What the data actually shows
Much of daily behaviour is habitual and automatic. Wendy Wood's research on habit formation estimates that a large share of everyday actions are performed in stable contexts that cue the same response, largely outside deliberate intention. The practical implication is that when the cue is present — the same place, mood, time, or trigger — the old response fires before a new plan can intervene. Intentions matter most when habits are weak; when habits are strong, the cue tends to win.
We also appear to learn surprisingly poorly from our own failures. Eskreis-Winkler and Fishbach (2019) ran a series of experiments and found that people learned less from failure feedback than from success feedback, even when failure was just as informative. A central reason is that failure is ego-threatening: it makes people feel bad about themselves, so they disengage from the information instead of absorbing the lesson. In other words, the very moments that should teach us the most are the ones we are most motivated to look away from.
On top of this sit two well-documented biases. Present bias is the tendency to overweight immediate rewards and costs relative to future ones, which is why the short-term pull of an old pattern routinely beats the long-term benefit of a new one. And the planning fallacy — the tendency to underestimate how long things will take and how much could go wrong — means we repeatedly walk into the same situations underprepared. Together these make repetition the path of least resistance.
Why this feels different from how it actually is
It feels like the problem must be willpower or self-discipline, because in the moment of failure we are intensely aware of having chosen badly. But that awareness arrives after the automatic response has already fired. From the inside, a cue-driven slip is indistinguishable from a deliberate choice, so we blame the chooser rather than the cue — and resolve to 'try harder' next time, which leaves the actual trigger untouched.
It also feels like we surely learned our lesson, because we remember the regret vividly. But remembering that something went badly is not the same as extracting why, and the research suggests the sting of failure often pushes us to stop thinking about the episode rather than to analyse it. The emotional memory survives; the usable lesson frequently does not.
And because we judge our past decisions knowing how they turned out, the repeat mistake can look more obvious and avoidable in hindsight than it ever was at the time. That makes the pattern feel like stubbornness or stupidity, when much of it is simply the ordinary mechanics of habit, emotion, and attention working as they always do.
What the research says to do about it
The most consistent lever is changing the cue, not just the intention. Habit research suggests that altering the context — removing the trigger, changing the environment, or inserting friction between cue and response — is more effective than willpower, because it interrupts the automatic chain before it starts. Disrupted contexts, such as a move or a change of routine, are also windows when old habits are unusually easy to replace.
Implementation intentions — concrete 'if-then' plans that pre-decide what you'll do when a specific cue appears — have reasonably good support for closing the gap between intention and action. They work by linking the new response to the same trigger that used to fire the old one, so the deliberate choice is made in advance rather than in the heat of the moment.
To actually learn from failure, the research points to deliberately reframing it as information rather than a verdict on yourself, which lowers the ego threat enough to engage with the lesson. Studying others' failures, or stepping back and analysing your own from a more detached, third-person perspective, can make the feedback usable precisely because it is less personally stinging.
What the research says does not help
Simply resolving to 'try harder' or 'be more disciplined' tends to do little, because it targets intention while leaving the cue and the habit intact. When the trigger reappears, the old response still fires faster than the resolution can. Effort aimed at the wrong link in the chain mostly produces guilt, not change.
Harsh self-criticism after a slip is not only unpleasant but appears counterproductive: it increases the ego threat that, in the research, makes people disengage from failure rather than learn from it. Punishing yourself for the mistake can make the lesson harder to absorb and the relapse more likely.
Assuming the regret itself will prevent a repeat is also unreliable. Vividly remembering how bad something felt is not the same as understanding what caused it, and present bias means the next time the immediate pull can still outweigh the remembered consequence. Feeling bad is not a mechanism for change; redesigning the cue and extracting the lesson is.
Real numbers in context
There is no clean statistic for how often people repeat their mistakes, and anyone quoting a precise figure should be treated with caution. What the research does support is more structural: Wendy Wood's work suggests a substantial fraction of everyday behaviour is habitual and context-cued rather than deliberately decided, which is why the same situations reliably produce the same actions.
The clearest experimental result is about learning, not frequency. Eskreis-Winkler and Fishbach (2019) found people learned measurably less from failure than from success across multiple studies — a robust direction even though the exact effect size depends on the task. The honest takeaway is qualitative: repetition is the default unless you change the cue or actively extract the lesson, and self-blame makes both of those harder, not easier.