The representation of S-cone signals in primary visual cortex

Abstract

Recent studies of middle-wavelength-sensitive and long-wavelength-sensitive cone responses in primate primary visual cortex (V1) have challenged the view that color and form are represented by distinct neuronal populations. Individual V1 neurons exhibit hallmarks of both color and form processing (cone opponency and orientation selectivity), and many display cone interactions that do not fit classic chromatic/achromatic classifications. Comparable analysis of short-wavelength-sensitive (S) cone responses has yet to be achieved and is of considerable interest because S-cones are the basis for the primordial mammalian chromatic pathway. Using intrinsic and two-photon imaging techniques in the tree shrew, we assessed the properties of V1 layer 2/3 neurons responsive to S-cone stimulation. These responses were orientation selective, exhibited distinct spatiotemporal properties, and reflected integration of S-cone inputs via opponent, summing, and intermediate configurations. Our observations support a common framework for the representation of cone signals in V1, one that endows orientation-selective neurons with a range of chromatic, achromatic, and mixed response properties.

Figure 1.

Intrinsic signal images of superficial V1 responses to cone-isolating stimuli. A, B, Single-condition images of superficial V1 of the same tree shrew (left) showing responses to S-cone-isolating (A) and ML-cone-isolating (B) sinusoidal grating stimuli of two different orientations (0°, top; 90°, bottom). Difference images (middle) show the cone-driven differential responses to these orthogonal orientations, with a region of interest (ROI) (enlarged on right) to show similarity in map structure for different cone-isolating stimuli.

Figure 2.

The functional mapping of orientation preference is similar for ML- and S-cone signals. A, B, False color orientation maps for one animal constructed from pixel-by-pixel Gaussian fits to intrinsic signal responses. Color indicates orientation preference. ML-cone-isolating gratings were used in A and S-cone-isolating gratings in B. C, False color map of absolute value of preference differences across the two maps (color scale in D). D, Cumulative histogram of preference differences for one animal. The median pixel exhibits less than a 20° difference in preference.

Figure 3.

Spatial and temporal frequency population responses are different for ML- and S-cone signals. A, Orientation difference images from the same tree shrew to S- and ML-cone sinusoidal grating stimuli of varying spatial frequency (0.1–1.6 cpd) and fixed temporal frequency (2 Hz). S-cone activation is more robust at lower spatial frequencies. B, Average normalized activation from S- and ML-cone stimuli of varying spatial frequency (n = 4 animals). Note the difference in magnitude of responses to ML- and S-cone stimulation. Error bars indicate SEM. C, Orientation difference images from same tree shrew to S- and ML-cone gratings of varying temporal frequency (2–32 Hz) and fixed spatial frequency (0.2 cpd). Note falloff of S-cone signals at higher temporal rates. D, Average normalized activation from S- and ML-cone stimuli of varying temporal frequency (n = 3 animals). Relative S/ML response modulation varied across individuals, but high-frequency S-cone falloff is consistent. Error bars indicate SEM.

Figure 4.

A subset of V1 neurons exhibit S-cone responses. A, Left, two-photon image of cells labeled with calcium dye Oregon green BAPTA-1. Middle left, Schematic of responses to an S-cone-isolating stimulus. Filled circles, Neurons that exhibited significant responses to color-exchange stimulus (see Materials and Methods) and also exhibited significant responses to properly oriented S-cone-isolating grating. Open circles, Neurons that exhibited significant responses to the color-exchange stimulus but did not exhibit significant responses to S-cone-isolating gratings. Middle right, Achromatic orientation responses. Cells were included if they gave significant visual responses to both color-exchange and oriented gratings compared with the blank stimulus (ANOVA, p < 0.05). Bars indicate cells that exhibited significant variation across the oriented gratings (ANOVA, p < 0.05), and the orientation of the bar matches the orientation preference of each cell. Right, S-cone-isolating orientation responses. B, Achromatic and S-cone-isolating orientation tuning curves for four cells indicated in A. Circular variance, tuning width (half-width at half-height), and orthogonal-to-preferred (O/P) ratio are shown for each curve. Cell 2 did not exhibit significant variation across the S-cone-isolating gratings so no fitting was performed. C, Scatter plots of circular variance, tuning width, and orthogonal-to-preferred ratio for all cells. Dashed lines indicate median values. Parameters were relatively similar for all conditions, although circular variance was slightly higher for S-cone-responsive cells (see Results).

Figure 5.

Visual cortical cells exhibit a diversity of S/ML chromatic tuning responses. A, Stimulation consisted of combinations of S-cone and ML-cone contrasts presented in-phase and out-of-phase. B, Predicted responses of model neurons to the stimulus suite in A. C, Responses to several stimuli in one animal. Left, Raw baseline image of cells; other images show change in fluorescence relative to baseline image. Note that different cells in field respond to different combinations of stimuli. D, Tuning curves from 12 representative neurons. Chromatic tuning ranges from opponent, summing, or ML-only to broadly tuned or S-cone selective. E, Plot of index values derived directly from stimulus responses (cone-summing index and S/ML response index) indicates a continuous distribution of chromatic tuning. F, Broad distribution of cone-summing index; few cells exhibited purely summing or purely opponent chromatic tuning.

Figure 6.

Models of cone interactions. A, Simple CC model; receptive fields are divided in half, but contributions from a given cone must be equal and opposite across halves. Responses depend only on relative S- and ML-cone contrast. B, The 2PCC model; receptive fields also divided in half, but cone weights of each half can take any value and either sign. C, Three cells that were well fit by the CC model. Tuning curve along with CC model and 2PCC model fits and parameters; * indicates best model as determined by the modified nested F test. Bottom cell could only be well fit if CC model was modified to include nonlinear contrast gain; gains shown (modified CC model). GOF, Goodness of fit. D, Cells that were poorly fit by CC models but well fit by the 2PCC model. Top and middle cell are well fit by mixed summing/opponent configurations, whereas the bottom cell is suppressed by ML-cone input, particularly when ML-cone input is out-of-phase from S-cone input.

Figure 7.

V1 cells exhibit a variety of S/ML-cone configurations. A, Cumulative histogram of goodness of fit for models. All models do well, but 2PCC model explains >90% of responses for median cell. B, Cumulative histogram of fraction of total cone input attributable to S-cones for the modified CC model, 2PCC model, and 2PCC model extended to include rods. Some 34–46% of cone input to median cell is derived from S-cones. C, Histogram of cone interactions in best model for mean fits and bootstrap simulations (see Results). Most common configuration is mixed summing/opponent, followed by opponent configuration, ML-only, and summing. D, Orientation tuning curves from cells with different cone input configurations. E, Orientation tuning curve parameters for the four most common configurations; no significant differences among these groups were observed (Kruskal–Wallis test p values shown). O/P, Orthogonal-to-preferred ratio.