Study tests 20-question screen for developmental prosopagnosia

A colleague’s face in a crowd is usually a shortcut to recognition. For people with developmental prosopagnosia, that shortcut can fail for life, even when vision is normal and there is no brain injury. A team led by Masaki Mori at the Japan Advanced Institute of Science and Technology evaluated the widely used questionnaire to screen for the condition, testing its 20 questions against responses from 599 Japanese adults.

The PI20 asks people to rate statements about everyday experiences: recognizing familiar people out of context, or remembering a face after meeting someone. Mori and former graduate student Midori Sugiyama of Keio University examined the questionnaire in two ways. Classical test theory checked whether it measured consistently; item response theory went further, testing how well each question distinguished different levels of face-recognition difficulty.

Most questions performed as intended. Their response patterns shifted step by step as the measured level of difficulty changed, with less overlap between groups. The analyses found the PI20 had excellent reliability and validity, supporting its use as a screening tool for a condition that often goes unidentified.

The study also showed where the questionnaire can improve. One question asks about recognizing people with distinctive facial features, which may be easier for most people and therefore less revealing of developmental prosopagnosia. Another concerns recognizing one’s own face in photographs, an experience that many people with the condition may not find especially difficult.

And so what, concretely? The PI20 is a screening tool for developmental prosopagnosia, while the item-level analysis points to specific questions that should be revised rather than treating the questionnaire as finished. Mori’s team has released the dataset and analysis code, creating a public basis for testing future versions across languages and cultures. The findings strengthen a screening instrument; they do not turn a self-report questionnaire into a treatment or a new clinical intervention.