Why do research institutions celebrate interdisciplinarity in their mission statements while their hiring committees, journals, and grant panels often quietly penalise it? The puzzle is not merely administrative. It points to something about how knowledge communities come to accept claims at all, and why claims that satisfy one community can strike another as unfounded, trivial, or not even claims.
The standard story treats interdisciplinary difficulty as a logistics problem. People sit in different buildings, use different jargon, and attend different conferences. Fix the scheduling, provide a glossary, and collaboration should follow. But this underestimates what a discipline actually is. It is not just a body of facts. It is a community of inquiry with its own tacit sense of what counts as evidence, what counts as an explanation, and which questions are worth asking.
Thomas Kuhn called these shared commitments paradigms, and he noticed that scientists working within different paradigms often talk past each other. Interdisciplinary work asks people to collaborate across exactly those boundaries. Understanding why that is hard, and when it nevertheless succeeds, tells us something important about how collective knowledge is built.
Trading Zones and the Pidgins of Science
In Image and Logic (1997), the historian of physics Peter Galison confronted a puzzle. Twentieth-century physics was split among theorists, experimentalists, and instrument builders, each with distinct practices and, in some sense, distinct concepts. If Kuhn was right that different frameworks are partly incommensurable, how did these subcultures ever manage to work together on a single experiment?
Galison's answer borrowed from anthropology. When cultures with no shared language trade, they do not first achieve mutual understanding. They develop a pidgin, a stripped-down contact language sufficient for exchange. If the trade persists, a pidgin may become a creole: a full language with native speakers. Galison argued that scientific collaboration happens in analogous trading zones, local sites where groups coordinate around shared objects and procedures while retaining very different understandings of what those things mean.
His examples include wartime radar work at MIT's Radiation Laboratory, where physicists and electrical engineers developed a hybrid vocabulary of circuits and fields. Another is the early Monte Carlo simulations at Los Alamos, which drew together mathematicians, physicists, and engineers who disagreed about whether computer simulations were experiments, calculations, or something entirely new. Agreement on how to proceed did not require agreement on what anything ultimately was.
This is both encouraging and sobering. It is encouraging because collaboration does not demand that a biologist become a physicist. It is sobering because a pidgin is thin by design. It enables exchange while leaving each party's deeper assumptions untouched. Many interdisciplinary projects stall at exactly this stage: participants trade outputs successfully but never examine the background commitments that shape what each side thinks those outputs mean. The sociologists Harry Collins and Robert Evans later described the next step as interactional expertise, which is fluency in another field's language without the ability to practise it.
TakeawayCollaboration can begin with a shared vocabulary long before it reaches shared understanding. Mistaking the first for the second is where many projects quietly fail.
When Standards Collide
If trading zones explain how disciplines coordinate, they also expose where coordination breaks down. The deepest obstacles are not terminological but normative. Disciplines differ in their standards of evidence, their standards of explanation, and their standards of significance. These differences are rarely stated aloud because, within each field, they simply feel like good judgement.
Consider evidence. A particle physicist typically requires a five-sigma result before announcing a discovery, while much of experimental psychology has historically accepted a p-value below 0.05. An epidemiologist may rank randomised trials above observational data, while an anthropologist may regard months of fieldwork in a single village as more revealing than any survey. Each standard is calibrated to the field's characteristic errors. Transplanted elsewhere, each can look either reckless or absurdly demanding.
Explanation diverges as well. A molecular biologist may want a mechanism. An economist may be satisfied by a model that predicts well even if its assumptions are unrealistic. A historian may want a narrative that makes an event intelligible in its context. Debates over integrated assessment models in climate policy illustrate the friction. Economists' damage functions, which translate warming into lost output, have been criticised by some natural scientists as poorly grounded in physical impacts. Economists have replied that tractable aggregation is precisely the point.
Significance may be the subtlest conflict. What counts as an interesting finding depends on what a community already believes and what it is trying to achieve, a point Philip Kitcher has developed in his work on well-ordered science. A result that is groundbreaking to a sociologist may strike a statistician as unremarkable, and the reverse is also true. When collaborators disagree about significance, they cannot even agree on what success would look like. These are not failures of goodwill. They are the residue of each field's hard-won learning.
TakeawayEvery disciplinary standard is a compressed history of the mistakes that field has learned to avoid. That is why one field's rigour can look like another's blindness.
From Stapled Reports to Genuine Integration
Not all interdisciplinary work stalls. The question is what separates genuine epistemic integration from superficial combination. In genuine integration, the contributions of different fields reshape one another. In superficial combination, what you get might be called the stapled report: parallel chapters, each written to its own discipline's standards, bound together under a common title. Funders often cannot tell the difference, and sometimes neither can the participants.
Several conditions seem to matter. First, integration tends to succeed when it is organised around a shared problem rather than a shared theory. Susan Leigh Star and James Griesemer studied the Museum of Vertebrate Zoology at Berkeley and showed how specimens, maps, and field notes functioned as boundary objects. These were things robust enough to hold a common identity across groups, yet flexible enough to serve each group's distinct purposes. Concrete objects give collaborators something to argue about productively.
Second, integration requires that disagreements about standards be surfaced rather than deferred. Here Helen Longino's account of objectivity is instructive. She argued that objectivity is a property of communities, not individuals. It depends on recognised venues for criticism, genuine uptake of that criticism, publicly shared standards, and a tempered equality of intellectual authority. Interdisciplinary teams often fail the last condition. One field's methods quietly become the default, and the others are relegated to providing data or context.
Third, institutions must make room for the slow work of building interactional expertise. That work is costly and frequently unrewarded. A 2016 analysis in Nature of Australian Research Council grants found that proposals spanning more disciplines had consistently lower funding success. When peer review is organised by discipline, integrative work is judged by standards it was designed to transcend, and researchers learn to keep their collaborations shallow. The result is a system that praises integration in principle while selecting against it in practice.
TakeawayIntegration is less about combining answers than about letting each discipline's questions be changed by the others. That requires communities willing to have their standards challenged.
The difficulty of interdisciplinary research is not a sign that disciplines are obsolete or that collaboration is futile. It reflects something true about knowledge itself. Standards of evidence, explanation, and significance are social achievements, refined within communities over long periods.
If that is right, then institutions seeking integration cannot rely on proximity and goodwill alone. They need structures that make tacit standards explicit and distribute intellectual authority across fields. They also need to evaluate integrative work on its own terms rather than through the lens of any single discipline.
Perhaps the most useful shift is to treat disciplinary friction not as noise to be minimised but as information. When two fields disagree about what counts as a good answer, they reveal assumptions that neither would otherwise see.