Strong surround antagonism in the dLGN of the awake rat

Classical center-surround antagonism in the early visual system is thought to serve important functions such as enhancing edge detection and increasing sparseness. The relative strength of the center and surround determine the specific computation achieved. For example, weak surrounds achieve low-pass spatial frequency filtering and are optimal for denoising when signal-to-noise ratio (SNR) is low. Balanced surrounds achieve band-pass spatial frequency filtering and are optimal for decorrelation of responses when SNR is high. Surround strength has been measured in the retina and dorsal Lateral Geniculate Nucleus (dLGN), primarily in anesthetized or ex vivo preparations. Here we revisit the center-surround architecture of dLGN neurons in the un-anesthetized rat. We report the spatial frequency tuning responses of N=47 neurons. We fit these tuning curves to a difference-of-Gaussians (DOG) model of the spatial receptive field. We find that some dLGN neurons in the awake rat (N=8/47) have weak surrounds. The majority of cells in our sample (N=29/47), however, have well-balanced center and surround strengths and band-pass tuning curves. We also observed several neurons (N=10/47) with notched or dual-band-pass tuning curves, a response class that has not been described previously. Within the space of circularly concentric DOG models, strong surrounds were necessary and sufficient to explain the dual-band-pass spatial frequency tuning of these cells. It remains to be determined what advantage if any is conferred by this novel response class, or by the heterogeneity of surround strength as such. We conclude that surround antagonism can be strong in the dLGN of the awake rat.