Purchase this article with an account.
Sirui Liu, Qing Yu, Peter Tse, Patrick Cavanagh; Neural Basis of the Double-Drift Illusion. Journal of Vision 2017;17(10):603. doi: https://doi.org/10.1167/17.10.603.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
When a Gabor patch moves along a path on an equiluminant background in one direction while its internal texture drifts orthogonally to the path, it can appear to deviate from its physical path by 45° or more (infinite regress illusion, Tse & Hsieh, 2006; curveball illusion, Shapiro et al., 2010, Kwon et al., 2015; double-drift illusion, Lisi & Cavanagh, 2015). Despite this remarkable perceptual effect, saccades to the Gabor are immune to the illusion (Lisi & Cavanagh, 2015). The present study investigated where and how this illusion is coded in retinotopic visual areas. We used fMRI and multivariate pattern analysis to classify the activation patterns driven by two illusory motion paths of the double-drift stimulus that shared the same physical path but had opposite directions of internal motion. We contrasted that to two physically different motion paths without internal motion that matched the perceived illusory orientations of the first two stimuli. Results showed that the classification accuracy was quite high (70 to 80%) for the two physically different motion paths in V1-V3 and area MT+, confirming that physical paths differing by 80° to 100° in orientation are decodable in these areas. In contrast, for the two illusory motion paths, decoding performance was poor in areas V1 and MT+ but was significantly above chance in V2 and V3 (about 60%), suggesting that the perceptual divergence of the double-drift stimulus emerged at V2 and V3. Given the distinct anatomical gaps across the meridians in retinotopic visual areas, our results are consistent with the previous behavioral findings that the magnitude of the double-drift illusion is degraded at both the vertical and horizontal meridians (Adamian & Cavanagh, 2015; Liu & Cavanagh, 2016), suggesting the involvement of quadrant-based visual areas V2 and V3 but not the hemifield-representing V1.
Meeting abstract presented at VSS 2017
This PDF is available to Subscribers Only