Coordination is usually described as an organisational problem. When several departments must contribute to the same outcome, institutions create committees, establish reporting arrangements, appoint coordinators, introduce shared procedures or build digital systems intended to connect their work. These mechanisms can be useful, but they can also encourage a narrow understanding of what coordination actually requires. An institution may succeed in bringing people together, exchanging information and clarifying responsibilities without becoming substantially better at understanding a problem that crosses organisational boundaries. The deeper challenge is cognitive. When relevant knowledge is distributed across specialised units, coordination determines whether those separate fragments can be combined into a form of judgement that no single part of the institution could have produced alone. In this sense, coordination is not merely an administrative arrangement around institutional cognition; it is itself an institutional cognitive capability.

This becomes visible whenever a public problem exceeds the perspective of the department responsible for one of its components. A housing problem may also involve employment, health, transport, education and local government; an environmental intervention may depend on economic regulation, scientific evidence, land management and community knowledge; a digital public service may require legal, technical, operational and behavioural understanding simultaneously. Specialisation allows institutions to develop expertise in each of these domains, but the existence of expertise does not automatically produce an integrated understanding of the problem. Each unit sees something that matters, yet none necessarily sees enough. Coordination becomes cognitively important because it provides the means through which partial perspectives can encounter one another and contribute to a more adequate institutional interpretation.

Simply moving information between departments is not sufficient. A report can be circulated widely without affecting anyone’s judgement, just as a meeting can bring together every relevant actor while leaving their underlying interpretations unchanged. Information becomes cognitively consequential only when it can enter the reasoning of another part of the institution. This may require translation between professional languages, recognition that different units classify the same situation differently, or an ability to understand why evidence that appears marginal from one administrative perspective is highly significant from another. Coordination therefore involves more than ensuring that information reaches the correct destination. It requires conditions under which distributed knowledge can become mutually intelligible and capable of modifying collective understanding.

The distinction is important because institutions often respond to coordination problems by increasing communication. More meetings are scheduled, additional reports are required, shared inboxes are created and cross-departmental groups proliferate. Sometimes these interventions solve the problem, particularly when relevant actors simply lacked access to one another. At other times, however, communication volume increases while collective understanding barely improves. Departments may exchange more information without changing what they consider relevant, how they interpret evidence or which consequences they perceive as important. An institution can therefore become highly connected administratively while remaining weakly coordinated cognitively. The question is not only whether its parts communicate, but whether communication enables them to think differently together.

This cognitive dimension helps explain why disagreement is not necessarily evidence of coordination failure. When specialised units encounter the same problem from different positions, they may reasonably reach different conclusions because they possess different knowledge, responsibilities and exposure to consequences. Effective coordination should not erase those differences prematurely. If integration simply forces everyone into a common interpretation before the reasons for disagreement have become visible, the institution may achieve apparent coherence by suppressing useful information. Cognitive coordination requires something more demanding: different perspectives must be able to remain distinct long enough to reveal what each knows, while the institution develops enough shared understanding to determine where those differences matter for collective action.

A health department and a finance department, for example, may interpret the same intervention through very different forms of evidence. One may focus on population outcomes and distributional effects, while the other emphasises expenditure, fiscal sustainability and opportunity costs. Neither perspective is automatically more institutional than the other. The cognitive task is to create a judgement in which both can become consequential without pretending that they are interchangeable. Coordination succeeds when the institution can reason across these differences rather than merely choose which department’s frame will dominate. The resulting judgement may still involve difficult trade-offs, but those trade-offs become more intelligent because the knowledge distributed across the organisation has actually entered the decision.

Authority complicates this process because the unit that possesses relevant knowledge is not always the unit authorised to act on it. Frontline organisations may recognise emerging problems long before central leadership does, while central bodies may understand strategic constraints that are invisible locally. Regulatory agencies may possess formal authority over decisions whose practical consequences are best understood elsewhere. Coordination therefore has to connect not only knowledge with knowledge, but knowledge with consequential judgement. An institution can have excellent information flows and still fail cognitively if the architecture through which authority operates prevents relevant knowledge from modifying decisions.

This is one reason why coordination cannot be reduced to goodwill. Public institutions are often populated by people who are entirely willing to cooperate but operate within structures that make intelligent collective judgement difficult. Performance measures may reward local optimisation, legal mandates may restrict information sharing, professional languages may make evidence difficult to translate, timelines may differ between organisations and accountability arrangements may encourage units to minimise risks that are then transferred elsewhere. Asking people to collaborate more enthusiastically cannot by itself resolve these structural conditions. Coordination becomes a capability when the institution develops reliable ways of making distributed knowledge consequential despite the organisational boundaries through which that knowledge is produced.

The same point applies to digital integration. Shared data platforms and interoperable systems can substantially improve institutional coordination because they reduce technical barriers to accessing information, but accessibility alone does not determine meaning. Two departments can observe the same dataset and derive different implications from it because they use different categories, thresholds or causal assumptions. A common dashboard can make evidence simultaneously visible without making it cognitively shared. Digital infrastructure therefore becomes valuable for coordination when it supports the interpretive work required to connect different forms of knowledge, rather than merely increasing the speed at which information moves between existing organisational perspectives.

Artificial intelligence intensifies this challenge. AI systems may combine data from multiple institutional domains, detect patterns that no individual department could identify and create new possibilities for cross-organisational analysis. Yet the ability to aggregate information computationally should not be mistaken for the ability to coordinate institutionally. Models themselves encode categories and assumptions, and their outputs must still become meaningful within organisations whose responsibilities, expertise and authority remain distributed. An AI system might reveal a pattern spanning health, employment and social services, for example, while no existing institutional arrangement possesses the capacity to interpret that pattern collectively or decide what it means for action. Greater computational integration can therefore expose the absence of cognitive coordination rather than automatically solve it.

Understanding coordination as cognition also changes what successful coordination looks like. The goal is not necessarily to produce consensus, eliminate boundaries or create a single institutional perspective. Complex institutions need differentiation because no organisation can hold every form of expertise in one place. The aim is instead to make differentiation cognitively productive: specialised units should be able to preserve the knowledge generated by their distinct positions while participating in processes through which those perspectives can alter wider institutional judgement. A well-coordinated institution is not one in which everyone thinks alike, but one in which different forms of knowledge can interact without becoming either isolated or prematurely homogenised.

This perspective makes interfaces especially important. Whenever responsibility, information or interpretation passes from one part of an institution to another, something cognitively significant can happen. Knowledge can be preserved, translated, distorted, simplified or lost; uncertainty can become visible or disappear; assumptions can be challenged or silently reproduced. Interfaces are therefore not merely administrative seams between otherwise complete organisational units. They are locations where institutional cognition is assembled. An institution that pays attention only to the quality of its departments while neglecting the quality of these interfaces may repeatedly improve its components without improving its capacity to understand problems that require those components to think together.

The practical implication is that coordination should be evaluated partly by the quality of collective judgement it makes possible. Counting meetings, shared systems, interdepartmental agreements or exchanged documents can reveal whether coordination mechanisms exist, but not whether they perform their cognitive function. More revealing questions concern whether important knowledge from one part of the institution can alter judgement elsewhere, whether conflicting interpretations become visible before decisions are made, whether local information can influence central reasoning and whether the institution can construct an understanding of cross-boundary problems that is richer than the perspective available to any participating unit individually. These are demanding criteria because they shift attention from coordination activity to coordination capability.

Institutions need specialised components, and those components will continue to possess different responsibilities, languages and perspectives. The challenge is not to eliminate this fragmentation but to prevent necessary differentiation from becoming cognitive isolation. Coordination becomes an institutional cognitive capability when distributed knowledge can cross organisational boundaries, remain intelligible enough to interact with other forms of knowledge and become consequential for collective judgement. The institution is coordinated not merely when its parts work alongside one another, but when their separate perspectives can combine into an understanding that none of them could have produced alone.