In this paper I develop a framework for relating dualities and emergence: two notions that are close to each other but also exclude one another. I adopt the conception of duality as 'isomorphism', from the physics literature, cashing it out in terms of three conditions. These three conditions prompt two conceptually different ways in which a duality can be modified to make room for emergence; and I argue that this exhausts the possibilities for combining dualities and emergence (via coarse-graining).

I apply this framework to gauge/gravity dualities, considering in detail three examples: AdS/CFT, Verlinde׳s scheme, and black holes. My main point about gauge/gravity dualities is that the theories involved, qua theories of gravity, must be background-independent. I distinguish two senses of background-independence: (i) minimalistic and (ii) extended. I argue that the former is sufficiently strong to allow for a consistent theory of quantum gravity; and that AdS/CFT is background-independent on this account; while Verlinde׳s scheme best fits the extended sense of background-independence. I argue that this extended sense should be applied with some caution: on pain of throwing the baby (general relativity) out with the bath-water (extended background-independence). Nevertheless, it is an interesting and potentially fruitful heuristic principle for quantum gravity theory construction. It suggests some directions for possible generalisations of gauge/gravity dualities.

The interpretation of dualities is discussed; and the so-called ‘internal’ vs. ‘external’ viewpoints are articulated in terms of: (i) epistemic and metaphysical commitments; (ii) parts vs. wholes.

I then analyse the emergence of gravity in gauge/gravity dualities in terms of the two available conceptualisations of emergence; and I show how emergence in AdS/CFT and in Verlinde׳s scenario differ from each other. Finally, I give a novel derivation of the Bekenstein–Hawking black hole entropy formula based on Verlinde׳s scheme; the derivation sheds light on several aspects of Verlinde׳s scheme and how it compares to Bekenstein׳s original calculation.