Gain control in the response of human visual cortex to plaids

Abstract

McDonald JS, Mannion DJ, Clifford CWG. Gain control in the response of human visual cortex to plaids. J Neurophysiol 107: 2570-2580, 2012. First published February 29, 2012 doi:10.1152/jn.00616.2011.-A recent intrinsic signal optical imaging study in tree shrew showed, surprisingly, that the population response of V1 to plaid patterns comprising grating components of equal contrast is predicted by the average of the responses to the individual components (MacEvoy SP, Tucker TR, Fitzpatrick D. Nat Neurosci 12: 637-645, 2009). This prompted us to compare responses to plaids and gratings in human visual cortex as a function of contrast and orientation. We found that the functional MRI (fMRI) blood oxygenation level-dependent (BOLD) responses of areas V1-V3 to a plaid comprising superposed grating components of equal contrast are significantly higher than the responses to a single component. Furthermore, the orientation response profile of a plaid is poorly predicted from a linear combination of the responses to its components. Together, these results indicate that the model of MacEvoy et al. (2009) cannot, without modification, account for the fMRI BOLD response to plaids in human visual cortex.contrast; spatial; functional magnetic resonance imaging; vision THE DYNAMIC RESPONSE RANGE of neurons is limited compared with the range of natural contrasts. Gain control allows neural sensitivity to be dynamically adjusted to suit the prevailing ambient contrasts (Crowder et al. 2008;Durant et al. 2007). This adjustment could be achieved through temporal sampling, i.e., adaptation (Gardner et al. 2005;Ohzawa et al. 1982), or spatial sampling in the area of the classic receptive field (Morrone et al. 1982) and extra classic receptive field (Levitt and Lund 1997) of a neuron. For two decades, divisive normalization has been the dominant model of contrast gain control (Brouwer and Heeger 2011;Heeger 1992). Principally intended to explain neuronal response saturation and crossorientation suppression-suppression of response to optimally oriented stimuli by an overlaid stimulus of a different orientation-the original divisive normalization model proposed that neuronal response is inhibited by a large pool of neurons tuned to different orientations and spatial frequencies. Although subsequent research has challenged the notion that inhibitory neural circuitry is responsible for the divisive gain behavior Freeman et al. 2002;Li et al. 2006;Priebe and Ferster 2006), the model retains considerable descriptive power and attractive theoretical properties.Two recent studies boast population, rather than isolated single-unit, measures of cross-orientation suppression (Busse et al. 2009;MacEvoy et al. 2009). Busse et al. (2009) found magnitudes of suppression in cat and human V1 consistent with traditional divisive normalization. However, MacEvoy et al. (2009) demonstrated that the response of tree shrew V1 to plaids is reliably predicted by the contrast-weighted average of the responses to the two components. In the case of two grati...

A Two-Stage Cascade Model of BOLD Responses in Human Visual Cortex

“…Responses increase with contrast and with number of orientations, consistent with recent fMRI measurements [16], [17]. This pattern of results is qualitatively reproduced by the CC model (Figure 3, red curve).…”

Orientation anisotropies in human primary visual cortex depend on contrast

“…In Experiment 1, the same display apparatus as that used in previous studies from this laboratory was used in order to facilitate comparisons McDonald et al, 2012), while in Experiment 2 the orientation response was measured at a very low contrast level not previously possible within the MRI scanning environment.…”

“…One further concern is that grating stimuli are almost always offset from the fovea by varying amounts in fMRI studies, due to the technical difficulties in reliably mapping the human foveal confluence (Schira et al, 2009). This study, and previous investigations from this laboratory McDonald et al, 2012) were careful to use gratings with a comparatively high spatial frequency (3.35 cycles/°; see Table 6). Initially, this was because it was within a spatial frequency range known to evoke strong BOLD responses in visual cortex (Singh et al, 2000;Kay et al, 2008).…”

“…Yacoub et al (2008) found a greater regularity or density of orientation columns tuned to vertical over other orientations when measured at a field strength of 7 T, although this was significant in only one of the two subjects (P b 0.05) and there was no difference in the overall fMRI signal change across all orientations. More recent studies however have indicated the opposite pattern of activity, with a larger activation to oblique over cardinal orientations Swisher et al, 2010;McDonald et al, 2012); something also measured with other noninvasive functional imaging techniques (Heinrich et al, 2008;Koelewijn et al, 2011;.…”

What we see changes how we see : analyzing the plasticity of the horizontal effect.

“…In humans, the differential representation of orientations in cortex has proved less conclusive; while some studies show a cardinal bias (Aspell, Wattam-Bell, Atkinson, & Braddick, 2010;Furmanski & Engel, 2000;Furmanski, Schluppeck, & Engel, 2004;Maffei & Campbell, 1970;Proverbio, Esposito, & Zani, 2002;Serences et al, 2009;Swisher et et al, 2010; others show no bias (Kamitani & Tong, 2005; McDonald, Mannion, & Clifford, 2012;Yacoub et al, 2008) or mixed results (Koelewijn et al, 2011). The amplitude of response to cardinal gratings does appear to be greater than that evoked by oblique gratings in human visual cortex.…”