Eric Hunsberger

PhD student

Just prior to the Mayan apocalypse, I transferred from the Master's program into the PhD program to continue my research with Chris Eliasmith and Jeff Orchard (my co-supervisors) on the human visual system.

For my PhD research, I will focus on the problem of depth-from-motion, which describes how people can infer the depth of objects in the environment from how these objects move relative to one another. A simple example is how, as you walk through a room, objects close to you will move quickly in your visual field and objects far away will remain mostly stationary. To infer depth information from motion, your visual system first determines the motion of objects in the incoming visual stimulus; this is called optical flow. Optical flow, along with other information that your brain gets from your vestibular (inner ear) and proprioceptive (muscular) systems, is used by your brain to infer your motion in the environment. Your brain can then combine your motion and the motion of the objects around you to infer the depth of those objects.

I plan to model this process from start to finish. My model will ultimately run in simulated spiking neurons, meaning that the model is constrained to use the same tools as the brain. By constraining the model like this, I will be able to select between the many various algorithms for performing optical flow and depth-from-motion, since any algorithm I choose must be implementable in neurons. Neurons are very flexible computational devices, but are better suited to some types of computations than others, a distinction which helps greatly in choosing between algorithms. Ultimately, if and when the model runs in spiking neurons, it will demonstrate not necessarily the way that the brain performs depth-from motion, but will present a way that the brain could perform depth-from-motion. Ideally, the model will make both behavioural and neuroscientific predictions that can be tested to help validate the model.

Publications