Personality and Individual Differences
A task you can actually do, and a dial that produces the same patterns without doing it. Speed and accuracy are always plotted together, because reading either alone is the mistake.
Simulated strategies — measures nothing about you
By the end you should be able to say why one outcome measure cannot separate ability from response caution, and read a joint speed–accuracy plot.
About 20 minutes. Nothing you do here is saved or sent anywhere.
Nothing here is timed out. Every trial waits as long as you need, practice comes first with feedback, and the keys are stated before anything begins.
You do not have to do the task at all. The strategy simulator further down reaches every conclusion on this page without it. And this measures nothing about you — not intelligence, not impulsivity, not attention or inhibition, and nothing clinical.
Two people do the same discrimination task. One is consistently faster than the other, and they are equally accurate.
Nine arrows appear. Decide whether most of them point left or right. Press the Left arrow or Right arrow key, or use the two buttons. Nothing is timed out and nothing flashes.
Four untimed practice trials with feedback, then sixteen scored trials with no feedback until the end.
Status
Three of them have identical ability and differ only in how much evidence they wait for. The fourth genuinely differs in ability. Everything below works whether or not you did the task.
Ability is held constant while you move this. Watch both numbers change.
| Respondent | Drift rate (ability) | Threshold (caution) | Expected accuracy | Expected RT |
|---|
Drift rate is how fast evidence accumulates — the ability term. Threshold is how much evidence is required before committing — the strategy term. The first three rows share a drift rate.
Response caution moves accuracy and reaction time together. A person can slide along the trade-off curve at will, without any change in how well they can actually discriminate. This is why instructing participants to "go as fast as you can while staying accurate" does not fix anything: it leaves each person to choose their own point on the curve.
Report accuracy alone and the most cautious respondent looks the most able. Report speed alone and the most impulsive one does. Neither ranking survives contact with the other measure, and both are routinely published. The fix is not a better single number — it is refusing to use one.
Evidence-accumulation models estimate drift rate and threshold separately from the joint distribution of choices and reaction times. That is why they are worth the trouble: they answer the question "is this an ability difference or a strategy difference?" which no summary statistic can.
RT distributions are skewed, so the mean and the median disagree and a handful of slow trials can move a result. Exclusion cut-offs, whether to analyse correct trials only, and how to handle anticipations are all researcher choices, usually made after seeing the data and rarely pre-registered. Your own block above shows the mean and median differing; on a real dataset that gap is where a good deal of flexibility lives.