What does this equation mean?

When the two error costs are equal, the cost ratio on the right collapses to one, and the rule reduces to the ordinary Bayesian classifier that a norm blind to consequences would prescribe — the classical, “unbiased” threshold heuristics-and-biases research uses as its comparison point. Error management theory’s substantive claim is that this special case almost never held over evolutionary history for fitness-relevant judgments: whenever $c_{miss}$ exceeds $c_{fa}$ , the right-hand side falls below one, and the cost-minimizing rule requires inferring H even when the raw likelihood ratio is unfavorable to it, which looks, from the classical vantage point, exactly like a bias. Haselton and Buss’s original 2000 paper tested a specific instance of this logic in cross-sex judgment: because a missed mating opportunity was plausibly costlier to male reproductive fitness across evolutionary history than the cost of a wrongly pursued disinterested partner, men should overperceive women’s sexual interest relative to women’s own self-reports, and the theory further predicts the converse asymmetry for a different judgment — because underestimating a partner’s commitment is the more costly error for a woman deciding whether to continue investing in a relationship, women should systematically underestimate men’s commitment. Two studies, with two hundred and seventeen and two hundred and eighty-nine participants respectively, documented the predicted commitment-underperception bias in women and replicated the sexual-overperception bias in men, including a condition in which the male bias was measurably reduced [ 10 ] . Dominic Johnson, Daniel Blumstein, James Fowler and Haselton’s 2013 review in Trends in Ecology & Evolution generalizes the same logic well beyond mate perception, arguing that biases across many species and decision domains should be understood as “design features rather than defects,” calibrated by the relative fitness effects of the two error types rather than by any organism’s access to an unbiased estimator [ 12 ] .