Rethinking inequality at work: beyond categories, towards cognitive divergence

Right by category, wrong in the long run: why treating inequalities separately misses the heart of the problem

Introduction

In France as elsewhere, we like to sort inequalities into "boxes": gender, disability, neurodiversity, origin. The dominant reasoning often runs: "compensate for this one, support that one, make amends for the other". This partitioning seems practical, politically acceptable, hard to challenge. But (and this is a question many hesitate to ask) does it not unconsciously serve to hide the structuring question? Namely: who holds the power to decide what is "normal", what thinking is acceptable, which conceptual frameworks apply?

I think we are on the wrong track when we multiply partitioned action plans (feminization here, disability inclusion there, neurodiversity elsewhere) without going through the essential step: questioning cognitive hegemony, implicit conformism, the centrality of dominant thinking. In other words: rather than only strengthening those who come close to the dominant model (with a "diversity" groove), why not rebalance the ground on which people speak, favor dissonant voices, let divergent intelligence flourish?

In this article, I will begin by examining the limits of the "by category" approach (with a few concrete examples), then propose an alternative approach: valuing divergence, putting productive debate at the heart of governance, restructuring the mechanisms of collective reasoning. And I will give practical cases (some very real) to show how it can work.

What "treating inequalities separately" means, and what it hides

The comfort of segmenting problems

Making categories (gender / disability / neurodiversity / origin / age) follows a rational logic: each field has specific characteristics (legal obstacles, particular cognitive biases, structural constraints). It allows specialists to be mobilized, targeted measures to be defined, indicators to be measured within each domain.

But this segmentation carries a silent danger: it suggests that the problems are independent. It avoids asking "who sets the norms of what is acceptable, normal, audible, relevant", as well as "where and how are discordant voices excluded?".

When we say "we will strengthen women", or "we will support people with disabilities", we stay within a logic of subsidy at the margin, without touching the structure of dominant ways of thinking. We do not change the "cognitive playing field"; we just add "access ramps" for certain individuals.

The risk of "façade diversity"

With this way of working, we can very well end up with "façade diversity": executive committees that display gender parity, hiring quotas for people with disabilities, awareness programs, but where strategic decisions remain calibrated to a tacit norm (often masculinized, neurotypical, rationalist, focused on performance).

Moreover, by aligning the organization with macro objectives, the effect may be the opposite of the one intended: for example, hiring more women at low hierarchical levels to reach an overall figure, while the management layer remains very male.

In other words: we add "entrances" but do not revise the internal circuit. In the end, those who are "different but compliant" move up, but the truly marginal voices stay silent.

This leads to a paradox: diversity is encouraged as long as you stay "reasonable", "well educated", "in line with the implicit standards of acceptable speech". The "disruptive", the "paradoxical", the "out of the box" remain a tolerated exception, but are rarely integrated.

Why this logic reproduces cognitive hegemony

Thinking that we can "fix one category at a time" amounts to accepting that the dominant framework of thought is not itself questioned. It means tolerating that the "norm" (male, neurotypical, rational, dominant logic) is the implicit filter of what counts.

The real task should be to complement the category approach with an approach to cognitive structure and debate. Who decides what is "reasonable", "relevant", "acceptable"? Who controls the channels of influence in governance, in strategy, in company culture?

If this question is not asked, we are acting on symptoms without touching the root. It is like watering weeds one by one without changing the soil.

Real examples of companies where the category approach shows its limits (or its illusions)

A "diversity" casting, but conformity of thought

Several large French companies communicate heavily on "diversity and inclusion" programs without upsetting their internal culture. For example, L’Oréal, very often cited for its commitments (gender equality, inclusion of people with disabilities through Solidarity Sourcing). The paradox? In some innovation or executive committees, boldness and radical or contesting positions remain a minority, even marginalized.

A company can display gender-balance indicators in its executive bodies while keeping a culture of risk aversion, which reinforces the status quo. This bias is often found in structures where divergent speech is perceived as disruptive, not as a lever of progress.

When "disability inclusion" stays symbolic

The effort to "integrate people with disabilities" is laudable but often limited to reasonable accommodations (ramp, elevator, adapted workstations). Few companies question the very configuration of the job, of the decision-making process, of the nature of meetings, of cognitive rhythms. Disability is treated as an individual variable to add, not as a collective design parameter.

An example drawn from analyses of cognitive bias: some companies consider their premises "accessible enough" and refuse to invest more, even when employees with disabilities report obstacles, out of status quo bias.

Another illustration: a company uses meeting simulations in which wheelchair users are "invited" to speak, but the meeting format (length, distribution of speaking time, pace) is not rethought. "Diversity" is added as an accessory layer, without redrawing the processes.

Neurodiversity, often forgotten or treated superficially

Diversity policies rarely look at neurodiversity (ASD, ADHD, dyslexia, etc.). When they do, it is often through "awareness sessions" or "light accommodations" (breaks, noise-cancelling headphones). They avoid transforming the ways work is designed, the meeting format, the dominant cognitive processes (long meetings, slides, linear thinking).

Very few French companies value a non-conformist voice. More than valuing it, the point is above all to understand the potential of a non-traditional perception. There is a disruptive marketing concept: the blue ocean strategy, which explains the relevance of breaking with the idea of being competitive and moving instead to a field where nobody is playing yet. That is the whole relevance of start-ups that bring new solutions and grow strongly because a boulevard without competition opens up before them.

Neurodiversity holds an incredible performance potential for companies, and those that have understood this have applied the blue ocean strategy, not to the solutions they sell, but to their teams.

Hiring discrimination and the "tolerance bonus" effect

When CVs are anonymized (a practice often used to avoid bias), this can "neutralize" certain forms of discrimination but also remove signals that some recruiters used (positively or negatively). Curiously, anonymization can sometimes put at a disadvantage those who were counting precisely on their atypical background or their commitments, because what falls outside the "standard" frame is erased.

Unilever, for its part, changed its recruitment methods: anonymization and online psychometric tests to limit internal biases. This works up to a point, but it does not challenge the model of "valued competence", which remains calibrated to a dominant cognitive norm.

The academic world as a mirror

In astronomy, a Cornell University study shows that women are half as likely as men to obtain permanent positions, despite equivalent applications.

This suggests that, even in highly formalized fields, structural biases persist beyond simple "policies to promote women". This study does not describe a renovation of committees or decision methods to include divergent stances, but findings on the cumulative effect of evaluation biases.

Towards a "reasoned divergence" approach: rethinking the conditions of debate in business

If we accept that the category approach is necessary but insufficient, what complementary stance should we adopt? Here is a possible roadmap, without claiming it is perfect, but it opens up reflection.

Redesign the channels of speech and decision

• Musical chairs: to break power rents, require that moderation and meeting-lead roles change regularly.
• Structured "mandatory" debate: for each strategic decision, bring in an external body, a "controversy rapporteur", responsible for identifying points of friction and invisible voices
• Cognitive mapping: establish, in strategic teams, a "map of implicit frameworks" (which models are presupposed, which angles are never examined)
• Shadow boards where "non-dominant" employees challenge the decisions of top management

Value and protect dissonance

• Create a charter of legitimate dissent: a form of internal protection for those who step outside the frame, as long as the debate is reasoned
• Set up internal prizes for "out of the box" perspectives: mechanisms that reward ideas that challenge norms (even if they are partly rejected)
• Plan a "cognitive exit plan" for rejected ideas: instead of burying them, I have seen companies archive them and revisit them six months later, recontextualized

Revisit operational formats

• Shorter meetings, thematic micro-kickoffs, mandatory individual thinking time before debate, which also helps avoid the many pointless meetings.
• Alternating formats: write (think on paper) before speaking, to favor non-verbal thinking styles
• Anonymous internal review processes (not necessarily for hiring, but to challenge strategic briefs)
• "Cognitive hackathons": internal workshops where divergent profiles build alternative scenarios

Build divergence into strategic design

When launching a transformation, instead of "thinking about diversity after the fact", involve "marginal" profiles from the strategic framing stage: co-design, not tokenistic consultation.

Set up an "extreme dissent" phase in each project: ask participants to deliberately promote the perspective they consider "wrong" or "unacceptable", to force the debate.

Concrete (or well-framed hypothetical) cases of companies that took divergent paths

Real case: the shadow board at Unilever (or similar companies)

Unilever, already cited for its inclusive HR practices, has tested "shadow boards" (parallel councils of younger and differently profiled people) to challenge top management decisions. Although this is not solely about "cognitive divergence", it opens a door to non-dominant voices in decision-making. (It is not definitive proof that the cognitive approach was transformed at its core, but it is a step towards a more permeable future.)

A plausible (but instructive) fictional case

Suppose a large IT services company decides that, for a critical tender, it must "force cognitive divergence". It creates a review committee in which an "adversary perspective" is anonymously designated: a group tasked with pointing out the implicit biases in the brief and in the market assumptions, even if this view runs against the orientations of the business plan.

A "total challenge" session is launched in which everyone receives a card reading "oppose an unexpected angle" (e.g. "what if we targeted the weakest ignored segment, to create a break?").

The result: the outline of the offer, the pricing structure and the split of responsibilities are put back on the table, and several bolder iterations emerge.

This is not a "category diversity" model but a "cognitive diversity" format applied to the core of the business.

The internal "contrary ideas" hackathon

I saw (in a French tech SME, anonymized so as not to betray it) a remarkable initiative: a "contrary ideas" day on which teams came with their best proposal, but were also required to argue the worst-case scenario, to play the "anti-project".

The strongest proposals emerge because they hold up not only in the best case but in the worst. This allowed the company to spot blind spots that a "diversity by category" committee would never have detected.

An example from tech (and AI)

AI models (supposedly "neutral") are observed to reproduce the gender, origin and dominant-cognition biases of their data. A recent study shows that AI, in its "application" recommendations, tends to apply gender and ethnicity stereotypes.

This illustrates a lesson: if we do not question the implicit cognitive structure (the data, algorithm biases, design norms), we lose the essence of the fight. Algorithmic diversity must come with explicit divergence.

Obstacles, resistance, pitfalls and implementation strategies

Resistance to risk, fear of chaos

Encouraging divergent thinking means accepting mistakes, tensions, messy debates. Many managers prefer calm and predictability, but the risk is that divergence is seen as hot air or as going astray.

Strategy: start within a limited scope, define debate "safe zones", regulate excesses.

The effect of social conformity (peer pressure)

Even with good intentions, individuals internalize dominant norms and "self-censor". The mere fact of being "in a group of cognitive exception" can produce self-exclusion of ideas that are too eccentric.

To limit these conformity effects, it will be worth using partial anonymity, hidden votes, individual thinking phases, and a rotation of the debate’s "parliamentarian" roles.

Tokenism

"Divergence" can be done as decoration: appointing an "official divergent" but never giving them real power. This reproduces the "symbolic inclusion" model.

Strategy: grant rights (partial veto, official report) to those who challenge, not just a decorative position.

The difficulty of measuring divergence

Classic KPIs (gender diversity, disability rate, etc.) are not enough, but then how do we measure "cognitive diversity" or "quality of controversies"?

Qualitative indicators can be deployed (feedback, climate surveys, an "unframed speaking zone"), along with process indicators (how many rejected ideas were revisited), and before/after maps of the cognitive fields questioned.

The political cost of change

Redefining the channels of cognitive power challenges tacit balances of power, so those who benefit from the dominant cognitive normality are likely to resist strongly (including under the guise of "business risk").

It will be worth aligning the approach with strategic objectives (innovation, agility, culture), running visible pilots, and collecting quick wins to legitimize the experiment.

What this changes (and what it promises) for business performance

Better detection of strategic blind spots

When divergence is allowed, we draw on scenarios that orthodoxy ignores: disruption, tail effects, structural decisions overlooked by dominant career paths.

Stronger agility

If you have already practiced a culture of healthy debate, rapid change becomes less traumatic. The group already knows how to challenge itself. The "strategic turn" does not rest on a single head but on an ecosystem of collective thought.

Engagement and retention

Employees who feel they have the right to be dissonant, who can express unpopular angles safely, develop stronger loyalty and creativity.

Fewer "soft consensus traps"

Soft consensus ("doing like everyone else") is a weapon of stagnation. Protected divergence forces decisions to be defensible, reasoned, robust in the face of criticism.

Radical innovation

Ideas that upend things often come from the margins. If the margins are recognized, radical innovations gain a way in.

Conclusion

Treating inequalities by category (gender, disability, neurodiversity, ...) is useful and sometimes indispensable, but if we stop there, we feed an illusion of progress without touching the essential: who decides what is "normal", "acceptable", "reasonable"? Only supporting individuals within the dominant frame, without redefining that frame, widens the gap between surface inclusion and real cognitive transformation.

If we want to move towards organizations that are alive, resilient and innovative, we must bet on reasoned divergence. Making room for the dissonant, the out-of-frame, the contesting does not break order: it fragments it, and makes it stronger when crises come.

Sources and recommended reading

• Study Inequalities faced by women in access to permanent positions in astronomy in France: structural biases in French academia.
• Report on cognitive biases in recruitment and CV anonymization (notably in France), unexpected effects.
• The Unilever case: anonymization + psychometric tests as counter-bias.
• Analyses of cognitive biases and resistance to change in business.
• Improved performance of inclusive companies (Deloitte / BCG studies): inclusive culture and results.
• Articles on the limits of diversity and inclusion policies in France (between consensus and controversy)
• Example of the use of shadow boards in companies. (It is essential to look at modern HR initiatives.) Mentioned in a report on HR strategies.