An Operative Account of Scientific Understanding — Epoche C2
The thesis, and a source that has to be withdrawn To understand a scientific theory, I argue, is to possess a particular ability: the ability to see what the theory implies about a case without working the case out, and to use it to intervene in the system it describes. Understanding on this view is not a propositional relation to a stock of true sentences that happens to be accompanied by skill; the skill is what the understanding consists in. That is a strong claim, it is contested by serious people, and the first business of this note is to say what would make it true and who denies it. Before any of that, a correction. The earlier version of this essay built its central apparatus — a distinction between understanding that , understanding how , and a third category called understanding through — on a 2021 Routledge monograph attributed to an author named Hernán Rena. That work cannot be located. Neither the title nor the author appears in the literature on scientific understanding, and no tripartite distinction of that shape is published under that name. The attribution is withdrawn in full. The distinction it was made to carry is nonetheless worth having, and the rest of this essay rebuilds it from sources that can be checked, which has the incidental benefit of showing what the first two members of the triad actually owe to Ryle and what the third has to be replaced by. The first two members are Ryle's. In The Concept of Mind (1949) he argued that knowing how to do something is not a species of knowing that something is the case, and the argument is a regress rather than an appeal to intuition: if every intelligent performance required the agent first to contemplate a proposition prescribing how to perform it, that contemplating would itself be a performance, requiring its own prior contemplation, and no action would ever begin. Ryle's target was a picture of mind, not a theory of science. But the regress transfers intact. If understanding a theory were exhaustively a matter of holding true propositions about it, then applying the theory to a new case would require a further proposition telling one how to apply it, and the regress restarts. What an ability-based account has to say to be non-trivial The trouble with "understanding is an ability" as a slogan is that it names no ability in particular, and an account that cannot be failed cannot be tested. Henk de Regt's Understanding Scientific Understanding (2017) supplies the missing specificity, and it is the position I am extending. De Regt separates two questions that are usually run together. The first concerns phenomena: on his criterion, a phenomenon is understood scientifically just in case there is an explanation of it that is based on an intelligible theory and that meets the basic epistemic values of empirical adequacy and internal consistency. The second concerns theories, and this is where the operative element enters: a theory is intelligible to scientists in a given context if they can recognise qualitatively characteristic consequences of it without performing exact calculations. That criterion has content because it can be checked case by case. Consider the heat capacity of gases. From the equipartition of energy among quadratic degrees of freedom, a physicist who finds the theory intelligible sees immediately that a diatomic gas at moderate temperature stores more energy per degree of temperature than a monatomic one, because the molecule has rotational degrees of freedom to share the energy with in addition to the three translational ones — the constant-volume heat capacity rises from $\frac{3}{2}Nk$ to $\frac{5}{2}Nk$. Nobody computes a partition function to see this. The exact calculation is available and would give the same answer, but the recognition precedes it and does not depend on it. The person who can only reach the result by grinding through the calculation has the same true propositions and does not have the same relation to the theory, and de Regt's criterion says what the difference is. The criterion is also contextual, and deliberately so. Which qualities make a theory intelligible varies with the training of the scientists concerned and with the period: visualisability was demanded of a mechanical model in the nineteenth century and not in the 1930s, and de Regt's historical chapters trace such shifts rather than legislating one standard. Intelligibility is a value scientists attach to a cluster of theory-properties that facilitate use, not a fixed property of the theory itself. The opposing position, stated as its holder states it The strongest opposition to all of this is not a defence of dry propositionalism but a reductive programme, and it is set out by Kareem Khalifa in Understanding, Explanation, and Scientific Knowledge (2017). Khalifa's explanatory model of understanding holds that understanding a phenomenon consists in grasping its explanatory nexus, that grasping is nothing over and above scientific knowledge of an explanation, and that the greater one's scientific knowledge of the phenomenon's explanatory nexus, the greater one's understanding. The abilities the operative account points to are, on his view, real but derivative: they are what possession of explanatory knowledge looks like from outside, and positing a further sui generis ability adds nothing that the knowledge did not already supply. This is not a straw position and it should not be answered by restating the thesis louder. What it demands is a case in which the propositional knowledge is held constant and the understanding varies. The heat-capacity example above is offered as one: two physicists who can both state and derive the equipartition result differ in whether they can see its consequences in an unfamiliar case, and it is not obvious that the difference is itself statable as a further piece of knowledge without reopening Ryle's regress. Khalifa would reply that the difference is a difference in how much of the explan