy of unitary characterizations of pain” (2014) in Philosophical Studies.


Pain is a real and ubiquitous part of our everyday lives. Many of these pains are transient; our heads ache, our stomachs hurt, and our feet throb. Some pains are longer lasting: chronic pain and the pains from disease, healing, and surgery. Paradigmatically, a pain is felt, unpleasant, located in our bodies, and motivates us to do something to relieve or minimise the pain.

Despite its ubiquity, I argued that neither pain, nor any type of pain, is a natural kind. ‘Natural kind’ is a term of art. As I use the term, a kind is natural to the degree to which it is usefully referred to in scientific generalisations for explanations and predictions.  My central claim, then, was that neither pain nor any type of pain are usefully referred to in scientific generalisations for explanation or prediction.

This may seem–as it initially seemed to me–incredible.  The nature of pain is both historically and contemporarily debated not only within philosophy, but across science and medicine, but there has nonetheless been implicit agreement that pain is a natural kind.  ‘Pain’ and terms for types of pain, e.g. ‘headache’ or ‘inflammatory pain’, have been assumed to be useful referring expressions for scientific generalisations. Whether pain is a particular kind of sensation, is controversial and has been subject to reasoned argument. Whether there are pain receptors, pain pathways, or a pain area in the brain are all debated questions that have been subject to rigorous empirical investigation. Underlying these disagreements, however, is the shared assumption that when the dust settles, pain and pain types will have been discovered to have a nature such that we can explain and predict them in sciences like biology, psychology, and neuroscience.

If pain or pain types were mechanistically explicable, we would have a strong reason to think that they were natural kinds. If there is a mechanism, or system, whose workings determined our pains, then the workings of this system, or mechanism–once we discovered it– could support our explanatory and predictive generalisations about pain.  The assumption that both pain and pain types will, ultimately ,prove mechanistically explicable underlies inquiries into pain across the disciplines and, typically, likewise underlies the agreement that these are natural kinds. I argue, however, that empirical inquiry reveals that there are no such mechanisms or systems.

First, consider pain. Notice that many of the pains we suffer are different, such that you engage different treatment options when you suffer them. This is appropriate. As encoded in standard models, distinct types of pain reports implicate distinct ranges mechanistic activity, such that distinct treatment options are more likely to be effective. Empirical inquiry into these different types of pain as reported reveals that there is no single pain mechanism, or system, whose workings determines them.

Consider now these different types of pain as reported. Notice that sometimes your pain does not get better, even when you pursue the best treatment option. While an aspirin will often help a headache, for instance, sometimes it is entirely ineffective. Why? Because even as there are multiple distinct ranges of mechanisms implicated in pains that feel different, so there are multiple mechanisms involved even across pains that feel similar. Aspirin alters the activity of inflammatory mechanisms. These are often involved when you have a headache. Other mechanisms, however, are also always involved. It is sometimes these other non-inflammatory mechanisms which are crucial in determining your particular headache. While we have a good understanding of inflammatory mechanisms, we have no reliable mechanistic explanation for headache pain. Or, indeed, for any other type of pain.

Scientific inquiry into pain has revealed two facts that undermine the prospects for any eventual identification of explanatory mechanisms for pain or pain types. First, each pain is the result of the activity of multiple mechanisms. Second, these mechanisms include those for cognition, memory, genetics, and more besides, such that their convergent activity is idiosyncratic. It is the idiosyncratic convergent activity of multiple mechanisms that undermines the prospect of mechanistic identification–for either pain or pain types.

Even if there are no pain mechanisms or systems whose workings determine pain, however, we may still think that pain is a natural kind–we may, that is, still think it is useful to refer to pain for scientific explanation and prediction. Idiosyncratic convergence, however, likewise undermines utile reference in scientific generalisations. If each particular pain is the result of idiosyncratic convergent activity, then our generalisations about pains will always remain poorer than those generalisations that we might instead offer concerning the activity of any of the mechanisms converging in a particular case.

In her commentary on my paper at the recent negative emotions conference, Giada Dirupo was in apparent agreement with all the empirical facts mentioned in the summary above. She agreed that we are currently lacking any mechanistic explanation for either pain or pain types and, moreover, that we now know that a wide range of mechanisms are involved, idiosyncratically converging in each token pain. She nonetheless expressed hope that we may, eventually, be able to identify mechanisms for pain or pain types, such that reference to them for scientific inquiry remains useful.

While I agree that there is much to learn about the many mechanisms involved in pain experiences, it is, I maintain, the empirical inquiries themselves that support the conclusion that this hope is misplaced. Learning more about any of the involved mechanisms will not change that their convergence is idiosyncratic for each pain experience. If that is right, and this idiosyncratic convergence entails that we always lose explanatory and predictive power when we generalise about the pains instead of the mechanisms, then it’s time to give up the assumption that pain is a natural kind.

We should not conclude from this that pains are unreal. We should instead, I urge, understand pains for the convergent, idiosyncratic experience that they are. Like a parent in response to a child’s report that they have an ‘owie’ or a ‘boo-boo’, a good doctor will take a pain report seriously. They will use that report to identify an appropriate treatment target and pursue a treatment option that is likely to be effective for their particular patient, as we do for our particular children. It is my hope that pain treatment will improve if we focus on identifying and treating the particular mechanisms implicated in token cases and targeting our scientific inquiries on improved understanding of their workings.