Cognitive Bias and Its Impact on Women’s Advancement in Tech #2
In this edition of the Cognitive Bias series, ACM-w-volunteer, Yildiz , reflects on the default prototypes of software engineers in computing, their effects on women, and what we can do to be more inclusive.
Fitting into the box
Picture a technical meeting where a software engineer interrupts a colleague: “Actually, that won’t scale. We need to refactor the architecture completely.” The engineer speaks confidently without softening their criticism.
Take a moment. Did you picture a man or a woman?
Now consider: How would you rate this person’s likability? Would your assessment change if you knew it was Sarah versus Steve?
Research suggests that many people unconsciously imagine this engineer as male. When told that the engineer is named Sarah rather than Steve, study participants tend to rate the same behavior differently, even though both engineers are equally committed to contributing to the project’s success. This illustrates prototypicality bias: the tendency to unconsciously compare individuals to a mental “prototype” of a category member.
What is Prototypicality Bias?
In computing, the prototype is male, young, technically obsessive, and socially awkward. Cheryan et al. found that both men and women associate computer science with masculine stereotypes, and these stereotypes actively deter women’s interest—even when women perform equally well in computer science courses [1].
The prototype extends beyond gender to personality traits. When researchers asked students to describe computer scientists, they consistently mentioned traits like “obsessed with computers,” “lacking interpersonal skills,” and “spending hours alone coding”—traits that students, particularly women, found unappealing and alienating [1]. Women displaying the assertiveness expected in tech leadership violate both gender norms and this “quiet geek” stereotype, facing a double bind where they’re penalized for being either too aggressive or too soft.
The Cognitive Load Tax: A Shared Challenge
The cognitive load tax refers to the mental resources anyone expends when navigating environments where they feel they don’t match the expected prototype. Spencer, Steele, and Quinn demonstrated this in controlled conditions: when women were told a difficult math test showed gender differences, their performance decreased (solving 4.65 problems correctly versus 6.97 in neutral conditions), while men’s performance remained stable at about 9 problems. Importantly, this gap disappeared entirely when the same test was described as gender-fair [2].
The tax manifests in everyday workplace interactions. Women in male-dominated STEM settings show increased physiological vigilance—literally scanning their environment more frequently for threat cues—and decreased sense of belonging compared to gender-balanced settings [4]. Every meeting where a woman calculates whether to speak assertively, every code review where she wonders if criticism is bias-driven, every presentation where she over-prepares to avoid confirming stereotypes—all represent cognitive resources diverted from actual work.
Evidence-Based Solutions
Environmental Changes
Small environmental adjustments can make spaces more welcoming to diverse talent. Heryan et al.’s research found that neutral decorations in computer science classrooms (nature posters, general magazines) compared to stereotypical ones (Star Trek posters, video games) increased women’s sense of belonging from 35% to 68%, while men’s comfort remained high in both settings [5].
Similarly, Gaucher, Friesen, and Kay’s analysis of over 4,000 job advertisements found that neutral language attracted more diverse candidates without deterring traditional applicants [6]. Companies using collaborative language (“work together,” “support team goals”) attracted both men and women equally, expanding their talent pool.
Institutional Interventions
GitHub researchers analyzing over 3 million pull requests found interesting patterns: when gender wasn’t identifiable, women’s code was accepted at 78.7%, and men’s at 74.6%. When gender was visible, these rates shifted [7]. This suggests that blind review helps evaluate work on its merits alone—benefiting anyone whose work might otherwise be subject to unconscious bias.
Professional development around bias can help everyone make better decisions. Carnes et al. tested a workshop that helped faculty recognize and address unconscious patterns. Departments receiving the training hired more diversely (32% women) compared to control departments (18%), while maintaining or improving overall hiring quality [8]. The training helped decision-makers focus on qualifications rather than prototype-matching.
Individual Strategies
Research has identified strategies that help anyone perform at their best. Miyake et al. found that brief values-affirmation exercises—having students write about their personal values for 15 minutes—helped students under stereotype threat perform better, with women in physics earning B grades versus C grades in control conditions, while non-threatened students’ grades remained stable [9].
These techniques work because they remind people of their multifaceted identities and values beyond any single stereotype. Organizations increasingly recognize that when employees can bring their whole selves to work, everyone benefits from diverse perspectives and approaches.
Amanatullah and Morris’s research on negotiation suggests that framing assertive behavior as benefiting the team helps everyone navigate workplace dynamics more effectively [10]. When actions are positioned as collaborative rather than self-serving, they’re received more positively regardless of who presents them.
Moving Forward
The cognitive load tax represents a measurable drain on productivity and innovation. Every moment a woman spends managing stereotypes is a moment not spent solving problems. The irony is stark: in trying to prove they belong in tech, women’s performance suffers—not from lack of ability, but from the burden of disproving biases.
The solutions are clear and evidence-based. Neutral environments, blind evaluations, and brief psychological interventions all show significant effects. The question isn’t whether we can fix prototypicality bias, but whether we will. Until Sarah’s and Steve’s interruptions are judged equally, until competence isn’t surprising in certain bodies, organizations will continue losing talent and innovation to an entirely preventable tax.