So much by way of surveying how cosmology, especially primordial cosmology, bears on scientific realism. Let us end by very briefly summarizing our position. We have espoused scientific realism as a modest thesis of cognitive optimism: that we can know about the unobservable, and that indeed we do know a lot about it. Cosmology causes no special trouble for this thesis: though there are systematic limitations, even in the form of theorems, about what we can know about the global structure of spacetime (Section 2). Besides, this thesis is well illustrated, we submit, by countless results of modern cosmology: astonishing though these results are, as regards the vast scales of distance, time (or other quantities, such as temperature, energy and density) that they involve (Section 3).
Of course, probing ever more extreme regimes of distance, time or these other quantities tends to call for more inventive and diverse techniques, as regards theory as well as instrumentation. So it is unsurprising that probing the very early universe involves intractable cases of under-determination of theory by data. In the second half of the paper, we saw this in inflationary cosmology: both for ascertaining the details of the inflaton field, for example its potential (Section 4); and for the problems of confirming a multiverse theory (Section 5). But as we said in Section 1, we do not see these cases of under-determination as threatening scientific realism. For it claims only that we can know about the unobservable, and indeed do know a lot about it - not that all the unobservable is knowable.
Acknowledgements: - FA's work is supported by the Wittgenstein Studentship in Philosophy at Trinity College, Cambridge. FA thanks Jim Hartle, Mark Srednicki, and Dean Rickles for conversations about some of the work reported in Section 5. For comments on a previous version, we thank: Anthony Aguirre, Bernard Carr, Erik Curiel, Richard Dawid, George Ellis, Michaela Massimi, Casey McCoy, John Norton, Martin Rees, Svend Rugh, Tom Ryckman, David Sloan, Chris Smeenk and Alex Vilenkin. We also thank the editor, not least for his patience.