No, a human cannot be programmed the way a computer is: there is no set of commands that is guaranteed to change any person in any situation. But psychology, behavioral economics, and clinical practice do have methods that raise the probability of specific actions, help habits stick, change how events are interpreted, and make people less vulnerable to certain manipulation tactics. The effects depend on the person, the context, the goal, and the quality of the research — and they are usually more modest than book covers promise. This article is a map of those methods with an assessment of the evidence, not a manual for controlling someone else's mind: its frame is changing your own behavior and protecting yourself from other people's pressure.
Key takeaways
There is no guaranteed control. No method described in the peer-reviewed literature gives a predictable result for every person. Studies report average shifts in groups, not guarantees for each individual. An average effect and one person's outcome are different things.
What changes is not "the person" but the probability of a specific action under specific conditions. A more useful question than "how do I remake them?" is "what makes this action easier, more visible, and more rewarding for the person themselves?"
Habits form through repetition in a stable context, but there is no magic timeline. The classic study by Lally and colleagues showed a huge spread: from a few weeks to many months. "21 days" remains a myth.
A specific "if–then" plan works better than a vague intention. The meta-analysis by Gollwitzer and Sheeran found a notable positive effect on average, although its size depends on the task and conditions.
The environment is often stronger than willpower. Cues, defaults, reduced friction, and visible progress are workable levers. But meta-analyses of nudge interventions show large heterogeneity, and some estimates may be inflated by publication bias.
Beliefs change more slowly and with more difficulty than actions. They require trust, good-quality information, and room to revise a position without losing face. Pressure and humiliation more often strengthen identity defenses.
Knowing how influence works is, above all, a tool for self-defense. A person who can name a tactic (social proof, false authority, manufactured urgency) responds to it slightly less often. There is evidence for "inoculation against manipulation," but it should not be overstated.
The line between help and manipulation runs along transparency and autonomy. If a person doesn't know they are being influenced and cannot decline without penalty, it is no longer help.
Why the "programming" metaphor is so appealing
Comparing a human to a computer is tempting because, in part, it works. A person has inputs and outputs: information, feedback, cues, actions. There is learning from mistakes, adaptation to repeated conditions, and automatic routines that run without deliberation. There is even an analogue of "debugging": notice the failure, change a condition, check the result. Cognitive psychology used such analogies for half a century, and they helped describe attention, memory, and problem solving.
The trouble begins when the metaphor is taken literally. A human has no single executable code that can be read and rewritten. Behavior arises from biology, prior experience, current goals, mood, relationships, cultural norms, fatigue, and chance events. The same "input" — say, the same reminder — produces a different "output" on a Monday morning and on a Friday evening. And unlike a program, a person also observes themselves: they can notice an attempt to influence them and resist, especially if they feel pressure. This is called reactance: when attempts are made to restrict their freedom of choice, people sometimes do exactly the opposite.
Hence a refinement of the question. "Can you program a human?" is a question with the wrong frame. A more productive one is: can the probability of a particular behavior be changed predictably, and by what means? Here science has something to say. The systematization by Michie and colleagues identified 93 behavior change techniques — from goal setting and self-monitoring to social support and changing the environment (Michie et al., 2013, Annals of Behavioral Medicine, DOI 10.1007/s12160-013-9486-6). The very fact that such a taxonomy was needed shows that "behavior change" is not one button but a set of different tools with different areas of applicability.
In this article we go through the main groups of methods, each time asking: what exactly changes (an action, a habit, a skill, a belief, an emotional reaction), how reliable the data are, where the method stops working, and how to tell help from manipulation. The tone is curiosity without sensationalism. If you want another angle on similar questions, the guides on critical thinking and procrastination and motivation sit nearby.
What exactly are we trying to change
Much of the confusion in popular talk about "reprogramming" comes from lumping different outcomes together. It helps to separate at least seven things.
An action — a person did a specific thing: opened a notebook, unsubscribed from a mailing list, went for a run. This is the most observable and most easily measured outcome.
A habit — an action has become easier to trigger in a familiar context and takes less conscious effort. A habit is not "forever" and not "ironclad": it is a tendency that can weaken when the surroundings change.
A skill — a person has learned to do something better. A skill is built through practice with feedback, not by a single decision.
A belief — a person has changed their assessment of whether a statement is true. This is a separate process that depends on trust, identity, and the quality of arguments.
An emotional reaction — the frequency or intensity of a reaction has changed, for example anxiety in certain situations.
Motivation and intention — a person has wanted or decided to do something. That is not yet behavior.
An enduring personality trait — a much slower and more complex target, for which honest researchers do not promise quick results.
The most important distinction is between intention and action. Studies that measure only questionnaire answers ("do I intend to exercise?") easily show shifts that do not turn into real actions. A person sincerely wants to, sincerely plans to — and does nothing. This is called the intention–behavior gap, and a significant part of modern behavioral science is devoted precisely to it. That is why in this article we give priority to work that measures actions and long-term outcomes rather than only stated intentions.
The second distinction is between changing yourself and changing someone else. Most of the methods discussed below were studied under voluntary participation: people themselves wanted to change something, or at least agreed to take part. These results cannot be carried over without caveats to "how to influence a colleague, a partner, or a voter." The conditions are entirely different: there is no shared goal, no consent, no guarantee of trust.
How to evaluate the evidence
To avoid turning the article into a list of "science has proven," let's introduce a simple four-level scale. It does not replace reading the studies, but it helps keep an honest distance.
A strong evidence base. There are systematic reviews or meta-analyses of controlled studies; the effect has been replicated, though not necessarily in all conditions and for all groups.
A moderate or context-dependent base. Results are generally promising, but effects differ noticeably across tasks, groups, and settings. The method is worth trying without expecting universality.
A limited or contested base. Small studies, mixed results, weak replicability, or an unclear mechanism. You can experiment, but should not build conclusions on it.
A popular promise without reliable support. The claimed universality or strength of the effect is not backed by good-quality data. Here it is more accurate to write "not supported by reliable evidence" than "impossible": absence of evidence is not the same as proven absence of an effect.
Four caveats go with this scale. First: statistical significance is not the same as practical importance. An effect can be "significant" and still so small that it is unnoticeable in real life. Second: an average effect does not guarantee a result for a particular person. If a method helps on average, for some people it will not help at all. Third: correlation does not prove causation — people who keep a habit diary may differ from those who don't in some other way. Fourth: a meta-analysis is only as good as the studies included in it; publication bias (when failed experiments go unpublished) can inflate the overall estimates.
For orientation, let's put the main methods into an "evidence matrix." This is a simplification: the quality ratings are provisional and tied to the outcomes measured in the respective studies, not to any life goals whatsoever.
| Method | What it changes | State of the evidence | Main limitations |
|---|---|---|---|
| "If–then" plans | Probability of a specific action | Meta-analysis: a noticeable positive effect on average | Depends on the goal and conditions; does not replace motivation |
| Self-monitoring and recording progress | Goal attainment | Meta-analysis: a small-to-moderate effect on average | Risk of fixating on the metric; anxiety |
| Repetition in a stable context | Habit automaticity | One key real-world study plus related work | Huge spread in timelines; limited sample |
| Nudges / choice architecture | Choice in a specific situation | Meta-analysis with large heterogeneity; dispute over bias | Effect depends on the domain; transparency and ethics |
| CBT | Symptoms of depression and anxiety | Meta-analyses for a number of disorders | Only for specific conditions; study quality varies |
| Social norms | Healthy behavior | Meta-analysis: a small effect on average | Requires accurate information; possible "boomerang" |
| Psychological "inoculation" | Recognizing manipulation | Systematic reviews; effect on recognition | Durability and impact on real-world sharing are weaker |
Methods that change behavior and habits
This section is the most "engineering" one. Here, change happens mainly through conditions, consequences, and repetition, rather than through persuasion or inner dialogue.
Reinforcement and the consequences of actions
The basic principle has been known for a long time: behavior followed by desirable consequences is repeated more often; behavior followed by undesirable ones, less often. But the terms are worth untangling, because they are constantly confused. Positive reinforcement is adding something pleasant after an action (praise, a visible result). Negative reinforcement is removing something unpleasant after an action (you closed the tab with the worrying email — you felt better; this is often how avoidance gets locked in). Punishment is adding something unpleasant or taking away something pleasant so that behavior becomes less frequent. Negative reinforcement is not punishment, although in everyday speech it is often called that.
The timing of consequences matters in practice. Immediate feedback ("I did it, and it shows") is usually stronger than an abstract reward in six months. That is why many working systems of behavior change break a distant goal into near-term micro-confirmations: a mark on a calendar, a short report, a check of the result right after a session. For more on the link between near-term rewards and putting things off, see the guide on procrastination and motivation.
The limitations are also significant. The response to consequences depends on context, type of task, and individual differences. External rewards do not always support intrinsic motivation: several studies have shown that paying someone for an activity they already enjoy can weaken their interest in it, although the effect depends on how the reward is presented. Punishment often produces temporary suppression of behavior rather than learning something new, and frequently leads to avoidance of the source of the punishment itself. A safe application is working with your own consequences: tracking progress, setting up structured feedback, reinforcing a skill gradually.
Prompts, cues, and context
Much of what we do is triggered not by a decision but by a cue: a place, a time of day, an object on the desk, a previous action. Coffee — and a hand reaches for the phone. Closing the laptop — and a thought of a snack arises. A cue does not compel; it raises the probability of an action.
A practical idea follows: sometimes it is more effective to prepare the environment than to fight the urge. A book is placed next to the breakfast spot. Distracting apps are moved off the home screen. A short workout is tied to an existing routine — after brushing teeth, after the first coffee. This approach doesn't require heroic willpower every time, because the decision is made once, when the conditions are set up.
The limit here is the same as everywhere: a cue raises the probability but does not guarantee follow-through. If fatigue, stress, or a conflict with other goals outweighs it, the prompt will go unnoticed. So the environment is one lever, not a panacea.
Repetition in a stable context and habit formation
It is useful to tell a habit apart from a deliberate decision. A decision requires attention: weigh, choose, begin. A habit is an automated response to a familiar context: the cue almost triggers the action by itself, and less conscious effort is needed. The practical principle of formation looks simple: repeat a small action in similar circumstances until it becomes more automatic.
The best-known empirical study on this topic is the 2010 work by Lally and colleagues. Participants chose a new daily behavior (for example, drinking water with lunch or running after breakfast) and rated every day how automatic it had become. The main result is not the average number but the spread: by the authors' estimate, the median time to reach the plateau of automaticity was about 66 days, but across people and actions it varied from roughly 18 to 254 days. Simple actions became established faster than complex ones, and missing a single day did not noticeably derail the process by itself.
The limitations should be kept in mind: the sample was small and made up of volunteers, and automaticity was measured by self-report. The study does not support a universal timeline like "21 days"; that number apparently grew out of a retelling of the observations of a 1960s plastic surgeon, not out of controlled experiments. The speed of formation depends on the complexity of the action, the stability of the context, the regularity of repetition, stress, and living conditions. The honest conclusion: repeat until it gets easier, and don't count days on a calendar.
Implementation intentions: "if X, then Y"
One of the most studied techniques is to phrase a plan not as a goal ("I'll exercise more") but as a situation–action link: "If situation X occurs, then I will do Y." For example: "If I finish breakfast, then I will open a book and read one page." The key difference from a vague promise is that the plan chooses the cue and the response in advance. At the critical moment you don't have to make a decision — it is enough to notice the condition.
The meta-analysis by Gollwitzer and Sheeran (2006) pooled about 94 independent tests and found, on average, a medium-to-large positive effect of "if–then" plans on goal attainment (the authors report d ≈ 0.65). For drawing confident conclusions, it matters that what was measured was goal attainment itself, not just intention. Even so, the size of the effect depends on the task, the conditions, and the person; for example, a plan helps little if the goal itself doesn't matter to the person.
A useful detail: a plan works better when the cue is specific and noticeable ("after breakfast") rather than blurry ("when I have time"). And when the action is small enough to be realistically carried out on an ordinary day.
Self-monitoring and visible progress
Records, trackers, diaries, and checkpoints turn a vague feeling ("I seem to be exercising more often") into verifiable feedback. The meta-analysis by Harkin and colleagues (2016) gathered experimental studies in which people were asked to monitor their progress toward a goal, and concluded that monitoring on average promotes goal attainment — the effect was estimated as small-to-moderate (about 138 studies, d+ ≈ 0.40 by the authors' estimate). The effect proved stronger when progress was not only tracked but also physically recorded or shared with others (Harkin et al., 2016, Psychological Bulletin, DOI 10.1037/bul0000025; PubMed).
Why this might work: recording makes the gap between "where I wanted to be" and "where I am" visible and lets you adjust the plan before disappointment accumulates. But there are risks too. Excessive monitoring can become a source of anxiety. The metric sometimes replaces the goal: a person starts "playing for the number" (steps for the step counter, not for health). So the record should serve decisions — change the plan based on observations rather than accumulate numbers.
Building skills and learning
A skill differs from a habit: it is about the quality of execution, not about automatic triggering. The basic toolkit here has long been known: deliberate practice, feedback on specific errors, breaking the complex into components, gradually raising difficulty. It is better to train the weak link than to repeat for the tenth time what already works.
A separate topic is sleep and distributed practice: many studies point to their role in memory, but a detailed treatment requires separate fact-checking, and we will not reduce learning to "brain hacks." To continue the topic, see the materials on discipline in learning and self-regulated learning.
Finally, it is worth distinguishing the learning of a real skill from a short-lived improvement on a single test. A person can pass a check brilliantly right after a session and remember nothing a week later. So the real criterion of a skill is transfer to a new situation and retention over time.
Thoughts, emotions, and interpretations
If the previous section was about actions and conditions, this one is about how a person understands a situation. Interpretations affect emotions, emotions affect behavior, and all of this can, at least in part, be studied and changed. But an important warning up front: the methods in this section were developed mainly in a clinical context, and their results should be carried over to everyday life with caution.
Cognitive behavioral therapy
CBT is a structured approach in which a person learns to notice automatic thoughts ("I'll fail at everything"), check them against the facts, and set up behavioral experiments (do what you've been avoiding and see what actually happens). The idea is not to "replace negative thoughts with positive ones" but to learn to test interpretations and change the behavior that maintains the problem.
The meta-analysis by Cuijpers and colleagues (2016), which updated the data on depression and anxiety disorders, showed that CBT is on average more effective than control conditions (for example, a waiting list), with effects differing across disorders and comparison conditions. The authors also discuss publication bias and the low quality of some of the studies — meaning that confidence in the estimates should be limited.
Three caveats follow. CBT has evidence for specific clinical problems, and it cannot be claimed to be equally effective for all problems and all people. It is not a technique of self-suggestion: its essence is reality testing and behavioral experiments, not repeating encouraging phrases. And it is not a way to manage other people: the method works because the person participates themselves and has a stake in the result.
Exposure and safety learning
Exposure is gradual, safe, and structured work with situations a person avoids out of fear. The logic is this: avoidance brings short-term relief (and is therefore reinforced — recall negative reinforcement) but gives no chance to learn that the situation is less dangerous than it seemed. Controlled exposure creates exactly that chance — the person accumulates new experience: "I managed, and nothing terrible happened."
The difference between controlled exposure within a suitable protocol and "throwing someone into the fear" on a sink-or-swim basis is fundamental. The latter can entrench a traumatic experience rather than weaken the reaction. That is why, with pronounced anxiety, phobias, and trauma, self-directed experiments should not create a real threat, and clinical conditions require assessment by a professional. For everyday "small fears" (calling a stranger, speaking at a small meeting), the principle of gradualness holds: start with a feasible step and raise the difficulty in stages.
Self-efficacy and feedback
Self-efficacy is the conviction that you are able to carry out a specific action in specific circumstances. It affects whether a person takes something on, how long they persist through setbacks, and how they interpret difficulties. It is not general "self-confidence" but a situational expectation: you can be confident in cooking and not believe you'll cope with public speaking.
What supports it? First of all, real experience of success on manageable tasks, then watching people similar to you who succeeded, then realistic feedback. Empty affirmations are not on this list. It is important to distinguish "supportive inner dialogue" ("this is hard, but I can break the task into steps") from the promise that a phrase repeated in front of a mirror will automatically change your life. The former is consistent with research on self-efficacy; for the latter there is no reliable support at that scale, and some studies have shown that positive statements that do not match a person's self-image may even turn out to be of little use. Conviction does not replace skills and resources: the belief "I can" without a plan and practice remains a hope.
Motivational interviewing and autonomous motivation
Motivational interviewing is a style of conversation whose aim is to help a person sort out their own reasons and decision. It uses open questions, reflective listening, clarification of ambivalence ("part of me wants to quit smoking, part doesn't"), and respect for autonomy. The observation underlying the method is simple: when a person is persuaded and argued with, they often start defending the opposite position. When they are asked questions and their right to choose is accepted, they more often formulate the arguments for change themselves.
It is crucial not to present this method as a covert technique of coercion. Its ethical meaning is the opposite: to help a person clarify their own values and make their own choice. If "motivational interviewing" is used to quietly steer a person toward a decision that benefits you, that is already manipulation, even if in soft packaging. For more on how autonomy and motivation are connected in learning, see the discussion of self-regulated learning, and for the role of emotions in judgment, the piece on intuition, emotions, and learning.
Social mechanisms of influence
We are social beings, and this makes us both susceptible to influence and capable of cooperating. Social mechanisms often work in the background — which is why it is especially important to understand their limits.
Social norms and the example of others
People take their cues from what others do and approve of. Researchers distinguish a descriptive norm ("what most people do") from an injunctive one ("what most people approve of"). A message like "most of your neighbors save water" relies on the first, "most people think water should be saved" on the second.
A recent systematic review and meta-analysis by Papakonstantinou and colleagues (2025), devoted to social-norm messages in health-promoting behavior in developed countries, estimated the effect as small and variable on average (Nature Human Behaviour). In other words, norm messages sometimes help, but one should not expect miracles from them. The effect depends on the type of behavior, the audience, the plausibility of the message, and the context.
There is also a "boomerang" risk: if you report that an undesirable behavior is widespread ("many people skip classes"), this can normalize it and even raise its frequency. False or distorted information about what the majority does makes the problem worse. So honest work with norms begins with accurate data.
Authority, trust, and the source of information
The competence, trustworthiness, and social role of a source affect how a message is perceived. The same piece of advice from a doctor, a neighbor, and a stranger online will carry different weight, and this is generally reasonable: the "trust experts" heuristic saves effort. But it is also a convenient point of attack for manipulation: it is enough to put on the appearance of authority (a white coat, a title, a confident tone, an "independent study").
The key point: authority is not proof of truth. Experts make mistakes, step outside their competence, and can be tied by conflicts of interest. A practical block for the reader: look at primary sources rather than retellings; ask who benefits from this wording; look for independent confirmation; distinguish an expert's opinion in their field from statements on adjacent questions. These are exactly the skills covered in the guide to critical thinking.
Persuasion, framing, and how information is presented
The same information can be presented in different ways: "surgery — 90% survival" or "10% mortality." Formally these are identical, but emotionally they are perceived differently. This is framing: emphasis and wording affect the judgment. Framing effects have been described in many studies, but they are not universal: a person's knowledge, values, and context can greatly reduce them.
What matters here is the line between honest and manipulative. An honest explanation emphasizes significant consequences but does not hide facts and alternatives. Manipulative framing conceals unfavorable data, picks only convenient statistics, and plays on fear. The criterion is simple: if the person saw the whole picture and learned why you chose this presentation, would they feel deceived?
Choice architecture and nudges
The idea of choice architecture: the environment of decision-making can be organized so that the desirable option is easier, without forbidding alternatives. Classic tools include defaults (for example, automatic enrollment in a pension plan with the option to opt out), simplifying the procedure, reminders, changing the order of options, and reducing "friction" for a useful action.
The largest meta-analysis — Mertens and colleagues (2022, PNAS) — pooled hundreds of effects from dozens of studies and found a positive average effect of choice-architecture interventions (about d ≈ 0.43 by the authors' estimate), which varied strongly by technique and domain; the effect was most noticeable in the area of food.
An important caveat is needed here. Soon afterward, the same journal published a reply by Maier and colleagues (PNAS, 2022): after correcting for publication bias, they found no convincing evidence of an overall nudge effect. The estimates thus depend on the statistical method, and heterogeneity across techniques and domains remains high — the dispute cannot be considered closed. A reasonable conclusion: defaults and friction reduction work in a number of situations, but the average effect may be lower than the first estimates reported and depends strongly on context.
Ethical criteria for nudges were developed in parallel with the idea itself: transparency (a person can find out what was done and why), the ability to opt out without effort, no hidden exploitation of vulnerabilities, and proportionality to the goal. Recall dark patterns in interfaces: "decline" is hidden in small print while "accept" is highlighted — that is a nudge that violates all four criteria.
How beliefs form and change
Beliefs are more complex than actions: they are tied to identity, belonging, trust in sources, and accumulated experience. So "changing someone's mind" is usually harder than helping them take a specific step.
Repetition and familiarity
Information we have encountered many times seems more familiar, and the familiar often seems more plausible. This is called the illusory truth effect, and it has been reproduced in various studies. Familiarity is a quick signal, and in ordinary life it is often useful: what we meet often more often turns out to be true. But it is vulnerable: disinformation repeated often enough can start to be perceived as a familiar fact, even when we know it is a disputed claim.
A practical recommendation: check claims regardless of how easily they come to mind or how familiar they sound. The feeling "I've heard this somewhere" is not proof. The same mechanism explains why smooth, much-retold texts seem more convincing than they really are; for a similar trap in connection with AI texts, see the piece "AI as an amplifier of human influence".
Psychological "inoculation" against manipulation
The idea of inoculation (prebunking) borrows the logic of vaccination: show a weakened example of a manipulative tactic in advance and give people practice recognizing it, so that its force is reduced later. This can be a game, a short video, or an exercise explaining techniques — emotional escalation, false dichotomy, substitution of authority, use of scapegoats.
The systematic review and meta-analysis by Lu and colleagues (2023, Journal of Medical Internet Research) evaluated such approaches across three groups of outcomes — assessment of credibility, willingness to share, and the ability to tell unreliable from reliable information; in a number of conditions positive effects were obtained (PMC). A more recent meta-analysis from the standpoint of signal detection theory (Simchon and colleagues, 2026, Current Opinion in Psychology) examines whether it is specifically the discrimination between reliable and unreliable news that improves, rather than just general skepticism (DOI 10.1016/j.copsyc.2025.102194).
But caution is needed. The effects were measured mostly as recognition or evaluation of information, often right after training and in laboratory or online settings. Durability over time and the impact on real behavior on social media (for example, on what a person forwards) may be weaker. There is also the risk of general distrust: a person taught to suspect everything may start believing reliable information less too. So it is sensible to regard "inoculation" as one way of defense, not as armor.
Cognitive dissonance and identity defense
When information threatens an important part of the self-image — a professional role, political or religious affiliation, the sense of one's own competence — a person often defends the position rather than revising it. The theory of cognitive dissonance describes the unpleasant tension from incompatible beliefs and actions and the ways of reducing it: change the belief, change the behavior, or discount the new information. Often the third is easiest.
It follows that we should give up the popular thesis "just catch a person in a contradiction and they'll change." More often it will strengthen the defense: public refutation turns a question of truth into a question of face. Practically, something else is both more ethical and more effective: respectful dialogue, questions instead of accusations, the chance to revisit a position in private and without loss of status, acknowledgment that there is a reasonable kernel in the other person's position. Humiliation and coercion, as a rule, produce hidden resistance and a drop in trust — and trust is the main condition for changing beliefs.
Bodily state and environmental conditions
Everything said above does not work in a vacuum. The body and the environment set the background against which any method becomes easier or harder.
Sleep deprivation, chronic stress, fatigue, pain, substance use, and overload affect attention, impulse control, mood, and the quality of decisions. After a sleepless night a person resists distractions less well, reacts more sharply to irritants, and more often chooses the nearest reward. This does not mean that the state of the body "programs" behavior or explains it entirely. It means that the same "if–then" plan is carried out differently in a calm week and in a crunch, and that behavior that looks like "weak willpower" often turns out to be simply a shortage of resources.
A practical conclusion follows: when assessing why a change isn't succeeding, look not only at the method but at the conditions. Sometimes it is wiser first to reduce the load, sort out sleep and routine, and only then add new habits. In other cases exactly the opposite: a small stable habit helps restore the routine.
But this section must not be turned into a set of medical advice. If you notice persistent insomnia, a prolonged low mood, dependence, chronic pain, or other alarming symptoms, you need relevant sources and, if necessary, a professional. Self-directed "behavior change" does not replace diagnosis and treatment. The environment can be designed, but the body has its own limits, and they should be respected, not "hacked."
What sounds scientific but isn't supported the way it's promised
Alongside genuine methods lives a great number of attractive promises. It is important that they be examined by their specific wording and by systematic reviews, not by rejecting whole fields on the strength of a single example. If good-quality data are lacking, it is more honest to write "not supported by reliable evidence" than "impossible." The table below is a working checking tool, not a final verdict: for each topic it is worth opening the primary reviews.
| Claim | What is known | Correct conclusion |
|---|---|---|
| "A person can be completely reprogrammed in 21 days" | In the study by Lally and colleagues, the time to form automaticity varied widely (from a few weeks to many months); there is no universal timeline | The timeline depends on the action, context, and person; count not days but repetitions in stable conditions |
| "Affirmations are guaranteed to change your life" | Supportive inner dialogue and reliance on real experience are linked to self-efficacy; guaranteed effects from repeating phrases have not been reliably shown | Affirmations without concrete steps and feedback are weak support; realistic wording and an action plan are more useful |
| "There are universal words that make anyone agree" | Framing, social proof, and requests influence decisions, but effects depend on context, audience, and trust | A universal "magic phrase" is not supported; influence is probabilistic and limited |
| "NLP has proven that eye movements reveal thoughts" | Controlled tests of the popular scheme "eye movements ↔ type of thinking or lying" have generally not confirmed the claimed link | Not supported as a reliable way to "read minds"; better to doubt than to draw conclusions from a glance |
| "Any behavior can be changed with a single reward or punishment" | Consequences influence behavior, but the effect depends on context, timing, individual differences, and motivation; an external reward can weaken interest | Consequences are one lever, not a universal off switch |
| "Microexpressions let you read lies and intentions infallibly" | Detecting lies from nonverbal cues in studies is on average only slightly better than chance; the reliability of microexpressions as an indicator of deception is disputed | Infallible reading is not supported; rely on verifiable facts, not an impression of a face |
| "Knowing a personality type, you can accurately predict all of a person's decisions" | Personality traits moderately predict behavior on average, but only weakly predict individual decisions in a specific situation | A typology describes tendencies; it does not predict an individual choice |
Notice the pattern: most promises take a real but limited effect and inflate it to a universal and guaranteed one. Habits really do form — but not in 21 days. Inner dialogue really does have an influence — but not instead of action. Context really does affect decisions — but not through incantations. Any thesis containing the words "infallibly," "guaranteed," "anyone," and "all" is suspect.
A practical protocol: change your own behavior without mysticism
Below is a compact sequence assembled from the methods discussed above. It is not a medical program and not a guarantee of results, but a way to organize an experiment on yourself.
- Define the behavior. Describe the action so that an observer could see it: "read for 10 minutes," not "become well-read."
- Find the context. When, where, and after what event does this behavior usually happen, or could it happen?
- Reduce friction. Make the desired action easier (a book within reach) and the unwanted one less automatic (the app removed from the home screen).
- Write an "if–then" plan. Tie a short action to a specific cue.
- Start small. Choose an amount you can realistically repeat even on a bad day.
- Record progress. Write down actual behavior, not just mood and intentions.
- Get feedback. Once a week, look at the records and change the plan based on your observations.
- Test durability. Assess whether the change holds when reminders are weakened and outside control is removed.
- Reconsider the conditions rather than blaming yourself. If the plan doesn't work, find out what gets in the way — time, fatigue, a poor cue, too large an amount — and rebuild the system.
Example: reading 10 minutes a day
Suppose you want to read books but usually "don't have time" in the evening.
Context. In the morning you have breakfast at the kitchen table; in the evening you're usually tired. Morning breakfast is a stable event, a suitable candidate for a cue.
Cue and plan. "If I've finished my coffee, then I will open the book at the bookmark and read at least one page." In the evening you put the book next to the mug and leave the phone in another room.
Minimum action. One page is the "entry ticket"; if you read longer, good, but the commitment is one page. Such a threshold reduces perfectionism along the lines of "I should read only when I have an hour."
Recording the result. On the bookmark or in a calendar, you mark any day when a page was read. Record the fact, not the quality of understanding.
Adjustment. After two weeks, look at the marks. Are there gaps? Look at what was happening on those days: sleeping in, a business trip, fatigue. Perhaps the cue needs to be moved (for example, to the commute or the lunch break) or a different minimum chosen. Don't write "I have weak willpower"; write "on days without breakfast at the table the cue doesn't fire — I'll replace it."
Durability check. After a month, remove the reminders and see whether you still read. If so, the habit is taking hold. If not, go back to steps 3 and 4. A date by which the habit "will definitely form" cannot be promised: it is individual.
If putting things off stops you at the very start, the detailed guide to procrastination will come in handy.
The line between influence, help, and manipulation
Influencing people's behavior is normal: teachers, doctors, parents, friends, and interface designers all do it. The question is not whether we influence, but how and for what purpose. To tell help from manipulation, a few questions are useful — a kind of checklist:
- Transparency. Does the person understand that someone is trying to influence them, and in which direction?
- Freedom to decline. Can they decline without disproportionate punishment, pressure, or loss of important goods?
- Alignment with interests. Does the intervention match the person's own interests and goals?
- Honesty of information. Are significant facts and alternatives being concealed?
- Vulnerabilities. Does the method exploit fear, dependence, fatigue, cognitive impairment, or a vulnerable position?
- Data and benefit. Who controls the data about the person, and who benefits from the result?
If the answers to most of the questions are alarming, you are most likely looking at manipulation — however attractive its packaging. If the person is informed, free to choose, and the outcome serves their goals, it is closer to help or teaching.
Knowledge of the mechanisms of influence is thus useful both ways. It helps you arrange your own changes, but it also lets you notice manipulation in advertising ("only two rooms left!"), in interfaces (a hidden decline button), on social media (feeds optimized for emotion), in political communication (false dilemmas, appeals to fear), and in personal relationships (pressure through guilt, "gaslighting"). How to respond to manipulation in an AI world, where a persuasive form has become cheap, is a separate topic, covered in the article "AI as an amplifier of human influence". If you want to practice dissecting other people's claims, try ThinkLens.
And one more rule, which we set as an editorial one: this article does not teach how to control someone else's mind. It describes the mechanisms, their limits, and ways to use the knowledge for your own changes and for protection against pressure.
FAQ
Can you program a human like a computer?
No. A human has no single executable code, and behavior depends on biology, experience, goals, surroundings, and chance. Scientific methods raise the probability of certain actions under certain conditions and work mostly with average effects in groups. They do not give guaranteed control over a specific person, and they work best when the person is themselves invested in the change.
How many days does it take to form a habit?
There is no universal timeline. In the study by Lally and colleagues, the median time to reach the plateau of automaticity was about 66 days, but the spread — from roughly a few weeks to many months — depended on the person and the complexity of the action. The "21 days" myth is not supported by research. It is more practical to go by how much easier the action has become to start in your usual context rather than by days.
Do affirmations work?
Guaranteed changes from repeated phrases alone have not been reliably shown. Supportive and realistic inner dialogue can help, especially combined with a specific plan, feasible steps, and feedback. Phrases that sharply diverge from how a person sees themselves may not work at all. Self-efficacy grows first of all from real experience of coping with tasks.
How does a nudge differ from manipulation?
A nudge is a change to the choice environment that makes a desirable action easier without forbidding alternatives. It becomes ethical with transparency, an easy way to opt out, no hidden exploitation of vulnerabilities, and proportionality to the goal. Manipulation conceals intent, makes declining harder, plays on fears, and benefits primarily the one doing the influencing. The same technical solution (for example, a default) can be either — context and honesty decide.
Does CBT help "reprogram" thinking?
CBT is a structured method with evidence for a number of conditions, for example depression and anxiety disorders (Cuijpers and colleagues, 2016). But describing it as "reprogramming" is inaccurate: it teaches you to notice automatic thoughts, test them, and try different behavior. The result depends on the task, the person's participation, and the quality of therapy. It is a clinical method: if you suspect depression or an anxiety disorder, it is better to see a professional than to try applying the technique on your own as a "hack."
How can you protect yourself from manipulation?
A few habits that work: slow down under time pressure ("today only"); ask who benefits from this message; check the primary source rather than the retelling; separate the claim from the tone and presentation; pay attention to emotional escalation. Training in recognizing tactics ("inoculation") improves, according to a number of studies, the discrimination between reliable and unreliable information, but the effect is not absolute and may fade over time. So it is more useful to combine it with regular practice in checking claims — see critical thinking.
Can you change another person?
Influence — yes; change on their behalf — no. Change in other people usually arises when there is trust, respect for autonomy, understandable reasons, and room for choice. Pressure, blame, and hidden tactics often provoke resistance. If you want to help, start with questions and a joint search for reasons, as in motivational interviewing, rather than with an attempt to "tune" the person.
Further reading
If you are interested in a systematic approach to self-change, start with the guide to critical thinking: it teaches you to distinguish a claim, evidence, and interpretation — that is, the basis for checking any promises about "programming." The most common barrier at the start — putting things off — is covered in detail in the guide to procrastination and motivation. How to turn intention into a durable practice is examined in the piece on discipline in learning, and the "plan — act — evaluate" cycle is described in detail in the guide to self-regulated learning.
To better understand how our judgments about our own abilities can be misleading, look at metacognition and learning. The role of emotions and intuition in decisions is covered in the piece on intuition, emotions, and learning. On how a cheap, persuasive form changes influence on people itself, see the article "AI as an amplifier of human influence". And you can practice dissecting other people's texts and persuasion tactics in ThinkLens.
Of this article's primary sources, the ones worth opening first are: the taxonomy of behavior change techniques (Michie and colleagues, 2013), the study of habit formation (Lally and colleagues, 2010), the meta-analyses of "if–then" plans (Gollwitzer, Sheeran, 2006) and progress monitoring (Harkin and colleagues, 2016), the nudge meta-analysis (Mertens and colleagues, 2022), the update of the data on CBT (Cuijpers and colleagues, 2016), the review of social-norm messages (Papakonstantinou and colleagues, 2025), and the reviews on psychological inoculation (Lu and colleagues, 2023; Simchon and colleagues, 2026). The bibliographic details of the key publications were checked against PubMed on October 13, 2026; before relying on specific figures, open the full texts and look at the studies' limitations.
Conclusion: not programming, but engineering conditions
None of the methods reviewed provides absolute control over a person, and this is not a shortcoming of science but a property of the subject. People are not systems that can be flashed with new firmware: they observe themselves, resist pressure, change their goals, and live in an environment that constantly reshuffles the probabilities of their actions.
What works in practice is less spectacular and more reliable. A clear goal described as an observable behavior. A suitable context and a noticeable cue. A specific "if–then" plan. A small repeatable action. Honest observation of real behavior and a willingness to adjust the system rather than blame yourself. For changing beliefs, arguments are not enough: you need trust, quality of information, a suitable context, and the person's readiness to revisit their views. And for protection against manipulation — the habit of asking about the source, the motive, and the freedom to decline.
The article's strongest conclusion is simple: behavior can be made more likely, learnable, and durable, but not guaranteed to be controllable. "Engineering conditions" is more accurate and more honest than the "programming" metaphor: we are not writing commands into someone else's mind, but designing a setting in which the desired action becomes easier, more noticeable, and more meaningful — above all for ourselves. Start with one observable action and one cue for the coming week — it is a small experiment that will tell you more about your own case than any universal formula.