Characterisation of receptive fields in the mouse retina

Abstract

Approaches to mapping the receptive fields (RFs) of neurons in the visual system have been in development since the mid-twentieth century. Traditional RF mapping techniques include small exploratory spots of light, spots of variable sizes, bars, drifting gratings, and the current most popular method, spatiotemporal white noise. In this thesis, a recently developed RF mapping technique is employed, which entails a flashing bar stimulus and the inverse Radon transform, and is termed ‘RNI’ (Radon – Non-negative matrix factorisation – Iterative Reconstruction). This approach presents several advantages over the aforementioned techniques. For example, the RNI technique can capture elliptical RFs, whereas spots of variable sizes and drifting gratings assume circular symmetry. Additionally, RNI is able to capture both linear and non-linear features of RFs. This thesis will explore the use of the RNI technique in extracting properties from the RFs of different ganglion cell types in the mouse retina. Chapter 1 provides an overview of visual processing in the retina, following which the classical and more recently-developed RF mapping techniques are described. Chapter 2 investigates the use of the RNI technique in extracting different features of ganglion cell RFs, including the concentric centre/surround structure, and temporal features. Chapter 3 extends on the characterisation of RF features described in chapter 2, using a gaussian model to quantify components and obtain RF parameters, which were compared to measurements obtained using a traditional spot stimulus. Finally, chapter 4 draws correlations between RFs and underlying morphological properties and mechanisms of ganglion cells.