Abstract
The predictive coding (PC) model of perceptual inferences (Rao and Ballard, 1999) proposes that perceptions are the result of a continuous comparison of perceived and predicted events. Recently, PC has also been proposed as a possible mechanism of repetition suppression (RS; the diminished neuronal signal for repeated stimuli when compared to non-repeated stimuli); it has been shown, that the repetition related reduction of the blood-oxygen level dependent (BOLD) signal, measured by functional magnetic resonance imaging (fMRI) techniques, is modulated by repetition probability (P(rep)) for faces (Summerfield et al., 2008). Surprisingly, however, other studies, using non-face stimuli, could not find P(rep) dependent modulations of RS, in either macaque single-cell activity of the inferior-temporal cortex (Kaliukhovich and Vogels, 2012), or in measuring the BOLD signal in the Lateral Occipital Complex (LO; Koávacs et al., 2013) of human observers, limiting the validity of PC explanations of RS to the category of faces. To address the issue of category selectivity of P(rep) modulations of RS, here we tested letters of the roman alphabet, as well as, unknown false-fonts as control stimuli. We modulated P(rep) for pairs of stimuli in individual blocks and observed a significant RS for both stimulus sets in the Letter Form Area (LFA), as well as, in the caudal-dorsal part of the LO. Interestingly, we found P(rep) modulations only on the RS for roman letters, but not for false-fonts. Our findings suggest that P(rep) effects on RS depend on the prior experience of the subjects with the applied stimulus category.
Meeting abstract presented at VSS 2014