Amygdala Hijack at Work: What the Metaphor Explains and Misses
Nour
Founder, Coachful
An amygdala hijack at work is a familiar explanation for snapping at a colleague, defending a weak answer or losing track of a difficult conversation. It can help someone recognize that their response became less considered. It cannot establish which brain circuits caused that particular event, excuse mistreatment or certify that a short pause restored sound judgment.
The useful coaching question is concrete: when a leader feels challenged, what happens to their listening, evidence checking and treatment of other people? This guide explains the metaphor’s history and limits, then follows a fictional forecast discussion through an outburst, an actual apology and a second attempt that still leaves work unfinished. The aim is accountable preparation, not a neurological diagnosis.
Start with Goleman’s own correction
Daniel Goleman popularized the expression in Emotional Intelligence. But his July 2022 article, Emotional Intelligence, Emotional Hijacks, and Systems Thinking, explicitly revisited it. He explained that the underlying science had developed beyond a single threat detector taking over a thinking brain. He described the amygdala working with widely distributed circuitry and suggested using “emotional hijack” instead. Presenting the older phrase as Goleman’s unqualified current anatomical explanation misses his own update.
That revision preserves an everyday observation: strong emotion can capture attention. It changes the confidence with which we assign a mechanism. A leader can say, “I became preoccupied with defending myself and stopped answering the question,” without claiming that a particular structure switched off. The description is available for examination even when nobody has measured brain activity.
Language also shapes responsibility. “My amygdala made me do it” places the behavior outside the speaker’s agency. “I interrupted you when I felt embarrassed” names an experience and an action. The second statement allows an apology, a factual correction and a different response next time. It does not deny that stress affects people.
The metaphor can nevertheless provide a useful opening when the client chooses it. Someone who previously treated every outburst as proof of a fixed bad character may become willing to examine a particular sequence: a question, an interpretation, an urge and a response. Work with that increased willingness while keeping the explanation provisional. Ask whether the phrase helps the person notice an earlier choice, or whether it makes the behavior seem inevitable. If the client prefers ordinary language, there is no reason to teach the label first. The useful output is a response they can attempt and review, rather than agreement with the coach’s preferred account of the brain.
What the neuroscience adds, and what it cannot identify
Defensive functions are not the same thing as a conscious feeling
In his 2012 theoretical review, Rethinking the Emotional Brain, Joseph LeDoux proposed distinguishing survival functions and their circuitry from the conscious feelings people describe. Defensive responses studied in animals do not, by themselves, establish the human experience of fear. In his account, bodily responses, environmental information, memory and cognitive processes contribute to how feelings arise. This is a conceptual framework for research, rather than a coaching assessment or a settled explanation of every emotion. Read LeDoux’s original review.
The practical implication is modest. An observer seeing a clenched jaw cannot read out “fear of losing status” from it. Ask the person what they experienced, allow uncertainty and examine the actual interaction. Their account may be useful without being a complete causal explanation.
Stress can impair particular cognitive functions
Amy Arnsten’s 2009 review describes how stress can impair prefrontal functions involved in working memory and flexible attention. Its mechanistic account includes primate working-memory networks: changing chemical signaling can weaken the connections that maintain a precise representation. The review also discusses human task studies and distinguishes effects on complex, flexible performance from effects on simpler or well-practiced tasks. This is richer than a blanket assertion that the entire thinking brain becomes unavailable. Read Arnsten’s review and working model.
These laboratory tasks and cellular mechanisms do not retrospectively identify what happened in a particular executive meeting. They do not establish that a leader’s wrong number was caused by stress, that a colleague’s question was dangerous or that counting for a fixed interval reverses a particular neural state. Treat them as relevant background, with a considerable gap between the research and the individual claim.
A feeling, a behavior and an explanation require different evidence
| Statement | What would support it? | What remains unproven? |
|---|---|---|
| “I felt my face getting hot and wanted to cut in.” | The person’s report of their experience. | The cause of that sensation and what others felt. |
| “I interrupted the analyst twice.” | A specific account or appropriately obtained meeting record. | The neural mechanism and whether all similar meetings follow that pattern. |
| “I did not check the revised forecast.” | The document history and actual checking sequence. | Whether stress, haste, an unavailable source or another condition explains the omission. |
| “My amygdala shut down my cortex.” | A coaching conversation cannot establish this. | The anatomical claim itself, as well as its relevance to the event. |
A coach can work productively with the first three rows. Turning the fourth into a confident diagnosis adds authority without adding usable evidence.
Define a behavioral purpose before choosing a calming technique
“Stay calm” is too broad to judge. Someone can speak quietly while dismissing a legitimate objection. Someone else can feel anxious and still listen accurately, acknowledge uncertainty and make a responsible decision. Choose a purpose that survives those differences, such as letting the question finish, checking the version being discussed and correcting any personal attack.
A pause may help create room for that response. Offer an ordinary choice: stop speaking briefly, look at the question in writing or ask the meeting chair for time to check. No breath holding, forced breathing or prescribed physiological reset is necessary. If the person does not want attention directed toward their body, keep the preparation entirely verbal or written.
The pause also needs a return. “I need a minute” can become avoidance if the question disappears afterward. Establish what happens next: repeat the question, identify what is known, state what is missing and ask the responsible person to agree any follow-up. During an immediate safety issue, use the organization’s appropriate response; do not delay necessary action to complete a coaching exercise.
Complete fictional case: a forecast question becomes a personal attack
Maren is a commercial director. Ivo is an analyst, and Dalia chairs their forecast meetings. All details below are invented. Maren books two private 50-minute coaching sessions with Rafi, on 3 and 17 November 2026, from 14:00 to 14:50 in Europe/London. The service includes those live sessions and no between-session coaching replies. The 100 minutes describe coach-client live time only; workplace preparation, messages and meetings are separate and are not measured here.
Reconstruct the event without choosing its cause first
At the 2 November meeting, Maren presents a forecast of £4.8 million. Ivo asks whether it includes a customer’s deferred £300,000 order. Maren interrupts him twice and says, “If you understood the business, you would not ask that.” Dalia stops the exchange. The meeting does not resolve which forecast version is current.
In the first coaching session, Maren calls it an amygdala hijack and says Ivo deliberately undermined her. Rafi asks what supports each part of that account. Maren reports heat in her face, an urge to stop the question and embarrassment about not having the answer. Those are her experiences. Ivo’s intention is unknown. His question concerns a material assumption and deserves an answer regardless of how uncomfortable it felt.
Maren also checks the documents she used. Her slides drew on version 3, prepared before the deferral was recorded. Version 4 contains £4.5 million, reflecting the £300,000 change. The subtraction reconciles, but the document sequence does not prove a neural cause for her behavior. It identifies a real preparation defect alongside a personal insult.
Rafi initially asks whether defending status was the main issue. Maren objects: she was also worried that changing the number would affect staffing plans. Rafi withdraws the suggested explanation and asks about both concerns separately. They agree that neither uncertainty about staffing nor embarrassment makes the insult acceptable.
Send an apology that does not demand reassurance
On 4 November at 09:00, Maren emails Ivo and copies Dalia because both heard the original exchange. She writes:
“I interrupted your question and said you did not understand the business. That was a personal attack, and I am sorry. Your question about the deferred order was relevant. My slides used version 3. Version 4 shows £4.5 million after the £300,000 deferral. I will correct the slide and its source reference. You do not need to reassure me or discuss the personal impact with me for the forecast question to be answered.”
At 11:00, Ivo replies that he accepts the correction to the record. He does not say that trust is restored. He asks to have the forecast file available during the next meeting because he still has a question about timing. At noon, Dalia agrees to chair the next discussion with the current file open and to let each question finish before asking for a response. The coach has not assigned work to either person; the actual recipients have accepted their own parts.
By 15:00, Maren replaces the slide and sends the corrected version and source reference to the original meeting attendees. Dalia acknowledges receiving it at 16:00. The record does not establish that every attendee opened it or that the staffing implications are resolved. The apology, document correction and remaining business question have different completion conditions.
Rehearse an accurate answer rather than a soothing performance
During the 3 November session, before sending that apology, Maren had chosen to practice a possible repeat question. Rafi asked permission to play the questioner and made clear that she could stop. In her first attempt she spoke more slowly and reported feeling less tense, but still answered, “There is no risk to the delivery timetable.” Nothing in the available documents supported that assurance. A smoother delivery had not made the answer accurate.
They stopped and revised the response:
“You are asking whether the deferred order changes the timing behind the forecast. The revised amount is £4.5 million. I have not checked the delivery-timing consequence. I need to identify who can confirm it before offering an answer.”
Maren tried again without a breathing exercise. She let the question finish, looked at its written wording and used the revised answer. She still felt uncomfortable. They recorded a successful rehearsal of those actions, not proof of calmness, a changed brain state or future performance. The difference between the two attempts was accuracy and acknowledgment of uncertainty.
Keep the second meeting’s partial result intact
The next forecast discussion takes place on 10 November. Dalia has version 4 open as agreed. Ivo asks about delivery timing. Maren lets him finish, says she has not checked that consequence and asks who holds the relevant information. No personal insult occurs in that exchange. That is a useful observed difference from 2 November, but it is one different meeting with changed chairing and source access.
Maren then slips into another overpromise: “Operations will confirm by tomorrow.” Operations has not accepted that deadline. Dalia asks who agreed it. Maren retracts the promise in the meeting and says the response time is unconfirmed. Her listening has improved in that moment; her handling of other people’s commitments still needs attention.
At 13:00 that day, Maren asks operations lead Oren whether he can check the delivery consequence and when he could respond. At 15:00, Oren accepts a review by 12 November at 16:00. Maren sends that accepted deadline to Ivo and Dalia at 15:30. Dalia acknowledges at 16:00. Receipt by Ivo is not recorded.
On 12 November at 15:00, Oren replies that the timing depends on a supplier confirmation that is still missing. He has completed the agreed review and reported its limitation; he has not confirmed delivery. Maren forwards that limited result to Ivo and Dalia at 16:00, making clear there is no confirmed delivery date. Dalia acknowledges it the next morning. No later supplier answer or Ivo acknowledgment is observed by the coaching review.
Review accountability without declaring a cure
On 17 November, Maren first says, “The hijack is fixed.” Rafi asks her to compare that claim with the record. She revises it: the second exchange contained no insult, she answered the actual question more accurately, and she corrected a new unauthorized promise. The business uncertainty remains, and so does the possibility of recurrence. Feeling less threatened was not measured across comparable situations.
Before the session ends at 14:50, Maren confirms that the coaching summary should retain the first rehearsal’s unsupported reassurance, the later promise error and the missing supplier confirmation. She chooses to keep using the written question and source-version check. There is no additional coaching support promised after the second session, no report sent to her employer and no claim that Ivo has forgiven her.
A completed incident-and-repair worksheet
Use the following seven fields for one event. Keep a short description that the client can inspect; do not request confidential business documents merely to make the worksheet look comprehensive. The completed entries show what Maren could actually distinguish.
| Field | Completed entry |
|---|---|
| Event | Ivo asked whether the £4.8 million forecast included a £300,000 deferral. |
| My interpretation and experience | I thought he was undermining me; that intention is unverified. I felt hot and embarrassed and worried about staffing implications. |
| My behavior | I interrupted twice and made a personal remark about his competence. |
| Impact and missing evidence | The question went unanswered in the meeting. Ivo’s later reply accepts the record correction, without reporting restored trust. My slide used an older version. |
| Repair actually completed | 4 November apology, corrected slide and source reference sent; Ivo and Dalia’s specific replies retained. |
| Practice and observed use | First calm-feeling rehearsal remained inaccurate. Revised rehearsal acknowledged uncertainty. On 10 November I listened, then made and withdrew a different unsupported deadline promise. |
| Conditions and next review | Current file open and chair’s turn-taking arrangement helped create a different situation. Supplier timing remains unknown; chosen preparation continues without further coaching booked. |
The worksheet does not need a field labeled “brain cause.” If the person wants to use the hijack metaphor, record it as their shorthand and return to the evidence. An empty impact field is also legitimate when no impact account has been offered. Do not manufacture forgiveness or turn a recipient’s refusal to discuss feelings into evidence of resistance.
Choose a response that fits the real problem
Different failures need different corrections. An outdated source calls for a version check. An insult calls for responsibility and repair. An unaccepted deadline calls for a request to the actual owner. A persistent pattern of aggression may need organizational action and professional support. One preferred calming tool cannot do all four jobs.
For listening practice, use a short real question with permission, stop on request and check whether the reply addresses it. Our guide to active listening in coaching develops that part of the work. Avoid provocative simulations designed to overwhelm the client, especially when a normal rehearsal already shows the decision error.
The NHS anger guidance suggests giving yourself time to think and seeking medical help when anger is difficult to cope with, self-help is not working or behavior is abusive or violent. Coaching is not treatment for those problems. Immediate danger requires appropriate urgent help and safeguarding, rather than rehearsal. Explain the service distinction using the coach and therapist comparison where useful.
Use Coachful to support preparation and accurate records
Coachful is our #1 recommendation for organizing this coaching workflow. We build Coachful, so that recommendation comes from the product’s maker. Its role here is to deliver preparation and retain chosen actions, not measure an amygdala response or establish that workplace conduct is safe.
- Deliver the worksheet deliberately. Open Clients, select the client, then Worksheets & resources and Assign. Delivered resources appear in My Content; an emailed link remains gated. Copying a private link alone does not grant access, and staff preview is not a client’s completed response.
- Separate reflection from the shared action. Unshared journal and takeaway pages stay private. Where Share with coach is enabled, the client can deliberately expose a particular page to authorized coach access. Keep sensitive personal words unshared if that is the intention; use a separate agreed action summary for the chosen question and version check.
- Record behavior without overstating the result. A task to check the forecast source can be completed even while delivery timing remains unresolved. A completed task does not prove that a colleague accepted a new deadline, restored trust or stopped feeling intimidated.
- Review the actual recap route. Recaps send automatically by default unless the applicable controls hold them. Check Sessions, settings gear, Session AI, Session recaps, the session type’s approval override and the actual text and audience. A generated summary saying “problem resolved” would contradict this case; saving it is not client approval.
Browse the Coachful features against the records your service needs. Keep commercially sensitive forecasts in their appropriate organizational system and use only a permitted minimal coaching summary. The ICF’s Code of Ethics resource emphasizes clear roles, confidentiality and accountability; a convenient workspace does not replace those agreements.
Frequently asked questions
Is an amygdala hijack a medical diagnosis?
It is a popular explanatory phrase, not a diagnosis established by a coaching conversation. Goleman himself revised the anatomical shorthand. Describe the person’s reported experience and actual conduct, and leave clinical assessment to appropriately qualified professionals.
How long does it take to reset?
This guide supplies no biological reset timer. A brief pause may create room to respond, but elapsed seconds cannot certify accurate thinking. Check whether the person can hear the question, distinguish known from unknown information and return responsibly to the discussion.
Does feeling threatened mean the colleague behaved badly?
No. A material question may feel uncomfortable and still be entirely legitimate. Examine its content and context. Also take actual harassment or discrimination seriously through appropriate channels; do not automatically reinterpret it as the client’s mistaken perception.
What if the leader is calmer but still gives a wrong answer?
Correct the answer and its preparation conditions. Maren’s first rehearsal felt easier but included an unsupported assurance. The more useful change was admitting the evidence gap and asking an actual owner to agree the next step.
Can this case show that coaching changed the brain?
No brain measures were taken, and the later meeting changed in several ways. It shows a fictional leader completing specific repairs, responding differently once and making another error. Neither that limited record nor a task-completion screen establishes neural change or lasting behavioral improvement.
