It is widely supposed that progress in understanding and treating neurological conditions requires collaboration across disciplines and sectors. Proponents of collaboration often argue for its necessity by appealing to the epistemic goods it is ideally supposed to deliver or to actual cases where such goods were the outcome. In the translational neuroscience space, for instance, interdisciplinary and transdisciplinary research teams are seen as a possible means of bridging the troubling gap between basic science and clinical research — the so-called “Valley of Death” — which stymies progress towards the development of effective treatments for neuropsychiatric and neurodegenerative diseases (e.g., Seyhan, 2019). Researchers themselves are often quick to bring up the value, or perhaps even the necessity, of interdisciplinary collaborations to their work. Recently, however, the challenges associated with interdisciplinary work have begun to garner more attention (Fam & O’Rourke, 2020; Fitzgerald et al., 2014; MacLeod, 2018).

In addition to the institutional and conceptual obstacles that large-scale, interdisciplinary collaborations face, researchers’ own expectations may play a significant, if underappreciated role. These expectations may be informed by a familiarity with what Fitzgerald et al. (2014) call “ideal-type accounts of cross-disciplinary labour” (p. 703), which may reflect a more general tendency in the literature towards metaphorical, abstract, and idealized models of interdisciplinary exchange (Boix Mansilla, 2010). That researchers employ idealizations to understand interdisciplinary collaborations is understandable. Large-scale collaborations of this kind involve multiple individuals, labs, and institutions, cut across disciplinary and institutional boundaries, draw on disparate research traditions, tools, and techniques, demand significant resource investments (e.g., time, money, and attention), and require sustained, coordinated action and communication across great distances and time. In short, the same reasons that underpin the use of idealization in science generally apply here as well; interdisciplinary collaborations are extremely complex, and given our cognitive and technical limitations, idealizations are necessary to make their investigation more tractable (Potochnik, 2019; Wimsatt, 2007).

By appeal to a case study of collaboration in translational neuroscience, we aim to show that the idealized models of collaboration that we find in the literature may serve to obscure the difficulties that researchers engaged in collaboration face on the ground. We argue that these models are ideal in two distinct senses of the word; they are ideal in terms of their structure and content (i.e., they are abstract and often include intentionally false features to serve their purposes), and they are ideal in the normative sense in so far as they describe highly successful cases (both in terms of outcome and process). Call these two senses idealizationcontent and idealizationnormative respectively.

Our positive proposal mirrors this distinction. Idealized models excel at making the complex dynamics of interdisciplinary collaboration more intelligible. Supplementing these abstract, analogical models with more detailed case studies of contemporary, large-scale collaborations, grounded in the actual experiences of participants — by incorporating material from semi-structured interviews or ethnographic work, for instance — may help mitigate the risks associated with idealizationcontent. At the same time, developing idealized models that focus on those factors that contribute to the failure, rather than success of such collaborations, helps address concerns associated with idealization.