My dissertation research concerns the development of an epistemology of data driven astronomy and astrophysics. This an area that has undergone relatively little philosophical work, despite being one of the largest and fastest growing areas of science. The models being developed are highly interdisciplinary endeavors that involve unprecedented engagement with some of the most advanced technological apparatuses (e.g., LIGO/Virgo/Kagra and the upcoming James Webb telescope) and the unprecedentedly large data sets they produce. There are many standards in testing and confirmation implicit in the scientific literature, but systematic attempts to engage these methodologies by philosophers of science are in the early stages. The dependence on “Big Data” for research in these areas raises substantial epistemological questions about how we acquire data, how we can reliably identify patterns and phenomena of interest, how we apply data to test and confirm models, and which methods are most efficient and productive. Philosophers of science are uniquely positioned to evaluate these questions and propose epistemic frameworks that are of direct benefit to the advancement of science. My research lays out an epistemic framework for A&A rooted in eliminative reasoning. Evaluation of these questions and proposals of corresponding epistemic frameworks are of direct benefit to the advancement of science.

It is becoming increasingly clear that not all of the traditional epistemic commitments of both philosophers and scientists can carry over into the new regime of data driven science. Philosophical examination of the way in which data-rich models in astronomy are developed and tested should produce beneficial results for both the sciences under examination and the continued development of practice-informed philosophy of science and epistemology. The advent of data-driven astronomy necessitates reconsideration of many epistemic commitments. The primary research interests I have are (1) the way in which different epistemic commitments in the methodologies of astronomy cohere or conflict, and (2) which methodologies are more epistemically productive and reliable. These questions intersect with one of the main problems currently facing data-driven sciences: how do we know that the algorithms we design to identify phenomena of interest in our data are truth-tracking? Many complex data machines are black boxes, meaning that the mechanisms within them are opaque to us. This requires us to adopt indirect means of evaluating the epistemic efficacy of our computational apparatuses. Developing a productive and justified set of epistemological criteria for our methods to meet can prove invaluable in this endeavor.

​