The human brain responses to Glass patterns: The effects of signal to noise ratio

Abstract

Purpose. A visual cortical neuron responds best to images that match its receptive field properties. Hence, increasing signal along one dimension should increase the responses most in neurons tuned to that dimension. We investigated the brain areas best respond to Glass patterns by measuring BOLD activation to Glass patterns at different signal to noise ratios. Method. The Glass patterns were consisted of randomly distributed dot pairs (5.4′ square dots) covering 1% of image. The distance between dots in a pair was 27′. The coherence of a Glass pattern is defined as the proportion of dots (signal dots) with orientation tangent to a concentric global form. The BOLD activation in 6 observers was collected on a Bruker 3T magnet (TR=3s, TE=60ms, flip angle=90o). The block design had a 18s test epoch, containing Glass patterns at one coherence level, alternating with a 18s control for 6 times. The test Glass patterns had coherence level 0.25 to 1.0 at 0.125 step. The control pattern had zero coherence. The V1–V4, hMT and LOC (the lateral occipital complex) were localized in each observer in separate sessions. Result. The Glass patterns activated V4, LOC, the inferior temporal gyrus (IT) near the temporal pole and an area in the dorsal middle occipital gyrus (dMOG). The slopes of the response to coherence functions were greatest in LOC. The V4 and dMOG showed a shallower slope. Conclusion. The result is consistent with a model of hierarchical processing of perceptual grouping in which curvature segments are combined into a form and then an object in V4 and LOC. The dMOG activation, similar to that of V4, suggests that there may be an area homologous to monkey V4d.

Footnotes

 Support by NSC-92-2413-H-002-027