Seeing Interest That Isn’t There: Four Routes to Sexual Overperception
The same social cues can be interpreted as friendliness or sexual interest (Photo by James X on Unsplash)
– by Iliana Samara
Someone holds your gaze, laughs at your jokes, and suggests continuing the conversation over a drink. You might interpret their behavior as sexual interest, but they are only enjoying your company. Research on sexual overperception has often shown this pattern: men tend to infer more sexual interest from ambiguous behavior than women do (Abbey, 1982; Haselton & Buss, 2000). However, the psychological process underlying this pattern remains less well understood.
In a recent article in Evolution and Human Behavior (Samara, 2026), I address this issue and argue that the measures we commonly use to measure this pattern do not allow us to identify the underlying process. Work on sexual overperception measures how interested we think a person is in us (e.g., with rating scales) or how often one mistakes an interaction for flirtatious or romantic. However, patterns of higher interest ratings or higher frequency of mistaken interactions can arise through different psychological processes. Therefore, using the term “overperception” does not capture which process produced it.
Error management theory (EMT; Haselton & Buss, 2000) provides one influential evolutionary account for this pattern. EMT posits that when the consequences of two errors differ, selection may favour a decision policy that reduces the more costly mistake, even if this increases the other type of error. Applied to sexual interest, the notion is that ancestral men may have faced greater reproductive costs from missing a mating opportunity than from making an unsuccessful advance. Under these conditions, a tendency to infer interest from relatively weak evidence could have been advantageous for men.
We can identify four different routes that lead to this error. The first is cue discrimination: how well someone distinguishes friendliness accompanied by sexual interest from friendliness alone. If the cues are difficult to tell apart, errors can arise in both directions (inferring interest when it was not present or missing interest when it was present). The second concerns expectations: someone who starts with the expectation that sexual interest is likely may require less evidence before judging it as such. Such prior beliefs are dependent on the individual and might for example reflect previous experiences or beliefs about how likely sexual interest is in a given context. The third route concerns the perceived consequences of being wrong. Two people could have similar expectations and read the cues equally well but require different levels of evidence before indicating that another person is interested in them. For example, someone especially concerned about missing genuine interest may use a lower threshold than someone especially concerned about mistakenly inferring interest. Finally, the fourth route concerns the answer someone reports may differ from their private judgment, their reporting policy. A difference in reported judgments could therefore partly reflect differences in what people feel comfortable disclosing.
These processes can also occur together. Signal detection theory helps by separately estimating discrimination and the decision threshold (see Brandner et al., 2021, for an application to sexual overperception). However, even after estimating that threshold, we still need to explain how it arises, as a lower threshold could reflect stronger expectations of interest, greater concern about overlooking interest, or differences in reporting. Bayesian decision theory makes the contribution of expectations and error costs clear (Green & Swets, 1966).
To examine how well existing designs distinguish between these possibilities, I conducted a targeted audit of 54 empirical papers. Only five reported signal detection measures. None directly elicited expectations as estimates of how often interest was present, and none experimentally manipulated the payoffs attached to different errors. Many of these studies addressed other questions, including when overperception occurs and which factors predict it. Overall, the audit showed that their designs generally leave the underlying processes difficult to separate.
Future work could address this point by for instance, varying how often interest is present in a task, asking participants what prevalence they expect, and testing whether their beliefs change. Studies could also vary who will see an answer while keeping the available cues the same. Each manipulation would help test a different explanation, given it changes the beliefs or incentives it is intended to change.
The distinction between mechanisms can inform future interventions. If someone struggles to distinguish relevant cues, practice with feedback might help. If their expectations are poorly calibrated, feedback about how often interest is reciprocated may be more relevant. If perceived consequences or disclosure conditions drive the pattern, those would need separate attention. These are different intervention hypotheses, and the observed error pattern alone cannot show us which to pursue. For evolutionary research, identifying these processes would clarify how a proposed function is implemented and why behavior changes across situations.
References
Abbey, A. (1982). Sex differences in attributions for friendly behavior: Do males misperceive females’ friendliness? Journal of Personality and Social Psychology, 42(5), 830–838. https://doi.org/10.1037/0022-3514.42.5.830
Brandner, J. L., Pohlman, J., & Brase, G. L. (2021). On hits and being hit on: Error management theory, signal detection theory, and the male sexual overperception bias. Evolution and Human Behavior, 42(4), 331–342. https://doi.org/10.1016/j.evolhumbehav.2021.01.002
Green, D. M., & Swets, J. A. (1966). Signal detection theory and psychophysics (1st ed.). Wiley.
Haselton, M. G., & Buss, D. M. (2000). Error management theory: A new perspective on biases in cross-sex mind reading. Journal of Personality and Social Psychology, 78(1), 81–91. https://doi.org/10.1037/0022-3514.78.1.81
