Imagine deciding to exercise before work. You buy the shoes, choose a program, and put the first session on your calendar. Monday goes well. On Wednesday, you sleep badly. On Friday, an early meeting gets in the way. By the following week, getting out the door has become another argument you have with yourself.
Wendy Wood would have you look beyond your resolve. Where is the gym? What happens immediately before you leave? Do you enjoy the activity once you get there? Have you chosen a time that your life will actually let you keep?
In Good Habits, Bad Habits, Wood argues that we put too much of the burden of change on conscious determination. We decide that something matters, expect that decision to carry us through the months ahead, and blame ourselves when it doesn't. Her alternative is to arrange the circumstances around an action so that repeating it becomes easier. A familiar setting can prompt the next step. A rewarding activity gives you a reason to return. Repetition can reduce how often you need to talk yourself into starting.1
There is a great deal to like here. Leaving your running shoes by the door may solve a problem that another promise to be disciplined does not. But a useful suggestion and a scientific explanation are different things. Putting the shoes there might help because they remind you to run, because you no longer have to find them, or because leaving them out was part of a more realistic exercise plan. Improvement alone doesn't tell us which explanation is right.
That distinction runs through my assessment of the book. Wood offers a valuable guide to the conditions that support repeated action. Her larger argument—that habits explain much of persistence and self-control, can be deliberately built across many areas of life, and help explain striking recoveries from addiction—is less secure. The book is strongest when it helps you redesign a recurring situation. It becomes less reliable when a result is made to carry a broader explanation than the research tested.2
A better question than “Why am I so undisciplined?”
Suppose you want to read in the evening but usually watch television. You could promise to try harder. You could also leave a book on your chair, move the remote, and choose something you actually want to read.
Those changes solve different problems. The book on the chair makes reading harder to forget. Moving the remote makes television slightly less convenient. Choosing a better book makes reading more appealing. You might eventually begin reading with little deliberation, but some of these changes could help on the very first evening, before a habit has had time to develop.
Wood gives this kind of preparation the attention it deserves. Her chapters on context, repetition, and reward ask readers to examine what makes an action easy to start and worthwhile to repeat. She discusses arranging food, preparing family routines, finding enjoyment in an activity, and replacing an unwanted response with something that fits the same situation. The phone appendix turns these ideas into specific suggestions about notifications, access, and alternatives to checking.1
This is not a book that simply declares motivation useless. Wood gives goals a role in choosing and establishing habits. She recognizes that everyday activities mix deliberate and habitual action, and that becoming excellent at something takes more than repeating it. Those qualifications belong in any fair reading of her argument.3
The same example also shows why the explanation matters. If you read for three weeks and then stop because the book is boring, choosing another book addresses a different problem from adding a reminder. Wood's emphasis on enjoyable rewards makes room for that response. What the repeated reading cannot tell us, by itself, is how much a habit contributed to keeping it going.
What counts as a habit?
By habit, Wood means a learned connection between a situation and a response, one that can prompt an action without a fresh decision each time.4
One of Wood's most memorable figures is that about 43 percent of everyday actions are habitual. The research behind that number asked people to report what they were doing, where they were doing it, and what they were thinking about. Actions performed almost daily in the same place were classified as habitual. In one of the studies, 43 percent of recorded actions met that definition.5
That is evidence of a lot of repetition in familiar settings. It isn't a direct measurement showing that 43 percent of everything people do happens without thought. The researchers separately examined people's thoughts and found differences consistent with reduced attention to habitual actions. Wood explains both parts of the research. Yet Wood also writes, “Fully 43 percent of the time, our actions are habitual, performed without conscious thought.” That sentence merges the classification and the separate thought findings into a much more definite-sounding claim.
Consider two people who make tea every afternoon. One reaches for the kettle almost before noticing what they are doing. The other pauses, decides that tea sounds good, and then fills it. Their diaries could show the same action in the same kitchen every day. Yet the action begins differently. Frequency and setting help identify a possible habit; they don't settle how the behavior is controlled.
Researchers therefore use several approaches. They can ask how automatically an action seems to start. They can change the setting and see whether behavior changes. They can also make an outcome less attractive and test whether someone continues responding for it. This last approach is called outcome devaluation: the action used to lead to something you wanted, and now that outcome is less valuable. Continued responding can suggest that a learned response is carrying on despite the change.4
The popcorn experiments Wood describes make the idea easy to picture. Researchers gave people fresh or stale popcorn and examined how much they ate in relation to their existing popcorn habits. Familiar movie-eating circumstances helped sustain consumption among habitual eaters despite the stale popcorn's reduced appeal. A further experiment found that disrupting the usual eating response by changing hands altered the pattern.6
That is more informative than merely discovering that moviegoers often eat popcorn. It changes something—the food's appeal or the usual way of eating—and observes the response. The results support a role for learned habits in that particular activity. Wood also warns that switching hands is not a weight-loss program. A useful experiment need not become a universal prescription to be worth knowing about.
These tests also have limits. In some laboratory tasks, a person may continue responding because they didn't understand what changed, rather than because habit overrode a goal. Or a familiar cue may prepare an old response that the person then catches and stops. In that case, measuring only the final action could miss the habit's influence. Wood and colleagues develop this second possibility in later scientific work.7
This helps explain why habit researchers can disagree about laboratory results. A 2018 series of experiments and a later four-laboratory study did not reliably produce the expected increase in habitual responding with more training. These findings weaken a simple account in which sufficient repetition predictably transfers control to habit. They also leave a measurement problem: how well do the tasks distinguish a weak habit from a strong habit that someone successfully overrides?7
There is positive experimental evidence for this distinction, too. In a study by Hardwick and colleagues, people practiced one response and then learned a different rule. When required to respond quickly, they were more likely to produce the old response; with more time, they could follow the new rule. A 2026 study likewise found that additional practice increased the time cost of switching away from a trained response. These findings support an influence of prior learning even when a person can ultimately act on the current goal. They do not establish how much of ordinary life is controlled by habits.8
The sensible conclusion is that habits can influence behavior while remaining difficult to measure cleanly. It is too easy to call every successful routine a habit and then treat that label as an explanation of its success.
Starting is different from doing
Wood makes a particularly interesting argument about self-control. Perhaps people who seem disciplined aren't continually winning battles against temptation. Perhaps their useful routines prevent some of those battles from occurring.
Research gives this idea support. In six studies involving 2,274 participants, Galla and Duckworth found that beneficial habits helped statistically explain links between self-control and positive outcomes. People reporting greater self-control also tended to report more useful habits, and those habits were related to better outcomes.9
But the researchers did not assign people to acquire habits and then establish that the new habits caused their advantages. Other differences could help produce both. A person with a predictable schedule, for example, might find it easier to maintain a routine and complete work on time. The studies support the habit explanation; they don't establish habits as the principal cause of the advantages associated with self-control. Wood goes further near the end of the chapter, saying that the preceding studies and stories have “established this mechanism.” Finding a pattern consistent with an explanation does not establish how much that explanation accounts for.
The practical distinction becomes especially important with demanding activities. Beginning a workout without a long internal debate can be a substantial improvement. It doesn't tell us how much attention or effort the workout itself will require. Sitting down at a piano can become familiar while learning a difficult passage still takes concentrated practice. The reliable start supports the work; it doesn't perform the work for you.
Wood recognizes that repetition and excellence are different, and her running example acknowledges difficult miles. Her point is that a regular runner need not keep debating whether to continue. But she extends that point into the assurance that “effort doesn't last,” adding that continued effort means “you're doing it wrong.” A habit can reduce repeated decisions without making continued mental effort a reliable sign of failure.3
Making a start easier is valuable, and familiar parts of an activity may also require less deliberation. But the studies discussed here do not establish that continued mental effort means habit formation has failed. Readers shouldn't be led to regard that effort as evidence that they are doing something wrong.
What the practical studies show
Wood is more careful about habit-formation time than the familiar “66 days” headline suggests. In the study behind that number, people chose a small eating, drinking, or activity behavior, repeated it in a consistent setting, and reported how automatic it felt. Researchers fitted curves to the reports to estimate how the sense of automaticity developed.10
The 66 days was the middle estimate in a selected set of well-fitting curves. It described the modeled time to get close to an individual's estimated plateau: the point where further repetition was predicted to add relatively little. It was not a deadline after which everyone had acquired a habit. Estimates varied widely, and some extended beyond the period the researchers actually observed. Wood herself rejects a single correct number of days.
For a reader, the limitation is straightforward. Feeling more automatic while repeating an action is encouraging, but it isn't the same as demonstrating that the action will persist for years or survive a major change in circumstances. Nor does a person's data failing to fit the researchers' chosen curve establish that the person failed to form any habit.
Trials that measure practical outcomes add something different. An English primary-care trial assigned 537 adults with obesity to a brief habit-based program called Ten Top Tips or to usual care. After three months, the program group had lost about 0.87 kilograms more—roughly two pounds—on average, after statistical adjustment. At two years, that early advantage was no longer present: participants in the program had maintained weight loss, but usual care had caught up. About 72 percent of participants attended the three-month follow-up and 58 percent attended at two years.11
That is a useful result for a brief intervention. It also illustrates why “the change lasted” and “the program remained better than the alternative” are separate claims. Both groups can improve while a difference between them disappears. And because Ten Top Tips combined advice, monitoring materials, and contact with a health professional, the trial cannot tell us how much of the benefit came specifically from habits becoming automatic.
A smaller trial of two habit-oriented programs found short-term weight-loss advantages over a waiting list. Longer follow-up continued for the intervention groups without retaining that untreated comparison. The later improvements therefore tell us how those participants did, but not how much better they did than they would have without the programs.12
A 2023 review of ten randomized physical-activity trials also found an average improvement in self-reported habit strength. That broadens the favorable evidence beyond one program. The result concerns how habitual activity felt, however; it is not a corresponding percentage increase in exercise or a direct measure of improved health.13
Together, these studies support using habit-oriented methods for some recurring behaviors. They provide less assurance about how reliably those methods reduce ongoing deliberation across the book's full range of applications, or how durable that change will be. Wood recognizes the conscious work needed at the start. But her suggestion that this investment eventually removes the need for continuing attention—yielding passive returns “for all time”—goes beyond what these studies establish.14
Even consistency requires judgment. A later workplace experiment compared incentives for gym visits at flexible times with incentives tied to a chosen two-hour window. Flexible incentives produced better attendance during the first ten weeks after the original incentive period; the difference was no longer statistically clear later in the follow-up. Daily reminders and a common attendance-linked lottery continued, so this was not a test of removing all support. A narrow window can exclude opportunities that would otherwise work. This doesn't undermine the usefulness of familiar cues; it shows why making a schedule rigid is not the same as making an activity dependable.15
Two examples where the details change the lesson
In the reward chapter, Wood describes a weight-loss trial with monthly payments for meeting goals. She suggests that a long delay between an action and its reward could explain the limited results. Perhaps participants dieted just before the weigh-in without building lasting habits. She explicitly presents the last-minute dieting as a guess.16
The missing detail is that the same trial included a more successful group-payment condition, also paid monthly. People in that condition were placed in anonymous groups of five. A monthly reward pool was divided among members who met their targets. At the end of six months, the group-incentive participants had lost an average of 4.8 kilograms, compared with 1.7 under individual incentives and 0.5 in the control group. An advantage over control remained at the follow-up three months later.16
Including that result changes the question. Monthly rewards did not perform uniformly poorly. Reward size, uncertainty, and the way the payments were organized could all matter. Some of those possibilities fit Wood's broader discussion of rewards. But the study did not measure habit formation, so it cannot establish that the less successful group failed because no habits formed—or that the more successful group benefited because habits did form.
A separate error concerns the scale of a meal. In Chapter 4, Wood describes a standard pasta-and-cheese serving as containing about 1,800 calories, with patrons eating about 1,700. Those figures correspond to kilojoules, a different unit of energy. The standard entrée actually contained about 422 food calories, and diners ate about 399. The larger serving led to average consumption of about 571 calories.17
The unit error makes the reported energy amounts more than four times their actual size. It does not erase the finding that diners ate about 172 additional calories from the larger entrée—a 43 percent increase.
Addiction asks more of the argument
The addiction chapter carries the highest stakes. Wood discusses heroin use among soldiers in Vietnam and the low rate of renewed addiction after they returned home. She contrasts a roughly 5 percent first-year re-addiction figure with a 40–60 percent relapse figure for conventional treatment, then asks why treatment should not aim for the lower number.18
The recovery finding deserves attention. It challenges the idea that addiction always follows the same course regardless of what happens around a person. But the two percentages are not interchangeable measures of treatment success. Becoming addicted again is different from using a drug again. A first-year outcome is different from an outcome measured over another period. Returning soldiers also differ from people entering treatment for persistent problems.
These differences are large enough to change the comparison. Wood acknowledges that many of the formerly addicted veterans used drugs again without becoming re-addicted. If a study counts any return to use, those people appear among its unsuccessful outcomes; if it counts renewed addiction, they may not. You cannot establish a superior recovery rate simply by putting the two percentages side by side.19
Coming home changed much more than exposure to familiar cues, and the study did not isolate the effects of those changes. In the original report, later republished in 2010, Robins and colleagues questioned whether changed availability and circumstances were a sufficient explanation for recovery. They also showed why treatment comparisons require care: the men with more severe addiction were more likely to receive treatment, so their outcomes could not by themselves establish whether treatment helped.19
Wood acknowledges that addiction has multiple causes, that it differs from ordinary habit, and that the Vietnam comparison was uncontrolled. She treats biological and environmental explanations as complementary. Those are substantial qualifications. They still don't make the 5 percent figure a sound benchmark for what a different treatment population should achieve.
The chapter's account of treatment also needs a correction. Wood describes twelve-step programs as performing no better than alternatives and cites a cocaine-treatment trial. Yet all four conditions in that trial included group counseling that encouraged twelve-step participation. The combination of individual drug counseling based on twelve-step principles and group counseling produced the strongest drug-use results. That comparison does not isolate the effect of community twelve-step groups, but it does not support the broad dismissal either.20
There is better support for the chapter's practical direction in human intervention studies. Treatments designed to help people build social networks supportive of abstinence produced more alcohol-free days, though not every drinking measure improved more than in the comparison treatment. These programs often used Alcoholics Anonymous to help build that network. Wood also describes employment-based programs in which access to paid work depended on drug-free test results. They produced benefits, but she notes that those benefits could recede when the condition ended.21
These interventions help explain what changing an environment can involve: access to supportive people, worthwhile activities, and continuing rewards for abstinence. They support giving those conditions serious attention. They do not establish how much of the improvement came from automatic habits rather than continuing incentives, social support, or deliberate choices. And needing continued support does not make an intervention worthless. Its value depends on the benefit and the burden of maintaining it.
The chapter is right to challenge the idea that recovery is merely a private contest of resolve. It is much less persuasive when it uses a striking historical percentage to judge what treatment ought to achieve.
Some evidence has changed since publication
A present-day review also has to account for research withdrawn after 2019. The clearest example in the rituals chapter is Brooks and colleagues' study of performance anxiety, which supplied Wood's karaoke experiment. The paper was retracted in 2024 after the authors lost confidence in the findings and an audit identified problems with the research record. The retracted article should no longer be treated as established evidence for that performance claim.22
Another retracted paper, by Tian and colleagues, appears in the chapter mainly as a source for a definition of ritual. Its retraction concerned discrepancies in the first study's condition coding; corrected analysis no longer supported that study's calorie-intake effect. Wood does not build her account around that experiment, so the withdrawal deserves less weight in assessing her argument than the karaoke paper does.23
Other ritual findings need to be assessed separately. A different study found a change in a brain response associated with making errors, without detecting an improvement in task performance or a change in self-reported emotional experience. Wood explicitly reports the absence of a performance improvement, then describes rituals as calming fears and anxieties. A changed brain signal alone does not establish that people felt calmer or performed better in everyday life.24
A further change occurred just before this review's cutoff. On September 2, 2026, Psychological Science retracted Ariely and Wertenbroch's 2002 paper on self-imposed deadlines, cited in Wood's first chapter. The editor could no longer rely on its findings after questions about the underlying data. Both authors agreed to the retraction.25
These developments weaken particular examples. They do not establish that Wood knew of problems in 2019, or that every ritual, deadline, or commitment device is ineffective. A fair review has to identify which source supported which claim. Counting retractions without examining their role in the book would be a poor substitute for that work.
Use the book to improve the situation—and keep examining the activity
The best reason to read Good Habits, Bad Habits is that it directs attention toward things you can often change: what is visible, what is convenient, what feels rewarding, and what repeatedly gets in the way. It also looks beyond the individual. The final chapter discusses policies, infrastructure, defaults, and feedback that can support behavior without asking everyone to supply more determination.26
Those changes can help through different routes. An automatic payment works because a system moves the money. An electricity display can help someone understand their use and decide what to change. Neither result requires the person to develop an automatic psychological response. The practical benefit can be real even when habit isn't the explanation.
Wood makes a related point in her chapter on disrupted routines: interruptions can expose better ways of doing familiar things.27 I would read her with that possibility in mind. Use her questions about context and repetition, and keep asking whether the activity itself fits. A convenient routine that you dislike may still be the wrong routine. Rearranging a kitchen may be feasible; changing a commute or social network can require resources and choices a reader doesn't have. Advice gains little from being clear if it overlooks the conditions needed to follow it.
Return to the missed morning workouts. Leaving shoes by the door could solve a problem with remembering or preparation. It won't move the early meeting. A shorter session, an evening class, a walk at lunch, or another activity might fit better. The goal is to find a way to exercise that you can sustain, not to prove you can make one particular schedule automatic.
Wood provides useful tools for that task. Their usefulness should not make us accept every explanation the book attaches to them. A habit can make the next step easier. You still need to decide whether it is the right step.
Editorial ratings
| Category | Rating |
|---|---|
| Scientific Accuracy | 50% |
| Reference Accuracy | 67% |
| Practical Value | 67% |
| Overall | 61% |
These ratings summarize nine editorial judgments on a 0–4 scale. They are not percentages of true statements or probabilities that the advice will work. Scientific Accuracy assesses three central claims; Reference Accuracy assesses source traceability, description, and inference; Practical Value assesses benefit, applicability and durability, and benefit relative to burdens and risks. The appendix explains each grade and the uncertainty around it.