Parallel processing in the corticogeniculate pathway of the macaque monkey

Abstract

Although corticothalamic feedback is ubiquitous across species and modalities, its role in sensory processing is unclear. This study provides a detailed description of the visual physiology of corticogeniculate neurons in the primate. Using electrical stimulation to identify corticogeniculate neurons, we distinguish three groups of neurons with response properties that closely resemble those of neurons in the magnocellular, parvocellular, and koniocellular layers of their target structure, the lateral geniculate nucleus (LGN) of the thalamus. Our results indicate that corticogeniculate feedback in the primate is stream specific, and provide strong evidence in support of the view that corticothalamic feedback can influence the transmission of sensory information from the thalamus to the cortex in a stream-selective manner.

Figure 1. Antidromic response latency and classification of corticogeniculate neurons

(A) Distribution of antidromic response latencies for 78 corticogeniculate neurons. Ten of 78 corticogeniculate neurons also received direct suprathreshold geniculocortical input (indicated by grey bars). Dashed line indicates the mean antidromic latency (10.3 ± 0.95 msec). (B) f1 to f0 ratio versus antidromic latency for 40 corticogeniculate neurons: 17 fast complex cells (FC, black diamonds), 10 simple cells (S, red circles), and 13 slow complex cells (SC, blue triangles). Four corticogeniculate neurons (all FC cells) that also received direct geniculocortical input are indicated by unfilled black diamonds. f1 to f0 values for complex corticogeniculate neurons (FC and SC cells) are significantly lower than those for simple corticogeniculate neurons (S cells; p=5×10^-6, Kruskal-Wallis test). (C) Average antidromic latencies of the fast complex (FC), simple (S) and slow complex (SC) cells. Error bars represent SEM. Asterisk indicates that all three classes of corticogeniculate neurons are significantly different from each other in terms of antidromic latency (p=4×10^-8, Kruskal-Wallis test). (D) Visual response latency versus antidromic latency for 17 fast complex cells (FC, black diamonds; corticogeniculate neurons with feedforward input indicated by unfilled diamonds) and 13 slow complex cells (SC, blue triangles). Dashed black line illustrates the linear regression fit to the data (R² = 0.53, p=1×10^-6). (E) Average visual response latencies of fast complex (FC) and slow complex (SC) cells. Error bars represent SEM. Asterisk indicates that the fast complex and slow complex visual response latencies are significantly different (p=4×10^-6, Mann-Whitney U-test).

Figure 2. Contrast response functions and temporal frequency tuning curves

(A) Contrast response functions for three corticogeniculate neurons: a fast complex cell (FC, black diamonds), a simple cell (S, red circles), and a slow complex cell (SC, blue triangles). Data were normalized to peaks and fitted with hyperbolic ratio functions. Error bars represent SEM. (B) Average, normalized contrast response functions for all fast complex (17), simple (10), and slow complex cells (13; conventions as in A). (C) C₅₀ versus antidromic latency for all fast complex, simple, and slow complex cells. Corticogeniculate neurons (all FC cells) that also received feedforward geniculocortical input indicated by unfilled diamonds. (D) Average C₅₀ responses for fast complex (FC), simple (S) and slow complex (SC) cells. Error bars represent SEM. Asterisk indicates that C₅₀ values for simple cells are significantly greater than those of fast complex and slow complex cells (p=6×10^-6, Kruskal-Wallis test). (E) Temporal frequency tuning curves for three corticogeniculate neurons: a fast complex, a simple, and a slow complex cell (conventions as in A). Data were normalized to peaks and fitted with smoothing spline functions. Error bars represent SEM. (F) Average, normalized, spline-smoothed temporal frequency tuning curves for 16 fast complex, 9 simple, and 11 slow complex cells (conventions as in E). (G) Comparison of the highest temporal frequencies to evoke a half-maximum response (TF high₅₀) versus antidromic latency for fast complex, simple, and slow complex cells (conventions as in C). (H) Average TF high₅₀ levels for fast complex (FC), simple (S) and slow complex (SC) cells. Error bars represent SEM. Asterisk indicates that TF high₅₀ values for simple cells are significantly lower than those for fast complex and slow complex cells (p=0.0012, Kruskal-Wallis test). (I) C₅₀ versus TF high₅₀ for fast complex, simple and slow complex cells (conventions as in C and G). Dashed line illustrates the linear regression fit to the data (R² = 0.16, p=0.017).

Figure 3. Extra-classical suppression

(A) Individual area summation tuning curves for three corticogeniculate neurons: a fast complex cell (FC, black diamonds), a simple cell (S, red circles), and a slow complex cell (SC, blue triangles). Data were normalized to peaks and fitted with difference of Gaussian functions. Error bars represent SEM. (B) Average, normalized area summation tuning curves for 14 fast complex cells, 9 simple cells, and 12 slow complex cells fitted with difference of Gaussian functions (conventions as in A). (C) Area suppression index values versus f1 to f0 ratios for fast complex, simple, and slow complex cells (conventions as in A; corticogeniculate neurons receiving feedforward geniculocortical input indicated by unfilled diamonds). Dashed line illustrates the linear regression fit to the data (R² = 0.11, p=0.05). (D) Average area suppression index values for fast complex (FC), simple (S) and slow complex (SC) cells. Error bars represent SEM. Asterisk indicates that simple cells (S) display less extra-classical suppression than complex cells (FC and SC combined; p=0.026, Mann-Whitney U-test).

Figure 4. Orientation tuning and direction selectivity

(A) Individual orientation tuning curves for three corticogeniculate neurons: a fast complex (FC, black diamonds), a simple (S, red circles), and a slow complex cell (SC, blue triangles). Data were normalized and plotted in polar coordinates such that the radial axis ranges from 0 to 1.0 normalized spike rate. Lightened lines represent SEM. (B) Orientation tuning bandwidth (peak half-width at half-maximum response) versus antidromic latency for 17 fast complex, 10 simple, and 13 slow complex cells. Corticogeniculate neurons (4 FC cells) that also received feedforward geniculocortical input are indicated by unfilled diamonds. Dashed line illustrates the linear regression fit to the data (R² = 0.36, p=4×10^-5). (C) Average orientation tuning bandwidth values (half-width at half-maximum response) for fast complex (FC), simple (S), and slow complex (SC) cells. Error bars represent SEM. Asterisk indicates that slow complex cells have significantly larger bandwidth values than fast complex and simple cells (p=0.0004, Kruskal-Wallis test). (D) Direction selectivity index versus antidromic latency for the fast complex, simple, and slow complex cells (conventions as in B). (E) Average direction selectivity index values for fast complex, simple and slow complex cells. Error bars represent SEM. Asterisk indicates that slow complex cells have significantly lower direction selectivity index values than fast complex and simple cells (p=0.05, Kruskal-Wallis test).

Figure 5. Firing rates and cone-contributions

(A) Average maximum and spontaneous firing rates for 17 fast complex cells (FC, black), 10 simple cells (S, red), and 13 slow complex cells (SC, blue). Error bars represent SEM. Asterisk indicates that simple cells have significantly lower maximum firing rates than fast and slow complex cells (p=0.0038, Kruskal-Wallis test). (B) Average L-cone (light red), M-cone (light green), and S-cone (light blue) contributions to the responses of 8 fast complex cells (FC), 8 simple cells (S) and 10 slow complex cells (SC). Error bars represent SEM. Left asterisk indicates that S-cones contribute significantly less to simple cell responses than L- and M-cones (p=0.027, Kruskal-Wallis test). Right asterisk indicates that slow complex cells receive significantly more S-cone input than simple cells (p=0.0075, Kruskal-Wallis test).

Figure 6. Cluster analysis

Dendrogram for a cluster analysis illustrating linkage distances across 34 corticogeniculate neurons based on 6 response parameters (antidromic latency, f1/mean, C₅₀, TF high₅₀, area suppression, and direction selectivity). Fast complex cells (FC) are indicated by black lines and mostly cluster to the left (n = 14), simple cells (S) are indicated by red lines and cluster to the right (n = 9), and slow complex cells (SC) are indicated by blue lines and cluster toward the middle (n = 11).