Immunity to Change: Overturning Big Assumptions

Immunity to Change: overturning big assumptions is work on predictions that can make a protective pattern feel necessary. The aim is to examine those predictions honestly, not persuade a client to announce the opposite belief. Sometimes the useful result is a more conditional understanding of a real risk.

Coachful is our #1 recommendation for organizing this reflective coaching workflow. An individual worksheet, the client's selected action and a review record can connect the map with actual learning. The platform does not discover hidden motives, validate an assumption or establish developmental change.

This guide explains the model's main distinctions and follows a complete fictional clarification case. The first attempted test does not test the prediction at all. A revised action produces mixed evidence, which changes the client's understanding without erasing a difficult workplace experience.

Attribute the model and understand its purpose

Robert Kegan and Lisa Lahey developed Immunity to Change. Harvard Graduate School of Education's overview identifies their approach to competing commitments and assumptions, distinguishing skill-focused change from work on how a person understands a situation.

In a Harvard interview, Kegan and Lahey distinguish technical and adaptive challenges. A technical problem may need instruction, information or practice. An adaptive element involves the person's interpretation of what is possible or threatening. Many situations involve both. Knowing a better communication technique does not necessarily settle what someone expects will happen if they use it.

The framework is not a claim that every obstacle is internal. Unclear authority, missing resources, discrimination or a genuinely punishing response require attention in their own right. A coach should not turn organizational failures into evidence that the client has the wrong mindset.

The immune-system language is a metaphor for a protective pattern. It does not identify a biological immune response, a clinical disorder or a measurable defect in the person.

Kegan and Lahey's four-part map connects an improvement goal, contrary behavior, protective competing commitments and assumptions that make those commitments feel necessary. Their published work also emphasizes testing and interpreting assumptions. The explanations and fictional working records below are original applications, not reproductions of their proprietary forms.

First distinguish the four kinds of information

The distinctions matter because each requires a different conversation. A goal needs ownership. A behavior needs a concrete description. A protective concern needs understanding. A prediction needs appropriate evidence. Moving directly from “I avoid questions” to “You fear failure” skips the client's own meaning.

Choose a goal the client recognizes

A useful goal describes what the person wants to do differently and why it matters. “Ask a necessary clarification before committing to a report design” gives the conversation a working situation. “Become fearless” offers no clear action or boundary.

Check whether the person actually wants the change. A sponsor may want faster decisions while the client wants more accurate decisions. These can overlap, but the coach should clarify the agreement rather than pretend the difference is resolved.

Keep the goal within an appropriate sphere of influence. The client can choose a question, request a process or change their own preparation. They cannot guarantee another person's welcome, approval or future opinion.

Ask what useful progress would look like without requiring a total transformation. One earlier clarification that improves the work can matter. It does not prove the client will now communicate confidently in every setting.

Describe the behavior before explaining it

Record actions an observer could recognize: drafting before checking a definition, not sending a question or rewriting work after discovering the requirement. “I am insecure” is an interpretation, not a description of what happened.

Include what the person does as well as what they avoid. Spending an evening preparing a finished answer can protect against asking a question. It may also consume time and make correction harder.

Do not assume every instance has the same cause. A missed question might reflect a deadline, lack of access to the owner or a genuine knowledge gap. The coach needs enough context to distinguish a protective pattern from an unavailable option.

Look for a specific recent example rather than a long inventory of failures. Ask what happened before the decision, what the client chose and what followed. Keep the account factual enough that later learning can be compared with it.

Understand the competing commitment without shaming it

Ask what the person would worry about if they acted differently. The response may reveal a protective aim: not being seen as unprepared, not upsetting an important relationship or not exposing an uncertainty before having an answer.

That aim can coexist with a sincere desire to change. The client may want accurate early questions and also want never to reveal a question. Naming the conflict is more useful than arguing that one desire is fake.

Use the client's language and allow correction. A coach's tentative observation should remain tentative. “Does this help you avoid showing uncertainty?” invites examination; “You are addicted to being competent” imposes an unsupported explanation.

Separate the protection from its present cost. The pattern may have made sense in a previous environment. The current question is where it remains useful, where it restricts choices and what evidence could inform a different decision.

The competing commitments guide develops the protective conflict. Here, the next step is to examine the prediction that makes that protection feel necessary.

Turn the big assumption into an examinable prediction

A value such as accuracy does not need to be disproved. A prediction such as “If I ask this owner a bounded question before finishing, they will stop including me in the next design review” identifies a possible consequence.

Specify the person, action, context and expected response. “Everyone will think I am useless” contains undefined people and a conclusion about private thoughts. “The owner will remove my review invitation after this question” offers something observable, although other explanations would still need consideration.

Ask what experience supports the prediction. It may include an actual adverse response. Do not force the client to dismiss that evidence just because the coach hopes for a more encouraging interpretation.

Then ask what the prediction leaves out. Does it apply to every colleague, every kind of question and every stage of the work? A more precise assumption can preserve important caution while making room for a different action.

A tentative formulation remains the client's formulation. It can be changed or left untested. The coach should not assign a developmental stage or diagnose the person's identity from the map.

Check whether the map explains one connected pattern

Read the entries together rather than treating four completed boxes as an assessment result. Does the described behavior work against the chosen goal? Does it plausibly protect what the client named? Does the prediction explain why that protection feels needed in this situation?

A connection can be missing. Someone may avoid a question because the owner is unavailable, while the suggested hidden commitment concerns popularity. The map should not disguise that mismatch through persuasive wording. Return to the actual event and let the client correct the account.

More than one protection may be involved, but that does not require testing everything together. Choose a specific proposition that matters to the client and can be examined appropriately. If the work is too exposing or the person does not want to continue, a map does not create an obligation to disclose more.

Choose evidence that can answer the actual question

Decide before acting what would be observed. Separate another person's words or actions from the client's interpretation and feelings. All can matter, but they answer different questions.

“I felt embarrassed” describes the client's experience. “The owner replied with the requested definition” describes the exchange. “The owner has lost confidence in me” is an inference unless supported by additional information.

A delayed reply can have several explanations. An impatient tone can coexist with a useful answer. Being invited to the next review gives information about inclusion, not complete access to the owner's private judgment.

Plan how to handle ambiguous evidence. No reply by the review date may leave the question unresolved. A changed meeting invitation might concern scheduling rather than exclusion. Do not decide in advance that every observation will disprove the belief.

Keep the learning separate from the desired work outcome. A technically successful report can follow an invalid test. A difficult exchange can still provide relevant information. Accuracy in the review matters more than producing a success story.

Consider how strongly a particular observation bears on the prediction. An explicit exclusion linked to the question is different from a canceled meeting with no explanation. A useful reply with a sharp opening is mixed evidence: it can challenge a predicted practical consequence while supporting concern about an uncomfortable interaction.

Do not average these into a single success score. State which part is supported, which is challenged and which remains unknown. This makes a revised belief more usable than a broad declaration that the test went well.

A counterexample can matter without proving universal safety. Equally, one difficult response does not establish that every future person will react identically. Keep the inference proportional to the actual context and the information available.

Design a bounded action with real responsibilities protected

Kegan and Lahey's 2009 book, Immunity to Change, in excerpts used in HarvardX teaching, distinguishes learning about an assumption from immediately trying to improve performance. They also emphasize comparing the actual test with the intended one before interpreting the result. That distinction is central to the case below.

For an original practical test record, specify the prediction, action, limits, evidence and review. Choose an ordinary situation in which an appropriate question can be asked without manufacturing a problem.

Check authority, deadlines and affected people. Do not withhold a required warning, allow unsafe work or misrepresent a colleague's behavior to create a learning opportunity. If the possible consequence is unacceptable, choose another way to obtain information or do not run that test.

Make the action honest. A clarification should address real uncertainty. A colleague does not have to participate in an undisclosed reassurance exercise or promise that the client is competent.

The guide to safe behavioral tests develops experiment planning in more detail. This article focuses on how the prediction changes when actual evidence is examined.

The demonstration worksheet builder illustrates reflection prompts. It does not identify a hidden commitment or validate a completed map.

Worked example: a report owner and an earlier bad experience

All people, workplace events, records and outcomes here are fictional. Kira prepares internal operational reports. She wants to ask relevant definition questions before spending time on a finished draft.

During the previous month, she produced four drafts. Two needed substantial rework because she guessed a category definition. Her technical writing skill is adequate for this task; the uncertainty concerns requirements she could ask the report owner to clarify.

Kira also remembers a previous manager, Don, saying that senior staff should not need basic questions. He once criticized her in front of others. The coach does not treat that event as imaginary or generalize it to every future colleague.

The current report owner, Greta, has not issued that instruction. Kira nevertheless prepares a full answer before asking Greta anything, which makes an early clarification less likely.

Complete the map in the client's own terms

Kira distinguishes two claims in the assumption. She cannot observe every private judgment Greta makes. She can observe the reply and whether she remains included in the next review. She also recognizes that tone could be uncomfortable without establishing exclusion.

Coach: “Does this record explain the choice you make?”

Kira: “Yes. I want the requirement right, but I try to protect myself by already having a complete answer.”

Coach: “What should remain protected if you ask earlier?”

Kira: “The deadline and accurate reporting. I am not going to submit an unsupported definition as a test.”

The map links goal, behavior, protection and prediction. It has not proved that the prediction is false or that the concern is purely internal.

Write the first test before acting

The report includes an unclear “active pilot site” category. Kira needs to know whether it means every site currently on the roster or only sites with completed training during the reporting period.

The draft is due on 14 February 2030 at 16:00. Greta is an appropriate owner to ask; Kira plans to send the question on 12 February at 10:00, leaving time for an answer and any necessary follow-up.

Completed fictional planned question: “Before I draft the category, does active pilot site mean every site on the current roster, or only a site with completed training in this reporting period? I will use the confirmed definition and note it in the report.”

The original test record specifies a real uncertainty, a bounded question before a finished answer, no change to the reporting deadline and a review of the actual reply and next design invitation. Kira will record her own response separately.

If an answer does not arrive in time, she will use the ordinary work escalation route rather than guess or let an experiment override delivery responsibilities.

Recognize that the first attempted test did not happen

On 12 February, Kira feels uncomfortable and changes the action. She attaches a finished draft using her chosen roster definition and writes, “Let me know if anything should change.” Greta replies that afternoon, “Looks okay.”

Kira initially says the assumption has been disproved because Greta did not criticize her. The coach asks what she actually did compared with the planned question.

Kira: “I did not show the unresolved choice. I used the finished answer to protect myself again.”

Coach: “What can Greta's reply tell us about asking the question before finishing?”

Kira: “Very little. She was responding to a different action.”

They record the attempted test as inconclusive for the chosen prediction. The draft's apparent acceptance is not used as evidence that revealing uncertainty is safe. The definition itself also remains insufficiently clarified.

This is useful process information without being the intended assumption evidence. Kira notices exactly where she substituted the protective behavior. She chooses whether to try again rather than being told she failed the framework.

Run the revised question and preserve the mixed response

Greta confirms she is available for requirement questions between 09:00 and 10:00 on 14 February. At 09:00, Kira sends the original two-option question without attaching a completed report.

At 09:20, Greta replies: “We covered this last month. Use completed training in the reporting period, not everyone on the roster. Put that definition in the footnote.”

Kira experiences the opening as impatient. She also receives a specific definition that differs from her initial choice. She revises the report and submits it by the existing 16:00 deadline.

Her record separates the observations: the question was actually asked before the revised draft; Greta answered; the definition changed the work; Kira felt embarrassed; no direct statement removed her responsibility or participation.

On 18 February, Kira receives the next design-review invitation. She remains included. That is relevant to the predicted exclusion, while not proving that Greta never thought the question basic.

Interpret the result without demanding the opposite belief

Kira's revised statement is conditional: “A bounded question to Greta can receive an impatient opening and still improve the report without excluding me. I should prepare the actual uncertainty and ask through the agreed route. I do not know that every relationship will respond this way.”

She does not replace the old assumption with “Questions are always safe.” Nor does she declare the past criticism irrelevant. The change is a more accurate distinction between contexts, consequences and choices.

The work outcome and learning outcome remain separate. Meeting the deadline matters operationally. The clarification matters because Kira asked a real question and recorded what followed. Its evidence needs the qualification below.

There is also a limitation: Greta had already seen Kira's finished first draft. The 14 February exchange therefore does not fully test revealing an unresolved question before any finished answer. It gives information about explicit clarification after a draft. Kira keeps that condition in the record rather than awarding the revised action the original test's conclusion.

Integrate the learning through another actual choice

On 21 February, a new report is at the requirements stage. Kira has not produced an outline or chosen a meaning for its “trained facilitator” category. Greta has agreed to receive requirement questions that morning. The outline is due on 22 February at 16:00.

Completed fictional new test: “Before I choose the category for the new outline, should trained facilitator include everyone invited to training, or only people who completed the course? I have not assigned a definition yet and want to use the agreed one.”

Kira records the same prediction about appearing unprepared and losing inclusion, but now identifies the next review as the 25 February meeting. The action exposes the actual unresolved choice before a finished answer; no previously completed outline accompanies it. The normal escalation route remains available if the definition is not confirmed in time.

Greta replies that morning: “Completed the course only. Add that to the category description.” Kira includes the definition in the outline and submits by the existing 22 February deadline. On 25 February she attends the design review as invited. Greta continues to ask for her analysis.

This later test directly examines the earlier-before-finishing condition. Kira's final revision is: “With Greta, two explicit clarification questions produced useful definitions and continued inclusion, although one opening felt impatient. In the new-report test, asking before preparing an answer did not produce the exclusion I predicted. I still need to consider the person, route and actual task.”

There are two explicit clarification exchanges, not two clean tests of the original timing condition. The distinction prevents a larger evidence claim than the record supports. It also leaves room for other workplace contexts to differ.

Kira retains a brief prompt before drafting: “Which requirement am I treating as settled without confirmation?” She plans to review whether that prompt helps over the next two reporting cycles and can stop using it if it becomes unnecessary.

The remaining question concerns how to handle Don's criticism if they work together again. The coach does not generalize Greta's replies into permission to ignore that history.

A contrasting case: the constraint is explicit

Another fictional client, Tomas, cannot publish a customer report until an authorized reviewer signs it. He says, “If I bypass review, the report may be withdrawn.” That concern reflects a stated work requirement, not merely a belief to overturn.

A coach should not propose skipping approval to test courage. Tomas can clarify the review process, request a faster decision or raise a resource problem with the relevant owner. The change might concern the process rather than his confidence.

If Tomas also predicts that even asking about the review timetable will end his career, that is a separate proposition. It still requires context and an appropriate route; identifying a prediction does not mean it is safe or useful to test through action.

The distinction protects both responsibilities: take real constraints seriously and remain open to examining assumptions beyond them.

Organize reflection and review in Coachful

1. Agree purpose and audience. Explain how the client-owned reflection will be used and what a sponsor may receive. A private map is not automatically a performance report.
2. Create a short individual worksheet. Use original prompts for the chosen goal, behavior, protection, prediction and evidence. Explain the intended coach-team audience; do not promise only one author can read every worksheet response.
3. Record the selected action. Create a client task only after agreement. Completing it records the action, not the truth or falsity of the assumption.
4. Keep actual and planned separate. Review what happened, what changed and which observations concern the prediction. An altered test should not inherit the original test's conclusion.
5. Review the next choice. Refine the statement, seek more information, address a real constraint or stop. The platform cannot certify a belief change.

The demonstration My Coaching view illustrates preparation and next actions. It is not Kira's completed map or evidence that any test has been carried out. Explore Coachful's features for the practical workflow.

Troubleshoot the learning problem

The statement is global: identify a person, situation, action and expected consequence without narrowing it merely to make disproof easy.

The client cannot name a competing commitment: stay with their experience rather than supply a hidden motive.

The action changes: assess what actually happened before interpreting it against the original prediction.

The evidence is painful: include it honestly and consider appropriate support or structural action.

The coach needs a success: return to the client's learning and choice. A more precise concern can be a useful result.

Frequently asked questions

Must a big assumption be overturned completely?

No. Evidence may support, qualify or challenge it. The useful result can be a more accurate conditional statement.

Can the coach identify the hidden commitment?

Offer tentative observations where appropriate, but let the client recognize, revise or reject the formulation.

Is every obstacle an immunity to change?

No. Skills, authority, resources and real environmental risks may need different responses.

Does one successful test establish lasting change?

No. It provides information for a next choice without proving a developmental stage or permanent transformation.

What if the test is not carried out as planned?

Record the actual action and assess whether it can inform the prediction. An inconclusive attempt is not evidence of disproof.