Purchase this article with an account.
Katrien Torfs, Sven Panis, Johan Wagemans; Priming of superordinate categorization of object pictures by spatial-frequency filtered versions. Journal of Vision 2011;11(11):834. doi: https://doi.org/10.1167/11.11.834.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Recent object recognition theories propose an initial fast feedforward sweep to high-level visual areas resulting in a first “interpretation” of the visual input which is then fed back to ongoing bottom – up analysis in low-level ventral visual areas (e.g., Bar, 2003). However, the factors mediating (the degree of) top-down processing necessary for recognition, the nature of this first interpretation, and the time-windows in which they are most effective are still unresolved questions. In the current study, we investigated changes in object recognition processing elicited by top-down interpretations based on spatial-frequency information in the visual input. Participants were asked to perform a superordinate categorization task (natural versus manmade) of grayscale pictures of everyday objects, which were preceded by a high- or a low-spatial-frequency filtered version of either the same or another object. We explored the influence of several factors on the formation of global shape candidates in this priming paradigm. Specifically, processing time of the prime (short versus long) and similarity in shape and/or category between the prime and target picture were manipulated. Discrete time survival analysis was applied to model the temporal dynamics of the effects (Singer & Willett, 2003). We found an early temporal advantage for categorizing objects when preceded by shape and category congruent primes compared to same category but different shape primes. Critically, low-pass filtered primes did not accelerate the categorization process compared to high-pass filtered primes. Results are interpreted within a dynamic, interactive processing framework. We are currently exploring the influence of these factors using a more demanding identification task, and degraded targets (fragmented object outlines as used in Torfs, Panis, & Wagemans, 2010).
This PDF is available to Subscribers Only