Evidence for a Classical Dissociation between Face and Object Recognition in Developmental Prosopagnosia

Abstract

It is still a matter of debate whether developmental prosopagnosia is a disorder selective to faces or whether object recognition is also affected. In a previous study, based on a small sample of developmental prosopagnosics (DPs; N = 10), we found impairments in both domains although the difficulties were most pronounced for faces. Importantly, impairments with faces and objects were systematically related. We suggested that that the seemingly disproportional impairment for faces in DP was likely to reflect differences between stimulus categories in visual similarity. Here, we aimed to replicate these findings in a larger, independent sample of DPs (N = 21) using the same experimental paradigms. Contrary to our previous results, we found no disproportional effect of visual similarity on performance with faces or objects in the new DP group when compared to controls (N = 21). The new DP group performed within the control range, and significantly better than the old DP-group, on sensitive and demanding object recognition tasks, and we can demonstrate a classical dissociation between face and object recognition at the group level. These findings are perhaps the strongest evidence yet presented for a face-specific deficit in developmental prosopagnosia.

Keywords: developmental prosopagnosia; dissociation; object recognition; selectivity; visual similarity.

Figure 1

Examples of the face and house stimuli used in the perceptual matching task. ‘Sim’ designates the similarity level with each stimulus differing from the one presented in the boxed area by one difference (Sim 3), two differences (Sim 2), three differences (Sim 1), and four differences (Sim 0).

Figure 2

Results from the simultaneous matching task. Performance with faces and houses when the similarity level was equal for the two categories for (A) second-order differences and (B) featural differences. Performance with faces and houses when the similarity level was higher for faces (Sim 3: see Figure 1) than for houses (Sim 0: see Figure 1) for (C) second-order differences and (D) featural differences. Performance with faces and houses when the similarity level was higher for houses (Sim 3: see Figure 1) than for faces (Sim 0: see Figure 1) for (E) second-order differences and (F) featural differences.

Figure 3

The mean correct RTs for the New-DP group and its control group in the four conditions of the simultaneous matching paradigm with faces and house. It is evident that the New-DPs only differed significantly from the controls in the second-order conditions with faces. For the three other conditions, there was a clear overlap in their associated 95% CIs.

Figure 4

The mean percentage correct across the four similarity levels of the simultaneous matching paradigm for the Old-DP and the New-DP groups. In all conditions the New-DP group is more accurate than the Old-DP group.

Figure 5

Scatterplots showing the correlation between the object decision task with silhouettes and the Cambridge face memory task for the Old-DP group (top) and the New-DP group (bottom).

Figure 6

The score obtained on the FEQ (the higher the score, the larger the degree of face recognition difficulty experienced) and the number of correct trials obtained on the CFMT for the members of the Old-DP and the New-DP groups.