Theories of choice, and their legal consequences, dramatically differ based on whether they are premised on rational or boundedly rational actors. This chapter describes the interactions between, and regulatory implications of, three types of uncertainties that the selection of an adequate theory of choice requires. First, it suggests that Knightian uncertainty concerning the distribution of degrees of rationality between regulatees obtains in many regulatory areas. More recently, this has been described as a ‘knowledge problem’ of behavioural law and economics. This chapter argues that the best regulatory response to the knowledge problem is to frame regulation as a problem of decision making under uncertainty. Second, with the rise of machine learning, it is arguably becoming ever more possible to estimate the level of bias, or even entire rationality quotients, of individual regulatees. This opens the potential for, but also the pitfalls of, personalised law. Third, even Big Data analytics generally only offers a snapshot of a distribution of rationality at one moment in time. Recent economic analyses have suggested behavioural heterogeneity can evolve over time in unpredicted ways that may lead to unforeseen consequences, leading to economic complexity. This calls for a greater role of standards, as opposed to rules, in regulating environments with dynamic behavioural heterogeneity. The chapter focuses on the normative implications of different types of uncertainty. In the end, theories of choice can aid more transparent normative trade-offs but they cannot replace the value judgments, and normative discourses, that balance the involved interests.