Cognitive reflection test scored on how often the first answer won

The idea behind these items is narrower than the phrase “cognitive reflection” suggests, and the narrowness is the point. Each question is built so that a wrong answer presents itself immediately, is easy to check, and is wrong for a reason anybody can see once they look. So the item does not test whether you can solve it — nearly everyone can — but whether the checking happens when an answer is already sitting there. That was the case made in Frederick (2005), Cognitive Reflection and Decision Making, Journal of Economic Perspectives, which reported a mean of 1.24 of 3 across students at US universities, several thousand respondents across university samples. That number is a landmark rather than a rank, and this page prints it only beside your score on those same three items: it is a student mean, not a population one, so being above it does not place you above the general public, and with no individual-level spread published alongside it there is no percentile to compute from it at all.

The problem the intervening twenty years created is exposure. These three questions are among the most reprinted on the internet, and a visitor who already knows the ball costs five cents is not overriding anything — they are recalling. Almost every version of this test online ignores that and reports a total anyway, which is the single biggest defect in the genre: the score of a well-read visitor and the score of a careful one are indistinguishable. This page handles it in two ways. Four additional questions are asked that are ours rather than anybody's published set, written on the argument that the construct survives outside the original arithmetic items — Thomson & Oppenheimer (2016), Investigating an alternate form of the cognitive reflection test, Judgment and Decision Making is where that argument was made and tested, and our four are not its four. And after the result is on screen, you are asked which items you had met before, with the count then repeated over the new ones. Asking that first would have been worse than not asking: a visitor primed to consider familiarity searches memory instead of the problem.

The second departure from how these items are usually presented is the scoring itself. Three outcomes are recorded per question rather than two, because the specific wrong answer and any other wrong answer mean different things — one says the question was never opened, the other says it was opened and something slipped. That is also why the answer box is empty rather than a list: seeing the correct answer beside the tempting one lets recognition stand in for reflection, and the resulting score measures neither. The nearest page to this one is the logical reasoning test, where the tempting answer is a conclusion that happens to be true, and the deductive reasoning test puts the same conflict inside a single rule. If it is the arithmetic rather than the override that interests you, the numerical reasoning test allows a calculator and measures whether you can find the right two numbers; the pattern recognition test asks the same question about sequences, and the mechanical aptitude test is where a first intuition about levers and gears gets checked against how they actually behave.

A reaction time here is the interval between the frame that painted the stimulus and the timestamp the browser attached to your key, both read from the same monotonic clock. What neither can see is the display pipeline behind it, so on a 60 Hz screen roughly 16 ms of every figure below is the machine rather than you. That is the timing floor: two numbers closer together than that are the same number, and this page reports no precision it cannot support.

Here the clock is doing real work rather than sitting in a footnote, because the difference between the time your correct answers took and the time your lure answers took is one of the two figures this page reports. Those durations run to tens of seconds, so the floor is thousands of times smaller than the effect, and both are printed to a tenth of a second so that no precision is claimed which the browser cannot deliver.

This is a measurement exercise, not a clinical assessment. It reports what you did on this page against a stated reference and nothing more — it cannot establish impulsivity, a thinking style, or how carefully you decide anything away from these seven questions. Only a qualified professional, working with more than a browser, can make that judgment.