Research

There are so, so many cell types in the retina. What is the optimal number of receptive field types for encoding natural videos?

Cell types with progressively larger spatial RFs and briefer temporal RFs form serially as more number of neurons are available. Efficient coding can explain cell type diversity in the retina as well!

Which retinal sensor architectures are optimal for encoding natural scenes?

banner

Simulated saccade movements in natural scenes
Receptive field optimization

Many sensory systems consist of ensembles or grids of ON and OFF detectors spanning sensory space. How should grids of ON and OFF detectors be arranged to optimally encode natural stimuli? It turns out that the preferred detector arrangements depends on sensor noise and the statistical structure of the natural stimuli.

How do multiple sensor types lead to more efficient neural encoding architectures?

banner

In the retina, bipolar cells are located between input photoreceptors and output RGCs, allowing multiple stages of nonlinear processing. There are multiple types of bipolar cells, with each type known to be activated by specific attributes of visual stimuli such as luminance and the presence or absence of colored inputs. What’s the role of these specialized sensor types in information processing? Can architecture and the neural diversity found in the retina make information processing more efficient?