Open AccessPolicy can change people's preferences. For example, how cities aredesigned has an impact on future preferences about modes of transport,even after people move. We examine the normative significanceof such preference formation for climate policy design, in which sucheffects are commonly ignored. Yet, on the long time scales that climatepolicy must consider, it is likely to change preferences. We arguethat this is of high relevance for the adequate evaluation of mitigationoptions. We further argue that the orthodox approach of addressingenvironmental concerns with marginal pricing and Pigouvian taxes ismissing a key element. Policies that change preferences can influence the cost of carbon mitigation, and hence change the optimal carbon taxrate. As an attempt to assist in policy making with endogenous preferences,we examine alternatives to preference satisfaction for normativeeconomics. Some new arguments for and against established positionsin welfare theory are considered. We conclude that substantive welfarecriteria, or perhaps a reinterpretation of the standard liberal approachto welfare as about fundamental preferences, may serve as a betterguide to policy where preferences are endogenous.