This paper examines three programmatic responses to the problem of cognitive bias: virtue epistemology, epistemic paternalism, and epistemic collectivism. Each of these programmes focuses on a single level of epistemic analysis: virtue theorists on individuals, paternalists on environments, and collectivists on groups. I argue that this is a mistake in light of the fact that cognitive biases arise from interactions between these three domains. Consequently, epistemologists should spend less time defending these programmes, and more time coordinating them. This paper offers empirically based arguments for the interactionist approach, and contends that its adoption is an essential step for minimizing bias in empirical science.