Abstract
This ERP study analyzes how certain facial features are integrated in order to create new face recognition units and, in particular, how visual agnosic individuals use "diagnostic" information to that end when they meet new faces. Our participants (2 prosopagnosic adults and 8 age-matched controls) carried out a face-feature matching task. In this task both external (E) and internal (I) facial features were presented in a sequence, followed by a complete unfamiliar face that could be an exact (congruent) combination or not (incongruent) of such features. We used two different sequences of features: in the E-I sequence the external features were displayed as the first stimuli in each trial while in the I-E sequence the internal features were the first to be displayed. Similarly to that found in younger people in a previous experiment, our controls exhibited an enhanced mismatch effect around 300-500 ms in the E-I sequence, suggesting a more efficient integration of piecemeal information at the beginning of the trial to create new face representations. Our acquired prosopagnosic also showed a larger mismatch effect in the E-I sequence but smaller, more anterior and right-sided than controls, suggesting the use of a different neural circuitry to solve the task. In turn, our developmental prosopagnosic showed a mismatch effect only in the I-E sequence, which was larger and left-sided when compared with healthy controls, suggestive of a predominant use of analytical strategies in unfamiliar face processing. Moreover, the lack of an expected P3 component evoked by features at the beginning of the trial in both patients indicates that task-relevant information is not optimally processed and kept in short-term memory to solve the matching. Thus, modulations in both amplitude and topography of long-latency ERPs concerning structural incongruences might constitute neural markers of cognitive dysfunctions associated to face processing.
Meeting abstract presented at VSS 2018