This paper explores the conceptual foundations of macroprudential policy. It does so within a framework that gradually incorporates and interacts two types of frictions (principal-agent and collective action) with two forms of rationality (full and bounded), all in the context of aggregate volatility. Four largely orthogonal rationales for macroprudential policy are identified. The first (time consistency macroprudential) arises even in the absence of externalities, not to prevent financial crises but to offset the moral hazard implications of (efficient but time inconsistent) post-crisis policy interventions. The second (dynamic alignment macroprudential) protects the less sophisticated (boundedly rational) market participants by maintaining principal-agent incentives continuously aligned along the cycle and in the face of aggregate shocks. The third (collective action macroprudential) responds to the socially inefficient yet rational instability resulting from uninternalized externalities. The fourth (collective cognition macroprudential) aims at tempering non-rational mood swings where credit-constrained rational arbitrageurs fail. Finding the right policy balance is complicated by the fact that the four dimensions face policy trade-offs and their relative importance is state-dependent, hence shifts over time.