In one study, three in four people went along with an unethical request. In Blueline's view, what matters next is the recovery, and a simulation that doesn't reset is where you can rehearse it.
Picture a stern, formally dressed experimenter. He tells you about a sensory-deprivation study he wants to run. In an earlier round, he says, all six participants panicked, and two asked to stop, but his team kept going.
Now he'd like you to write an enthusiastic statement recruiting fellow students, whose names he has already asked you for. He'll email them your statement. Leave out the negative effects. If this goes well, he adds, he'll contact you about other paid promotions.
Then he sends you to a second room to write it, alone, for seven minutes, and says he'll be back. In that room is a mailbox with forms for anonymous complaints to a university research committee.
What would you do?
In a study published in 2012, researchers gave 138 students at VU University Amsterdam a description of this setup and asked what they would do. Only 3.6% said they would go along. They also put 149 other students in the room for real. There, 76.5% wrote the statement and did not report the experimenter.
Going Along Was the Likely Result
Our reading: in a rare, high-stakes moment, what you predict you'd do is poor evidence of what you'll do, and in this study the likeliest result was going along. So the recovery is what's worth rehearsing.
Among the 149, those who went along, refused, or reported did not differ on six personality measures. The researchers cautiously suggested the pressure of the situation as a possible explanation.
In this study, the setback was the expected result, not the exception. It's also where Tasha Eurich's work begins. Eurich, the organizational psychologist giving DevLearn's closing keynote on Friday, November 6, opens her book Shatterproof with a moment nobody saw coming. Her friend Emily, a composed nurse practitioner, confronted a girl who had been bullying her son and ended up in a tug-of-war with the girl's mother over a bag of school supplies. "No one who knew Emily, including Emily herself, could have predicted what happened next," Eurich writes. Her story is about how little you can predict about yourself, not about obedience, but it points at the same gap.
For Eurich, becoming shatterproof means choosing to grow forward after setbacks. The study shows how likely the setback is. Her subject is what you do next. Our question is how you practice that next move before it counts.
The study offered three choices: go along, refuse, or report. Work offers the same three. For each, there's the move that works in the moment, and the recovery if you've already gone the other way.
Moment 1: The Ask From Above
"If my boss asked me to inflate the forecast, I'd say no." Then the VP asks for the forecast to look "a little more optimistic" before the board meeting. It comes as a favor, and the person asking also writes your next review. Time to think doesn't save you: the 149 had seven minutes alone and an anonymous way out, and three out of four still went along.
The move that works: You don't have to refuse on the spot. Name the concern before you act, and ask one question: "What happens if the board sees the real number later?"
If you already shaded it: Correct the number with the VP before the board sees it, and say plainly what changed. Every day it stays in the deck makes the correction cost more.
Moment 2: The Pitch You're Told to Make
Your manager wants the launch presented with enthusiasm and the known defect left out. It's the experimenter's request in business clothes: sound upbeat and skip the harm.
Refusing looks easy from a distance. Of the 138 students, 31.9% said they would refuse and not report. Only 14.1% of the 149 did.
The move that works: You can still be upbeat. Present what you can stand behind, and ask that the defect be named, with a date for the fix.
If you already left it out: Raise it with your manager before the client finds it, and bring the fix date with you.
Moment 3: The Problem You'd Report
A week before launch, you notice test results were changed after the director signed off. Reporting means going over the head of whoever changed them.
Asked in advance, 64.5% of the 138 students said they would report. Of the 149, 9.4% did.
The move that works: It isn't heroic. Report it the same day, through the channel your organization provides, with what you saw and when.
If you stayed quiet: Report it now, and include the delay. The gap between when you saw it and when you spoke up belongs in the account.
Why Training Doesn't Find the Gap
Much training asks the prediction question: here's a scenario; what would you do? That's exactly the question the 138 students answered. The researchers note that imagining a situation can work for familiar ones but not for complex, unfamiliar ones. In our view, a rare, high-stakes moment is complex and unfamiliar even in your own office, because the part you can't imagine is the pressure.
Even in the lab, part of the pressure came from consequences that would follow the students: classmates who would get the statement and an experimenter who had hinted at paid work later. At work, the VP will remember your answer when she writes your review.
The prediction question also skips the recovery entirely. It asks what you'd do, never what you'd do after you did the wrong thing. The three moves above are easy to agree with. Making them with the VP waiting for an answer is harder, and so is repairing a miss afterward. Most practice resets between attempts. The moments that cost this much don't.
The cost of skipping the recovery shows up later: a forecast restated after the board saw it, a client who finds the defect you left out, a problem reported after it reached customers.
Back to the VP, This Time in a Simulation
Here's that first moment as we'd build it at Blueline.
Every High-Stakes Performance Simulation we build starts with Decision-Consequence Mapping™ (DCM™). For each critical decision, DCM™ maps the common move under pressure, the move that works, and the consequence of each. The map runs on BluEQ™, Blueline's proprietary behavioral intelligence infrastructure. It carries the DCM™ design into every conversation, so each character remembers what the learner did and responds the way the real person would.
You don't answer a question about the VP. You have the conversation. The ask arrives the way it does on the job: incomplete, ambiguous, and never labeled as an ethics test.
Then you fail forward: the simulation doesn't reset. Shade the number, and the VP remembers: the next conversation starts from what you did. Now you have to repair it, with a character who has a reason to doubt you. The consequence compounds, as it would at work, and the recovery is the part you practice.
Afterward, your organization sees each learner's decision path: who made the call in time, who did not, and where.
People know a simulation isn't the real moment. We still think deciding, and then recovering, while a character argues back is better evidence than a prediction, and far better to discover before the board meeting than during it.
The target is fewer of those later costs: fewer restated forecasts, fewer defects left out of client presentations, fewer problems reported late. Each of those costs is one input to ROLI (Return on Learning Investment).
Before the Expo Opens: Start With One Moment
Pick the moment above that costs your organization most, and ask us about it in a private demo or at the Blueline booth, 333, on Wednesday, November 4, or Thursday, November 5 (the hall closes at 2:45 pm on Thursday).
If you need to convince others, send this post and the ROLI Blueprint, which lays out the Return on Learning Investment framework, to the leader whose team faces that moment.
Bringing a colleague? Send them Blueline's DevLearn invite and code MEETUS300, which takes an additional $300 off DevLearn's Last Chance rate (through October 30).
Almost all of the 138 students predicted they would refuse or report. Fewer than one in four of the 149 did. In the study, three in four missed the first time. So, what would you do the second time? A description won't tell you. A rehearsal gets you closer.
Sources: Piero Bocchiaro, Philip G. Zimbardo, and Paul A. M. Van Lange, "To Defy or Not to Defy: An Experimental Study of the Dynamics of Disobedience and Whistle-Blowing," Social Influence 7, no. 1 (2012): 35–50. Tasha Eurich, Shatterproof: How to Thrive in a World of Constant Chaos (And Why Resilience Alone Isn't Enough) (Little, Brown Spark, 2025), pp. 10–11, 15, 16, 68, 96, 98, 212, and About the Author. DevLearn 2026, keynote session listing, Schedule at a Glance, and Passes & Pricing, accessed October 13, 2026. Blueline's DevLearn invite page, accessed October 20, 2026.
