Exploring the Mechanisms Driving Temporally Dynamic Spatial Frequency Processing in the Primary Visual Cortex of Mice

Neurons of the primary visual cortex (V1) process spatial frequency information in a sequential fashion whereby low spatial frequencies (SFs) are processed prior to high SFs. This phenomenon, known as coarse-to-fine processing, has been well documented across several mammalian species and is involved in several important visual computations such as feature-linking, core object recognition, and the efficient coding of natural scenes. Despite the importance of coarse-to-fine processing in visual computation, a complete description of the mechanisms behind this phenomenon has yet to be established. Prior studies have questioned whether coarse-to-fine processing stems from the primary feedforward visual pathway or from local cortical interactions and feedback from higher visual areas. This study attempts to investigate this question by determining whether a purely feedforward model can produce coarse-to-fine-like responses and whether such a model is biologically plausible. I have found that a biologically plausible feedforward model can only produce coarse-to-fine-like responses that are limited to the lower band of SFs; that is, the model responded to progressively higher SFs with progressively longer time lags but did not respond to SFs beyond a certain value. This result indicates that a purely feedforward model cannot explain the high SF portion of the coarse-to-fine response but may explain the low SF portion. However, because recorded V1 neurons still show prominent responses to high SF, high SF information is likely present in the feedforward pathway and is potentially enhanced via intracortical interactions or feedback from higher visual areas.