Abstract
Categorizing objects in the visual environment is an effective strategy to enhance processing efficiency. The development of this ability has rarely been investigated using similar methods across age. We addressed this issue by means of fast periodic oddball visual stimulation, an approach that provides an implicit, objective, and robust measure of categorization in adults (Liu-Shuang, et al., 2014; Rossion, et al., in press) and, potentially, in infants. Adults (N = 11; 2 males), 4- (N = 30; 14 males) and 7-month-olds (N = 39; 18 males) were presented with short sequences of animal and furniture pictures. 4 items of the same category were presented consecutively (standard category); every fifth stimulus belonged to the other (oddball) category. Base stimulation frequency F was 6.033 Hz. The oddball response (1.21 Hz; F/5) and its harmonics (nF/5) were used to measure discrimination. In adults, oddball responses were observed in the animal and furniture oddball conditions (SNR/Z-scores: furniture oddball = 4.05/5.39, p < .0001, animal oddball = 4.72/5.91, p < .0001). Low-level visual cues contributed only minimally to the response: Discrimination was substantially decreased for phase-scrambled images with identical power spectra as the original images (SNR/Z-scores: animal oddball = 1.52/1.12, p = .13, furniture oddball = 2.53/1.97, p < .05). In 7-month-old infants, animal (SNR/Z-score = 1.83/1.83, p < .05) and furniture oddballs elicited discrimination (SNR/Z-score = 1.54/1.51, p = .07). The discrimination response for animal oddballs was erased by phase-scrambling images (all SNRs/Z-scores < 1.51/1.40). In 4-month-olds, animal oddballs elicited discrimination (SNR/Z-score = 1.67/1.69, p < .05), but furniture oddballs did not. Further data collection on phase-scrambled conditions is under way. Our findings suggest that both infants and adults are capable of categorizing animal and furniture items even when the category is represented broadly and diverse images are displayed in a rapid fashion.
Meeting abstract presented at VSS 2015