By Tiago Forte
The holy grail of self-improvement is a framework for self-directed experimentation and learning that can be used by the average person. The key question such a framework would have to answer is “How do people change?”
In this post, I will suggest possible answers to this question by looking at the recent history of behavior change, the main obstacles this framework would have to address to be feasible, and a few promising directions from research and practice.
The fall of Neo-Pavlovian behavior change
Every book, seminar, and workshop on self-improvement (very broadly defined) ends the same way: “Do X.”
The topic could be leadership or public speaking, correct running or posture, relationships or spirituality, creativity or productivity. If it operates even partly on the assumption that people can change, period, the “takeaway advice” won’t seek to merely change your thoughts or beliefs. It will seek to change what you DO.
And they never want you to do X just once. They want you to do X…wait for it…repeatedly. They want you to make it a habit. As if doing that is trivial; just a matter of setting the autopilot in the right direction. Somewhere between doing it once and doing it many times, this daily action, however mundane, is supposed to transform your life. Assigning a (preferably odd) number like 21, 47, or 99 to the number of days it takes to “make it” a habit lends both scientific credibility and mystery to the advice.
Thus the current thinking on behavior change is that it boils down to habit formation. I’ll call this theory Neo-Pavlovian Behaviorism (NPB), a throwback to Pavlov and his salivating dogs. We’re back to setting up simple one-way triggering mechanisms, linking cues and rewards to cut out all the stuff in between. You know, the human. This era has reached its peak with the launch of the Pavlok self-shocking bracelet, with its promises to shock you out of bad habits and into good ones.
But I believe the first cracks are starting to form in the NPB edifice, pointing to a new and more fruitful direction: behavior as an emergent pattern, which can only be changed by challenging, shaping, and creating mental models on an individual basis.
The enduring power of mental models
Historically, NPB’s rise is closely tied to the rise of cognitive-behavioral therapy (CBT) in recent decades, which shares a similar operating principle — that much of the complexity of the human psyche can be short-circuited or at least ignored, in favor of “retraining” unproductive recurring thought patterns. In other words, changing mental habits directly.
The problem with this approach is that the pesky human resists short-circuiting. It’s becoming increasingly clear that underlying thoughts and beliefs do matter. They matter quite a bit. Specifically, how thoughts and beliefs are structured in the mind to explain how things work — mental models — I believe are the key to understanding not only academic theories of behavior change, but also how to effect it on the level of a single individual.
There’s a sort of meta example of the power of mental models of behavior within the history of psychology itself. As the excellent Guardian article Therapy Wars explains, Freudian psychoanalysis is today widely viewed as debunked. Scientists denounce it with unusually strong words: “Arguably no other notable figure in history was so fantastically wrong about nearly every important thing he had to say” than Sigmund Freud, the philosopher Todd Dufresne says. Nobel prize-winning scientist Peter Medawar describes psychoanalysis as “a terminal product as well — something akin to a dinosaur or a zeppelin in the history of ideas, a vast structure of radically unsound design and with no posterity.”
Yet as the fantastic BBC documentary The Century of the Self tells it, there is a different version of this story. Freud’s ideas were channeled through his nephew, Edward Bernays, who became the father of American advertising. Bernays invented the art and science of public relations, pioneering its use to sell cigarettes (aka “torches of liberty”), recasting consumption as a form of status and self expression, and developing the techniques of engineered consent so familiar to us today. Even if you reject all of Freud’s specific claims and conclusions, his basic premise — that we are driven by deep, conflicting forces that must be controlled or satiated — lived on through his nephew, evolving to underpin modern consumerism.
A traveling exhibition called Disney and Dali: Architects of the Imagination unapologetically traces Freud’s influence on Dalí’s art, and Dalí’s influence and collaboration with Walt Disney, of which the film Fantasia is the most obvious product (it’s coming to San Francisco in July). Freud set the stage of the subconscious mind, Dalí provided the fantastic imagery, and Disney figured out monetization and distribution. This gives new meaning to the Disneyfication of society at large.
The mental model that Freud created as a mere backdrop for his theories outlived those theories, becoming a force unto itself. The resilience and adaptability of this model is highlighted by recent findings suggesting the “fall” of cognitive-behavioral therapy: a 2015 meta-analysis found its effect size across studies declining by half since 1977. Another study soon after found psychoanalysis dramatically more effective in treating serious depression than any other method.
The pendulum swings the other way. Or the tables turn, depending on which model you prefer.
Let’s look at a technique that, in my experience running habit formation workshops, is among the most effective in changing behavior: Small Wins. It does this by working not at the level of external cues and rewards, but internal mental models.
Drawing from self-verification theory and BJ Fogg’s Tiny Habits program, and popularized by blogger James Clear, the technique starts with the criticism that most attempts at behavior change make the mistake of creating “high-stakes” situations. People tend to choose difficult goals, like going on a run every morning, meditating an hour per day, or studying a new language every evening. They choose these sexy-sounding, impressive habits as a way of maximizing the initial boost of energy and excitement that most rely on to get started. The problem with continuing to rely on sustained momentum is that any hiccup or missed day is interpreted as a failure. High stakes means, sooner or later, a long way to fall.
Small Wins is the technique of replacing this binary win/loss outcome with a series of progressively easier versions of the habit. If you can’t run 5 miles, run 3. If you can’t run 3, run 1. If you can’t run 1, run around the block, or walk around your house, or pace the living room, or if necessary, put your running shoes on, declare victory, and then take them off. You do only as much as you can, or feel like, or have time for. Any excuse is valid, as long as you do something. You get as much credit for the easiest version as the hardest version.
This explanation always elicits incredulous laughs. It sounds like a blank check for laziness. But something interesting happens when you try it: by eliminating the “barrier to entry” of even getting started, it calls your bluff that some external force (lack of time, money, or energy) is the true constraint. By reframing the black-and-white choice as a menu of options tailored to any level of effort you’re willing to expend, it calls attention to the fact that the most difficult step is 0 to 1, not 1 to n. Doing something, anything, instead of nothing. The rest is just optimization. This method shines a spotlight on the ways we twist logic into arbitrary “behavioral rules” to justify almost any decision. When you feel resistance to flossing even one tooth (which will inevitably happen), you have to come to terms with the reality that the friction is not between you and the world. It is between your conflicting motives.
We set up false choices: “I can’t exercise. Too busy with this project at work.” We use moral licensing: “I saved so much by not buying X, I deserve to buy Y.” We make questionable assumptions, and then treat them as laws of nature: “I can’t work out if I’ve already showered.” We pretend like we have control over situations we can’t influence (“If I spend a lot of time worrying, the plane is less likely to crash”) while denying control when we actually can (“If there’s free doughnuts in the break room, I can’t resist eating them”).
All the familiar cognitive biases are recruited to tame the cognitive dissonance of following our impulses in a conformist world. These behavioral axioms eventually come to seem so real that we can feel trapped by mutually contradictory rules without realizing that they are self-imposed.
The goal of this exercise, and where I strongly differ from mainstream authors, is NOT to steamroll your subconscious urges into submission. This shouldn’t become an abstract high-modernist fantasy, rationalizing the unruly child from above. Bad habits often serve equally useful functions as good habits. The goal is to overcome “introspection illusion” — the (often mistaken) assumption that we understand our own motives and fears. Importantly, and unlike behavioral economics exercises like “aversion factoring,” it treats this process as experiential and embodied, instead of yet another intellectual game hobbled by the very beliefs you’re trying to expose.
The point here is that personal mental models are not incidental, or purely theoretical. The model you use to understand a given change powerfully influences the outcome.
Take meditation. It’s fascinating to me to hear how people conceive of what they’re actually doing sitting on the mat in silence. In my experience, the longer they stick to the unhelpful beliefs that they need to “empty the mind” or “find peace,” the less likely they are to stick with it. A more useful metaphor is bicep curls. Every time your mind wanders, the act of bringing attention back to the breath is a repetition, strengthening the muscle of focus. This allows you to greet the inevitable with a sense of progress: just as you need gravity to build muscle, you need distraction to build focus.
Or consider the ego depletion vs. non-depletion debate, which has recently been in the news. Voluminous research has shown that willpower is like a muscle — using it too much makes it tired, leading to poor performance and poor choices. Equally voluminous research indicates that willpower is unlimited, or actually increases with use, or depends on your beliefs about the nature of willpower.
But my favorite model is that willpower is a story. People do what they enjoy, and then narrativize it as self-discipline after the fact. When we see someone with high performance we desire, we extrapolate from the immense amount of effort it takes for us to perform even at a low level, and conclude that if they perform at 10x our level, it must require 10x the willpower. But this ignores the critical fact that they enjoy doing it. It doesn’t take willpower for a hard-core runner to get up at 5 in the morning. It takes willpower for them not to. The hard truth is that no one really does anything they don’t enjoy for long. At most, they focus their efforts on finding the elusive intersection between what they enjoy and what they must do.
I once went to a health conference, where a professor presented his research into the relative efficacy of different forms of exercise. Interrupting the mind-numbing charts and graphs, a bewildered soul raised his hand and asked, “But how do I know which type of exercise is right for me?” “That’s easy,” the professor said. “The one you like.” Apparently the average exercise regimen lasts about six weeks, so sticking to it in any form is far more important than how you go about it, for everyone but elite athletes. Other studies have found that the most important factor in selecting athletic footwear is comfort, and a major reason interval training is more effective is that it is simply more enjoyable.
These conclusions seem radical because they contradict one of the deepest models many of us seem to hold in common: that the more positive and impactful the change, the more painful it must be.
Curiosity arbitrage, via dopamine
You could conclude, based on the above point, that habit formation is simply a matter of finding ways to make unenjoyable activities more enjoyable. There is a limited sense in which this is true. Getting a dog may encourage you to walk more. Listening to podcasts may make cooking more tolerable.
But there is a serious flaw (besides being factually untrue) in treating human motivation as a simple choice between pain and pleasure, with our mind presumably always pushing us toward the latter: there are diminishing returns to pure pleasure-seeking, and a finite number of ways to make something enjoyable. The most durable habits are inherently rewarding, a quality that tends to arise because they are difficult or challenging, not despite it (think puzzles).
A possible answer comes from understanding the role of dopamine, a neurochemical that pops up in any discussion of novelty, satisfaction, and risk-taking in humans. It’s been labeled the “Pleasure Neurochemical” by popular media, but this isn’t quite accurate. As Kelly McGonigal explains in her book The Willpower Instinct, the feeling that dopamine imparts is more like anticipation or arousal: “We feel alert, awake, and captivated. We recognize the possibility of feeling good and are willing to work for that feeling.”
McGonigal cites studies showing that “…you can annihilate the entire dopamine system in a rat’s brain, and it will still get a goofy grin on its face if you feed it sugar. What it won’t do is work for the treat. It likes the sugar; it just doesn’t want it before it has it.” It’s easy to see how natural selection could have favored the almost happy (thus motivated) individual over the totally content one.
In other words, the voice of dopamine is not “That felt good!” It is “If you do this, then you’ll feel good.” Which explains that “one more…” feeling you get when eating dopamine-triggering foods like chips and fries, which persists right up until you’ve eaten so much that you’re sick. It also explains why you can still crave “junk” — from fast food to action movies to vacuous social media — even when you feel terrible while consuming it and after. The “reward” associated with anticipating the experience is far more powerful than the experience itself. By the time you sit down to enjoy that hamburger and soda, the brain has already received enough of a reward in the form of anticipation to reinforce the craving for next time, even if you don’t actually enjoy more than the first few bites.
Our brains’ ability to trick itself into doing things it doesn’t even like can be a bit discouraging, but also presents us with a promising possibility: making new habits enjoyable not through visceral pleasure, but through finding new ways to be curious about them.
Curiosity is a powerful phenomenon: it is a motivated state of mind that taps directly into our need for novelty, which Gregory Berns claims in Satisfaction is the most fundamental source of life satisfaction. Like anticipation, and using the same dopamine pathways, it points to “what’s next,” focusing our attention and efforts on the “information gaps” between what we know and don’t know. But curiosity has another quality that makes it a much more sustainable, potentially even addictive, source of motivation: the more gaps you fill, the more you see. The more you know, the more you know you don’t know, the fractal nature of your ignorance unfolding endlessly before you.
I call this curiosity arbitrage: how many different ways can you discover to be curious about something?
Habits as emergent patterns
So far we’ve talked about models as being more or less useful, or having different effects. But is there anything I’m willing to take a stand on? Not much, but I think we can understand behavior change a bit more accurately using the idea of emergence.
Emergence comes from the study of complex adaptive systems (ant colonies, networks of neurons, the immune system, the Internet, the global economy). It describes systems whose complex behavior is more than the sum of its simple parts. Think of board games like chess — a couple dozen rules governing just 64 squares somehow produces possibilities that we’re still discovering after two centuries of study. Or a seed — somehow it contains specifications for unlimited variations of structures thousands of times its size. Many of the most important and least-understood phenomena in the world exhibit emergent properties, from consciousness to intelligence to ethics to life itself.
Let’s start by asking “What exactly is a habit?” It has a physical manifestation, but is defined primarily in the mind. It exists in time and through time, but depends equally on past performance and future intention. Thus at least 50% of a habit is immaterial and doesn’t exist according to science. A habit implies continuity but clearly doesn’t require it — you can miss a day and still “have” it. Heck, you can abstain for years and find its strength undiminished (as in alcoholism, or riding a bike), meaning it has no half-life and doesn’t decay, at least in some cases. Which brings up a problem of definition: you can perform a habit 100 different ways and still consider it the same one. And categorization: 100 different habits we still categorize using this one word. You could say that a habit is the purpose it serves, regardless of how you go about it. But that would seem to imply there are no habits, only intentions. Which again, aren’t measurable and thus seem to be outside the scope of science.
Many of the above features can be explained by thinking of a habit not as a “thing,” but as an emergent pattern. An emergent pattern arises from the interactions between simpler parts, but is also distinct from those parts. Think of the curve made by a row of balls being rolled down an incline. The curve doesn’t exist without the balls, but removing any given ball will not destroy it. And curves in general continue to exist even if you have no balls. Emergent patterns persist even with continual turnover in their constituents — think of a standing wave behind a rock, or the human body replacing all its cells every few years. This, of course, accurately describes a habit. It not only tolerates but requires that each day’s manifestation begin and end and pass away, to make room for the next oscillation.
An important feature of emergent patterns of all types — creativity, love, culture, theories — is that they cannot be programmed or dictated to meet direct objectives, like a computer program. They have to be grown, not built, because at every stage of development there is a different system of forces to be balanced. Imagine trying to “build” a human as a fully-formed adult: you’d presumably start with a beating heart because that is a “core” feature, but the blood would pour out on the ground without a functioning circulatory system. Starting with the circulatory system wouldn’t make sense without a respiratory system to bring in oxygen. And so on. You can’t even understand an already existing emergent pattern by analyzing its components — because it is more than the sum of its parts, disassembling the parts will not reveal the essence, the “more.” Thus the pointlessness of all the books and websites chronicling the habits of successful people in tedious detail: success is an emergent pattern of emergent patterns, even more resistant to imitation.
The upshot is that by accepting that habits, as emergent patterns, cannot be directly programmed to achieve certain goals, we are led to an uncomfortable conclusion: that despite being partially defined in terms of future intentions, habits cannot be designed and executed strictly according to upfront intentions. This conclusion neatly sums up the current state of affairs in our understanding of behavior change, as summarized by a 2006 meta-analysis: “Across dozens of studies on behavior change interventions, researchers have found that the conscious mind’s sincere, concerted intention to change behavior has little relationship to actual change in behavior.”
It is this grim conclusion that laughs in the face of our best efforts to “convince” people out of bad habits using rational arguments.
An emergent theory of behavior change seems to imply that we have no control, that we’re just along for the ride. If you can’t impose a goal on the system, what’s the point of trying?
But there is a glimmer of light in remembering that the “system” we are discussing here is us, and that of all the “objects” interacting to produce emergent habit patterns, at least a few of those things we have some influence over — attention, intention, free will. There is one idea from the study of emergence that may provide clues as to how we can act within this model: disturbance propagation.
Disturbance propagation describes one way that emergent systems can change — by using an external disturbance as the “seed” of a new pattern, and propagating this new pattern across the rest of the system in a cascading sequence. Think of how the body heals itself after sustaining a wound. It would be difficult or impossible for the healing process to be directed from a centralized source like the brain. There are too many variables, too many ways it can happen, and the transit time for sending updates and receiving instructions is too long for something so potentially life-threatening. Instead, the “disturbance” of the wound activates a locally-directed sequence of steps, coagulation followed by homeostasis followed by repair. This decentralized mechanism has numerous advantages, including that it works when the central executive is offline, and the steps can occur at different speeds in different places as needed. At a higher level, this leaves the brain free to determine the source of the wound, and to integrate this new information into the pattern of how it perceives the environment.
I would argue that a similar mechanism explains how habits are broken and formed. As any product designer will tell you, your main competition is not another new product. It is the status quo. Our existing habits are so stable that it takes an outside force — a disturbance — to destabilize the system just long enough for new solutions to establish themselves. This study reported that 36% of successful changes in behavior were associated with a move to a new place (nearly three times the rate associated with unsuccessful changes). I’m sure you’ve experienced this — moving to a new city or traveling, suddenly all your behavioral axioms become unmoored from their context, and it takes all your energy to fulfill basic needs like food, water, and a place to sleep. It is during these in-between times that new habits are formed and old ones broken almost effortlessly. From the chaos emerges a new pattern of habits to resolve a new set of forces.
There’s even evidence that emergent systems need chaos in order to form stable patterns. Habits seem to be little bubbles of structure in the midst of our chaotic lives; thus many habit formation theories emphasize controlling the environment and reducing variability in a bid to help these bubbles survive. The problem is that it takes a tremendous amount of willpower to control the environment and reduce variability in the first place — the energy required is neither created nor destroyed, just shifted around a little. There is another model of how habits survive and grow in a chaotic environment that highlights a potential alternative: dissipative structures.
The idea is that sometimes the most efficient way for energy to be dissipated is through organized structures. Think of the whirlpool in a bathtub. No one designed it and it doesn’t require any extra energy to be maintained. If you disturb it, it gravitates naturally back to its previous form. This describes the ideal habit — it saves you energy not by preserving some conserved quantity of willpower, but by “sucking” disorderliness from the environment into a stable structure. The more chaos, the more order is created, like crystalline diamonds being formed under intense heat and pressure.
Lastly, disturbance propagation seems to imply that change only comes from outside. Which paradoxically would confirm behaviorism. But computer modeling of emergent systems in recent years has led to a firm conclusion, as described in Growing Artificial Societies: “Bottom-up models suggest that certain cataclysmic events — like extinctions — can be brought on endogenously, without external shocks (like meteor impacts) through local interactions alone.”
We can rest assured that there will be no shortage of “disturbances” to propagate. They can come from the outside (life events), from inside (mid-life crises), and can possibly even be created on purpose, like Josh Waitzkin’s “self-created earthquakes” that high performers call upon for creative inspiration.
Self-efficacy and self-compassion: two sides of the same coin
On the surface, it would seem that an experimental framework for individuals is just around the corner. The military has OODA, companies have lean, teams of programmers have agile and scrum. The trend seems to be going smaller and more decentralized, as the need to innovate diffuses from C-suites to living rooms. But translating such a tool to the level of a single individual is an altogether different challenge.
The main reason is that most people’s risk tolerance is very low, because self-efficacy (defined as “a person’s conviction or confidence about his or her abilities to mobilize the motivation, cognitive resources or courses of action needed to successfully execute a specific task within a given context”) is remarkably fragile. When it comes to trying and learning new things, people have difficulty transferring success in one arena to even highly related ones. Even small failures lead to learned helplessness so quickly, we learn to protect against that eventuality by not trying new things unless success is guaranteed.
The primary risk of entrepreneurship and other free agent lifestyles is not financial or even social — it is the risk to a person’s very self-concept as someone who does what they set out to do. In entrepreneurial endeavors that depend just as much on luck and timing as intelligence and hard work, this feels like a terrible gamble. And it’s a gamble with odds you can’t improve through careful preparation and planning (in fact, too much planning will probably worsen your odds). This sort of risk is not any less threatening in cultures that are relatively tolerant of failure, like the U.S. If anything, it’s more threatening, since these cultures also tend to be more individualistic, with your actions reflecting more directly on your character and abilities. Stripped of risk-mitigating social structures, we are faced with the terror of total accountability. The correlation between individualism and suicide rates in developed countries speaks to the risks of this sort of attribution.
One of the necessary aspects of any experimentation framework is the atomization and disposability of the things being tested. The Scientific Method doesn’t work if hypotheses can’t be discarded. Testing different business models won’t do any good if you’re unwilling to pivot. The reason lifestyle experimentation is so risky for individuals is that, unlike a company or a product, you can’t just fail fast, walk away, and try again. There is no exit — this is your life. Self-concepts are not disposable.
As we reach new speeds of technological and social change, people’s self-concepts and routines get disrupted simultaneously. Adapting to these changes requires more fundamental behavior change. But as we’ve learned through painful experience, traditional approaches to learning such as formal education are simply ineffective at changing behaviors at this level.
Take the example of health. This 2012 AON Hewitt healthcare survey reports that 80% of healthcare costs are accounted for by 15 conditions, which are driven by just 8 risks that are at least partly behavior-dependent (poor diet, physical inactivity, smoking, lack of health screening, poor stress management, poor standard of care, insufficient sleep, excessive alcohol consumption). We pour billions into education and training, but ignore the fact that most health-related decisions are based on habits, intuitive response or assessment, self concept, or heuristics, not rational cost-benefit analyses. When smokers are shown anti-smoking videos, it just triggers their urge to smoke. Adding healthy items to a fast food menu makes sales of hamburgers skyrocket.
What we need if we want to change behavior at this fundamental level is to replace predictive models of behavior change — do this and you’ll get that — with exploratory models. The purpose of an exploratory model would be to guide further inquiry, to help formulate relevant questions, and to identify repeating patterns, whether in the form of habits or new, more helpful self-concepts. Just as importantly, it would also look backward, helping people tell new stories and reinterpret old ones.
Stories may actually be a more accurate way of describing how people think about and use mental models of behavior change. Stories, like emergent systems, only move in one direction. They cannot be rolled back and played again. This irreproducibility suggests the importance of another form of psychological capital that is also highly correlated with successful behavior change: self-compassion. They are two sides to the same coin — you need self-efficacy to believe you can do it, but you equally need self-compassion to be ok when you don’t. Self-compassion aids change by removing the veil of shame and pain that keeps you from examining the causes of your mistakes (and often, leads you to indulge in the very same bad habit as a way of forgetting the pain). Self-forgiveness is the first step in fostering an invitational attitude that is open to feedback, from yourself and others.
There is something about the turning of this coin — between efficacy and compassion — that I believe lies at the heart of the experimentation framework I’m envisioning. And the more I think about it, the more I suspect compassion is the far more radical and important side.