The empirical part of the study focuses on the evaluation of research proposals in a national research funder in Finland, the Academy of Finland. Drawing on analyses of research proposals and peer review deliberations, the study explores the various ways in which scholars coordinate, negotiate, and modify different disciplinary regimes in the pursuit of high-quality scientific knowledge. Based on the empirical findings and a review of the literature on interdisciplinarity, social epistemology, and science policy, the study emphasizes the importance of considering epistemic accountabilities in a contextsensitive, open-ended manner in knowledge production and evaluation. The study makes both a theoretical and a practical contribution. First, it provides a complementary perspective on the changing governance of science by articulating the notion of interdisciplinary accountability. While recent debates have emphasized problem solving and public accountability as important indicators of legitimate science today, this study argues that accountability across academic disciplines holds an equal promise of more relevant and reliable knowledge. Interdisciplinary accountability is thus a socioepistemic mechanism for responsible science, and serves as a counterforce to disciplinary autonomy as well as the "tyranny" of political or economic forces over epistemic values. Second, the study makes a practical contribution to the evaluation of interdisciplinary research. To this end, it articulates a framework for conceptualizing interdisciplinary accountability in research proposals, and considers ways to include interdisciplinary accountability in peer review. The framework helps to identify the relevant epistemic stakeholders, the functions and benefits of proposed research, as well as the methodogical procedures for accomplishing the stated goals, which constitute the prerequisite for any evaluative act. As for the evaluative act itself, the study suggests using interdisciplinary dialog between reviewers as a type of epistemic standard. A reasonable strategy is to mix experts from different but not disparate fields, and select generalist panel members who possess a broad knowledge beyond any one academic field.
