Classical receptive field (CRF) stimulation in single V1 columns.

(A) Neuropixels probe (384 channels/3.84mm) penetrations made into the lateral surface or underlying calcarine sulcus of macaque V1. Bottom: Diagram of designated laminar compartments. (B) Current-Source density (CSD) profile from an example recording session (M1, pen1), showing the early current sink (red) indicative of layer 4C. Laminar compartment boundaries (dashed lines) were determined using histological data and the CSD profile. (C) Top: 71 well-defined classical receptive fields (CRF) vertically stacked along the cortical depth, mapped with a 10 x 10 array (4 x 4 dva) of probe stimuli (gratings, 0.4 dva in diameter). Bottom: Drifting Gabor grating (1.5 dva in diameter) used for CRF stimulation, positioned within the joint receptive fields of recorded neurons. (D) Heatmap of visual responses from 159 neurons to 36 drift directions of Gabor gratings, vertically stacked along cortical depth (M1, pen1). Top: Distribution of differences in preferred orientation for all pairwise combinations of neurons.

Nonclassical receptive field (nCRF) stimulation and border ownership.

(A) Border ownership test stimuli for nCRF stimulation. Stimuli consisted of uniformly white or black objects (8° squares) on black or white backgrounds, respectively. One border of the object fell within the CRFs (red circle) of the recorded neurons, with orientation varying in 45° steps. Within the CRF, depending on local contrast polarity (LC), the border could either be a dark-light edge (LC1, Left column) or a light-dark edge (LC2, Right column). For the same LC condition, the border within the CRF could be the bottom-left edge (Side 1, top row) or top-right edge (Side 2, bottom row) of the object. (B) Enlarged view of stimulus configuration within the CRF, showing identical stimulus regions between Side 1 and Side 2 for each LC condition (8 x 16 dva). (C) Four example Bown neurons recorded from V1. For each neuron, dot raster plots (top) and instantaneous firing rates (bottom) show responses to the four stimulus conditions depicted in (A), respectively. Magenta represents Side 1; green represents Side 2. (D) Bar plots showing the mean spike rates of the four neurons across the four stimulus conditions. (E) Top: Cartoon depicting the preferred border orientation for each of the four recording sessions (pen1-4). Arrows indicate the preferred side of objects relative to the border in the CRF. Bottom: Bown modulation index for each neuron plotted at its recorded cortical depth across the recordings. Color bars on the ordinate represent laminar compartments. Black dots denote statistically significant Bown neurons, determined by ANOVA with object side and local contrast as factors (p < 0.05). Middle: Marginal distribution of the Bown modulation index, red arrows indicating the population median. *p < 0.05; ***p < 10-5.

Interlaminar cross-correlations during CRF and nCRF stimulation.

(A) Example recording (M1, pen1) showing 159 visually responsive neurons (circles) recorded simultaneously during CRF stimulation, plotted at their relative cortical depth. The abscissa is magnified for visualization. Also shown are eight example neuronal pairs with significant CCGs during CRF stimulation. Neuronal pairs with zero-time lag CCGs are connected by dashed lines, while pairs with nonzero-time lag CCGs are connected by arrowed lines, with the arrow pointing to the lagging neuron. (B) Eight corresponding CCGs from (A). Neuronal pairs are separated by their laminar compartment combinations. The reference neuron (1st neuron in the CCG function) is designated as j, and the target neuron as k. PL (peak lag, ms) is defined as the relative time delay in peak correlation between the two spike trains. (C) Diagram illustrating feedforward and feedback (and horizontal) projections within the designated laminar compartments. Feedforward inputs arrive within laminar compartments 4C and 6, while feedback (and horizontal) projections principally target superficial and deep layers. (D) Heatmap of all significant CCGs during either CRF stimulation (Gabor gratings, top row) or nCRF stimulation (border ownership, middle row), for the four key interlaminar combinations (columns), calculated from one example recording (M1, pen1). Individual CCGs are normalized by their absolute maximum for visualization, and different CCGs are vertically stacked. Bottom row: Distribution of CCG asymmetry for each of the four key interlaminar compartments as in (A), across CRF stimulation (gray) and nCRF stimulation (magenta). Arrows indicate the population median. **p < 0.002; ***p < 10-5.

Comparison of interlaminar CCG asymmetries during CRF and nCRF stimulation.

Combined dataset for each of the four key interlaminar combinations. Histograms and cumulative distributions of asymmetry for all significant CCGs during CRF (gray) and nCRF (magenta) stimulation. Arrows indicate the population median. ***p < 10-5.

Dependence of nCRF modulation on interlaminar signal flow.

(A) Scatter plot of the Bown modulation index versus CCG asymmetry for all significant CCGs, aggregated across the four key interlaminar combinations and all recording sessions. CCG asymmetries were calculated during nCRF stimulation. The geometric mean of the Bown index for each neuronal pair was used. Red lines denote the fitted regression line and the 95% confidence interval in linear regression. (B) Similar to (A), but using the local contrast (LC) index instead of the Bown index. The LC index was calculated as the difference in neuronal responses to borders with opposite contrast polarities, normalized by the maximum response.

Number of neurons selective to Bown (LC) across recordings.

Comparison of CCG asymmetries during CRF and nCRF for each recording.

GLMs for dependence of border ownership and local contrast on CCG asymmetry.

Histological data for Neuropixels recordings in macaque V1.

(A) Neuropixels probe (384 channels/3.84mm) penetrations made into the lateral surface or underlying calcarine sulcus of macaque V1. Bottom: Diagram of designated laminar compartments. (B) Histological reconstruction of probe tracks based on DiI derivative staining for two example recordings (pen3-4). Here, Neuropixels probes passed through the lateral surface and reached the calcarine sulcus of V1. Only neurons in the calcarine V1 from these two recordings were tested and included in the analysis. Designated boundaries are shown. Scale bar, 500 µm.

Comparison of interlaminar cross-correlations during CRF and nCRF stimulation under different CCG criteria.

(A) Histogram distributions of CCG peak lags for each of the four key interlaminar combinations in the combined dataset during CRF (gray) and nCRF (magenta) stimulation. The shift (ι1) in median CCG peak lags during nCRF stimulation relative to CRF stimulation is shown. Significant CCGs are defined as those with peaks exceeding 7 SD above the noise level (flank). (B) Similar to (A), but comparing CCG asymmetry across all CCGs, including both significant and non-significant ones. (C) Comparison of CCG peak lags using a stricter criterion, where significant CCGs are defined as those with peaks exceeding 10 SD above the noise level (flank). (D) Comparison of CCG asymmetry using the stricter criterion of 10 SD.