Abstract
Spatial phase information is essential for image perception, because important visual features such as edges are perceived at locations of maximal phase congruency (Morrone and Burr, 1988). Step edges, which require phase-sensitive pooling of different spatial frequencies, are common in natural images (Griffin, Lillholm, and Nielsen, 2004). A study on natural image statistics suggests that outputs from the primary visual cortex (V1) are optimally analyzed by pooling different spatial frequency bands together to extract broadband edges (Hyvärinen, Gutmann, and Hoyer, 2005).
Here we studied responses to changes in relative phase alignments in grating stimuli and selectivity to phase congruency using functional magnetic resonance imaging (fMRI). In experiment 1, fMRI adaptation design comprised compound grating stimuli with two spatial frequency components (f = 0.4 cpd and 3f = 1.2 cpd at 7.6 deg eccentricity). Subjects were adapted to one phase alignment, and then they were shown compound gratings with different relative phase alignments. Areas V1, V2, V3/VP, V4v, V3AB, and LOC showed an increase in the fMRI response as function of the relative phase difference compared to the adaptation stimulus (Page's L test, p [[lt]]0.01 for each area). We controlled that these responses were not explained by differences in local contrast or position.
In experiment 2, we compared fMRI responses to compound grating stimuli (f, 3f, 5f, 7f, 9f) with congruent and random phase structures. Areas V1, V2, V3/VP, V4v, V3AB, and LOC showed stronger responses for the congruent stimuli (Wilcoxon's signed rank test, p [[lt]]0.05 for each area) with a gradual increase in the phase congruency selectivity along the ventral stream from V1 to LOC (Page's L test, p [[lt]]0.001).
Our results indicate that in the human visual cortex, phase relationships across spatial scales are extracted from the visual stimulus to locate broadband features like boundaries of an object.
Funding acknowledgements: Finnish Graduate School of Neuroscience, Finnish Cultural Foundation, Academy of Finland (National Programme for Centres of Excellence 2006 â€″ 2011 grant 213464, NEURO-program grant 111817, and grants 105628, 210347 and 124698).