Neuroscientists have added fresh insight to the observation that people are surprisingly willing to hurt others if they are ordered to do so.
This was famously shown by psychologist Stanley Milgram in the 1960s.
In the new study, subjects in pairs were paid to deliver mild electric shocks to one another.
If they were instructed to administer the shocks, they sensed more of a delay before the jolt was delivered, compared to when they made their own decisions.
Researchers regard this timing judgment as an indicator of how responsible we feel for our actions.
When we switch on a light, for example, we know we are in control and we usually perceive the effect as instantaneous, even if there is a lag.
By contrast, the new findings suggest that if we are following orders, that joined-up perception drifts a little and our sense of "agency" is genuinely reduced.
"A useful marker of the sense of agency... is the subjective compression of the interval between what I do - and what I make happen," said Patrick Haggard, senior author of the study, which appears in Current Biology.
"Most previous work had been based on just asking people whether they felt responsible; that's a little bit tricky, because people tend to report what they think they should say."
He and his colleagues wanted to test whether being bossed around produces a real, measurable change in how people perceive their own actions.
What is the psychological underpinning, if any, for the claims of Nazi defendants at the Nuremberg trials that they took no responsibility because they had orders to follow?
It was in the wake of those trials that Dr Milgram's experiments achieved notoriety. Volunteers obediently ramped up the shocks they were giving to a "learner" in an adjoining room - actually an actor - whom they could hear protesting and, eventually, in obvious distress.
Those findings were taken as a powerful illustration of our propensity to decouple ourselves from our choices, if someone else - in that case, a commanding man in a white coat - is giving the orders.
They are a touchstone of psychology courses worldwide and were the subject of a recent film.
"Milgram's interest was really focused on whether people will obey an instruction or not. But he did not really focus on what it feels like when people do follow instructions," Prof Haggard, from University College London, told BBC News.
Effects on the brain
In Prof Haggard's key experiment, subjects could choose to give their co-participant a "painful but tolerable" electric shock - or not - by pressing one of two keys.
Each shock would add 5p to their fee for taking part, and each pair took turns so that both subjects knew exactly what the shocks felt like.
"Pressing either key on the keyboard produces a tone, and the participant's task is to report, in milliseconds, how long they think the interval between the key-press and the tone was," Prof Haggard explained.
That interval, in reality, varied randomly between 200, 500 and 800 milliseconds.
"That gives us our implicit measure of sense of agency."
Crucially, a scientist was also in the room and for half the trials, she firmly told the participants which key to press.
"The interesting result is that people perceive the interval between the action and the tone as longer, in the condition where they've been given a coercive instruction, than in the condition where they decide for themselves what to do," said Prof Haggard.
"Coercion produces some subjective experience of distancing. Instructions really can change the way we feel about what we're doing."
The team also recorded brain activity during some of the experiments, using electrodes on the top of the scalp (an electroencephalogram or EEG).
These measurements revealed subtle differences between how the brain responded to the sound of the tone, depending on whether the subject had been ordered to act or not.
"We take this as showing that coercion has a surprisingly powerful effect on the brain, reducing the extent to which the brain processes the consequences of our actions," said Prof Haggard.
Dr Molly Crockett is a social neuroscientist at the University of Oxford. She said the new findings were interesting and novel - particularly because previous studies of coercion had largely relied on people self-reporting their feelings of responsibility.
"You can imagine that explicitly, if someone wants people to think they're a better person, they would report, 'Oh yeah, I totally didn't feel responsible when someone asked me to do this bad thing,'" Dr Crockett said.
"But on this implicit measure, which people are unlikely to be able to manipulate, you still see a signature of reduced agency - which is really cool."
Asked about the implications of his findings, Prof Haggard said there was no reason to give any credibility to the so-called "Nuremberg defence".
"We always need to be sceptical of that defence - of somebody who says, I was only obeying orders," he said.
"Because very often, people have a secondary motive for saying that. They think they can avoid punishment.
"Secondly, just because somebody feels they are not responsible, doesn't mean they are not responsible. Society might still want to hold them accountable."
Instead, he suggested, the results emphasise the power - and the responsibility - of those who are giving orders.
"I think we need to hold people who give orders even more accountable. Because the people who execute those orders may not feel as responsible as one might like."
Follow Jonathan on Twitter