There's a moment every L&D professional recognizes, even if they never talk about it out loud.
You've designed the program. The content is solid. The facilitators are good. The participants show up, engage, maybe even get emotional. Someone fills out a feedback form and writes "life-changing." You mark it a success.
Three months later, nothing has changed.
The manager who needed to stop micromanaging is still micromanaging. The team that needed to speak up in meetings is still silent. The culture that was supposed to become psychologically safer is somehow less safe than before — because now people resent having sat through a training about it.
"You tried to change behavior without touching the beliefs and values that generate it. That's not a tactical error. It's a structural one."
And until the field is honest about it, we will keep building the same beautiful programs that produce the same beautiful nothing.
Three layers. They're not metaphorical. They are the actual sequence by which human action is produced.
Beliefs come first. Aaron Beck defined them as the central evaluative schemas through which people interpret their experience — not what they think, but the lens through which thinking happens. Stable neural networks, formed through years of experience, hardened by repetition. When the brain encounters new information, it doesn't process it neutrally. It processes it through what already exists. The new is always filtered by the old.
Below beliefs, there are deeper ones — Core Beliefs. "I am not enough." "Authority cannot be trusted." "Effort is rewarded." "Failure is permanent." These are not thoughts you have. They are the architecture that generates the thoughts you have.
Values come second. But not arbitrary values. The values that emerge from the beliefs you already hold. Shalom Schwartz mapped values across more than 80 countries and found something that changed how the field thinks about this: values aren't random. They form a structure — a circle, with opposing values in genuine tension. You cannot fully maximize Security and Openness at the same time. What you actually value is only visible when values conflict — in the moment where you choose one and sacrifice the other.
But what you value is constrained by what you believe. If you hold a Core Belief that people are fundamentally untrustworthy, the value of Benevolence is available to you in theory but blocked in practice. Douglas McGregor understood this. Two radically different belief systems about human motivation — Theory X and Theory Y — produce radically different organizational realities. Not different policies. Different people, shaped by different foundational assumptions.
Then behavior. Behavior is last. It is the output. Organizations try to change it without touching the first two levels — and it is why hundreds of billions per year in corporate training produces, by the most generous estimates, a 10 percent transfer rate.
One in ten training experiences actually changes what people do. Nine out of ten leave no behavioral trace. (Baldwin & Ford, 1988 — replicated many times since.)
The reason is not that the training is badly designed. The reason is that behavior is downstream of values, and values are downstream of beliefs, and neither of those layers is being touched.
The B→V→B chain has a direction. Most organizations run it backwards.
They start with behavior. They identify the behaviors they want — in a competency model, a performance review framework, a leadership program curriculum. They train those behaviors. They measure those behaviors. They reward and punish based on those behaviors.
And the behaviors don't change. Or they change superficially and revert. Or they change for some people in some contexts and not others. Because behavior is the output. You cannot sustainably engineer the output without touching the input.
I have worked inside some of the most admired companies in the world. Here is what I saw.
Amazon's Leadership Principle "Earn Trust" articulates something real: trust is built through candor, accountability, transparency, and genuine respect. The language is clean. The intention is serious. Jeff Bezos quotes are attached to it.
And simultaneously — in the same organization, during the same years — a system existed that mandated the annual removal of five to six percent of the workforce, regardless of actual performance. Stack ranking. Fixed exit quotas.
You cannot build trust as an organizational value inside a system architecturally designed to produce annual casualties. Trust requires safety. Mandatory elimination requires fear. Fear and trust are not in conflict — they are mutually exclusive.
This is not a criticism of individuals. The people who built and ran that system were not villains. They were operating inside a deeper belief — a Basic Underlying Assumption, in Edgar Schein's language — about what drives human performance: that competition within the organization is a performance accelerant. That the threat of elimination sharpens people.
That is a belief. About human nature. About motivation. And from that belief, a value emerged: sustained performance above all else. And from that value, behavior was produced — the annual ritual of ranking, calibrating, and cutting. Consistently. Regardless of what the principles board in the lobby said.
Schein described three layers of organizational culture. Artifacts — the visible things, the posters and the slogans. Espoused Values — what shows up in the strategy decks and the onboarding materials. And Basic Assumptions — the deep, unconscious beliefs that actually drive behavior. Most organizations live at layers one and two. They have never touched layer three.
This is what Chris Argyris spent his career trying to explain, and almost no organization has genuinely listened.
Argyris made a distinction that cuts through everything: Espoused Theory versus Theory-in-Use. What an organization says it does versus what its behavior actually demonstrates it does.
Detect an error and correct it within the existing framework of values and assumptions. Fix the behavior without questioning the beliefs that generated it. Comfortable. Keeps the fundamental assumptions intact. The most common form of organizational response to failure.
Detect an error and ask why the governing variables produced it. Question what you believe, not just what you do. Threatening — because examining governing values means questioning the entire framework of decisions made inside them. Rare. Necessary.
"Single-Loop correction of behavior without Double-Loop examination of governing values is not change. It is the appearance of change in service of the maintenance of the status quo."
— Chris Argyris
The company had genuine aspirations around sustainability, human-centered design, meaningful work. People who worked there believed in those things. The culture had a real warmth to it.
And then the quarterly numbers would arrive. And you could watch the values reorganize themselves in real time. Not consciously. Not cynically. The beliefs underneath — about what survival required, what the organization would ultimately reward — would surface and rearrange what actually mattered.
This is what Jonathan Haidt showed that corporate L&D has never fully reckoned with: moral decisions don't begin with reasoning. They begin with intuition. The emotional response fires first. Then the rational mind follows, constructing a justification for what the gut already chose.
He called it the emotional dog and its rational tail. The tail doesn't wag the dog. It just explains the dog's movements in language that sounds like reasoning. The values people claim are often post-hoc rationalizations of decisions already made at the level of belief and intuition.
The training wasn't about content. It was about identity. The question it kept asking, in a hundred different forms, was not "what do you know how to do?" It was "who are you when you're at your best, and how do you become that person more consistently?"
That is a beliefs intervention. Not a values workshop. Not a behavior change program. A deliberate and sustained challenge to the beliefs people held about their own capabilities.
Albert Bandura spent his career showing that behavior is less determined by actual capability than by the belief about capability. Self-Efficacy — the belief that you can do a specific thing — is a better predictor of performance than measured ability. The belief comes before the action. Without the belief, the action is never attempted.
Apple Training was, among other things, an extended Self-Efficacy intervention. It built mastery experiences — Bandura's most powerful source of belief change — into the daily work. Small wins. Visible competence. The accumulation of evidence that you are, in fact, capable of this. And from those beliefs, new values became accessible. Not values delivered through a presentation. Values that felt real because the person now had evidence that they could live up to them.
That is the B→V→B chain working in the right direction.
Kahneman gives us the clearest mechanism for understanding why behavior-first fails. System 1 — fast, automatic, unconscious — generates most of human behavior. It runs on patterns, on what has always worked, on habituated responses built through years of reinforcement. System 2 — slow, deliberate, effortful — can override it in the moment. But System 2 is expensive. Under pressure, when the cognitive load is high, System 2 withdraws and System 1 takes over.
This is why behavior under pressure reveals values more accurately than behavior in a training room. System 1 is running on beliefs — on the deep, automatic, neurally encoded assumptions about what is real and what is required.
Training delivers information to System 2. It doesn't touch System 1.
Neuroscience reinforces this. LeDoux showed that the amygdala responds to challenges to existing beliefs as if they were physical threats. Before the cortex has processed the new information, the defensive system is already active. Genuine belief change requires what Mezirow called a disorienting dilemma — an experience, not an argument, that makes the existing belief structure impossible to sustain. And it requires safety. The kind of safety where examining what you believe doesn't feel like annihilation.
"The B→V→B chain is not a theory. It is the architecture of human change. Most organizations will not do this work. Because doing it requires them to examine their own Basic Assumptions — about human nature, about what performance means, about whether people are primarily resources to be optimized or humans to be developed."
Development Washing is not a single event. It is a sequence — a predictable progression that unfolds in most organizations across the same five stages. Recognizing where your organization sits in this sequence is the first step toward addressing it.
A CLO is hired. An LMS is launched. Press releases reference investment figures. LinkedIn posts celebrate "our people are our greatest asset." Every Fortune 500 company currently makes this claim. The announcement is the deliverable — a signal to boards, investors, and candidates that the organization invests in people.
Average corporate spend: $1,400 per employee (Josh Bersin, 2025). Global market: $400B+. Zero correlation with behavioral change at organizational level.
Compliance training dominates. Optional courses exist but have no connection to performance outcomes. Content is purchased from vendors, rarely custom-built for actual business problems. The field adopts identical practices — LMS, 70:20:10, leadership academies — through institutional isomorphism (DiMaggio & Powell, 1983): mimicking what other organizations do, regardless of evidence of effectiveness. Completion is tracked. Impact is not.
Amazon internal data: Safety Compliance 99% completion vs. Leadership Development 10% — same month, same organization. The 70:20:10 model has no robust empirical validation (Clardy, 2018).
Satisfaction scores. Enrollment numbers. These metrics protect the L&D budget but measure activity, not impact. Kirkpatrick's Level 1 (reaction) dominates: 90% of organizations measure it, only 35% reach Level 4 (results). Goodhart's Law applies: when a measure becomes a target, it ceases to be a good measure. Courses become shorter, simpler, easier — optimized for metric, not learning.
Only 12% of employees apply new skills from L&D programs (24x7 Learning, 2015; cited in Glaveski, HBR, 2019). Smile sheets show "negligible benefits" and do not correlate with learning (Thalheimer, 2016).
Managers are the #1 factor in learning transfer — and they are systematically excluded from the development system. L&D operates in a silo, disconnected from workflow, performance management, and business strategy. Saks & Belcourt (2006) showed that manager follow-up explains the greatest variance in transfer. Without it, the design of the training program is almost irrelevant.
20% of employees successfully apply new skills without active manager support. 75% of 1,500 managers expressed dissatisfaction with their company's L&D function (Glaveski, HBR, October 2019). External hires face 61% higher risk of being fired compared to internal hires — yet organizations systematically prefer external hiring (Bidwell, ASQ, 2011).
L&D is the first budget cut. The same companies that announced "learning culture" execute mass layoffs without capability investment. The development promise was never structural — it was performative. Brunsson (1989) named this: the Organization of Hypocrisy. The gap between talk, decisions, and actions is not a bug. It is the system working as designed, satisfying conflicting stakeholder demands simultaneously. This is not failure. It is the system revealing its actual priorities.
2025–2026: HP, IBM, Klarna, Amazon, PwC — all with public L&D commitments, all executing mass layoffs. PwC simultaneously publishes L&D best practices and cuts back-office staff. 55,000 AI-attributed job cuts in 2025 — in companies that practiced Development Washing, workforces were unprepared.
Organizations measure what is visible to stakeholders — not what drives performance. This is not negligence. It is a rational response to a broken incentive structure. L&D's survival depends on metrics that are easy to report. Reporting actual behavioral change rates (12%) would defund the function. Reporting completion rates (85%+) keeps the lights on. The Visibility Trap is not a cognitive error. It is the only viable strategy inside a system that rewards signals over outcomes.
Goodhart's Law (1975) and Campbell's Law (1979) both describe the same dynamic: when a measure becomes a target, it ceases to be a good measure. The field has known this for fifty years. It has not changed what it measures.
The McNamara Fallacy operates at full force: count what's easy, ignore what matters. Body count logic applied to human development. Training hours equal success. Actual capability change equals unmeasured.
Bjork & Bjork (2011) identified the deepest problem: the conditions that produce the best satisfaction scores — smooth delivery, immediate confirmation, low challenge — are precisely the conditions that produce the worst retention. High ratings are evidence of poor design. The field has optimized for exactly the wrong signal.
Every quarter of Development Washing accumulates Development Debt: the growing gap between the capabilities an organization needs and the capabilities its people actually have.
Like technical debt, Development Debt accumulates silently, compounds over time, becomes exponentially more expensive to address, and eventually forces a crisis — mass layoffs, failed transformation, talent exodus. The organization that spent five years on Development Washing cannot reverse that in one quarter. The debt is structural.
Seligman's learned helplessness applies here (1967): after repeated uncontrollable negative experiences, organisms stop attempting to improve — even when escape becomes possible. Employees who experienced repeated broken development promises stop engaging. Organizations interpret this passivity as "lack of ownership." Cause and effect are reversed.
Indicators of high Development Debt:
The counter-model is not a methodology. It is a set of non-negotiable conditions. Huselid (1995) and Combs et al. (2006) both showed in large-scale research that training works when it is part of an integrated system — incentives, selection, autonomy, manager support, psychological safety. Isolated training programs, which are the norm, don't meet this condition. The following five principles are the minimum viable system.
Start with a business problem, not a course. If you can't name the specific decision or behavior that needs to change, you don't have a development need — you have a signaling need. Catalog-first design guarantees Stage 2 of the Development Washing sequence.
Support at the moment of decision, not in a classroom two weeks before. Ebbinghaus (1885) showed 75% retention loss within 6 days without spaced reinforcement. Massed practice (classroom events) produces forgetting. Spaced, contextual practice produces retention. The delivery model must match the neuroscience.
Track what people do differently on the job, not what they clicked or how they rated the session. Brinkerhoff's Success Case Method (2005) reveals what standard metrics hide: in most programs, only a small minority apply what they learned, and the differentiator is always the organizational environment — not the training design.
Managers are the #1 lever for learning transfer. Saks & Belcourt (2006) showed manager follow-up explains the greatest variance in whether training transfers. Without active manager involvement before, during, and after any intervention — briefing, coaching, reinforcement — the investment is structurally designed to fail. No program design compensates for manager absence.
55,000 AI-attributed job cuts in 2025 — in organizations that practiced Development Washing, workforces were unprepared for the transition. AI adoption requires human foundations: self-knowledge, judgment under uncertainty, the capacity to unlearn. Without addressing beliefs about competence and identity first, AI tools become another compliance training — high adoption, zero transformation.
Development Washing is not a polemic. It is the predictable outcome of organizational systems that have been documented in peer-reviewed research since 1885. The conditions for effective adult learning are established. The gap between what organizations do and what the research requires is quantified. Six disciplines converge on the same conclusion.
Most organizations will not do this work. Because doing it requires examining their own Basic Assumptions — about human nature, about what performance means, about whether people are primarily resources to be optimized or humans to be developed.
That examination is uncomfortable. It is slow. It does not produce a quarterly result. But every quarter without it adds to the Development Debt. And Development Debt compounds.
Viktor Frankl wrote — after surviving conditions that had stripped away everything except the capacity to choose one's inner stance — that what you value is not contingent on whether the organization deserves it. The B→V→B chain is yours. It was built by your history. But it is yours to examine, yours to question, yours to deliberately reconstruct where it has been built on false foundations.
That is what genuine development looks like. Not the acquisition of competencies. The examination of what you believe.
← Back to Overview