Abstract
Humans are extremely efficient at detecting object categories in briefly presented photographs of natural scenes. Recent work investigating the neural mechanisms underlying this ability (Peelen et al., 2009; 2011) found that response patterns in object-selective visual cortex carried information for the relevant but not for the irrelevant category. This suggests that rapid category detection in natural scenes is supported by an attentional search template that pre-activates category-specific neurons, enabling efficient target processing. In the current study we aimed to investigate the contents of this template using a subliminal priming paradigm. We reasoned that, if a subliminally presented stimulus facilitates target detection this may be indicative of overlap between the representation of the prime and the attentional search template. Participants viewed centrally presented monochromatic photographs of natural scenes containing (1) people, (2) cars, (3) cars and people or (4) no category. On separate runs, participants were instructed to detect either cars or people. Scenes were preceded by a black and white random noise mask that was presented for 200ms. A prime was shown on two thirds of the trials for 35ms following the mask. The prime depicted one of four exemplars of either a car or a person. Invisibility of the primes was confirmed for each subject in a second experiment that required categorization of the primes. In the main experiment target detection was significantly faster for scenes preceded by a prime matching the target category compared to scenes preceded by a non-matching prime. Simple prime-response associations cannot account for this detection advantage as the prime condition did not differentially affect reaction times for target-absent scenes, thus suggesting overlap between the representation of the employed primes and the attentional search template. In future work systematic variation of the primes will allow for more detailed characterization of the search template’s content.
Meeting abstract presented at VSS 2012