Figure 1.
Contrasting face and shoe recognition for frontal images.
(A) In experiments 1–4 participants were first presented with 4 target stimuli, and were trained during at least 16 feedback trials prior to the test. During feedback training, participants with congenital prosopagnosia (CP) on average made more mistakes during initial learning than controls only for faces but not for shoes (B, solid line for equality). Reaction times during later training trials with unlimited viewing were strongly influenced by participants' age (C, shown for faces only, solid lines represent linear model fits). Comparison of residuals, which account for age related differences, revealed longer reaction times for CPs compared to controls for face stimuli but not for shoe stimuli (D, RT, inverse transformation i.e.
/RT see Methods). In face and shoe recognition, the presentation time needed to achieve 80% correct recognition (PT
) increases with age (E shown for faces only, solid lines represent linear model fits). CP participants needed longer presentation times than controls in tests for face recognition (Exp. 1) but not for shoe recognition (Exp. 2). (F comparing residuals PT
). Group differences in mean reaction time for faces stimuli (G left boxplots, RTs centered around control mean) vanished after subtracting PT
presentation time ((G right boxplots, values centered around control mean): CP participants needed to inspect face stimuli longer than controls. Boxplot shows group distributions (whiskers:
CI); significance values according to a Wilcoxon rank sum test.
Figure 2.
Contrasting face and shoe recognition for rotated images.
In the recognition of rotated images, CP participants performed worse than controls for for faces (Exp. 3) but not for shoes (Exp. 4). This was tested for every rotation angle separately (A with significance values according to a Wilcoxon rank sum test). A comparison of individual residuals, which account for differences in the experimental setup, revealed that the deficit is selective for all CPs (B). Note that the magnitude of group differences observed for face recognition didn't change with rotation angle (Likelihood-ratio test, ).
Figure 3.
Face recognition under the constraint of limited presentation time.
(A) In experiment 5, participants were shortly presented with a target stimulus for 50ms, 150ms, 450ms or 750ms, and after a short blank had to recognize the target in a display of four face images. For all presentation times used in experiment 5, performance of CP participants differed from controls (B), while there was no significant difference in reaction times (C). However, while controls responded faster with increasing presentation time, there was no significant influence of presentation time on CP reaction times (Likelihood-ratio test, ). (D) In experiment 6, participants were repeatedly presented target faces, which - after a total of 24 feedback trials - had to be recognized in a 2-alternative forced choice paradigm. During the test, faces were presented for variable durations (50ms, 150ms, 450ms or 750ms). Independent of the duration, CP participants performed worse than controls (E), without significant differences in reaction times (F).
Table 1.
Description of CP participants.