As part of my research I regularly used machine learning techniques like Fisher's linear discriminant to discern what a human subject was seeing from their brain activity during a behavioural task. This established a solid understanding of machine learning principles such as data labelling, model training and best practices to avoid overfitting experimental data.

I worked with a custom data pipeline and software tools developed by our lab head in matlab. This meant working in matlab to perform matrix calculations and apply machine learning and statistical analysis techniques. It also meant regularly debugging the pipeline for issues which came about due to changes in the methods used to acquire experimental data.

My work gave me first hand experience of performing fMRI experiments with visual stimuli and human subjects, as well as measuring human behaviour using visual psychophysics. My early focus was on how large draining veins affect fMRI data recorded from the visual cortex. During my research I made great use of signal detection theory, which despite being pioneered for radar, has long been used to study cortical activity in the visual system. My later research was particularly focused on how bias, which is central to signal detection theory, might be represented in the human visual system.