Range-frequency effects can explain and eliminate prevalence-induced concept change.

Why would concepts seem to grow when their instances become rare? Human observers can respond to decreases in stimulus prevalence by expanding their conceptual boundaries of those stimuli. This prevalence-induced concept change may have serious social consequences, since many real-world detection tasks demand consistent judgments over time. The current work aims to identify the computational process that produces prevalence-induced concept change. I review some plausible models from the cognitive and social sciences that could account for this phenomenon, and then use trial-level computational modeling to see how well each model predicts actual human data, finding that they are best explained as a range-frequency compromise in judgment. Finally, I test an intervention that successfully eliminates prevalence-induced concept change by making stimuli more intense as they become rare.