Institutions often respond to failure by reforming themselves. Structures are redesigned, responsibilities are redistributed, procedures are rewritten, new units are created, reporting lines change and technologies are introduced, sometimes through relatively modest adjustments and sometimes through transformations extensive enough to alter the visible architecture of the organisation. Because reform so often follows the recognition that something has gone wrong, it is natural to interpret this sequence as evidence of learning: a problem occurred, the institution recognised it, and the institution subsequently changed. Yet this apparently straightforward progression conceals an important distinction, because an institution can be reformed without having revised the way it understands the problem that made reform necessary in the first place. Organisational change tells us that something has been altered, but it does not by itself establish that experience has changed the cognition through which the institution will encounter similar problems in the future.

This distinction is fundamental because organisational reform can arise for many reasons that have little to do with institutional learning. A government may restructure an agency because a minister demands a different distribution of authority, a regulator may impose new procedures, a budget settlement may force consolidation, a technological programme may redesign workflows, or a crisis may generate emergency arrangements that later become permanent. Each of these developments can produce genuine and consequential reform while leaving many of the assumptions, interpretations and judgement patterns of the institution substantially unchanged. The organisation may operate differently because the conditions under which it operates have changed, rather than because it has developed a different understanding of what happened. For that reason, REFORM OCCURRENCE ≠ LEARNING OCCURRENCE: the existence of reform demonstrates that the institution has been changed, but it does not yet tell us what the institution itself has learned.

Consider an organisation that repeatedly fails to recognise early signs of implementation problems. Following a major breakdown, leadership creates a new monitoring unit, introduces additional indicators and requires more frequent reporting. The reform may be entirely sensible and may even improve performance because information now travels more quickly, responsibilities are clearer and senior managers receive more regular updates. Yet if the institution continues to interpret success through the same narrow indicators that previously concealed emerging problems, the structure of monitoring has changed without a corresponding revision in the model through which reality is understood. More information is moving through a new architecture, but the institution may still be asking the same questions of that information and interpreting the answers through the same assumptions. Reform has therefore occurred, perhaps successfully, while the deeper cognitive problem remains substantially intact.

The distinction becomes especially visible when institutions copy reforms from elsewhere. Governments frequently adopt organisational arrangements associated with successful institutions in other jurisdictions: innovation laboratories, delivery units, digital transformation offices, central analytics teams, cross-functional structures or new governance committees. Such borrowing can be entirely rational, particularly when another organisation appears to have solved a comparable problem, but reproducing the visible form of a successful arrangement is not the same as reconstructing the reasoning that made that arrangement useful in its original setting. An institution may import an answer without having developed a sufficiently different understanding of the question. In these circumstances reform can create the conditions under which learning might subsequently occur, but the adoption of the new structure cannot itself demonstrate that learning has already taken place.

This does not diminish the importance of organisational reform. Existing structures can themselves prevent institutions from learning because reporting lines fragment information, incentives discourage candour, procedures preserve obsolete assumptions or authority is located too far from the knowledge required to exercise it intelligently. Changing such arrangements may therefore be indispensable, and in some circumstances deep institutional learning will necessarily produce structural reform because the institution’s previous architecture no longer fits what it has come to understand. The analytical point is simply that enabling learning and demonstrating learning are different things. A reform may remove barriers to institutional cognition, create better channels for evidence or make revision easier without yet proving that the institution has actually changed the way it perceives, interprets or judges.

Reform also has political and symbolic functions that can make the distinction harder to preserve. Following visible institutional failure, pressure for action can become intense, and doing nothing may reasonably be interpreted as denial, complacency or irresponsibility. New structures, procedures and programmes demonstrate responsiveness in ways that changes in interpretation are rarely able to do because organisational reform produces visible evidence that authorities have acted. This signalling function can be legitimate and even necessary, but the visibility of the response can obscure a more difficult question: what exactly has changed in the institution’s understanding of itself and of the problem? A new organisational chart can be published immediately, whereas a revised causal model or a more sophisticated interpretation of evidence may be much harder to observe. The result is a persistent temptation to use the visibility of reform as indirect evidence that cognitive revision has occurred.

That temptation helps explain why institutions can repeatedly reform familiar problems without necessarily escaping them. A coordination failure may produce a new committee; when coordination remains weak, additional reporting requirements are introduced; when delays persist, responsibilities are centralised or redistributed. Each intervention can alter the formal architecture while leaving the underlying conception of coordination largely untouched. Perhaps the institution continues to treat coordination primarily as a problem of communication when the deeper issue concerns conflicting incentives, incompatible categories, fragmented authority or different interpretations of the same evidence. If the diagnosis remains stable, repeated reforms can reorganise the answer without reconsidering the question, producing substantial administrative activity while preserving the cognitive mechanisms that continue to generate similar difficulties.

Cognitive continuity can survive structural reform in more subtle ways as well. A department may be reorganised, but the same professional communities, routines and interpretive habits migrate into the new structure. People acquire different titles and report through different channels while continuing to understand problems through familiar categories. New procedures may be layered onto old assumptions, and new technologies may accelerate processes whose underlying logic has never been reconsidered. From outside, the institution can appear substantially transformed because its formal arrangements have changed; internally, much of the cognition through which it interprets reality may remain remarkably stable. Organisational novelty and cognitive novelty therefore need to be examined separately rather than treated as different expressions of the same transformation.

The same distinction matters when reform is evaluated. Conventional assessments reasonably ask whether new structures were established, whether procedures were adopted, whether staff complied and whether measurable outcomes improved. These questions reveal whether the reform was implemented and whether it produced useful effects, but they do not fully reveal whether institutional learning occurred. A reform can be implemented successfully while leaving the organisation no better able to diagnose the next variation of the problem it was designed to address. Conversely, a reform can encounter implementation difficulties while still producing valuable learning if the experience exposes a false assumption, reveals an overlooked dependency or changes the institution’s understanding of why previous arrangements failed. Reform success and institutional learning can therefore diverge because they describe different dimensions of change.

The more demanding evidence of learning lies in whether experience modifies future-generating cognition. Did the institution revise an assumption that previously shaped its decisions? Did it change which evidence it considers relevant? Did it develop a distinction that allows a problem to be recognised earlier? Did it alter how uncertainty affects judgement, or change who is treated as a legitimate source of knowledge? These kinds of changes indicate that experience has done more than trigger a response: it has modified some part of the cognitive architecture through which future situations will be perceived and interpreted. Organisational reform may embody those modifications, but the modifications themselves are what make the process learning rather than simply change.

This distinction is also important when reform depends heavily on particular individuals. A new leadership team may redesign an organisation and produce significant performance improvements, but if the effectiveness of the new arrangements depends primarily on the judgement, relationships or tacit knowledge of those leaders, the institution may have changed without fully learning. Years later, the formal structures may remain while the capability that made them work disappears with the people who originally created them. A learning institution would therefore need to preserve not merely the new arrangement but some understanding of why it works, which distinctions matter within it and how future actors should recognise when its assumptions no longer hold. Reform becomes institutionally meaningful when its cognitive rationale can survive beyond the particular people who initiated it.

Digital transformation makes the same issue especially visible. A public organisation can redesign workflows, digitise services, centralise data and introduce artificial intelligence into decision processes, producing changes that are extensive in both organisational and technological terms. Yet technological transformation does not automatically imply cognitive transformation. An institution can adopt sophisticated tools while preserving the same assumptions about what matters, which evidence counts and which problems deserve attention; it can automate existing judgement without reconsidering whether that judgement remains appropriate. Technology may make the institution faster, more consistent or more efficient while leaving its interpretation of reality largely untouched. Conversely, digital tools can support genuine learning when experience with them leads the institution to revise categories, expose previously hidden uncertainty, reconsider decision criteria or develop a different understanding of where human judgement is required.

The distinction therefore should not be used to oppose reform, but to make reform more intellectually demanding. Institutions often need new structures, procedures and technologies, and learning itself may reveal that existing arrangements are no longer adequate. What should be resisted is the assumption that the existence of reform constitutes sufficient evidence that the institution has learned. If organisational activity becomes a substitute for cognitive revision, institutions can expend enormous political and administrative capacity rearranging structures while leaving the mechanisms that generated previous failures substantially intact. The relevant question after reform is consequently not only what has changed, but what the institution now understands differently because of the experience that produced the change.

A cognitively mature reform process would therefore be able to explain not only which units were reorganised, which procedures were rewritten or which technologies were introduced, but also which assumptions were revised, which distinctions became clearer, which forms of evidence acquired new significance and how these cognitive changes are expected to affect future judgement. Sometimes the answer will reveal that profound learning has indeed occurred and that structural reform was its necessary expression. At other times, reform may have been valuable primarily because it responded to political, operational or technological requirements rather than because it reflected a deeper revision of institutional cognition. Both forms of change can be legitimate, but they should not be confused.

Reform is not automatically learning because an institution can redesign structures, redistribute authority, introduce new procedures and replace technologies while preserving much of the same underlying interpretation of reality. Reform demonstrates that the organisation has been changed; institutional learning requires something further, namely that experience has modified the cognitive architecture through which the institution will perceive, interpret and respond to the next problem it encounters.