This poster will present a general framework for evaluating the epistemic merits of forms of empirical inquiry, with a consequence that undermines a view found in the philosophy of science literature.

We start with this question: given different methods for empirically studying a natural system, what general criteria ought to motivate a choice of one method over others? Drawing on prior work in the philosophy of experiment, we identify three “parameters” central to such a decision, which we call signal clarity, background characterization, and variability of precipitating conditions. The first concerns how clear a signal can be extracted from this system by the method in question, the second refers to means for identifying and subtracting irrelevant features from recorded data, the third refers to researchers’ ability to distinguish between the different conditions that may affect them and to track how they co-vary with these conditions.

For each parameter, we claim that a method A that performs better than another B on these lines is ceteris paribus epistemically superior to B. This claim is justified by reference to the following notion of epistemic superiority:

(ES)   An empirical method X is epistemically superior to Y, with respect to a system of interest S, if X produces results that reliably discriminate between more relevant hypotheses about S than Y.

That is, we argue that each parameter contributes to the epistemic superiority of one method over another that is otherwise equivalent.

The remainder of the poster will be devoted to spelling out and justifying a surprising consequence of this framework: the received distinction between observation and experiment in philosophy of science does not track epistemic superiority. This distinction is typically drawn with respect to whether a method involves a manipulation of the system under study. If it does, it is experimental; if not, observational.

We will show that our three parameters cross-cut the observation-experiment distinction. Further, we identify the typical assumptions about manipulative methods that underlie the claim that they are epistemically superior to observational methods—i.e., that manipulation entails greater control over data generation, or that only manipulation can produce causal knowledge.[1] We argue that manipulation is neither necessary nor sufficient for control or causal knowledge. This challenges the claim that manipulation is a distinct parameter contributing to the epistemic value of a method.

References

Currie, A. and Levy, A. (2019): “Why experiments matter”, Inquiry, 62, pp. 1066-1090.

Hacking, I. (1989): “Extragalactic Reality: The Case of Gravitational Lensing”, Philosophy of Science, 56, pp. 555-581.

Herschel, J. (1831): A Preliminary Discourse on the Study of Natural Philosophy, London: Longman, Rees, Orme, Brown, Green & Taylor.

Okasha, S. (2011): “Experiment, Observation and the Confirmation of Laws”, Analysis, 71, pp. 222-232.

Zwier, K. (2013): “An Epistemology of Causal Inference from Experiment”, Philosophy of Science, 80, pp. 660-671.

[1] Instances of these assumptions arise in Hacking (1989), Okasha (2011), Zwier (2013), Currie and Levy (2019), along with historical sources like Herschel (1831).