Skip to main content
Abstract soft mint-green flowing forms suggesting non-linear cognitive pathways intersecting with geometric structures
Synthetic Minds at Work

Organisations Still Built for Linear Minds

On structure, sequence and the limits of alignment

April 14, 2026

There is an assumption embedded in most organisational structures that is rarely examined directly, although it shapes almost every aspect of how work is coordinated, evaluated and brought to completion. It is the assumption that thinking progresses in a largely linear manner.

Tasks are defined, broken down into discrete steps and assigned in sequence. Information is expected to move in a controlled flow, from input to processing to output. Decisions are framed as points of closure that follow from sufficient analysis, and communication is organised in ways that privilege clarity, order and progression. This model has proven remarkably durable. It does not require that all individuals think linearly, but it requires that their thinking can be translated into a linear form before it becomes actionable within the organisation. What cannot be expressed as sequence struggles to be integrated, and what cannot be integrated tends not to influence outcomes.

What cannot be expressed as sequence struggles to be integrated, and what cannot be integrated tends not to influence outcomes.

The Sequential Filter

The consequence is subtle yet pervasive, because cognitive variation is never excluded directly so much as filtered through the requirement that it conform to a structure that was never designed to accommodate it. For individuals whose thinking already aligns with sequential processing, this translation occurs with relatively little effort, as ideas can be stabilised early, structured in advance and communicated in a form that moves smoothly through existing systems.

For others, the process is considerably less straightforward, since thought may develop through association rather than sequence, insight may emerge through iteration rather than through prior structuring, and understanding may depend on holding multiple possibilities in tension rather than resolving them into a single line of reasoning at the outset. When such processes are forced into linear form prematurely, what tends to be lost is less a matter of correctness than of dimensionality.

This is not merely a theoretical observation. When Henry Mintzberg studied how senior managers actually spent their time, he found that the prevailing model of the executive as a reflective, sequential planner bore almost no resemblance to reality. Managers worked in brief, fragmented episodes, constantly switching between unrelated topics, relying on verbal communication and intuitive judgement far more than on formal analysis. The linear model of management, Mintzberg argued, was a post-hoc rationalisation imposed on a fundamentally non-linear process.1

When such processes are forced into linear form prematurely, what tends to be lost is less a matter of correctness than of dimensionality.

How Structure Shapes Thought

The organisation, however, has limited capacity to engage with dimensionality that cannot be reduced to sequence, given that meetings are structured around progression, reporting systems require defined outputs, and decision processes are built on the expectation that ambiguity will decrease over time rather than remain an active component of the work. Under these conditions, what matters is less whether non-linear thinking exists than whether it can survive the act of translation.

Often, it survives only partially. Ideas are simplified in order to be communicated, exploratory threads are collapsed into single narratives, and unresolved tensions are presented as conclusions, less because they have been resolved than because the structure demands resolution. This creates a form of distortion that is difficult to detect from within the system, since outputs appear coherent, decisions appear justified, and the process gives every impression of functioning well, while the extent to which the structure has shaped the thinking it receives remains largely invisible.

Karl Weick's research on sensemaking in organisations demonstrated precisely this mechanism, showing that organisations do not simply process pre-existing information so much as actively construct the reality they then respond to. The structures through which information flows determine which signals are amplified and which are suppressed, which interpretations gain traction and which are discarded before they can be articulated. In Weick's account, organisational structure functions less as a neutral container for thought than as a filter that determines what counts as thought in the first place.2

The consequences of this selection extend well beyond individual contribution, since over time it influences how problems are approached in the first place. If only certain forms of reasoning can be integrated effectively, individuals learn, often implicitly, to shape their thinking in ways that fit those forms, with the result that exploration becomes constrained by anticipated translation and questions are framed in ways that lead more readily to acceptable answers, gradually narrowing the range of variation the system is able to express.

Organisational structure functions less as a neutral container for thought than as a filter that determines what counts as thought in the first place.

The Stability Trap

In stable environments, this narrowing can be advantageous. Sequential processing supports coordination. Predictable structures reduce friction. Decisions can be made and implemented without continuous renegotiation. Toyota's production system, widely regarded as one of the most effective organisational models of the twentieth century, demonstrated the extraordinary efficiency that becomes possible when every process is defined, sequenced and continuously refined within a linear framework.3

The limitations emerge when the nature of the work itself becomes less linear. Problems that involve uncertainty, interdependence and evolving conditions do not lend themselves easily to predefined sequences. They require forms of engagement that can tolerate ambiguity, revisit assumptions and allow for partial understanding to coexist with action. In such contexts, the expectation that thinking should be fully structured before it is acted upon becomes a constraint rather than a support.

The pharmaceutical industry offers a particularly instructive illustration. Drug development follows one of the most rigorously sequential processes in any industry, moving from target identification through lead optimisation, preclinical testing and phased clinical trials to regulatory submission. Each stage has defined entry and exit criteria, and each depends on the successful completion of the one before it. Yet the actual scientific thinking that drives discovery is profoundly non-linear. Researchers at Roche have described how promising compounds are frequently abandoned less because the science fails than because the organisational process cannot accommodate the iterative, back-and-forth nature of the inquiry. The sequential structure, designed to manage risk, simultaneously constrains the cognitive processes most likely to produce breakthrough results.4

In stable environments, this narrowing can be advantageous. The limitations emerge when the nature of the work itself becomes less linear.

The Persistence of Misalignment

Yet the structures themselves tend to persist. Organisations rarely redesign their underlying models of coordination in response to changes in the nature of the problems they face. They adapt incrementally, adding layers of process, introducing new tools, refining existing workflows, while leaving the core assumption intact.

The result is a growing mismatch between the cognitive demands of the work and the organisational infrastructure through which that work must be expressed. Work becomes more complex, less predictable and more dependent on forms of thinking that do not unfold in sequence, while the organisation continues to operate as though sequence remains the primary mode through which value is produced.

This mismatch is often experienced as inefficiency, manifesting in processes that take longer, decisions that require more iteration, and communication that becomes increasingly demanding. The intuitive response is to improve alignment, increase visibility and impose clearer structure, yet these responses address symptoms while leaving the underlying assumption untouched. If the structure itself privileges a form of thinking that is no longer sufficient for the problems at hand, then improving the efficiency with which that structure operates may in some cases reinforce the very limitation it seeks to resolve. The Decision Architecture series examines this dynamic from the organisational side, and in When Decisions Refuse to End the consequences become particularly visible, as organisations designed for sequential closure encounter problems that resist it.

Clayton Christensen's work on disruptive innovation documented a version of this dynamic at the level of entire industries. Established companies failed to respond to disruptive technologies less because their managers lacked intelligence or foresight than because their organisational processes and resource allocation systems were optimised for the existing business, rendering the structures that made them efficient in one context structurally incapable of responding to another.5 The parallel to cognitive architecture runs in the same direction, since organisations optimised for linear processing are configured in ways that go beyond merely being slow to accommodate non-linear thinking, because they actively suppress it.

Organisations optimised for linear processing are configured in ways that go beyond merely being slow to accommodate non-linear thinking, because they actively suppress it.

The Generative Layer

Generative systems introduce an additional layer to this dynamic. By enabling thinking to be externalised and iterated upon, they reduce the cost of translating non-linear processes into more structured forms. Ideas can be explored through interaction, partially stabilised and then presented in ways that align more closely with organisational expectations.

This can be experienced as a form of progress. The organisation receives more coherent outputs. Individuals are better able to express their thinking. The gap between cognitive process and organisational structure appears to narrow. Microsoft's 2025 Work Trend Index found that 75 per cent of knowledge workers were already using generative AI at work, with a significant proportion reporting that it helped them structure and communicate ideas more effectively than they could unaided.6

What changes, however, is not the structure itself. The underlying assumption remains. Thinking must still arrive in a form that can be processed linearly. The difference is that the effort required to produce that form is redistributed, often from the individual to the system. This creates the appearance of alignment without necessarily resolving the mismatch.

The Agile movement attempted something more fundamental. By replacing sequential planning with iterative development cycles, Agile explicitly acknowledged that the best outcomes often emerge from processes that cannot be fully specified in advance. Yet even Agile, as it has been adopted at scale, has frequently been re-linearised. Spotify's widely cited "squad model" was abandoned by Spotify itself after it became clear that the framework had been interpreted as a fixed structure rather than as a set of evolving principles.7 The tendency to impose sequence on non-sequential processes appears to be deeply embedded in how organisations understand coordination.

What changes is not the structure itself. The underlying assumption remains, namely that thinking must still arrive in a form that can be processed linearly.

What Remains Invisible

The organisation becomes more capable of handling outputs that conform to its expectations, while the diversity of underlying cognitive processes continues to be shaped, and in some cases constrained, by those very expectations. What matters here is less whether organisations can accommodate different ways of thinking than whether they are prepared to reconsider the structures through which thinking becomes work.

As long as the requirement for linear expression remains central, non-linear cognition will continue to be translated rather than integrated, and the system will function within a narrowed range of possibilities. What appears as efficiency may therefore be, in part, a reflection of what has been made invisible, and what remains invisible is by no means absent so much as unable to take a form the organisation knows how to process.

Amy Edmondson's research on psychological safety at Harvard Business School has shown that the organisations best able to learn from failure and adapt to novel conditions are those that create explicit space for incomplete, provisional and contradictory thinking to be expressed without penalty.8 This is, in effect, a structural accommodation for non-linear cognition, a recognition that the path from uncertainty to insight does not always pass through the orderly stages that organisational processes assume.

The distinction matters because the AI-augmented workplace risks reinforcing precisely the illusion that linear presentation equals linear thought. If generative tools make it easier than ever to package non-linear thinking in sequential form, organisations may conclude that the accommodation problem has been solved. In reality, the thinking has merely been made palatable to a structure that has not itself changed. The same dynamic, examined from the perspective of internal evaluation rather than work design, is the subject of the capstone essay on the cognitive meritocracy illusion. The cost of that misunderstanding will not be visible in quarterly reports. It will be visible in the ideas that were never pursued, the questions that were never asked, and the cognitive diversity that was present in the workforce but absent from the work.

Questions this essay answers

What is meant by an organisation built for linear minds?

A linear organisation is one whose design assumes that cognitive work proceeds in predictable steps that can be summarised in standard formats and reviewed at standard intervals. The assumption is embedded in the architecture of the firm rather than in any single policy, which is why it is difficult to revise.

Do generative systems remove the linear bias?

Generative systems reduce the friction of producing linear-looking output from non-linear cognition, which makes the bias less costly for the individual employee. The bias itself, however, remains in the design of the organisation and continues to shape who is heard and on what terms.

What is the practical implication for leaders?

Leaders who want to capture the contribution of non-linear cognition will need to redesign at least the immediate evaluation and communication structures around their teams. Generative systems alone do not perform that redesign.

Sources

  1. Mintzberg, H., The Nature of Managerial Work, Harper & Row, 1973. See also Mintzberg, H., "The Manager's Job: Folklore and Fact," Harvard Business Review, 1975.
  2. Weick, K. E., Sensemaking in Organizations, Sage Publications, 1995.
  3. Liker, J. K., The Toyota Way: 14 Management Principles from the World's Greatest Manufacturer, McGraw-Hill, 2004.
  4. Scannell, J. W. et al., "Diagnosing the Decline in Pharmaceutical R&D Efficiency," Nature Reviews Drug Discovery, 11, 2012, pp. 191-200.
  5. Christensen, C. M., The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail, Harvard Business School Press, 1997.
  6. Microsoft, 2025 Work Trend Index Annual Report, May 2025.
  7. Jeremiah Lee and Kniberg, H., "Failed #SquadGoals: Spotify Doesn't Use the Spotify Model," Jeremiah Lee Blog, 2020. Confirmed by multiple former Spotify employees.
  8. Edmondson, A. C., The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth, Wiley, 2018.

About this Series

This article is part of the Synthetic Minds at Work series exploring how generative AI is reshaping the role of cognition in organisations. The structural questions raised here, regarding linearity, evaluation and the design of work itself, are taken up at the level of organisational architecture in the capstone essay, The Quiet Reordering of Talent.