The Behavior Control System

The people who come to Phalanx Labs are, for the most part, unusually capable. They run companies, lead teams, trade, operate, build and manage under pressure, and they do it well. Then, in a small number of predictable moments, usually private and usually late in the day, they lose to the same habit again. The habit varies: alcohol, cannabis, nicotine, stimulants, pornography, gambling, compulsive phone use, food. The shape of the problem does not. A person who can hold a difficult line in a negotiation cannot hold it against a drink at the end of a hard week, and cannot explain the difference to themselves.

Nearly all of them have tried to stop. Of the 290+ clients we have worked with across 12 countries over 7 years, 98.4% had already tried and failed to stop before they came to us. Those attempts were sincere. They included cold turkey, apps and blockers, books, moderation rules, gym streaks, retreats and, for some, conventional treatment. The failure rate is consistent with the published record. In a systematic review of people who tried to stop smoking without treatment, most relapse occurred within the first eight days, and only 3 to 5% of such attempts produced prolonged abstinence at six to twelve months¹. Across alcohol and tobacco treatment studies, the curve that describes relapse has a common shape: losses are steepest soon after stopping and slow as abstinent time accumulates².

The dominant explanation for repeated failure, inside and outside the person, is a shortfall of willpower. We think that explanation is wrong in a specific and useful way. Willpower, in the sense of deliberate prefrontal control over action, is real, but it is the system most easily taken offline by exactly the conditions in which these habits occur: stress, fatigue, poor sleep and the presence of a learned trigger^{3, 4}. A person who relies on it is relying on the one resource the situation is most likely to remove. At the same moment, several other systems are working in the habit's favor, and none of them responds to resolve.

Modern addiction neuroscience describes those systems in detail. Repeated use changes how the brain's reward circuitry learns, how it values the habit relative to everything else, how it responds to stress and how strongly the settings of past use pull behavior back toward the habit⁵. Behavioral science adds a second account that matters just as much for people whose habit has never been severe enough to call an addiction: any behavior repeated in a stable context becomes a habit in the technical sense, triggered by the context with little reference to what the person currently wants⁶. The two accounts reach the same conclusion. A habit like this is maintained by a set of interacting systems, and an attempt that addresses only one of them leaves the others to restore it.

The Behavior Control System was built from the opposite starting point. It treats the habit as the output of conditions and changes the conditions in a deliberate order. It starts with the physiological ground the habit runs on, because a dysregulated stress system and a disordered sleep pattern make every later skill harder to use. It then works on the triggers and the moment of temptation, because that is where lapses actually happen. It ends with identity and social context, because a person who has stopped but still sees the habit as part of who they are, and still lives among people and settings organized around it, is exposed to the slow pull back that most relapse research describes.

The rest of this paper sets out the model behind the method (Section 2), how the loop beneath one person's habit is identified before the program starts (Section 3), and the four stages of the program itself: commitment (Section 4), restoring the substrate (Section 5), triggers, cravings and permission thoughts (Section 6), and identity and social context (Section 7). We then describe aftercare and relapse prevention (Section 8), how outcomes are tracked and what our figures do and do not show (Section 9), and the limits of the method, including when someone needs a doctor first (Section 10). The evidence for each choice is placed beside the choice. Where the program is described day by day, we say what that day targets and why. We do not describe how the exercise is done.

We use the word habit for the behavior our clients want to end, whether or not it would meet a clinical threshold for addiction. The program spans that range. It takes in the person who drinks too much three nights a week and the person who cannot go a day without cannabis, and the systems described below operate across it with different intensity. The model has seven parts. None of them is new; each is well established in its own literature. The contribution of the method is to treat them as one interacting system and to address them in an order that makes each later step easier.

2.1Dopamine, prediction error and incentive salience

Dopamine neurons in the midbrain do not simply signal pleasure. In the work that defined the modern view, their output tracked the difference between the reward an animal received and the reward it predicted⁷. A reward that was better than expected produced a burst; one that was fully predicted produced little response at the moment of delivery, and the response moved instead to the earliest reliable signal that the reward was coming. This is a learning signal. It teaches the brain which situations lead to reward and assigns them value. A habit's triggers acquire their power through this process: the setting, the time of day and the feeling that reliably preceded the habit become the moment the reward system responds.

A second distinction explains why people continue a habit they no longer enjoy. Berridge and Robinson separated the brain systems for wanting a reward from the systems for liking it⁸. Wanting, or incentive salience, is generated by large and robust circuits that include mesolimbic dopamine. Liking, the pleasure of consumption, depends on smaller and more fragile systems and does not depend on dopamine. Their incentive-sensitization theory holds that repeated drug use can amplify wanting, especially wanting triggered by associated stimuli, without amplifying liking. In practice this produces the experience our clients describe most often: a pull that arrives with full force when a trigger appears and a payoff that is smaller, shorter and less satisfying than the pull promised.

Repetition also changes the system's baseline. Brain imaging studies summarized by Volkow, Koob and McLellan show that people with addiction tend to have lower dopamine D2 receptor availability and blunted dopamine release to ordinary rewards, while drug-associated stimuli continue to produce strong responses⁵. The person is left with a reward system that responds less to the ordinary rewards of life and more to the habit and its triggers. Recovery of these markers with sustained abstinence has been documented, though it is slow and incomplete. In one small imaging study, dopamine transporter levels in the striatum of people who had used methamphetamine rose substantially between short abstinence of under six months and protracted abstinence of twelve to seventeen months⁹.

2.2Glutamate and the learning of triggers

If dopamine marks what matters, glutamate carries much of what is learned about it. Glutamate is the brain's main excitatory transmitter and the substrate of most synaptic learning, including the connections between prefrontal cortex and nucleus accumbens that let goals regulate reward seeking. Kalivas proposed that chronic drug use produces a lasting imbalance between synaptic and non-synaptic glutamate, which he called a disruption of glutamate homeostasis, and that this imbalance impairs communication between prefrontal cortex and nucleus accumbens¹⁰. The result is a system in which drug-associated stimuli drive seeking strongly while prefrontal regulation of that seeking is weakened.

The behavioral expression of this learning is trigger reactivity. In a meta-analysis of studies using the cue-reactivity paradigm, exposure to drug-related stimuli reliably increased self-reported craving in people with addiction, with smaller effects on physiological arousal¹¹. The learning is also bound to context. In animal experiments, a dose of heroin given in a setting the animals had not used it in was more often fatal than the same dose in the familiar setting, because the familiar setting had come to trigger the body's anticipatory, protective responses¹². Bouton's review of extinction explains why this matters for anyone trying to stop: stopping does not erase the original learning. It lays new learning over it, and the new learning depends more on context than the old. When the person returns to the setting, time, mood or company in which the habit was learned, the original response can return¹³.

Two practical conclusions follow, and the program is organized around both. First, the triggers a person has learned are specific to them and can be listed. They fall into a few broad groups that recur across the trigger-reactivity literature: internal states, situations and times, people and places, and objects. Second, abstinence alone does not remove them. Triggers that can be removed should be removed, and the ones that cannot be removed need a new, rehearsed response attached to them, practiced in the contexts where the old response was learned.

2.3Habit and goal-directed control

Behavior is controlled by two systems that normally work together. Goal-directed control chooses actions by their expected outcomes and adjusts quickly when an outcome is no longer wanted. Habitual control runs well-practiced responses directly from context, with little reference to current goals⁶. Everitt and Robbins describe addiction as a progressive shift from the first to the second, from prefrontal to striatal control and from ventral to dorsal striatum, with drug-associated stimuli gaining direct influence over seeking and relapse and with top-down prefrontal inhibition weakening along the way¹⁴.

Stress moves the balance toward habit. In a laboratory study, people stressed before learning a task kept performing an action for a food they had just eaten to satiety, while unstressed people adjusted; stress made their behavior insensitive to the outcome's current value⁴. This is the experience of a client who decides at nine in the morning that tonight will be different and finds at nine at night that the decision has no grip. The morning decision was made by goal-directed control. The evening behavior is being run by a context-triggered habit, under accumulated stress, and the system that made the decision is not the one in charge.

2.4Prefrontal control under stress and fatigue

The prefrontal cortex supports what psychologists call executive functions: inhibiting an impulse, holding a goal in mind and switching flexibly between options¹⁵. It is also the brain region most sensitive to the effects of stress. Even mild acute stress that feels uncontrollable can cause a rapid loss of prefrontal cognitive abilities, and prolonged stress produces structural changes in prefrontal dendrites³. Diamond's review lists stress, lack of sleep, loneliness and lack of exercise as conditions that each impair executive function, and notes that executive functions can be improved with practice¹⁵.

The implication is uncomfortable for anyone relying on resolve. The faculty a person uses to refuse the habit is the one most degraded at the hours the habit tends to occur: after a demanding day, when tired, when stressed and often after poor sleep. People are also poor at predicting this in advance. In a calm state, people consistently underestimate how much a future craving or arousal will change what they want and do, a pattern Loewenstein called the hot-cold empathy gap¹⁶. A plan made in the morning is written for a person who will not exist at the moment it is needed, unless the plan accounts for that person's state.

2.5Allostasis and the stress system

Koob and Le Moal proposed that addiction is best understood through allostasis: the brain maintains apparent stability in reward function by shifting its set point, and the counteradaptive processes that normally limit the effects of a reward fail to return to their usual range¹⁷. The person ends up in a chronic deviation from normal reward function, an allostatic state, which they argue explains the persistent vulnerability to relapse long after use has stopped. In later work Koob described how brain stress systems, including corticotropin-releasing factor in the extended amygdala, are recruited during the development of addiction and come to drive use through negative reinforcement: the habit is repeated to relieve a negative state that the habit itself has deepened¹⁸.

Stress is also among the most consistent predictors of relapse in prospective research. In a study of people leaving inpatient treatment for cocaine dependence, greater craving in response to a personal stress script was associated with a shorter time to relapse over the following 90 days, while craving in response to drug-related stimuli was not; stress-induced hormonal responses predicted how much was later used¹⁹. For our clients the stress route is usually the main one. The habit arrived as relief from pressure, became the default way of ending a day, and now fires whenever pressure rises.

2.6Sleep and the circadian system

Sleep and substance use are linked in both directions. Substances disturb sleep, and poor sleep is a risk factor for relapse; the circadian system that times sleep also regulates reward processing, which suggests a link to substance use that runs partly independently of sleep itself²⁰. Alcohol is the most common example among our clients. At all doses studied it shortens the time taken to fall asleep and disrupts the second half of the night, and it delays the first period of rapid eye movement sleep, with total REM sleep reduced at moderate and high doses²¹. The person experiences the drink as helping them sleep and wakes less restored.

Sleep loss acts directly on the systems described above. One night of sleep deprivation reduced the availability of dopamine D2 and D3 receptors in the human ventral striatum, a change the authors interpreted as receptor downregulation²². In another study, a night without sleep amplified the amygdala's response to negative images and weakened its functional connection with the medial prefrontal cortex, which normally regulates it²³. A short or irregular night therefore produces, the next day, a reward system less responsive to ordinary reward, an emotional system more reactive to stress, and prefrontal regulation less able to hold either. The habit offers fast relief from all three.

The body clock is set mainly by light. Light acts on the master clock in the suprachiasmatic nucleus and on other regions to regulate circadian rhythms, sleep and mood²⁴. Bright light late in the evening delays the clock and daylight early in the day anchors it. Because most of our clients work indoors, keep late hours under artificial light and use screens in bed, their circadian timing is often weakly anchored before the habit is considered at all. This is the first system the program addresses.

2.7Interoception: how the urge is felt

A craving is felt in the body: restlessness, tension, a pull in the chest or stomach. Interoception, the sense of the body's internal state, is represented in the insula, and the anterior insula in particular has been proposed as a basis for moment-to-moment awareness of how one feels²⁵. Its role in addiction was shown unusually clearly in a study of smokers who had suffered brain damage. Those whose damage involved the insula were more likely to stop smoking easily and immediately, without relapse and without a persisting urge to smoke, than those whose damage spared it²⁶. The finding suggests that the conscious urge depends heavily on the brain's representation of bodily state.

Kavanagh, Andrade and May's elaborated intrusion theory describes how such a feeling becomes a craving episode²⁷. An intrusive thought about the target is triggered automatically by external or bodily stimuli. If it promises pleasure or relief, the mind elaborates it, especially as sensory images, and the elaboration occupies working memory and heightens awareness of the deficit the habit would relieve. The theory gives the program two targets: the physical state that feeds intrusion, and the elaboration that turns an intrusion into an episode. Both can be trained. The person can learn to read the bodily state as information rather than an instruction, and to interrupt elaboration early.

2.8The loop as one system

Volkow, Koob and McLellan organize the neurobiology of addiction into three repeating stages: binge and intoxication, in which the basal ganglia and dopamine learning dominate; withdrawal and negative affect, in which the extended amygdala and stress systems dominate; and preoccupation and anticipation, in which prefrontal control fails to hold back craving⁵. The program teaches clients a closely related five-stage version of the same cycle in their own terms: an emotional state, a craving, a ritual sequence, the act itself and the guilt that follows. The guilt matters as much as the act: it is a negative state that can itself become the trigger for the next round.

Two people with the same habit rarely have the same loop. One drinks to end a working day that never otherwise ends; another drinks only in company and only when a particular group is present; a third drinks alone at night after the household is asleep, and the drink is the only part of the day that belongs to them. The neuroscience in Section 2 applies to all three, but the entry points differ, and an intervention aimed at the wrong entry point fails even when it is carried out well. This is the most common pattern in the histories our clients bring. Their earlier attempts were reasonable responses to the wrong loop.

Every program at Phalanx Labs therefore begins with a Root-Cause Diagnostic: an application describing the habits involved and what has already been tried, followed by a thirty-minute, one-on-one conversation with our team. Its purpose is narrow. It identifies the loop beneath the habit and the reasons earlier attempts broke down, and it decides whether we can help. If we cannot, we say so.

The structure of the diagnostic parallels a framework proposed in addiction research for exactly this problem. The Addictions Neuroclinical Assessment argues that people who meet the same diagnostic criteria differ in cause, course and response to treatment, and proposes assessing three functional domains drawn from the addiction cycle: incentive salience, negative emotionality and executive function²⁸. Our questions are practical rather than neuroclinical, and we take no measurements of the brain. But each maps onto one or more of those domains, and together they produce a working account of the person's loop.

1. What the habit is doing for the person: the relief, reward or escape it provides, and the state it most often ends.
2. The earliest recurring trigger: the first point in the sequence where the outcome starts to become likely.
3. The conditions that make the habit likely: times, places, people, states of fatigue or stress, and other habits that travel with it.
4. How earlier attempts were structured, and which part of the loop each one left in place.
5. What is missing for change to hold: the routines, prepared responses, sleep, structure or support that are absent now.

The emphasis on earlier attempts is deliberate. Given the share of our clients who have tried and failed before, the history of those attempts is the richest data the diagnostic has. Real-time monitoring studies of people trying to stop smoking show why. First lapses, compared with ordinary moments and with temptations the person resisted, were more likely to occur during negative mood, when cigarettes were easy to get, where smoking was permitted and when other smokers were present; people were far more likely to report using a coping strategy during temptations they resisted than during lapses²⁹. A lapse, in other words, usually happens where the trigger, the state and access coincide and no response is ready. Reconstructing the last few lapses in detail shows where those coincidences occur for this person.

The diagnostic also looks for habit stacks. Many clients have a core habit and one or more that travel with it: drinking and smoking, cannabis and late-night food, stimulants and pornography. Each use of one reactivates the reward learning shared with the other, and lapses on the secondary habit often begin with the core one. The program later asks the client to identify their core habit explicitly (Day 15), but the diagnostic is where the pattern is first named.

3.1From diagnosis to program

Every client follows the same program, in the same order. What the diagnosis changes is emphasis: which triggers are mapped first, which stress responses matter most, where the daily structure needs to be strongest and which permission thoughts are likely to appear. The program's structure is summarized in Figure 8 and Table 2 and described phase by phase in Sections 4 to 8. It asks for about thirty minutes of focused work a day: a short lesson each morning, a brief log each night, and a set of daily practices, which the program calls north stars, built into the routine the person already has.

Most attempts to stop begin on impulse, the morning after a bad night, when motivation is high and nothing else has changed. Phase 0 exists to separate the decision from the start. It comes before Day 1, does not count toward the 90 days, and has one purpose: to make sure the person enters the program with a commitment that is specific, witnessed and already tested in a small way. It also includes an onboarding call with our team, where the diagnostic findings are turned into the emphasis the program will carry for this person.

Specificity matters because the competition between the habit and the future is uneven. People discount delayed rewards steeply, and the discounting is steeper in addictive behavior: a meta-analysis found reliably greater delay discounting in people with addictive behaviors than in controls⁵¹. A future the person can picture in detail is one they discount less, which is part of why the program returns to it in Phase 2 (Section 6.5). In Phase 0 the purpose is simpler. The client has to know, in their own words, what they are working toward.

Commitments that are made public or recorded hold better than private intentions. In a meta-analysis of 138 studies, interventions that increased monitoring of goal progress improved goal attainment, and the effect was larger when progress was physically recorded or made public³². A commitment made to a specific person also changes what a lapse would cost, because it is no longer a private event.

This single day draws on Bandura's account of self-efficacy, which identified mastery experiences, the direct experience of succeeding at the behavior in question, as the most dependable source of a person's belief that they can do it³¹. People who come to us after many failed attempts tend to have little of that belief left. One deliberate day does not prove much about day forty, but it moves the question from whether stopping is possible at all to how to keep doing it.

The choice of start date rests on research into what Dai, Milkman and Riis called the fresh start effect. Temporal landmarks, such as the start of a week, a month, a year or a meaningful personal date, were followed by increases in aspirational behavior, including searches about dieting, gym attendance and the creation of commitment contracts³⁰. The authors suggest that landmarks create a psychological separation from a past self and its failures. A start date chosen for its meaning uses that separation on purpose.

Phase 1 rests on a simple argument from Section 2. Prefrontal control, the faculty people try to use against the habit, is degraded by stress, sleep loss, circadian disruption and the absence of other rewards^{3, 15}. If those conditions are left in place, every later skill is learned and used by a brain working at a disadvantage. The first month therefore works on the substrate: the physiological and behavioral ground the habit has been running on. The client stops on Day 15, after two weeks in which their sleep, light exposure, training and daily structure have been rebuilt.

The order is deliberate, and it differs from the conventional order in an important way. Most attempts start with abstinence and hope that the rest of life improves. The program starts with the rest of life, so that by the time abstinence begins, the stress system is quieter, sleep is more regular and the hours the habit used to fill already have something else in them. The relapse curve in Figure 1 shows why that matters. The first days after stopping carry the most risk, so the support for those days has to be in place before they arrive.

5.1Days 1 and 2: the model and the log

The log begins on Day 1 because monitoring is itself an intervention. In Harkin and colleagues' meta-analysis, prompting people to monitor their progress improved goal attainment, with a stronger effect when the monitoring was recorded³². The log is also the program's main source of evidence. It turns the diagnostic's hypothesis about the person's loop into data, and it is what the weekly review, the phase reviews and later the trigger map are built from. Teaching the model on the same days serves a related purpose. A person who understands why a craving happens is more able to watch it happen, and watching is the first skill Phase 2 depends on.

5.2Days 3 to 5: the body clock

The body clock comes first because it governs so much of what follows. Light is the main signal that sets the master clock in the suprachiasmatic nucleus, and through that clock and more direct pathways it influences sleep, alertness and mood²⁴. The circadian system also regulates reward processing directly, which is one reason disrupted rhythms and substance use are so often found together²⁰. A fixed wake time anchors the clock from one end; morning daylight strengthens that anchor; reducing light and screens in the evening stops the clock from being pushed later at the other end.

The effects on the habit loop are direct. One night of sleep deprivation reduced dopamine D2 and D3 receptor availability in the ventral striatum, which the authors interpreted as downregulation²², and sleep loss amplified the amygdala's response to negative stimuli while weakening the prefrontal connection that regulates it²³. For clients whose habit is alcohol, the relationship runs both ways: the evening drink disrupts the second half of the night and suppresses REM sleep at moderate and higher doses²¹, producing the poor sleep that makes the next evening harder. Restoring sleep therefore removes one of the conditions that keeps the loop running.

5.3Days 6 to 9: training, meditation, nutrition and structure

Exercise is the most broadly supported of the lifestyle interventions in the addiction literature. Lynch and colleagues' review concluded that exercise facilitates dopaminergic transmission and may normalize the glutamatergic and dopaminergic signaling altered by drug use, with effects that depend on the stage of addiction and on the type and timing of the exercise³³. It also supports executive function¹⁵ and offers a reliable, effortful reward that arrives on a schedule the person controls. The program makes training progressive for a second reason. Visible, measurable progress in one area of life is a mastery experience in Bandura's sense, and the self-efficacy it builds carries into the attempt to stop³¹.

Meditation enters early because it trains the two capacities the rest of the program depends on: attention to internal states, and the ability to observe an urge without acting on it. In a study by Tang, Tang and Posner, smokers recruited for a stress-reduction program who received two weeks of meditation training, five hours in total, reduced their smoking by 60%, while those given relaxation training showed no reduction; the meditation group also showed increased resting activity in the anterior cingulate and prefrontal cortex³⁴. The participants had not been recruited to stop smoking. The change came from training self-control, not from a goal to quit.

The anchoring routine applies the habit research directly. Wood and Rünger describe habits as responses triggered by stable contexts, and note that much of daily behavior runs in this mode⁶. The hours in which a habit runs are usually unstructured: the gap after work, the late evening, the weekend afternoon. A day with fixed anchors does two things. It fills those gaps with something already decided, and it gives the new practices a stable context in which they can themselves become habitual. Lally and colleagues found that the automaticity of a new daily behavior built gradually over weeks, with a median of 66 days to reach 95% of its plateau and wide variation between people, and that missing a single occasion did not materially affect the process⁴⁹. A routine started on Day 9 is therefore expected to become automatic around the end of the program, which is the point at which the program hands it over to the client.

5.4Days 10 to 14: the case for stopping completely

The case for complete abstinence follows from the extinction literature. Extinction lays new learning over the old rather than erasing it, and the old learning returns readily when the original context or the reward itself reappears¹³. Each permitted use is a reminder of the original association, and every reminder is an occasion for the old response to return. Moderation also requires the person to make a judgment about quantity at exactly the moment their judgment is least reliable.

The reward inventory comes from Bickel and colleagues' account of addiction as a reinforcer pathology: an excessive valuation of one commodity, combined with a short time horizon, at the expense of other rewards³⁵. One route out is to build alternative sources of reward that compete with the habit. The inventory makes the competition visible. Most clients find that their target habit sits among several fast, high-intensity rewards, often including the phone, that together keep the reward system tuned to effortless stimulation and away from slower, earned rewards. Day 12 names them. Day 15 removes them.

5.5Day 15: the client stops

Three features of Day 15 are deliberate. The first is timing. Two weeks of circadian, physical and structural work are designed to reduce the stress and sleep loss that drive craving, so the client starts abstinence on firmer footing than in their earlier attempts. The second is the distinction between a slip and a relapse. Research on the abstinence violation effect, discussed further in Section 8, showed that the way a person explains a lapse matters: people who attributed a lapse to internal, stable and global causes were more likely to go on to full relapse⁵⁰. The program defines a single slip in advance as an event to record and analyze, so that it is not read as proof that the attempt has failed.

The third is the instruction to keep attention on the daily practices rather than on resisting the habit. Wegner's research on thought suppression found that trying not to think about something can produce a rebound in which the thought returns more often, and that giving people a specific alternative to think about reduced the later preoccupation⁴². So the program gives the client a full day of specific alternatives to attend to, instead of a thought to fight.

5.6Days 16 to 23: building other rewards and early control

Brief voluntary stressors are included because they produce a strong physiological response the person chooses and controls, the opposite of the uncontrollable stress that most damages prefrontal function³. Cold water in particular produces a large catecholamine response. In one study of head-out immersion in cold water, plasma noradrenaline and dopamine rose substantially while cortisol did not⁵². The program's exposures are far shorter than that study's, and we do not claim they produce the same response. What we take from it is that cold produces a sharp rise in arousal and alertness without the cortisol response of a stressful event, and that for many clients it offers a strong physical reset at the times of day the habit used to fill.

Day 17 is, in our experience, one of the most consequential days in the program. Animal research on environmental enrichment, in which animals are given novelty, exercise, social contact and challenge, found that enrichment can both prevent the development of drug addiction and reduce relapse-like behavior, and the authors describe enrichment as acting as a functional opposite of stress³⁶. In humans, the equivalent is a pursuit that is effortful, absorbing and rewarding in proportion to the work put into it. It is the plainest example of the alternative reinforcers Bickel and colleagues describe³⁵. It also fills time, which matters as much as anything else in the early weeks of abstinence.

Awareness training is introduced in Phase 1 because every response the client learns in Phase 2 depends on noticing the trigger in time. Quinn and colleagues compared strategies people used to control strong and weak habits and found that vigilant monitoring, watching for the signs of the habit and telling oneself not to do it, was the most effective strategy for strong habits, and that it worked by heightening inhibitory control⁴⁰. Real-time monitoring of lapses shows the cost of not noticing. Lapses occur when the trigger, the state and access coincide and no coping response is used²⁹. The training makes noticing routine weeks before the full response system is built.

These days build the stock of things the person can do instead. Slow breathing practices reduce physiological arousal (Section 6.4). Absorbing, cognitively demanding tasks compete for the working memory on which craving imagery depends: the elaborated intrusion theory predicts that a task occupying the same mental resources should weaken a craving episode²⁷, and executive functions improve with practice¹⁵. The two practice days exist because a skill learned once is rarely available under pressure. By the end of Day 23 the client has a full day of practices and has used each of them several times.

5.7Days 24 to 30: perspective, evidence and the turn to triggers

The evidence reviewed on Day 25 comes from imaging studies of people in abstinence. In a cross-sectional study of people with cocaine dependence at different points in abstinence, grey matter volume in the anterior and posterior cingulate, insula and prefrontal regions was greater in those who had been abstinent longer, with some regions exceeding control levels after roughly 35 weeks³⁷. The study was cross-sectional and the authors note that recovery appears to involve several distinct processes, so it does not show what will happen in any one person. Its use in the program is to replace the belief that the damage is permanent with an accurate, cautious expectation that function recovers over months.

Phase 1 ends by explaining the mechanism Phase 2 will work on. The disease model the client learns on Day 29 is the one set out by Volkow, Koob and McLellan⁵, presented so that the client understands their difficulty as the predictable result of changes in the brain rather than a defect of character. Day 30 explains how those changes attach to particular settings: glutamatergic learning between prefrontal cortex and nucleus accumbens binds the habit to the details of where and when it happened¹⁰, and exposure to those details reliably raises craving¹¹. A client who arrives at Day 31 knows that their triggers are learned, specific and identifiable. Phase 2 identifies them.

By Day 31 the client has been abstinent for about two weeks, their sleep and daily structure are more stable, and they have a set of practices that compete with the habit. What they do not yet have is a prepared response for the specific moments in which a lapse becomes likely. Those moments are predictable. Real-time studies show lapses concentrating where a trigger, a negative state and easy access coincide and no coping response is used²⁹. Phase 2 is built to make sure that, for each such moment, a response exists, has been rehearsed and is ready to use.

The phase follows the order in which a lapse unfolds. It begins with the triggers themselves and the environment that presents them. It then moves to the moment of temptation: noticing, interrupting and redirecting the sequence. It covers the thoughts that give the person permission to continue, the emotions and stress that most often start the sequence, and the bodily experience of the urge. It ends by binding each trigger to its own response in a written plan and practicing that plan daily until Day 60.

6.1Days 31 and 32: the trigger map and the environment reset

Mapping triggers by type follows from what is known about how they are learned. Trigger memories encode the full context of past episodes^{10, 13}, so a person's triggers are rarely confined to one category. A client who believes their trigger is stress usually finds, once they look at a month of logs, that it is stress on a particular day of the week, at a particular hour, in a particular room, after a particular kind of conversation. Each of those elements is a separate trigger and may need a separate response. Mapping across every group makes it difficult to miss one.

Removal comes before any in-the-moment skill because it is the cheapest and most reliable form of control. Duckworth and colleagues' review of self-control strategies argues that situational strategies, which change the situation before the impulse begins, are more effective than strategies that try to resist an impulse already under way³⁹. Verplanken and Wood describe the related idea of habit discontinuity: changes to the context in which a habit runs disrupt its automatic triggering and open a window in which new behavior is easier to establish³⁸. Both point in the same direction. If the object that starts the sequence is not in the house, the sequence does not start there.

6.2Days 33 to 37: why early abstinence is hard, and the response sequence

Day 35 applies the allostatic model directly. If repeated use has shifted the reward and stress systems away from their original set points, stopping removes the habit before those systems have returned, and the person feels the gap as low mood, irritability, restlessness and craving¹⁷. Those states are among the strongest drivers of relapse¹⁸. Explaining this in advance changes how the person reads them. A hard evening in week five is an expected stage of recovery, with a known cause and a known direction, rather than evidence that stopping is not working.

The sequence is designed around the time course of a lapse. The habit sequence, once started, is run by habitual control and does not consult current goals^{4, 14}. The response therefore has to come early, as soon as the trigger is noticed, which is why awareness training began on Day 18. Breaking the automatic chain opens a short window in which deliberate control can engage. The response used in that window is chosen in advance, because the person in the hot state is poor at choosing one¹⁶. Each episode is recorded, so the plan is revised with evidence and improves with use.

6.3Days 38 to 43: situations, interruptions and permission thoughts

Positioning extends the situational strategies of Day 32 into settings the client cannot remove, such as a work dinner, a family event or an evening at home. Duckworth and colleagues' framework distinguishes selecting a situation from modifying it, and both act before the impulse is at full strength³⁹. Pattern interruption addresses the case where the sequence has already begun. Because habitual behavior runs as a chain triggered by context, breaking a link early in the chain returns control to deliberate choice at a point where it is still available.

Permission thoughts are the most consistent feature of the lapse histories our clients describe. They are usually arguments for an exception: this occasion is special, today was hard enough to earn it, one will not matter, the plan can restart tomorrow. They sound reasonable because they are produced in the hot state, by a person whose valuation of the immediate reward is temporarily inflated¹⁶. The program therefore identifies the person's recurring permission thoughts in advance, while they are in the cold state, and prepares a response for each.

Defusion comes from acceptance and commitment therapy, where it describes changing a person's relationship to a thought, observing it as a thought rather than treating it as a fact or an instruction, without trying to change its content⁵³. Hayes and colleagues' review reported encouraging evidence for the approach while noting that the literature was still developing, and we use it with that caution. It matters here for a specific reason: the alternative, suppression, tends to backfire. In Wegner's experiments, people asked not to think of a white bear later thought about it more than people who had been allowed to think about it all along⁴². A permission thought that is pushed away returns. One that is observed, named and set down has less hold.

Reappraisal of this kind has a measurable effect on craving. In an imaging study by Kober and colleagues, people who smoked were shown smoking-related images and asked either to focus on the immediate effects of smoking or to think about its long-term consequences. Thinking about the long-term consequences reduced reported craving, increased activity in dorsomedial, dorsolateral and ventrolateral prefrontal regions, and reduced activity in regions associated with craving, including the ventral striatum, amygdala and ventral tegmental area⁴¹. The written review on Day 43 prepares that long-term frame in advance, so that it is available in the moment rather than constructed under pressure.

6.4Days 44 to 49: emotion, stress and the felt urge

Naming an emotion is itself regulating. In Lieberman and colleagues' imaging study, putting a feeling into words, affect labeling, reduced the amygdala's response to emotional images and increased activity in the right ventrolateral prefrontal cortex, a region associated with inhibitory control⁵⁴. For our clients, emotions are the trigger type most often present in a lapse and the one most often missed, because the emotion is felt as a general sense that the evening needs something. Precise naming turns it into a specific trigger with a specific response.

Day 45 brings the stress research of Section 2 to the client directly. Koob's account of the recruitment of brain stress systems in addiction¹⁸ and Sinha's finding that stress-induced craving was associated with faster relapse after treatment¹⁹ together explain why so many lapses follow a difficult day, and why the stress response needs its own set of tools rather than being left to willpower. The next four days provide them.

Slow breathing is the most direct voluntary lever on the stress response. In a study of 40 healthy adults, a program of diaphragmatic breathing practice over eight weeks improved sustained attention, reduced negative affect and lowered cortisol⁵⁵. In a randomized study by Balban and colleagues, five minutes a day of a breathing practice emphasizing extended exhalation improved mood and reduced resting respiratory rate more than mindfulness meditation over the same period⁵⁶. These are studies of general populations, not of people stopping a habit, and we cite them for the mechanism: a short, learnable practice that lowers arousal within minutes and can be used at the moment a stressful trigger appears.

The practice on Day 48 draws on research into how working-memory load affects emotional memories. In a study by de Voogd and colleagues, eye movements made while recalling an aversive memory transiently deactivated the amygdala and improved extinction of fear. The authors noted that the effect is probably not specific to eye movements: other tasks that load working memory may act in the same way⁵⁷. That fits the elaborated intrusion account of craving, in which a craving episode depends on vivid imagery held in working memory and is weakened by tasks that compete for it²⁷. We present the practice to clients as one way of loading working memory while the emotional weight of a trigger is high, not as a specific therapeutic technique.

Urge observation, often called urge surfing, uses the interoceptive account of craving in Section 2.7. The urge is a representation of bodily state^{25, 26}, and it can be watched. In an experiment with 123 people who smoked, a brief instruction in urge surfing did not reduce the strength of the urges participants reported, but it did lead to fewer cigarettes smoked over the following week⁴³. The result is useful because it is modest and specific. The practice changes what the person does with an urge rather than removing the urge. That is the purpose it serves in the program: by Day 49 the client has been abstinent for five weeks and will still have urges, and what matters is that an urge does not have to end in the habit.

6.5Days 50 and 51: delay and the future self

Delay discounting is the tendency to value a reward less the further away it is. It is measured in most people and is greater in people with addictive behaviors: MacKillop and colleagues' meta-analysis found a reliable difference of moderate size between people with addictive behaviors and controls⁵¹. In practice, a moderate difference in discounting is enough to make the drink tonight worth more than a month of better mornings. Bickel and colleagues' account of reinforcer pathology describes the short time horizon as part of what maintains addiction, and identifies lengthening it as one route out³⁵. Day 51 works on that horizon. A future that is vivid and personally specific is easier to hold in mind against an immediate reward.

6.6Days 52 to 60: binding each trigger to its response

Day 52 is where the second phase comes together. The form of the plan follows the research on implementation intentions: plans that specify when, where and how a person will act in response to a particular situation. Gollwitzer and Sheeran's meta-analysis found that forming implementation intentions had a medium-to-large effect on goal attainment, and that they work in part by making the critical situation easier to notice and by triggering the planned response automatically when it arrives⁴⁴. That is the property this phase needs. A response that has to be chosen in the moment is chosen by the weakest version of the person. A response bound in advance to its trigger runs with far less deliberation.

The final week is practice, because a plan has to be used to become reliable. Each craving in this period is an occasion to run the prepared response against a real trigger, and each run is recorded. Responses that worked are kept. Responses that failed are revised the same week, with evidence of where they failed. The craving record also shows the client something that is hard to see from inside a single evening: that cravings arrive, peak and pass, and that their frequency and strength change over weeks. The relapse curve flattens with accumulated abstinent time². By Day 60 the client has lived through several weeks of that flattening, and has their own record of it.

By Day 61 the client has been abstinent for six or seven weeks and has a rehearsed response for each of their triggers. Two sources of risk remain, and both are slow. The first is identity. A person can stop a behavior while still regarding it as part of who they are, and in that position every difficult evening reopens the question of whether they are really someone who has stopped. The second is social context. Most habits are embedded in relationships, routines and settings that were organized around them, and those continue to exert pressure long after the trigger plan is in place. Phase 3 addresses both.

7.1Days 61 to 66: rebuilding the self-concept

Identity work comes late in the program for a reason. A new self-description is only credible when there is evidence behind it, and by Day 61 the client has weeks of it. Oyserman's identity-based motivation model holds that people act in ways that feel congruent with the identities salient to them in the moment, and interpret difficulty according to whether the goal feels like one that people like them pursue⁴⁵. A goal felt as part of who one is gets more effort and survives difficulty better. The effect of identity framing can be seen even in single sentences. In field experiments by Bryan and colleagues, asking people about being a voter, rather than about voting, increased turnout as measured in official records⁴⁶. The noun invited the person to claim an identity, and the identity carried the behavior.

The structured, game-like framing of Days 63 and 64 serves the same purpose by different means. It gives the new identity concrete, measurable attributes, so that progress is visible in the same way that progress in training was visible in Phase 1. Motivation is matched to the person because what sustains effort differs between people: some are moved most by mastery and achievement, others by connection or by influence over outcomes. Goals that draw on the person's own source of motivation are more likely to be pursued when the original reasons for stopping have faded from view.

7.2Days 67 to 70: self-image and outside pulls

Social influence on habits is large and well documented. In Christakis and Fowler's analysis of a large social network followed over three decades, people stopped smoking in clusters, and a person's chance of smoking fell substantially when a spouse, sibling, friend or coworker stopped⁴⁷. The influence runs in both directions. A network that still drinks, uses or stays up late with the phone exerts a steady pull back. Litt and colleagues tested a way of changing this in a randomized trial with people being treated for alcohol dependence: a treatment designed to shift the person's social network toward support for abstinence changed their networks in that direction and improved drinking outcomes at follow-up⁵⁸.

Recovery research describes the same process at the level of identity. Best and colleagues' social identity model of recovery proposes that recovery is a social identity transition: people move away from groups whose identity is bound up with use and toward groups whose identity supports recovery, and the new group memberships support the new self-description⁵⁹. On Days 68 to 70 the client looks at which relationships, events and feeds pull toward the old identity and which support the new one, and adjusts their time accordingly. Removing the feeds that present triggers on demand extends the environment reset of Day 32 into the client's phone.

7.3Days 71 to 80: rehearsal

Mental rehearsal is a well-studied performance technique. Driskell, Copper and Moran's meta-analysis found that mental practice reliably improves performance, though less than physical practice, with effects that depend on the type of task and the time since practice⁴⁸. Rehearsal suits high-risk situations in particular because they are often rare, such as a wedding, a work trip or a difficult anniversary, and cannot be practiced live often enough. By imagining each in detail and rehearsing the response, the client meets the situation the first time with a response that has already been used. The self-statements connect the rehearsal to the identity work. The client rehearses not only the action but who they are when they take it.

7.4Days 81 to 90: replacement and review

Replacement is the last step because it is what makes stopping self-sustaining. A habit that served a function leaves a gap when it stops, and a gap left empty is filled by whatever is most available, often the old habit. Wood and Rünger's review describes how habits are maintained by stable contexts and how new behaviors become habitual when they are repeated in those same contexts⁶. A replacement placed at the same time and in the same setting as the old habit can take over its context. Lally and colleagues' finding that automaticity builds over weeks to months⁴⁹ is the reason the program does not end with the replacement merely chosen. By Day 90 it has been in place long enough to have begun to run on its own.

Relapse research gives two reasons to continue beyond Day 90. First, the risk falls with time but does not vanish: the relapse curve flattens as abstinent time accumulates, but it does not reach zero within months². Second, the systems that sustain the habit change slowly. The allostatic shift described by Koob and Le Moal persists after use stops¹⁷, trigger learning remains available to return in the right context¹³, and structural recovery in control regions continues over many months³⁷. A person on Day 90 has a working system and three months of evidence, but some of the changes the program set in motion are still under way.

Self-forgiveness opens aftercare because guilt is the stage of the loop that most often restarts it (Figure 6). Carried forward, it is a negative state of the kind that drives use through negative reinforcement¹⁸. It also shapes how the person will respond if a slip occurs. Curry, Marlatt and Gordon studied people who had lapsed after stopping smoking and found that those who went on to full relapse attributed their lapse more to internal, stable and global causes, such as a lasting personal weakness, than those who lapsed and recovered⁵⁰. This is the abstinence violation effect. A person who has forgiven their past is less likely to read a single slip as proof of a fixed defect.

The program's model of relapse follows Witkiewitz and Marlatt's reconceptualization of relapse prevention, which describes relapse as a dynamic, nonlinear process in which risk factors such as stress, craving, self-efficacy, coping and mood interact, and in which small changes at one point in time can produce sudden shifts in behavior⁶⁰. In that model there is no single cause of relapse to remove, and no point at which the person is permanently safe. What protects them is a set of coping responses that remain available as conditions change. That is what the client has built, and aftercare is the period in which they keep using it without the daily structure of the program around them.

The combination of awareness training and relapse prevention that runs through the program also has direct support in the literature. In a randomized trial by Bowen and colleagues, 286 people who had completed initial treatment for substance use disorders were assigned to mindfulness-based relapse prevention, standard relapse prevention or treatment as usual. Both relapse prevention approaches reduced the risk of relapse compared with treatment as usual at six months, and at twelve months the mindfulness-based approach was associated with fewer days of substance use⁶¹. The program is not either of those protocols, and the trial does not test it. But the elements the trial found useful, sustained awareness of urges and triggers combined with prepared coping responses, are the ones the client carries into aftercare.

Aftercare runs for six months after Day 90. In that period the client runs the system they built, and we follow up on whether they have stayed off the habit. The six-month window is also the period over which our success rate is defined (Section 9).

Phalanx Labs has worked with 290+ clients across 12 countries over 7 years. Of those clients, 98.4% had already tried and failed to stop before they came to us. Our success rate is 97%, defined as the share of clients who ran the program and stayed off the habit through the program and the six months of aftercare that follow it. Because the definition matters as much as the number, we set out below exactly what each figure means, how it is tracked and what it does not establish.

9.1How outcomes are tracked

Throughout the program, the client keeps a daily log and completes a weekly progress update. These are qualitative records of what happened, what was difficult and what was handled; they are not scored instruments. At the end of each of the first two phases the client has a progress call with our team, the first around Day 28 and the second after Day 60. During aftercare we follow up with clients on whether they have stayed off the habit. A client counts as a success when they have run the program and have stayed off the habit through the program and the full six months of aftercare.

9.2What these figures do not show

Our outcomes come from client follow-up in the ordinary course of our work. They are not the results of a clinical trial. There was no randomization and no control group, the outcomes were not assessed by anyone independent of Phalanx Labs, and the method has not been published in a peer-reviewed journal. We state this plainly because it limits what the figures can show, and because readers with a research background will want to know.

• Self-report. Whether a client has stayed off the habit is known from what the client tells us, not from biological testing.
• Selection. Clients choose to apply, are accepted after a diagnostic in which we decline people we cannot help, and are counted only if they ran the program. People who decide to work with us may differ in motivation or circumstance from people who do not.
• No comparison group. Without one, we cannot say how many of the same people would have stopped through some other route over the same period.
• Time window. The success rate covers the program and six months of aftercare. It does not describe what happens after that.
• Variety. Clients differ in their habits, severity and circumstances, and a single rate across all of them does not predict the outcome for any one person.

The research cited in this paper has the opposite set of limits. Most of it comes from controlled studies, but of single mechanisms, often in one substance, sometimes in animals, and almost never of a program like ours as a whole. It supports the reasoning behind each element of the method. It does not show that the combination works, or how much each element contributes. The figures in Table 3 are our evidence for the combination, with the limits stated above. A controlled evaluation of the method would need independent outcome assessment, a comparison condition and follow-up beyond six months. We would welcome one.

Phalanx Labs is a behavior and performance engineering firm. The Behavior Control System is a structured method for changing behavior. It is not therapy, clinical care or medical treatment, and it does not diagnose or treat any medical or psychiatric condition. Our team does not act as the client's doctor or therapist, and the program does not replace either.

The safety note is placed at the start of the program and repeated wherever it applies, because the program asks people to stop, and for some substances stopping is the dangerous part. A person physically dependent on alcohol or benzodiazepines can experience severe withdrawal when they stop abruptly, and that requires medical assessment and, in some cases, medically supervised withdrawal. Where the diagnostic suggests this may apply, we tell the person to see a doctor before they begin, and the stop date on Day 15 follows medical advice rather than the other way round.

Some of the program's physical practices also carry ordinary risks. Cold and heat exposure, and starting a training program, are not suitable for everyone, and people with heart conditions, high blood pressure or other medical conditions should check with their doctor first. The program gives general information about supplements on Day 15 and tells clients to consult a doctor before taking any; it does not prescribe or recommend medication or supplements.

The program is also not designed for every situation. It is not a crisis service. It is not suitable as the only support for someone with a serious mental health condition, someone at risk of harming themselves or others, or someone whose substance use needs medically supervised treatment. The diagnostic exists partly to identify these situations, and when we cannot help, we say so and suggest the person seek appropriate care.

Finally, a note on what this paper discloses. It states what each part of the program targets and why, with the evidence for each choice, so that readers, including clinicians and researchers, can evaluate the reasoning. It does not publish the exercises, their steps or their wording. That is a commercial choice, and it means a reader cannot reproduce the program from this paper. It does not change the evidence cited, which is public and can be checked through the references that follow.