Two leadership archetypes dominate the modern workplace, and neither one works the way people think it does.
The first archetype yells. Berates in public. Takes credit for other people’s ideas. Throws tantrums that clear conference rooms. Demands devotion in exchange for proximity to greatness. The second listens. Admits mistakes. Asks for input. Shares decision-making. Treats employees like thinking adults whose opinions matter.
Management consultants sell the second approach. Bestselling authors make careers explaining why empathy wins. The research appears to support them.
But we should ask what kind of research.
Most of the data on psychological safety and performance comes from self-reported surveys. Employees are asked whether they feel safe speaking up. They are asked whether they believe their team is effective. The correlation between these two self-reports becomes evidence that safety produces performance.
When researchers compare what people say they do to what they actually do, the gap ranges from embarrassing to catastrophic. Americans say 22% attend church weekly. Cell phone location data shows 5%. Across 55 countries, voters overreport turnout by margins that would swing elections. The phenomenon has a name: social desirability bias. People answer surveys with who they believe themselves to be rather than what they actually did.
So when employees report high psychological safety and also report high team effectiveness, we may be measuring something real. Or we may be measuring people who feel good about their workplace telling researchers they feel good about their workplace. The causal arrow remains genuinely unclear.
The behavioral data tells a messier story. Companies with tyrannical founders sometimes outperform their empathetic competitors for years before the dysfunction catches up to them. And companies with collaborative cultures sometimes drift into stagnation that only shows up in missed market opportunities, which are harder to measure than employee satisfaction surveys.
Meanwhile, something strange keeps happening in the real world.
The employees who work for tyrants describe their experience with a confusing mixture of resentment and awe. They tell war stories about surviving their boss’s wrath the way soldiers describe combat: terrible, meaningful, bonding. And the employees who work for empathetic leaders describe their experience with warmth and appreciation while somehow also seeming less galvanized, less certain, less willing to run through walls.
The research on this paradox goes back decades.
Three Tyrants and Their Believers
Travis Kalanick built Uber by cultivating chaos. His leadership style emphasized pushing boundaries, stifling internal dissent, and rewarding results above all else. He encouraged what he called a “warrior mentality,” comparing running Uber to being in battle. His infamous remarks to regulators made clear that rules existed for other people.
In 2017, Uber’s culture collapsed into public view. Susan Fowler, a former engineer, published a blog post describing systemic harassment and HR departments that protected high performers regardless of their conduct. Female representation at the company had plummeted from 25% to just 3%. An internal investigation led by former Attorney General Eric Holder confirmed what whistleblowers had been saying: Uber’s leadership had enabled a toxic environment.
And yet. Employee testimonials after Kalanick’s forced resignation praised his ability to connect with people in ways that made them want to follow him. Many employees were incredibly inspired and personally touched by Kalanick’s leadership. The social media response to his ousting was not unanimous relief. It was, in many cases, mourning.
Elizabeth Holmes operated Theranos with similar intensity and similar results. At one company event, she told employees she was “building a religion” and that if any among them didn’t believe, they should leave. At another, she claimed Theranos’ technology was “the most important thing humanity has ever built” and that everyone needed to work as hard as humanly possible to deliver it.
Her technology never worked. The blood tests were fraudulent. Patients received incorrect diagnoses for conditions including diabetes and cancer. Holmes is now serving an eleven-year prison sentence.
Employees were particularly susceptible to her charms. Company whistleblower Tyler Shultz recalled, “Elizabeth had this way of locking you in, and when she was talking to you, she made you feel like you were the most important person in her world right now and you were so important to achieving this vision that you were dedicated to.” Even after discovering the fraud, employees struggled to reconcile what they had witnessed with the person they had believed in.
Lower-level employees at Theranos described working at the organization as being involved in a “South American dictatorship or drug cartel.” They were fearful of questioning upper management. They signed non-disclosure agreements. They watched colleagues get fired for raising concerns. And many of them, even now, speak of their time there with something resembling nostalgia for the mission.
This is not rational behavior. But it is predictable behavior, and understanding why requires going back further than any HR textbook.
The 200,000 Year Old Problem
If you’ve read my previous pieces, you know I keep circling back to the same unsettling truth: we’re running modern lives on ancient hardware.
Homo sapiens spent 95% of our existence in small nomadic bands where the wrong leader meant death and the right leader meant survival. The neural circuits that helped our ancestors evaluate tribal chiefs didn’t disappear when we invented corporations. They just kept running in the background, responding to modern stimuli with stone-age logic.
The brain hasn’t updated. The circuits running your response to your CEO are the same circuits your ancestors used to respond to tribal leaders on the African savanna.
Evolutionary leadership theory, developed by researchers including Mark van Vugt at VU University Amsterdam, proposes that humans evolved what he calls “cognitive ancestral leader prototypes.” These are mental templates, built into our psychology through natural selection, that automatically activate when we evaluate potential leaders.
In times of conflict, the prototype favors a physically strong, risk-taking individual willing to dominate. In peacetime, the prototype shifts toward someone with social skills and relationship-building capacity. These prototypes developed when following the wrong leader could get you killed, when survival depended on quick judgments about who could protect the group.
The modern workplace triggers these same circuits without the same survival stakes.
A study spanning more than 140,000 participants across 69 countries found that under conditions of economic uncertainty, people escalate their support for dominant leaders. This effect was mediated by participants’ psychological sense of lack of personal control.
When you feel uncertain about your job, your industry, your company’s future, your brain interprets this as a survival threat. And survival threats activate the ancient preference for the dominant leader who projects certainty, even when that certainty is disconnected from competence.
The narcissistic CEO keeps getting selected despite all the evidence about long-term organizational damage. The selection happens at a level below conscious reasoning. The board members who choose the confident narcissist over the humble expert are not being foolish. They are being human, running cognitive software optimized for a world that no longer exists.
The preference is not for men. The preference is for dominance.
When researchers ask people to design their ideal leader by selecting traits from a list, both men and women view assertiveness and competence as necessities while treating warmth and collaboration as luxuries. The evolutionary preference for the alpha persists regardless of who embodies it.
63% of women and 71% of men say they prefer male bosses. They are not preferring maleness. They are preferring the dominant traits culturally associated with male leadership. And when women reach leadership positions, they do not rewrite the template. They adopt it.
Researchers at Yale and MIT studied promotion decisions at a large retail chain and expected female managers to evaluate female employees more fairly. The opposite happened. Female managers gave lower potential ratings to their high-performing female subordinates, just as male managers did. Assigning women to be managed by other women does not solve the promotion gap. The gatekeeping operates independent of the gatekeeper’s gender.
Women in leadership positions are no more likely than men to sponsor, mentor, or advocate for junior women’s advancement. The evolutionary architecture that favors dominant leadership and penalizes communal traits runs in women’s brains too. The expectation that female leaders will automatically support other women assumes gender solidarity trumps the ancient circuitry. It does not.
The successful female leaders who break through often display the same dominance architecture as their male counterparts. Research on Queen Bee syndrome describes women in leadership who are less overtly confrontational than domineering male leaders but bullying just the same. They distance themselves from other women. They succeed by fitting the existing template rather than rewriting it.
The double bind facing women leaders makes this dynamic visible. Women who conform too closely to feminine stereotypes are perceived as lacking leadership qualities. Women who adopt masculine traits are viewed as lacking warmth and likability. The solution that works, according to decades of research, is displaying enough dominance to trigger the ancient leadership prototypes while adding just enough warmth to avoid backlash for violating gender norms. The successful female leader must perform alpha competently while also performing femininity competently. The male leader only has to perform one role.
When an organization claims to value empathetic, collaborative, feminine leadership styles, yet its most successful women display assertiveness and dominance, the stated preference and the revealed preference are telling different stories. The gender of the leader may change. The evolutionary expectations do not.
Physical traits still predict leader selection in ways that make no sense for knowledge work. Taller candidates win more elections. Deeper voices correlate with perceived leadership ability. Facial features associated with testosterone predict career advancement in the military. These traits predict who gets selected, not who leads effectively. But the ancient brain doesn’t distinguish between selection and effectiveness. It responds to dominance cues that would have mattered when the leader’s physical strength determined whether your tribe survived the next conflict.
The tyrannical boss, male or female, triggers circuits designed to respond to the tribal chief who could protect you from the neighboring tribe. The fact that your neighboring tribe is now a competitor’s sales team, and the relevant skills are strategic thinking rather than physical intimidation, hasn’t updated your neural architecture.
The Pairs
Narcissistic female CEOs actually outperform their male counterparts. The double bind may explain it: women who rise to the top while displaying narcissistic traits have been forced to calibrate those traits more carefully than men who face no such constraint. The bright sides of narcissism get expressed. The dark sides get moderated.
Consider the pairs.
The Demanding Performers
Jeff Bezos built Amazon into a dominant force while sending emails at 2am and expecting immediate responses. His “Day 1” philosophy kept the company in permanent startup mode. Former employees describe the culture as brutal, with frequent public criticism in meetings and a stack ranking system that pitted workers against each other. Amazon’s market capitalization under his leadership grew to over $1 trillion.
Lisa Su took over AMD when it was near bankruptcy, with a stock price around $3. She sends memos to executives late at night, often after midnight, followed by morning calls to discuss strategy. She demands chips that are 40% faster than previous generations and challenges her teams with ambitious goals. AMD’s stock price has increased nearly 50 times under her leadership. She was named TIME’s CEO of the Year in 2024. Her employee approval rating sits at 95%.
Both leaders are demanding. Both expect excellence. Both delivered exceptional results. The difference is that Su, operating under the double bind, had to balance dominance with enough warmth to avoid backlash. Bezos faced no such requirement.
The Cult Builders
Steve Jobs demanded devotion. He berated employees in public, took credit for others’ ideas, and created a reality distortion field that convinced people the impossible was possible. He was fired from his own company and returned to build it into the most valuable corporation in the world. Employees who survived describe the experience as the most meaningful work of their lives.
Elizabeth Holmes studied Jobs obsessively. She lowered her voice. She wore black turtlenecks. She demanded complete loyalty and punished dissent. She told employees they were building a religion. The cult of personality was indistinguishable from Jobs’s except for one detail: her technology never worked. She is now serving an eleven-year prison sentence.
Same archetype. Same dominance displays. Same employee devotion. Radically different outcomes. The narcissistic leadership style produces results when paired with genuine competence and vision. It produces fraud when the confidence is disconnected from capability.
The Autocrats
Travis Kalanick ran Uber with a warrior mentality. He stifled dissent, pushed boundaries with regulators, and created a culture where high performers were protected regardless of conduct. Female representation dropped from 25% to 3%. He was eventually forced out after a cascade of scandals, but employees still speak of his ability to inspire.
Marissa Mayer ran Yahoo with similar intensity. She banned remote work, micromanaged details down to fonts and colors, and made employees feel they were perpetually disappointing her. The culture became toxic. Talent fled. She was described as robotic and stuck up. The company was eventually sold to Verizon.
Both autocratic. Both created fear-based cultures. Both lost the confidence of employees and boards. The gender of the autocrat did not change the outcome.
The leadership archetype predicts results more than the gender of the leader does. Dominant women and dominant men create similar cultures, inspire similar devotion, and face similar consequences. The main difference is that women face more backlash for the same behaviors, which forces the ones who succeed to regulate more carefully.
This regulation may be why narcissistic female CEOs outperform their male counterparts. The double bind, which penalizes women for being too dominant and too soft, inadvertently trains successful women to calibrate their narcissism. Men face no such training. Their narcissism goes unmoderated.
The Hostage Layer
On top of this evolutionary foundation sits a more immediate psychological mechanism.
In 1973, four bank employees in Stockholm were held hostage for six days. When the ordeal ended, they defended their captor at trial. They raised money for his legal defense. They visited him in prison. Psychologists coined a term for what they had witnessed: Stockholm Syndrome.
The workplace version operates on the same mechanics.
Corporate Stockholm Syndrome can be defined as employees of a business beginning to identify with, and being deeply loyal to, an employer who mistreats them. The parallel works because the psychological conditions are similar. Your boss controls your income, your status, your professional reputation, your health insurance, and your family’s stability. When that control comes paired with intermittent reinforcement, a cycle of criticism and praise, punishment and reward, the human brain responds in ways that look irrational but follow predictable rules.
Because so much of our self-worth in modern times is defined and derived by work, we are at risk for experiencing Corporate Stockholm Syndrome when put into a certain work environment for long enough.
We select dominant leaders because of the evolutionary layer. We stay loyal to them because of the Stockholm layer. The two mechanisms reinforce each other. Your ancient brain chose the alpha. Your modern brain, trapped in economic dependency, rationalizes the choice.
The abusive boss creates something the empathetic boss struggles to replicate: a container for meaning. When work feels like survival, it also feels important. When your team bonds through shared suffering, those bonds carry weight. When you earn a tyrant’s rare approval, that approval means something precisely because it was so difficult to obtain.
Stanford’s Charles O’Reilly has spent years studying narcissistic CEOs. There must be 20 or 30 studies that demonstrate this: if you gather a group of strangers and give them a task, those who are more narcissistic are much more likely to be selected as leaders. They look like leaders. They sound like leaders. Grandiosity reads as confidence. Self-absorption reads as vision. Complete certainty reads as conviction.
In anxious times people are looking for a hero, a confident person who says, ‘I have a solution.’ Narcissists are the only ones who seem confident in such times because they are the only ones whose confidence is disconnected from evidence.
Employees can simultaneously resent their abusive boss and feel something like reverence. The resentment comes from the conscious experience of being mistreated. The reverence comes from circuits older than language, responding to dominance displays the way your ancestors responded to the alpha who might keep them alive.
The Humble Leader’s Dilemma
Satya Nadella took over Microsoft in 2014 and proceeded to demonstrate what empathetic leadership looks like at scale. He replaced the company’s internal competitiveness with what he called a “learn-it-all” culture. He placed psychological safety at the center of his management philosophy. He led with vulnerability, sharing how raising a son with cerebral palsy had reshaped his understanding of empathy.
A survey of nearly 13,000 Microsoft employees showed 92% CEO approval and 83% willingness to recommend Microsoft as an employer. The company’s market capitalization tripled. Microsoft became the second company in history to reach a $3 trillion valuation.
By every measurable standard, Nadella’s approach worked. Employees reported feeling valued. Innovation accelerated. Retention improved. The company transformed from a stagnating giant to an industry leader.
But something in the Nadella model points toward its own limitation.
When teams expected egalitarianism, having a humble leader increased knowledge and information sharing and helped those teams be more creative. But the same effect didn’t hold when teams accepted that power should be unevenly distributed. On teams where members expected leaders to be dominant and powerful, humble leaders were met with doubt and team members felt unsafe to speak up and take risks.
The humble leader made people feel less safe. Not because humility is bad, but because it violated expectations in ways that created uncertainty. People say they want collaborative leadership. When they get it, some part of them wonders who is actually in charge.
Radical Candor empowers everyone on the team to lead. Teams that embrace it flatten hierarchy around ideas and behavior. This is the selling point. This is also, paradoxically, the problem. When hierarchy flattens, clarity often flattens with it. The candid leader gives people input into decisions and, in doing so, removes something people didn’t know they needed: the psychological container of someone else being responsible for direction.
Your ancestral brain evolved to follow a clear leader because unclear leadership got people killed. The modern workplace rarely involves mortal stakes, but your neural circuits don’t know that. When your boss says “what do you think we should do?” part of you appreciates the respect. Another part, the older part, wonders why the leader isn’t leading.
The Exhaustion Equation
There is another cost to candor that rarely gets discussed, and it falls entirely on the leader.
Every genuine interaction requires emotional labor. Every admission of uncertainty, every solicitation of feedback, every moment of authentic engagement with an employee’s concerns draws from the same psychological reserve. Emotional labor has been shown to detract from well-being through emotional exhaustion and contribute to burnout.
Research on leadership vitality describes emotional dissonance, defined as the conflict between actual emotions experienced and organizationally or situationally required emotions. When the gap between who you naturally are and who your role requires you to be grows too large, the result is exhaustion.
For the narcissistic leader, there is no gap. The required self and the natural self are identical. Both want attention, credit, control. The performance of leadership costs nothing because it is not a performance.
For the empathetic leader, every interaction involves a choice about how to show up, how to regulate, how to balance honest reaction with constructive response. When employees adhere to organizational emotion norms and adjust their emotions accordingly, they may expend considerable resources. Excessive emotional demands lead to job burnout.
The candid leader must care enough to be honest, regulate enough to be kind, and repeat this process across hundreds of interactions per week, month after month, year after year. The unsustainability of this approach explains why many empathetic leaders eventually either burn out, become less available, or develop boundaries that employees interpret as distance.
Tim Cook runs Apple with a more collaborative model than his predecessor. It has reduced employee turnover and burnout. But Cook doesn’t inspire the same reverence. He runs a more sustainable operation that generates less mythology.
The Tradeoff
Neither archetype wins cleanly.
The tyrant delivers visible wins and creates a mythology people want to belong to. Short-term output goes up. Employees tell war stories for decades. But the tyrant burns through humans, creates cultures that can’t outlive founders, and optimizes for results at the cost of everything else.
The empath builds something durable with healthier people inside it. Long-term organizational health improves. Retention stabilizes. But nobody tells war stories about adequate work-life balance. The empath optimizes for sustainability at the cost of mythology and sometimes speed.
The problem is that these outcomes don’t carry equal weight in human psychology. People will tolerate suffering if they feel like they’re part of something. They’ll accept burnout if the mission feels important. They’ll stay loyal to someone who damages them if that person also makes them feel chosen.
The tyrant trades longevity for intensity. The empath trades intensity for longevity. Neither configuration is stable across all conditions.
What This Means
If you lead people, you are operating inside a contradiction that cannot be resolved, only managed.
Your employees want accessibility and decisiveness, humility and confidence, collaboration and clarity. These are not complementary traits. They are tensions that require constant calibration.
The tyrant’s path requires less emotional regulation. It also creates cultures that cannot outlive their founders, organizations that hemorrhage talent, and workplaces that damage the people inside them. Once you create these cultures, it’s very hard to change them. There are long-term consequences.
The candid path requires more emotional regulation. It creates sustainable organizations with healthier employees but also leaders who are at higher risk of burnout, and teams who sometimes interpret openness as lack of direction.
The employees who describe their abusive boss with a mixture of hatred and reverence are not confused. They are experiencing something real, a form of meaning-making that comes from surviving pressure and earning approval from someone who withholds it. The psychological machinery that produces loyalty through fear is the same machinery that produced tribal cohesion for 200,000 years.
The employees who love their transparent, collaborative boss but somehow don’t quite feel the same fire are also not confused. They have gained safety but lost a certain kind of story. They have a better workplace and a less interesting one.
Neither approach removes the fundamental tension between what people say they want from leaders and what they actually respond to when they get it. The mismatch between ancestral psychology and modern organizational life isn’t going away. Your employees’ brains are still scanning for the alpha who might save them from the neighboring tribe.
The boss who makes you cry might make you do your best work.
The boss who respects you might make you wonder if anyone is steering the ship.
Both things can be true. Neither is sustainable indefinitely.
The only path forward is understanding what you are trading when you choose an approach, and being honest about which costs you can afford to pay.