Panayiota Siskos ’23
While mice are useful for investigating components of visual perception, this research is limited by insufficient knowledge of the organization of the visual cortex of the mice. Visual information is processed via computations while traveling from the retina to lateral geniculated nucleus and visual cortices. The early visual system processes complex visual stimuli by encoding various stimulus attributes by individual neurons while higher order visual cortices process nonlinear features. While global topographic organization of the mouse visual cortex is known, neural population code and how information is represented in neural activity is unknown. Due to neural responses from a population of over one hundred visual cortical neurons, linear classifiers have high accuracy decoding two decoding tasks, one with six stimulus classes with complex spatiotemporal features and one with eight drifting grating directions. There is differential decoding accuracy between primary, lateral, anterolateral, anteromedial, posteromedial, and rostrolateral visual areas implying differential information representation in visual areas, as well as differences between populations from different cortical depths with the superficial layer populations having more information than deeper layers. In addition, there is evidence that directional turning in neurons does not predict population decoding accuracy, indicating distributed representation of information. The aim of this study is to characterize population neural code associated with cortical organization of visual information processing.
For every one of 186 mice, one of six genetically tagged cell types and one of six visual areas were targeted while they were viewing visual stimuli consisting of static images and movies with complex long-range correlations. Linear classifiers were trained to decode one of six visual stimuli categories with distinct spatiotemporal structures from population neural activity.
Neurons in primary visual cortex and secondary visual areas have various levels of stimulus-specific decodability. Neurons in superficial layers are more informative concerning stimulus categories. Further decoding analysis of directional motion is consistent with the findings. Synergy was also observed in population code of direction in visual areas indicating area-specific organization across neurons of information representation. Such differences in decoding capacity shed information on specialized organization of neural information processing across anatomically different populations and establishes the mouse as a model to understand visual perception. This study is significant because it gives evidence of functional and anatomic organization of the mouse visual cortex and corroborates trends in visual information processing and supports the existence of information processing streams. Further directions for this study include seeing if anatomic trends in stimulus-specific decoding is due to certain input pathways from the primary visual cortex.
 K. Esfahany, et al., Organization of Neural Population Code in Mouse Visual System. eNeuro 5, (2018) doi: https://doi.org/10.1523/ENEURO.0414-17.2018
 Image retrieved from: https://www.pexels.com/photo/close-up-photo-of-white-mice-4052861/