Content area
Full Text
Introduction
fMRIa, the attenuation of blood oxygenation level-dependent (BOLD) signal which follows repeated presentation of a stimulus, is considered the neuroimaging equivalent of neuronal repetition suppression. Despite the popularity of this technique, its underlying neural mechanisms are still unclear (Grill-Spector et al., 2006; Malach, 2012 ).
Several theories view fMRIa as an outcome of bottom-up processing, thus attributing the signal reduction to more efficient processing of the repeated stimulus (Grill-Spector et al., 2006 ). In contrast, fMRIa has been associated with top-down processes related to perceptual expectations. Specifically, stimulus repetition leads to increased expectations and reduction of prediction errors (Rao and Ballard, 1999 ). This account is supported by studies demonstrating that fMRIa is modulated by the likelihood of stimulus repetition, and that this effect is independent of bottom-up repetition effects (Summerfield et al., 2008; Larsson and Smith, 2012; Grotheer and Kovács, 2014a,b )
These latter studies induced expectations by increasing the probability of repeated versus alternating trials (i.e., short-term expectations). However, perceptual expectations may also be instantiated based on real-world statistical regularities (i.e., long-term expectations). Here we examine whether such regularities could modulate fMRIa by utilizing possible and impossible objects (Penrose and Penrose, 1958 ). These stimuli are visually similar, but the perceived 3D structure of impossible objects violates real-world statistical regularities (Fig. 1 , Elber, 2011 ). Thus, the comparison between these object categories offers a unique test-case for examining the role of long-term expectations in fMRIa.
To examine the role of such expectations in fMRIa, we compared the activation profiles of the alternated trials that were composed of possible and their matched impossible objects, and manipulated the order of stimulus presentation (Fig. 2 A). We assumed that in the context of an experimental trial, the visual system generates a representation based on the first object within a pair (prime) and this representation is then compared to the second object (probe). Importantly, the likelihood of alternated and repeated trials was identical in order to avoid short-term expectation effects.
The manipulation of stimulus order could provide informative insights regarding the nature of stimulus representation. The bottom-up account predicts a similar activation profile in the LOC in alternated trials, regardless of stimulus presentation order, since the visual information...