What neuroscience and psychology actually reveal about challenge, recovery, and the limits of the “hardship creates greatness” narrative.
What the science of desirable difficulties, the steeling effect, and allostatic load actually says about hardship and capacity — and where the popular narrative overstates it.
Startup culture has a foundational myth: that the founders who make it are the ones who had it hardest — that scarcity, instability, or a childhood without a safety net forged something their more comfortable peers never developed. The myth borrows its credibility from real science, but it does so by fusing together two distinct research traditions that describe different mechanisms entirely. One is “desirable difficulties,” a concept from cognitive psychology explaining why effortful practice produces deeper learning than easy practice. The other is the “steeling effect,” a developmental psychology finding about how manageable adversity can build coping capacity for future stress. Neither claims that hardship is a shortcut to achievement, and neither extends to severe or uncontrollable adversity. The neuroscience of chronic stress, in fact, points the other way.
This essay separates three literatures — cognitive science, developmental psychology, and neurobiology — and takes each in turn.
Desirable Difficulties: A Finding About Learning, Not About Suffering
The term “desirable difficulties” originates with the UCLA psychologist Robert Bjork, who coined it in 1994 to describe a specific and well-replicated phenomenon in memory research. It has no connection to childhood trauma or entrepreneurial survival stories; Bjork was studying how people acquire and retain information under different practice conditions.
His central finding, confirmed across decades of subsequent work, is that the conditions which make learning feel easiest in the moment are frequently the conditions that produce the weakest long-term retention. Massed practice — studying one topic in one sitting, in a consistent format, until it feels mastered — generates a sense of fluency that is often illusory. The practice conditions that produce durable learning tend to feel more effortful while they’re happening: spacing sessions apart in time rather than concentrating them, interleaving related topics rather than blocking them individually, and retrieving information from memory rather than passively re-reading it. Bjork called these difficulties “desirable” precisely because they are difficulties a learner can overcome through effort, and because that effort is the mechanism by which memory consolidates more durably.
Consider a founder learning distributed systems. Re-reading documentation for six hours produces a comforting sense that the concepts are clicking into place. Building a small cluster, deliberately breaking it, and diagnosing the failures from memory over several weeks feels slower and considerably more frustrating. But it is the second approach — not the first — that produces someone capable of diagnosing a production outage at 2 a.m. without the documentation open in another tab.
This is a genuine, well-supported claim about the architecture of skill acquisition. It says something directly useful to anyone learning to code, read a balance sheet, or negotiate a term sheet: effortful, spaced, self-tested practice will likely outperform smooth, repetitive study. What it does not say is that job loss, poverty, or a chaotic childhood operates by the same mechanism. Bjork’s research concerns deliberate, structured learning tasks — not the experience of surviving hardship. Popular writers, Malcolm Gladwell among them in David and Goliath, have since extended the phrase to cover life disadvantages more broadly. That extension makes for a compelling narrative, but it is worth being precise that it is an extrapolation from Bjork’s work, not a finding within it.
The Steeling Effect: A More Cautious Story Than It Sounds
If desirable difficulties don’t apply to lived hardship, does adversity itself build resilience? Here the developmental literature offers something genuinely interesting — but considerably more conditional than the founder-hardship narrative suggests.
Beginning with work by Michael Rutter in the 1980s and refined in subsequent decades, developmental psychologists have described a phenomenon known as the steeling effect, or stress inoculation. The proposal is that a moderate, manageable degree of adversity — stress that a person can face, work through, and recover from, typically with some form of support available — can leave that person better equipped to handle future stress. This pattern has appeared in both animal models and longitudinal human research: individuals who encounter some early difficulty and successfully navigate it sometimes show better coping capacity later than individuals who encountered almost none.
The relationship, however, is explicitly not linear. Rutter and later researchers describe it as curvilinear — closer to an inverted U than a ramp, with both the complete absence of adversity and heavy, sustained adversity associated with worse outcomes than a moderate, bounded amount in between.
This is the point at which the developmental literature directly contradicts the founder-hardship narrative. Outcomes in this body of work vary considerably by individual, by the type of adversity, and by whether support was available at the time it occurred, and trauma researchers are careful to note that many people who experience severe or chronic adversity do not emerge with enhanced coping ability — they emerge with worse mental and physical health outcomes. The “what doesn’t kill you makes you stronger” framing describes a real pattern under specific conditions, but it is far from a guarantee, and treating it as one risks minimizing the actual cost of what people who live through severe hardship experience.
What the Stress Response Is Actually Doing
The neurobiology of stress is often summarized as an “alarm system,” which is a reasonable shorthand but obscures the distinction that matters here. When the brain registers a demanding situation, a signaling cascade — running through the hypothalamic-pituitary-adrenal (HPA) axis — triggers the release of hormones, cortisol among them, that mobilize metabolic energy and sharpen short-term attentional focus. This is adaptive activation: a person facing a genuine, time-limited demand benefits from it.
Acute stress of this kind helps direct physiological resources toward an immediate challenge. Chronic activation is a different phenomenon, not a more intense version of the same one. The neuroscientist Bruce McEwen, in an influential 2007 review, described the cumulative physiological cost of repeated or unrelenting stress-response activation as “allostatic load” — the wear that accrues when the body’s stress systems are engaged without adequate recovery between episodes. In animal and human studies, this cost has been associated with structural and functional changes in the hippocampus, and McEwen’s own framework treats allostatic load as exactly what the term implies: a cost. Not a performance upgrade, and not a metaphor — a quantifiable biological toll.
This is where the cognitive-science and neurobiological literatures actually converge: adaptation depends on oscillation between challenge and recovery. Whether the system under discussion is a muscle, a memory trace, or an HPA axis, adaptation occurs during the recovery interval, not during continuous load. A founder who never allows the pressure to let up is not accumulating desirable difficulties. They are accumulating allostatic load.
That distinction has direct consequences for how this research gets applied to founder behavior. Someone who has, through a bounded and manageable difficulty, learned to regulate a specific kind of pressure may well carry a genuine advantage into that specific domain. But someone carrying ongoing chronic stress — financial precarity, unresolved trauma, unstable housing — is, according to the same body of research, more likely carrying allostatic load that impairs functioning over time, not a hidden reserve of composure. At the level of the HPA axis, both are initially processed in similar ways: the system does not pre-sort adversity into “narratively useful” and “merely depleting” before it responds.
What Actually Follows From This Research
Three implications follow from this distinction.
The first concerns how practice should be structured, and it is the piece of this research with the most direct application regardless of anyone’s personal history. Spacing practice on a difficult skill, testing recall instead of re-reading, and interleaving related problems instead of drilling one type repeatedly are all empirically supported methods for building durable competence more efficiently. Someone learning to model financial projections, debug a system, or run a negotiation will likely get more from effortful, spaced, self-tested practice than from passive repetition — independent of background.
The second concerns the kind of stress worth seeking out. The steeling-effect literature does suggest a meaningful difference between challenges never encountered and challenges encountered, worked through, and recovered from. That points toward something specific and achievable: deliberately taking on problems that are difficult but survivable, with adequate support and recovery built in, rather than either avoiding discomfort entirely or treating constant crisis as evidence of capability.
The third concerns how chronic hardship should be interpreted. If someone is experiencing genuine, ongoing adversity — financial strain, health problems, an unstable environment — the allostatic-load research suggests this is more likely a drain on cognitive and physiological resources than a hidden advantage. That distinction matters both for how people interpret their own histories and for how hardship gets used, often uncritically, as a heuristic in hiring or mentorship.
A final point is less about mechanism than about inference. For every well-known founder whose difficult early circumstances are cited as the source of their advantage, there are considerably more people who faced comparable or worse circumstances and did not go on to found anything — in some cases precisely because the adversity did what chronic adversity often does, and made things harder rather than easier. Visible success stories are a small, non-random slice of everyone who lived through similar conditions, which is reason enough to be cautious about drawing general lessons from them.
Conclusion
Bjork’s desirable difficulties describe how effortful, well-structured practice produces durable learning. The steeling effect describes how bounded, recoverable stress can build coping capacity, with the same evidence cautioning against extending that logic to severe or chronic adversity. And the neuroscience of stress, including McEwen’s own foundational model, is largely an account of the cost of a stress response left running without recovery, not a case for seeking that state out.
What connects the three is a single mechanism, not a coincidence of terminology. A memory consolidates in the gap after retrieval, not during it. A person’s coping capacity is strengthened through successfully processing and recovering from a manageable stressor, not simply through exposure to its acute phase. The body restores physiological balance during recovery after activation, not while cortisol is still circulating. Learning systems, developmental stress adaptation, and physiological recovery are three different substrates running the same underlying rule. Capacity is built in the interval between a demand and its resolution — never in the demand alone, and never in its absence.
Read this way, the founder-hardship narrative doesn’t just overstate the evidence. It locates the source of adaptation in the wrong place. Hardship is not the engineer of resilience. Recovery is.
Further Reading
- Bjork, R. A. (1994). Memory and Metamemory Considerations in the Training of Human Beings.
- Bjork, R. A. & Bjork, E. L. (2011). Making Things Hard on Yourself, but in a Good Way: Creating Desirable Difficulties to Enhance Learning.
- McEwen, B. S. & Stellar, E. (1993). Stress and the Individual: Mechanisms Leading to Disease. Archives of Internal Medicine.
- McEwen, B. S. (2007). Physiology and Neurobiology of Stress and Adaptation: Central Role of the Brain. Physiological Reviews.
- Rutter, M. (1987). Psychosocial Resilience and Protective Mechanisms. American Journal of Orthopsychiatry.
- Rutter, M. Research on stress inoculation and the “steeling effect” in resilience.
