Social & Critical Psychology · Simplified
The Frame You Take With You
Working out an answer with other people is not the same as agreeing with them, and it lasts longer.
What this is, and what it is not
You will watch a point of light in a dark field and judge how far it moved. It really does move, by a small amount that changes from trial to trial and is never revealed. There is no ruler and no scale, so the judgement is genuinely yours to make and no answer is correct.
This is a browser analogue of the ambiguous movement judgement used in Sherif's studies, not a replication of them. Sherif used the autokinetic effect, which needs a darkened room and will not appear reliably on a lit screen; the drift here is real movement put there on purpose. What is reproduced is the structure of the task, not the perceptual phenomenon, and there is more on the difference at the end.
The three other participants are generated by a rule that is printed further down, from your own opening judgements. Nothing you enter is saved or sent anywhere, and nothing here measures anything about you.
Step 1 of 3
First, a prediction
In the original studies of this kind, people judged an ambiguous display alone, then in a small group over several rounds, and then alone again. What happened on those final solo rounds?
Step 2 of 3
Eight trials
1 = barely at all, 10 = a great deal. Whole numbers from 1 to 10 only. There is no correct answer, and you will not be told what the display did.
Step 3 of 3
What you did
The vertical scale runs from 1, barely at all, at the bottom to 10, a great deal, at the top. Your judgements are the heavy line. The shaded trials are the ones with other people in them, and the three thin dashed lines are theirs, labelled B, C and D at the point where they leave the task. Each line is labelled at its own right-hand end. The thin lines appear here for the first time: they were on screen only while the others were part of the task.
| Group | Opened near | Settled near |
|---|
What this shows
A norm is not an agreement
Key idea: Going along with a group because it would be awkward not to is a decision, and it ends when the group does. What happens on an ambiguous task is different in kind. With nothing to judge against, the other people become the thing you judge against, and the figure that emerges is not a compromise anybody agreed to but a frame for reading the display at all. That is why it can still be there on the final solo trials, when the others are gone and there is nothing left to copy, and why the three other groups in the table settled somewhere else entirely from the same rule. A norm, on this account, is not what people have agreed. It is what has stopped being a question.
Eight trials is short, three simulated others is small, and their rule is a single convergence weight with a little noise rather than anything resembling a person. The display is a browser analogue rather than the autokinetic effect, and the difference is worth stating plainly. In Sherif's studies the light did not move at all: a stationary pinpoint in a fully darkened room appears to drift because the eye cannot hold still and has nothing to anchor against, so every participant was judging an illusion. That illusion needs darkness, dark-adapted eyes and a controlled viewing distance, none of which a web page can rely on, and a display that only might move for some readers would leave them estimating something they never experienced. So the point of light here genuinely moves, by a small amount that changes on every trial. The ambiguity is therefore real but differently produced, and it is ambiguity of measurement rather than of existence. Two further cautions. Not converging is not a failure and not a sign of independence of mind; it is one of the things this task can show, and a single short run could not tell those apart anyway. And the historical studies this borrows its logic from involved deception and conditions that would not be approved today, which is a reason to treat their findings as a starting point rather than as settled facts about people.
The longer version runs more rounds in each phase, lets you compare your own convergence against three full simulated groups round by round, and adds an explicit contrast with compliance. It is at Sherif Norm Formation Laboratory in the main collection.