| |
Abstract:
There is anecdotal evidence that prosopagnosic patients may
have different biases when recognizing faces as compared to common
objects. Patients often report that although they are impaired at
face recognition, their non-face recognition abilities have been
spared -- a belief that may be tied to the fact that common objects
are typically recognized at the basic level, but faces are
recognized at the relatively more difficult individual level. Some
patients also respond much faster when recognizing faces as
compared to common objects. In earlier research we found that
prosopagnosic patients exhibited similar recognition abilities for
faces and objects when the level of categorization was equated and
a bias-free dependent measure (A') was used. Here we systematically
investigated the sensitivity and bias patterns for one
prosopagnosic patient (SM) using both faces and non-face objects
across different discrimination levels (easy / difficult) and
recognition tasks (simultaneous matching / sequential matching).
ROC curves, measuring sensitivity across different levels of bias,
were derived by having SM provide a confidence rating along with
his response on each trial. Two results stand out. First, SM was
poorer at face recognition than either Greeble or common object
recognition. Second, in comparison to age-matched controls, SM was
far worse at recognizing all object classes, including faces.
Overall, we conclude that when more stringent dependent measures
are adopted in studying prosopagnosic patients, category-specific
impairments are less evident.
|
