Figure 1.
Synthetic first- and second-order contour stimuli.
LG, sine-wave luminance gratings. CM, contrast modulated contours. PR, phase-reversal defined contours. Each column depicts one type of contour stimuli with an orientation of 90°. Arrows superimposed on each stimulus type in the top row represent the bidirectional motion of the global contours. The contours move leftward for 2 seconds and then rightward for another 2 seconds, as depicted below by the traces in the space-time plots. The square brackets and black arrows point to the second-order contours.
Figure 2.
Orientation preference domains were activated by different first- and second-order contours in macaque V1 and V2.
(A), Picture of the cortical surface taken from the left hemisphere of macaque 766 with region of interest (ROI) indicated by red box. This image was obtained under 550 nm green-light illumination. The broken white line indicates the border between V1 and V2. LS, lunate sulcus. L, lateral. A, anterior. (B), Differential orientation maps of 0° minus 90° in V1 and V2. Blood vessels were masked gray on all the maps. White boxes represent regions of V1 and V2 that were further analyzed and compared. (C), Response strength comparison for first- and second-order stimuli from ROIs of V1 and V2 in B. (D), Representative areas of V1 and V2 as boxed in B. Pixels covered by blood-vessel masks as shown in B were interpolated for clarity. Iso-orientation contours, derived from the orientation preference map generated using LG stimuli, were superimposed on each map. Colors of contours indicate orientation preferences as indicated by the color code below figure. (E), Normalized orientation response profiles for first- and second-order contour stimuli calculated from V1 and V2 areas in D. Error bars represent S.E.M. Scale bar: 1 mm.
Figure 3.
Orientation preference domains in V1 and V2 activated by phase-reversed contours.
(A), Picture of the surface vasculature with the region of interest (ROI) in V1 and V2 as outlined by the red box. LS: lunate sulcus. A, anterior. L, lateral. (B), Differential orientation maps of V1 and V2 derived from LG and PR stimuli with orthogonal orientation pairs of 45° and 135°. Blood vessels were masked gray on all the maps. (C), Differential orientation maps from the representative areas of V1 and V2 (white boxes in B). Both pairs of the grayscale images were displayed based on the intensity range of the PR map and were superimposed with iso-orientation outlines derived from the orientation map generated by LG stimuli. (D), Normalized response profiles for LG and PR stimuli. The two pairs of curves were closely matched in the orientation preference for both V1 and V2. Scale bar: 1 mm.
Figure 4.
Orientation differential and preference maps in V1 and V2 for contrast modulated contour stimuli.
(A), Picture of the surface vasculature of the left hemisphere of macaque 740 with the region of interest (ROI) in V1 and V2 outlined in red. Diagrams of the 0° and 90° oriented LG and CM stimuli were shown on top. (B), Differential orientation maps of V1 and V2 derived from LG and CM stimuli with orthogonal orientation pairs of 0°–90° and 45°–135°. Blood vessels were masked gray on all the maps. (C), Color coded orientation preference maps generated by LG and CM stimuli with blood vessels masked gray. (D), Orientation preference maps from representative areas of V1 and V2 as boxed in C. (E), Histograms produced by pixel-wise subtraction of the two pairs of orientation preference maps in D. The distributions of angular differences of preferred orientations peak around 0°. The percentages of pixels with angular differences less than 30° were 82% and 84% in V1 and V2, respectively. Note that orientation ranges from 0° to 180°, so the difference between two orientation values will be in the range of −180° to 180° by direct subtraction. Scale bar: 1 mm.
Figure 5.
Spatial correlation coefficients (SCC) and response strengths across all animals studied.
(A), Spatial correlation coefficients for both V1 and V2 between the first-order stimulus and the two second-order stimuli. SCC represents spatial similarity between the orientation differential maps activated by LG stimuli and those by CM and PR stimuli. The numbers of orientation differential maps computed were 6 and 4 for CM and PR stimuli respectively in both V1 and V2. The small solid black squares represent mean values while the red lines represent median values. (B), Comparison of population response strengths (ΔR/R) in V1 and V2. Error bars represent S.E.M. Data from 6 monkeys.
Figure 6.
Energy model simulations of population responses to all stimulus types across V1 and V2.
(A), Illustration of the receptive field of the modeled neuronal populations in the spatio-temporal frequency space. RFs with different tuning properties are assumed to be Gabor wavelets of different shapes, resulting in Gaussians in the frequency space, whose elliptic outlines are displayed. RFs illustrated in the same color correspond to a same orientation preference lying on radial sections in the frequency space (e.g. red: horizontal orientation; light blue vertical orientation). (B), Population response magnitudes produced by the energy model to different types of stimuli. (C), Normalized response profiles from V1 and V2 simulated by the energy model to the first- and second-order contour stimuli (0°–90°). As for the optical imaging experiment, the noise-texture carriers used for the simulation were static and were not filtered. (D), Normalized response profiles simulated by using dynamic and high-pass filtered (eliminate components with SFs below 9 cpd) noise textures as the carrier for CM and PR stimuli (0°–90°). The error bars indicate S.E.M over 256 trials and are smaller than the data marker.
Figure 7.
Distribution of differential power between stimuli with orthogonal orientations in the frequency space.
(A, D, G), Diagrams of pairs of stimuli with orthogonal orientations of 90° and 0°. White arrows indicate the bidirectional motion of the stimuli. The stimuli moved 0.51 second in each direction to reduce the computational time. We then used a Fourier transform (matlab function ‘fftn’) to compute the power of each stimulus. (B, E, H), Power distribution difference (90°−0°) in the two dimensional spatial frequency space corresponding to each stimulus pair. Note that to reduce dimensions, the power was integrated over all temporal frequencies. 32 pairs of the CM and PR stimuli were used for the computation. (C, F, I), Differential power distributions in the orientation, spatial frequency, and temporal frequency dimensions for each stimulus pair, respectively. The gray shades in the left panels of F and I represent SEM over 32 pairs of stimuli.
Figure 8.
Simulated differential population responses with different preferences in V1.
(A–C), Details of differential responses of neuronal populations preferring different orientations, spatial, and temporal frequencies to each stimulus pair (0°–90°). N = 10 trials.
Figure 9.
Simulated differential population responses with different preferences in V2.
(A–C), Details of differential responses of neuronal populations preferring different orientations, spatial, and temporal frequencies to each stimulus pair (0°–90°). N = 10 trials.