Description
It is a widely accepted view in the philosophy of science that what counts as evidence for a theory depends on the theory itself. However, the question of how a theory selects its evidential base remains largely underexplored. This paper addresses this gap by analyzing a case study in cosmology: the evolving role of galaxies as evidence for cosmological models, particularly those involving dark matter, from the 1970s until now. While galaxies were central to testing and constructing cosmological models, they have, over time, lost their status as decisive evidence. This shift raises questions: Why and how did this happen? What criteria determine whether galaxies count as evidence for particular phenomena and our theories of them?
I trace the historical role of galaxies, showing how phenomena like galactic tidal tails and galaxy morphology provided critical data for evaluating cosmological scenarios of structure formation in the 1970s–1980s. Since the 1980s, advances in the precision, depth, and scope of galaxy observations have been remarkable, but the current standard cosmological model, Lambda Cold Dark Matter (ΛCDM), struggles to account for them (so-called small scales problems).
Despite these challenges, the cosmological community largely does not interpret galactic discrepancies as evidence against ΛCDM (De Baerdemaeker & Boyd 2020). Instead, galaxies are considered “too complicated” to provide clean tests of the model. This attitude marks a shift: galaxies, once crucial, are now often dismissed as unreliable sources of evidence. I ask: How did galaxies become “too complicated”? What changed in the epistemic practices of cosmology to account for this shift?
To address these questions, I engage with Harper’s (2011) claim that “converging measurements of parameters across diverse phenomena” solidified ΛCDM’s status in the early 2000s. I argue that this convergence was selective, as galactic-scale challenges known since the 1990s (e.g. Moore 1994) were excluded from the body of evidence. This suggests that the very notion of evidential convergence is contingent on the theory’s criteria for selecting evidence.
To analyze these criteria, I employ Curiel’s (2024, forthcoming) framework grounded on the idea of epistemic control – an understanding of how a theory connects with empirical data, including its regime of applicability and conditions under which its formalism can be considered physically meaningful. I argue that the loss of epistemic control at galactic scales—due to the complex, multiscale nature of baryonic physics—has rendered galaxies “epistemically untrustworthy” within the ΛCDM framework.
My main argument is that the selection of evidence by a theory is governed by the level of epistemic control over the phenomena in question. When the users of theory lack control over the regime of applicability, e.g. in the case of ΛCDM and galaxies, the phenomena are dismissed as too complex to provide reliable evidence. Curiel’s framework not only explains the historical trajectory of galaxies in cosmology but also provides general criteria for understanding how theories determine their evidential bases.
