Abstract
Why do we act on habit even when we intend to do something else? The answer lies in habit memories, or context-response associations, that form when people repeat rewarding actions in stable contexts. Although habits can form as people pursue goals, once habits develop, the perception of the context directly activates the response in mind. Because habit activation does not depend strongly on motivation, changing intentions has limited impact on habit memory. Instead, successful habit-change interventions directly impact the behavior itself: Along with classic behavior therapy interventions, habits change with (a) reward systems that form new habits, (b) disruption of context cues to forestall activation of the habit in mind, and (c) friction that makes the habitual response difficult and alternatives easier. Despite the strong evidence that habits are activated by contexts, people tend to believe that their own habits are a product of goal pursuit. This subjective reality might also explain why some researchers continue to maintain that habit performance depends on goals.
What makes us enter an online password that we know has expired? Drive straight home from work when we meant to stop at the store? Accidentally add the desired amount of sugar to our coffee twice? In classic accounts, such action slips occur because people absentmindedly respond habitually without intending to do so (e.g., James, 1914). People thus slip up and use old passwords, head home, or add more sugar because the familiar context (login screen, well-traveled road, sugar bowl) triggers the practiced but currently unwanted response.
Action slips illustrate a basic feature of habit: The response is activated in mind by familiar context cues—and is often performed—with limited input from current response outcomes such as whether it meets a goal or yields a reward. Thus, habits can be beneficial when they achieve desired goals but detrimental when they do not.
Habits are commonly defined as context-response associations that form in memory when people repeat rewarded responses in given contexts (Verplanken & Orbell, 2022; Wood & Rünger, 2016). With sufficient repetition, the context comes to signal the practiced action. In essence, habits are mental shortcuts that offload decisions about action to regularity in daily environments.
This article begins by analyzing the current status of habit research. The common view that habitual responses are driven directly by context cues (i.e., activated by context-response associations in memory) contrasts with a recent repurposing of the habit construct as goal-directed (i.e., activated by context-outcome-response associations; Buabang et al., 2022; Kruglanski & Szumowska, 2020). Although this theoretical distinction might seem trivial, the article concludes by showing its important implications for behavior-change interventions.
Habit Mechanics
Throughout human evolutionary history, our brains evolved with multiple memory systems designed to solve somewhat different problems (Amodio, 2019; O’Doherty et al., 2017). The habit system is engaged when people repeat responses in stable contexts, such as how to make coffee at home with their own coffee maker. In contrast, goal-directed, declarative processes flexibly drive choices in changing contexts or given new response options, such as what to order at an unfamiliar coffee shop.
People tend to repeat actions and form new habits when an action is more rewarding than expected. For example, if you try out a new coffee maker and like the taste of its coffee better than your old one, then you experience a reward prediction error, or a greater reward than expected given your past experience (Perez & Dickinson, 2020). Positive prediction errors signal that you made the right choice. These errors sum each time you respond, thereby strengthening habit memories, or context-response associations.
Many habits form as we pursue goals in daily life and repeatedly choose the same means to achieve that goal. Yet the formation of habits through goal pursuit is but one of the many ways goals and habits interact (see Wood & Rünger, 2016). Furthermore, any one action can draw on a combination of habits, goals, and other processes. As an illustration, posting on social media involves a set of repeated actions of logging in, accessing the posting function, and starting to type along with more thoughtful decisions about content and expression. Despite this interplay between habits and goals, for simplicity, the current article treats this distinction as binary and contrasts habitual with goal-directed actions.
Once habits form, they are activated by context cues relatively independently of goals. This feature is linked to the architecture of the basal ganglia, in particular the lack of reward-based modulation of neural activity in the sensorimotor loop (Yin & Knowlton, 2006). As a result, habitual responses show limited sensitivity to rewards, meaning that they tend to persist even when rewards change.
In contrast with this generally accepted view of habits, a few researchers have argued that habits depend on goals. Their claim is that human habits are driven by stable goals that function automatically, outside of conscious awareness (e.g., Buabang et al., 2022; Kruglanski & Szumowska, 2020). In support of this view, greater task repetition has not always yielded stronger habits that persist regardless of a change in goals (de Wit et al., 2018; although see van Timmeren et al., 2022). 1
Research measuring the potential drivers of repeated behavior, however, casts doubt on goal-directed accounts. One challenge is demonstrating the role of automated goals. Gawronski and De Houwer (2014) concluded that overlearned goal representations are not automatically triggered by stimuli in ways that could drive habitual responses (for failure to find that implicit attitudes drive repeated choices, see Lin et al., 2016). Instead, evidence is accumulating that performance contexts bring habitual responses to mind without depending on goals. For example, Neal et al. (2012) found that habitual runners accessed thoughts of running faster after being primed with the locations where they typically ran (e.g., track)—but not after being primed with their own personal goals for running (e.g., weight loss). For occasional runners, however, who presumably had to motivate themselves to exercise, thoughts of running were activated by their own running goals. Further suggesting that habits are directly activated by contexts, Labrecque et al. (2024) manipulated how often participants repeated an experimental task and found that greater repetition yielded stronger context-response associations. Furthermore, the stronger the habit associations, the more successful participants were at repeating the task in the future.
Especially strong evidence that people act on habits without automatically activating a goal comes from a recent experiment intended to show the reverse—that habits depend on goals (Buabang et al., 2022). The study procedure inadvertently created confusion about correct goal outcomes, but participants continued to act on habit (see Wood & Mazar, 2023).
Specifically, Buabang et al. (2022) repeatedly trained participants to choose one door from a pair, with the correct choice depending on the color (e.g., for red doors, choose left; for yellow doors, choose right). After forming habits in this way, participants were told that the correct choices had shifted, but only for half of the door colors (e.g., for red doors, now choose right). Finally, participants (a) made door choices again and (b) reported their goal expectations about which door was correct. This final assessment involved a strict time pressure to ensure that participants gave accessible, automated responses. Most interesting are the half of door colors that did not change. Participants’ initial habit training continued to be correct, but the new instructions were confusing (see Fig. 1). Over 25% of the time participants reported that the wrong door was correct (e.g., for yellow doors, expecting left to now be correct). However, participants’ choices, which were based on habit, were largely insulated from this confusion. As a result, participants continued to choose the correct doors most of the time (Fig. 1, solid bars). Furthermore, even moderate training at this task seemed sufficient to form choice habits that persisted (Fig. 1, leftmost bars). Thus, participants given little time to deliberate responded correctly out of habit regardless of their expectations about the correct goals.
Percentage of errors when correct door locations did not change (Buabang et al., 2022). Depicted are errors in choice of doors and in reported goal expectancies. Data are from Figures 4 and 5 in Buabang et al. (2022) and are summarized in Wood and Mazar (2023).
Although it might seem surprising that habits are repeated with limited input from goal outcomes, offloading response choices to the environment is functional when environments are stable and goal outcomes do not change. When goal outcomes do change, however, habits often fail to accommodate. Social media use is illustrative. In a study by Anderson and Wood (2023), others’ recognition and social rewards (e.g., likes, loves) on posts increased posting rates of new or occasional users. However, these rewarding outcomes had little influence on more habitual users. Instead of engaging more when others respond favorably, habitual users continued their regular posting. Yet habitual users seemed largely unaware of this pattern and reported being even more concerned about others’ reactions than were less habitual users (see end of article for a discussion of lay explanations of habit).
The inflexibility of habit to current experiences extends to social influence, even behavioral forms of influence. For example, in a study by Mazar, Itzchakov, et al. (2022), students with stronger habits to drink water in their dining commons mimicked less an interviewer’s drinking (see Fig. 2). Interestingly, they later reported that they were aware of how much the interviewer drank. Furthermore, goals such as thirst and intentions to drink did not explain why students persisted with their old drinking habits. Similarly, students with stronger habits for other behaviors were also less influenced by variations in social norms in these actions, despite recognizing the normative influences. Thus, habitual responses persist even when we recognize changes in the social forces around us.
Number of drinks as a function of habit strength and frequency of other’s drinking (Mazar, Itzchakov, et al., 2022). Higher numbers reflect greater participant drinking. Low, moderate, and strong habits represent mean and ± 1 SD habit strength. This pattern does not reflect a ceiling effect, given that interviewers who drank frequently took 13 sips of water, which could readily influence strong-habit participants to drink more than their three sips on average.
Habitual responses may also explain why intergroup biases persist despite people’s best intentions. After repeated rewarding interactions with members of a group, people may start to habitually engage with that group over alternative ones (Hackel et al., 2022). As a result, people may discriminate against members of certain groups, continually choosing not to interact with them, even when they wish to be impartial and fair (Brauer, 2024).
Habit-Change Interventions
Behavior-change interventions often target attitudes and beliefs with the logic that a change in attitudes naturally flows to behavior (see Mazar, Tomaino, et al., 2022). These strategies might be effective in the short term, especially when people have a motive and the opportunity to control the unwanted habit (Quinn et al., 2010). However, when stressed, tired, distracted, or under time pressure, people have little capacity to control habitual responding (see Wood et al., 2022). Furthermore, habit memories decay slowly and may persist long after people abandon or forget their desire to change.
Accordingly, people persistently act on habit even after changing attitudes (Itzchakov et al., 2018) or intentions (Webb & Sheeran, 2006). These failures were echoed in Albarracin et al.’s (2024) systematic review of behavior-change interventions that altered knowledge, attitudes, beliefs, or emotions but produced only minimal behavior change. There is simply little evidence that habits can be enduringly changed by changing goals (contrary to Buabang et al., 2022).
Social media use, especially the spread of misinformation, illustrates the challenges of controlling habit performance. Despite common claims that people share misinformation on social media because they are too lazy or too biased to evaluate its validity, Ceylan et al. (2023) found that habit drives much of the spread of misinformation (see Fig. 3). In their research, 15% of the most habitual users shared nearly 40% of the misinformation in a simulation of Facebook use. Habitual users shared false headlines even after having rated the accuracy or political position of each news item. Thus, participants’ sharing of misinformation was not due to a failure to consider accuracy or detect bias. Instead, the habitual spread of misinformation persisted seemingly blind to response outcomes such as misleading others.
Predictors of the amount of misinformation shared on social media (Ceylan et al., 2023). Effect sizes reflect the number of falsehood posts shared by participants higher versus lower in each attribute, as reflected in stronger habits, lower need for cognition (reasoning ability), and more conservative political orientation (motivated reasoning). Habit strength was assessed using the self-report scale of behavioral automaticity (Gardner et al., 2012) or reports of the past frequency of sharing information on social media.
Greater success at changing habits has been shown with classic behavioral therapy interventions (see Albarracin et al., 2024). For example, in habit reversal training, clients identify a cue to an unwanted habit such as a behavioral tic and then repeatedly practice an incompatible response. Similarly, repeated food-specific inhibition training has been shown to modify people’s approach to particular foods (Turton et al., 2016).
In addition to therapeutic interventions, structural changes can successfully alter habit performance (see Chater & Loewenstein, 2023). Such strategies directly leverage the psychological mechanisms that activate and maintain habitual responses, as shown in Figure 4.
Strategies to change habitual responses. Habitual responses can be changed by (a) changing cues in performance contexts so that habit associations are not activated in memory, (b) increasing friction on the response, or (c) altering reward structures so that people are continuously motivated to perform a different response.
New reward systems
Once formed, habits tend to be repeated with minimal consideration of the outcomes or consequences of the response, including whether it is rewarding. However, when rewards shift sufficiently to attract attention, people may try to exert control over their behavior instead of relying on habit. In this way, new rewards, when experienced immediately following a response, can be a continuous source of motivation to inhibit an old habit and repeat a new action.
Demonstrating habit change with a new reward system, Ceylan et al. (2023) instituted monetary payments to alter the amount of misinformation shared online. Some participants received a monetary reward each time they shared truthful information, whereas others were rewarded for sharing false information. After extensive training with the new rewards, participants shared primarily whatever content (true/false) had been rewarded. They persisted even when the rewards were no longer provided, suggestive of habit. Additionally, the new reward training did not change people’s broader goals for using social media. Regardless of what they shared, most participants reported that they wanted to spread truthful information. Thus, instead of changing goals, the rewards were a salient behavioral incentive to repeatedly inhibit old habits and form new habits to share certain types of information.
Thus, people can learn to share various kinds of information online depending on what gets rewarded. Social media’s current popularity-based reward and algorithm systems encourage sharing attention-getting information, and falsehoods spread faster than truths (Vosoughi et al., 2018). However, new structures can be instituted to reward accuracy, and thereby promise to establish new habitual sharing patterns.
Change context cues
Another strategy is changing the cues that activate habitual responses in mind. This is the idea behind habit discontinuity, in which everyday habits are disrupted by naturally occurring context changes such as moving, starting a new job, or changing schools (Verplanken & Orbell, 2022). Discontinuities provide a window of opportunity for people to consider alternatives and potentially decide to act in new ways. Yet such discontinuities work only when the specific cues that activate habits are disrupted. For example, students’ gym-exercise habits persisted after transferring to a new university if they could continue to exercise in similar locations (e.g., a gym in their apartment building at the old and new schools; Wood et al., 2005).
The disruptive effects of cue change are evident with social media use. Platform updates often redesign the online context cues that support habitual use. Anderson and Wood (2021, 2023) found that habitual users decreased posting after Facebook design updates, despite that new or occasional users showed increased engagement. The design change altered how users posted online, and habitual users were apparently slowed by old habits interfering with use of the new platform features.
Impose friction
Another way to control habit performance is to increase the friction, or the difficulty, of performing a behavior. Much as in the physical world, psychological friction impedes motion. Friction is reflected in the distance traveled to perform a behavior, the time required, or the effort involved. As Mazar, Tomaino, et al. (2022) showed, even seemingly negligible amounts of friction can impede important behaviors such as voting. 2 Friction is also tied to downstream consequences: Registered voters who underestimated friction were more likely to support restrictive voting policies.
A noteworthy example of the use of friction comes from an early study to reduce elevator use in a four-story office building (Houten et al., 1981). The researchers first tried influence. They posted signs by the elevator, such as “Save energy. Use the stairs for short trips.” The signs had no effect. Houten et al. (1981) then engaged friction by slowing the closing of the elevator door by 16 s. This delay reduced elevator use by a third. Even more impressive, after a month, when the doors were returned to their original speed, people kept taking the stairs. They had apparently formed a stair-climbing habit during the month of elevator friction, and they continued to repeat that behavior without testing the elevator again.
Making it easier to perform alternatives to existing habits can also reduce habit performance. For example, reducing friction on environmental behaviors can help people follow through on desires to change lifestyle habits in order to reduce greenhouse gases and pollution (Mazar et al., 2021): To reduce friction on recycling, bins can be placed in easily accessible locations, close by office desks, or in classrooms. The use of mass transit becomes easier when transit stops are close by and arrivals are signaled in real time with electronic signs. Even restaurant choices can be shifted by restructuring restaurant menus so that meat-heavy choices are in less prominent locations than vegetarian options. In Albarracin and colleagues’ (2024) meta-analysis, increasing access to a new behavior was the most effective social structural intervention to promote behavior change.
In summary, changes in intentions, beliefs, and attitudes have little long-term impact on habitual responses. Enduring changes require directly influencing behavior performance through intensive therapeutic interventions or structural changes of new reward systems, context shifts, and friction on unwanted responses.
Conclusion
Finally, for many readers, the current description of habit will conflict with their subjective experience. People often overlook habits and attribute their own repeated responses to goal pursuit. For example, Mazar and Wood (2022) found that habitual coffee drinkers overestimated the extent to which their actions were due to the desire for a pick-me-up as opposed to their habit. Even novel habits recently formed through repetition in an experimental task were misinterpreted as driven by goals.
This tendency to overlook habit is understandable: Habit representations in memory are not accessible to introspection. Unlike our beliefs and desires, we can’t feel the strength of our habits. Moreover, people have a strong tendency to interpret even ambiguous actions as goal-directed (Rosset, 2008), and this bias is especially marked in the United States (see Mazar & Wood, 2022). Thus, habit does not register in subjective experience, and it is further obscured by the belief that we are in control of our own behavior.
Perhaps psychologists’ recent attempts to repurpose habit into a goal-driven concept reflects the lure ofphenomenology, or tendency to overvalue salient, motivational determinants of behavior (Duckworth et al., 2016). Although this belief might seem benign, it can have detrimental effects such as motivating attempts to change habits through attitudes and intentions at the expense of more effective direct influences on behavior.
Recommended Reading
Poldrack, R. A. (2021). Hard to break: Why our brains make habits stick. Princeton University Press. An engaging analysis of the neuroscience of habit, outlining how our brains are “habit-building machines.”
Verplanken, B., & Orbell, S. (2022). (See References). A research-based review of the role of attitudes and habits in driving human behavior.
Wood, W. (2019). Good habits, bad habits. FSG Press. A book distilling the basic findings from habit research in ways useful in daily life, highlighting the importance of the environments around us in creating our everyday habits.
