A simple behavioral evaluation test of human face identity recognition with natural images validated with the case of prosopagnosia PS

Abstract

Abstract Behavioral evaluation of face identity recognition (FIR) ability has been performed with images for decades but remains subject to challenges regarding the stimuli and tasks used as well as the measured variables. Here we validate two simple nonverbal tests of FIR with ambient natural face images in neurotypical individuals and the well-documented brain-damaged case of prosopagnosia PS. In the first task, participants must select a famous identity among 3 simultaneously presented pictures. While neurotypical individuals scored almost at ceiling, PS was at chance level. In contrast, she was flawless at the same task with written names. The second task requires matching different views of familiar or unfamiliar faces for their identity against one distractor face, at upright or inverted orientation. While performing well above chance level, PS was significantly impaired and slowed down compared to neurotypical controls. Moreover, unlike controls, she showed no advantage for familiar or upright faces. Altogether, these simple behavioral FIR tests present with numerous advantages in terms of validity (i.e., natural images, lack of response biases, familiarity and inversion effects) and sensitivity (i.e., high contrast between patients and controls), making them particularly advantageous to rapidly evaluate the integrity of this function in normal and pathological populations.