Why does a near-miss on a short, timed task make people start it again almost immediately, while a clean failure often sends them off for a cup of tea? The question matters because the shape of an outcome, not just its valence, seems to govern whether we retry. And on tasks measured in seconds rather than hours, that retry decision happens fast, often before any deliberate reasoning has kicked in.

The anatomy of a near-miss

A near-miss is an outcome that falls just short of a success threshold. In a 90-second task, the threshold might be a score, a time, or a count of correct responses. What makes near-misses psychologically distinctive is that they carry two contradictory signals at once: the outcome was a failure, but the margin was small enough to imply that success is close at hand.

Behavioural research on this goes back further than most people assume. B. F. Skinner's work on schedules of reinforcement established that intermittent reward produces more persistent behaviour than consistent reward. A near-miss is a particularly potent form of intermittency, because it isn't random noise — it looks like information. The margin tells you something about your own capability and about the task's difficulty, and both of those readings point toward "try again now."

The critical word is now. Near-misses compress the retry interval. A clean failure at 40 per cent of target reads as a gap in competence; a failure at 95 per cent reads as a rounding error. The first invites reflection or abandonment. The second invites another attempt before the feeling fades.

Why the second attempt arrives before the reasoning does

Dual-process accounts of cognition, popularised by Daniel Kahneman though with roots in much earlier psychology, separate fast automatic responses from slow deliberative ones. Retry-after-near-miss sits firmly in the fast lane. There is no cost-benefit calculation happening in the moment; there is a small spike of arousal, a sense of unfinished business, and a hand already moving.

This is where the Zeigarnik effect is genuinely useful. Bluma Zeigarnik's observation, from work in the 1920s, was that interrupted or incomplete tasks are remembered better than completed ones. The mechanism is usually described as a kind of cognitive tension that persists until closure. A near-miss is an incompleteness signal of an unusually sharp kind: the task is not merely unfinished, it is almost finished. That combination is difficult for the mind to leave alone.

There is a second layer. Self-determination research, particularly work by Edward Deci and Richard Ryan, distinguishes intrinsic motivation from external pressure. Short timed tasks with a visible personal best engage something closer to mastery motivation than to reward-seeking. The near-miss doesn't dangle a prize; it dangles a slightly better version of your own performance. That is a more durable hook than an external incentive, and it explains why the retry feels self-directed rather than manipulated.

Three retries before lunch: what the pattern reveals

The specific pattern in the title — two near-misses, then three retries in quick succession — is worth taking seriously as a behavioural signature rather than an anecdote. It suggests a short, self-limiting burst of persistence rather than an open-ended loop.

Consider a concrete illustration. A staff member at a Manchester software firm uses an internal typing-and-accuracy drill: 90 seconds, a fixed passage, a score combining speed and errors. Her best is 94. She scores 91 twice in a row, both times with the errors clustered in the final fifteen seconds. She then runs the drill three more times inside twenty minutes, without a break, before stopping for the day. The three retries are not identical in character: the first is a direct response to the near-miss, the second is a response to the first retry's own near-miss, and the third is the one where attention visibly degrades and the score drops.

That degradation is the important part. Near-miss-driven retries tend to cluster tightly and terminate abruptly, because the same arousal that fuels the first retry erodes the fine motor control or sustained attention the task requires. The pattern is self-limiting, but not because the person decides to stop. It stops because performance collapses.

Loss aversion versus the near-miss pull

Kahneman and Amos Tversky's work on loss aversion established that losses loom larger than equivalent gains. It is tempting to read near-miss retries through that lens — the near-miss is a loss of a success you almost had. But the fit is imperfect. Loss aversion predicts caution and risk-aversion; near-miss retries often look like the opposite, a willingness to re-engage immediately with the thing that just defeated you.

A better fit is the distinction between counterfactual and actual outcomes. Research on counterfactual thinking, associated with Neal Roese among others, shows that near-misses generate vivid upward counterfactuals — "if I had just paced the last section better" — and that these counterfactuals are motivating precisely because they are specific and actionable. A clean failure generates vague counterfactuals, or none. You cannot easily simulate a path from 40 per cent to 100 per cent, but you can simulate one from 91 to 94.

Designing for the retry you actually want

For anyone building or supervising short timed tasks — training drills, assessment instruments, practice tools — the near-miss effect is a design variable, not a fixed feature.

Set thresholds where near-misses are informative. A scoring system that produces near-misses at a rate of roughly one in five attempts sustains engagement without tipping into frustration. Too few, and the task feels arbitrary. Too many, and the near-miss stops carrying information and becomes noise.

Cap the burst. Because the third retry in a cluster tends to be the one where performance degrades, there is a case for building in a short enforced pause after two consecutive near-misses. Not a lockout — a pause. The arousal fades in a couple of minutes, and the fourth attempt is usually better than the third.

Make the counterfactual visible. If the near-miss is worth acting on, the person needs to see where the margin came from. Error clustering by time segment, as in the example above, turns a vague sense of "so close" into a specific adjustment. Without that, the retries are just repetition.

Watch for the shift from mastery to compulsion. The healthy version of this pattern is curiosity about your own limit. The unhealthy version is retrying to discharge tension rather than to improve. The tell is whether the person can articulate what they will do differently on the next attempt. If they can't, the retry is a reward-loop artefact, not learning.

What to watch for next

The near-miss retry is one of the clearest windows we have into how people convert ambiguous feedback into action, and short timed tasks make the mechanism unusually visible because the whole cycle — attempt, outcome, appraisal, retry — fits inside a few minutes. If you run any kind of repeated short assessment, the most useful thing you can do this quarter is log the margins, not just the pass rates. Record how close each failure was, how quickly the next attempt followed, and whether performance improved or degraded across the cluster. Within a few weeks you will have a picture of your own near-miss dynamics that no general principle can give you, and a much better basis for deciding when to prompt a retry and when to prompt a pause.