The nature of anisotropy in gain control pools.

Suppression;Natural scene perception;Contrast normalization;Orientation;Temporal processing;Horizontal/oblique effect

Yeon Jin Kim

University:University of Louisville

Department:Psychological and Brain Sciences

关键词: Suppression;    Natural scene perception;    Contrast normalization;    Orientation;    Temporal processing;    Horizontal/oblique effect;   

Others  :  https://ir.library.louisville.edu/cgi/viewcontent.cgi?article=1749&context=etd

美国|英语

来源: The Universite of Louisville's Institutional Repository

【 摘 要 】

When observers view a naturalistic (l/f) broadband image, the various spatial components present in the image stimulate many detecting-mechanisms that suppress each other. This suppression is anisotropic, being relatively greater for mechanisms that detect horizontal components and least for oblique-detecting mechanisms (c.f., Essock et al., 2009; Haun & Essock, 2010; Kim et al., 2010), and leads to better sensitivity to and greater salience of obliquely-oriented structure when viewed in the presence of broadband background (the 'horizontal effect', Essock et al., 2003). It is thought that anisotropic suppression reflects a gain control mechanism whose output is intended to equalize the neural response to the orientation content biases in natural scenes. Here we further investigate the dependence of this gain control anisotropy on temporal and spatial frequency by measuring tuning properties of these pools. Experiment 1 evaluates (1) the shape and the peak suppression of temporal frequency tuning functions for grating targets at 10 fixed spatio-temporal combinations by varying the temporal frequency of the l/f broadband masks, and (2) whether the observed suppression show the horizontal effect. Experiment 2 evaluates the same properties for spatial frequency tuning. The results showed that multiple local gain control pools across the spatiotemporal locations are tuned in temporal frequency and spatial frequency, and all of these pools are stronger at horizontal. Surprisingly, results showed that multiple temporally tuned pools are revealed with a broadband mask, whereas prior studies using very narrowband masks have shown only 2 (or 3) temporally-tuned channels (e.g., Lehky, 1985; Hess & Snowden, 1992; Fredericksen & Hess, 1998; Cass & Alias, 2006; Cass et al., 2009). We proposed that to drive these pools, the use of an appropriate broadband mask which can rise above a certain detection threshold of each of the underlying many, unknown, and lesser sensitive temporal detectors is crucial, and thereby can create strong suppression between detectors tuned to similar temporal rates which form local (' tuned ') pools.