Theories of predictive coding (PC; Rao & Ballard, 1999) have dominated neurocognitive research in explaining thought and perception processes in various domains. The basic principle is that perception relies not only on bottom-up processing of sensory input but also on top-down predictions. The current thesis describes several neuronal response alterations in cortical visual areas measured with neuroimaging methods. The so-called repetition suppression (RS) effect was connected to predictive coding as repetitions make stimuli more expected, which results in a smaller prediction error and therefore attenuated neuronal activity. Still, it is questioned whether RS reflects the PE or is a local process by neuronal populations that occurs without top-down influences (Grill-Spector et al., 2006). Another often investigated effect is the reduced neuronal response to expected or predicted visual input called expectation suppression (ES). A considerable body of research on contextual response changes, such as RS and ES, relates to the visual system and the face-processing network in particular. Overall, we demonstrate the importance of stimulus predictability for studies using RS to uncover expectancy-related effects. Furthermore, we suggest that the influence of sensory precision on measures of RS and ES needs more attention in future research. Concerning the stimulus material in the presented studies - unfamiliar, visually familiar, and famous familiar faces - we also emphasize the importance of thoroughly considering the characteristics of faces in terms of prior belief and sensory input precision and predictability when using them for testing prediction-related effects.
copy citation link