It is tempting to imagine institutional intelligence as a property of what an organisation contains. We look for knowledgeable people, reliable databases, strong analytical teams, well-designed procedures, effective technologies and access to external expertise, and then assume that the institution’s cognitive capability will somehow emerge from the quality of those components. Yet two institutions can possess remarkably similar resources and still differ dramatically in what they are able to understand, decide and do. The difference may lie not in the components themselves but in the relationships that connect them. Who can reach whom, which signals cross organisational boundaries, where authority and expertise intersect, how evidence travels, which perspectives are brought together before a decision and which forms of knowledge remain separated all help determine what the institution is cognitively capable of becoming.

This means that relationships are not merely social features surrounding institutional cognition. They are part of its architecture. A highly capable expert who cannot reach the decision in which their knowledge matters may contribute far less to institutional cognition than their individual competence would suggest. A rich dataset that remains isolated from the teams able to interpret it may have little practical cognitive value. A local unit that repeatedly detects an emerging problem may still fail to influence institutional understanding if the relationship between observation and central attention is weak. In each case, the institution contains a potentially valuable cognitive resource, yet the configuration through which that resource might become part of collective judgement is insufficient.

The distinction extends the principle that a resource is not the same as a capability. Resources become capabilities only through configurations that allow them to be mobilised appropriately, and relationships are among the most important elements of those configurations. An institution with experienced staff, extensive records and sophisticated technology does not automatically possess the capability to integrate evidence, recognise emerging problems or make decisions under uncertainty. Those capabilities depend partly on whether the relevant people, information and authority can be brought into relation when the situation requires it. The same components arranged through different relationships can therefore produce different institutional capabilities.

This helps explain why apparently small changes in organisational connection can sometimes produce disproportionately large changes in institutional performance. Creating a direct relationship between an operational team and an analytical unit may shorten the path through which weak signals become visible. Establishing a regular interface between legal, technical and policy specialists may allow a problem to be interpreted in several dimensions before a proposal hardens into a decision. Giving local offices a reliable route for escalating anomalies may reveal patterns that no central dataset captures quickly enough. None of these changes necessarily adds new knowledge to the institution. They alter how existing knowledge can encounter other knowledge.

For the same reason, relationships should not be reduced to communication channels. Communication matters, but institutional relationships also carry expectations, permissions, dependencies, authority and patterns of attention. Two units may exchange information frequently while still failing to form a useful cognitive relationship if neither can influence how the other interprets or acts on what is shared. Conversely, a relationship used only occasionally may be cognitively critical if it gives specialised knowledge access to a high-consequence decision at the moment it is needed. What matters is not simply whether information flows but what the relationship makes possible.

Authority is especially important here because institutional cognition does not end with understanding. An institution often has to convert interpretation into action, and the relationship between expertise and authority shapes whether that conversion can occur. A technical team may recognise a serious vulnerability while lacking the authority to change a process. A policy unit may have decision rights but insufficient access to operational knowledge. A central body may possess strategic visibility while local actors hold the contextual information needed to judge whether an intervention is workable. The cognitive quality of the institution depends partly on whether these different forms of knowledge and authority are connected in ways that permit informed action without collapsing legitimate distinctions between roles.

Relationships also influence what the institution notices. Attention is not distributed evenly across organisations. Some actors are routinely consulted, some signals travel through established reporting systems, some problems fit categories that already have owners and some perspectives remain peripheral until a crisis makes them difficult to ignore. Relational architecture therefore shapes visibility. If certain communities, local units, professional groups or external partners occupy weak positions in the institutional network, the information they hold may enter institutional awareness late or only in distorted form. The institution’s cognitive map of reality can consequently reflect not only what evidence exists but which relationships allow that evidence to become institutionally salient.

This is one reason collaboration should not be treated as a synonym for relational cognition. Collaboration usually implies people working together toward a shared task, and it can undoubtedly strengthen institutional capability. But many cognitively important relationships do not involve close teamwork. An independent oversight body may contribute precisely because its relationship to an institution preserves distance. A frontline service may provide important evidence without participating in central decision-making. A scientific advisory mechanism may need enough autonomy to challenge assumptions rather than becoming embedded in operational consensus. The relevant question is not whether everyone collaborates more, but whether the architecture of relationships allows the right forms of knowledge, challenge and authority to meet without destroying the distinctions that make them valuable.

Relational architecture can also create vulnerabilities. If too much institutional knowledge moves through a small number of intermediaries, those actors can become hidden cognitive bottlenecks. If a decision depends on informal personal relationships rather than durable institutional interfaces, capability may disappear when particular individuals leave. If information can travel upward but not horizontally, different units may repeatedly rediscover the same problem without recognising that others are seeing it too. A dense network is therefore not automatically a better one. More connections can create duplication, overload and ambiguity as easily as they can improve understanding. The question is whether relationships are configured around the cognitive demands the institution actually faces.

Technology can change this configuration, but not necessarily in the way organisations expect. Digital platforms may make it easier to circulate information, discover expertise or connect units that previously worked in isolation, yet merely increasing access does not guarantee that institutional cognition improves. A shared platform may expose everyone to more information while making it harder to distinguish what matters. Artificial intelligence may help route questions toward relevant expertise or identify patterns across organisational boundaries, but it can also reinforce existing relational structures if the data, permissions and workflows on which it relies reproduce the same separations already present in the institution. Technology changes relational possibilities; it does not decide which relational architecture will support better cognition.

The design of institutional relationships is therefore inseparable from the design of cognitive capability. If an institution needs to recognise emerging risks earlier, it may need relationships that connect weak signals from dispersed locations to places where they can be compared. If it needs better judgement under uncertainty, it may need relationships that bring different forms of expertise into contact without forcing premature consensus. If it needs to act coherently across organisational boundaries, it may need relationships that connect authority, knowledge and implementation more reliably. In each case, cognitive performance depends not only on improving individual components but on arranging connections among them.

This relational perspective also clarifies why institutional knowledge can be distributed without necessarily being integrated. Knowledge may exist across people, documents, systems and routines while the relationships among those elements determine whether the institution can assemble a meaningful picture when required. The integration problems that follow from specialisation and fragmentation deserve analysis in their own right, but they rest on a more basic proposition: distributed cognition is possible only through some architecture of connection. Before asking whether that architecture works well, we need to recognise that it is part of the cognitive system itself.

Once this is visible, organisational charts begin to look different. A chart can show formal reporting lines, but institutional cognition also depends on relationships that may cut across those lines: advisory relationships, escalation paths, informal translation roles, professional networks, external partnerships, data-sharing arrangements and temporary cross-functional structures. Some of these may be highly formalised, while others remain tacit. Together they create a relational topology through which the institution senses, interprets and acts on its environment.

Institutions therefore do not think simply because they contain intelligent people, useful information and capable technologies. They think through the relationships that allow those elements to encounter one another in institutionally meaningful ways. Knowledge must reach attention, evidence must encounter interpretation, expertise must connect with authority and local observation must sometimes become collective awareness. The components matter, but so does the configuration among them. To understand institutional cognition, we have to ask not only what an institution has, but how what it has is related.