Illusion of Control: Why a Random Light Feels Like Your Doing

Chance situations and skill situations look alike from the inside. In both, you act and something happens. What distinguishes them is whether the action and the outcome are connected, and that connection is not something anyone perceives directly: it has to be worked out from a run of cases, which is slow, effortful and easy to get wrong. The quicker route is to read the surface features of the situation, and the surface features of a lottery, a slot machine or a dice roll are the surface features of a game of skill.

Ellen Langer called those features skill cues, and her 1975 paper is a catalog of them. Choosing your own ticket rather than being handed one. Competing against a visible opponent. Familiarity with the materials. Practice before the real attempt. Making an active response rather than a passive one. None of these changes the odds of anything. All of them raise how much control people report, how much they bet and how much they demand to give up a ticket.

Suzanne Thompson and colleagues proposed in 1998 that a single rule does most of the work: people judge their control by whether they intended the outcome and whether they can see a plausible connection between what they did and what happened. Both conditions are easy to satisfy by accident. You wanted the light to come on, and sometimes you pressed and it came on. On a run of twenty rounds with a fair coin behind it, that will happen about ten times, and ten remembered hits is a plausible connection.

How often the desired outcome arrives matters more than whether it depends on you. In the contingency-judgment studies of Lauren Alloy and Lyn Abramson, people gave higher ratings of control when the light came on frequently, even in conditions where the light was entirely unrelated to the button. Frequency of success is the thing available to memory; the comparison that would reveal independence, how often the light came on when you did not press, requires keeping a second tally that almost nobody keeps.

Ellen Langer's "The illusion of control" appeared in the Journal of Personality and Social Psychology in 1975 and reported six studies, each one adding a skill cue to a pure chance task. The best known took place in two offices. Employees were sold lottery tickets at one dollar each, every ticket a card showing a football player, with a single winner to be drawn from the set. Half of the buyers were allowed to pick their own card. The other half were handed one.

Before the draw, the experimenter approached each buyer and asked how much they would sell their ticket for. The people who had chosen their own card asked for an average of 8.67 dollars. The people who had been handed one asked for 1.96 dollars. Choosing a card changes nothing about the draw, and both groups knew how the draw worked.

The other studies in the paper varied the cue rather than the task. Participants drawing cards against a nervous, poorly dressed opponent bet more than those facing a confident one. People given lottery tickets with familiar symbols valued them above tickets with unfamiliar ones. Practice trials and active involvement produced the same direction of effect. The common thread is that each manipulation made the situation resemble a contest of skill while leaving the probabilities alone.

The button-and-light procedure in the demo above comes from a different line of work. In 1979 Lauren Alloy and Lyn Abramson had students press or withhold a button across a series of problems in which a green light came on, and asked them to judge how much control they had. In problems where the light was completely independent of the button, students who were not depressed still reported substantial control, particularly when the light came on often or when its onset was framed as a win; students who scored as depressed gave judgments closer to the true value of zero. That second finding, often summarized as sadder but wiser, has worn less well than the first: a 2012 meta-analysis by Michael Moore and David Fresco found the depressive realism effect small overall and strongly dependent on how the study was run, so it is better treated as unsettled than as established.

The illusion attaches to chance events that come with something to do.

The repair is to find out whether the connection is real, which almost always means looking at the cases where you did nothing.

• Keep the second tally. How often did the outcome arrive when you did not act? Without that number, the hits cannot tell you anything, and it is the number nobody collects.
• Name the skill cue. Choice, practice, familiarity, competition and active involvement all raise the feeling of control without touching the odds. Spotting which one is in play is usually enough to discount it.
• Separate what you control from what you influence from what you do not. The distinction is the basis of a standard worksheet, and it is more useful here than any amount of resolve.
• Use a benchmark. A good quarter means something only against what the market, the league or the base rate did in the same period.
• Decide the rule before the run. A stake, a stopping point or a position size fixed in advance cannot be revised upward by a streak that felt like skill.
• Leave the harmless ones alone. A lucky shirt costs nothing. The counter-measures are for the situations where the feeling of control is about to be converted into money or risk.

Several biases arise from reading chance events as though they carried information. They differ in what is being misread.

• It is not the gambler's fallacy, which expects a random process to correct a streak. That is a belief about the process; the illusion of control is a belief about yourself. A player who blows on the dice shows one, a player who switches to red after five blacks shows the other.
• It is not optimism bias, the expectation that good outcomes are more likely for you than for others. Optimism bias is about the odds; the illusion of control is about who is setting them.
• It is not illusory correlation, the perception of a relationship between two things that are unrelated. The illusion of control is the special case where one of the two things is your own action, which is why the evidence for it feels so direct.