I do brain science. That’s the briefest version. A slightly more elaborate description of my research interests follows below. You may also be interested in my publications and the Paper Decoder blog.
Research Interests
I’m interested in the principles of information processing in the brain and work under the assumption that uncertainty is an important component of it, because in most tasks the brain experiences only incomplete and noisy information about the world. Because Bayesian inference provides a principled means of processing uncertain information in all kinds of situations, my working hypothesis is that the brain implements Bayesian inference (the Bayesian brain hypothesis). I believe that in its most general formulation the Bayesian brain hypothesis is not falsifiable, i.e., I could argue that we already know that the brain does, or approximates Bayesian inference in one form or another. It is all the more important then to constrain the theory by applying it to particular tasks and situations and see whether it fits or predicts the observations we make from real brains.
I have found that perceptual decision making is an ideal testbed for investigating the nuances of Bayesian inference in the brain, because it is the simplest, established paradigm which allows us to control the uncertainty about a stimulus and to observe the outcome of information processing in form of behaviour. Importantly, there is a straightforward theoretical link between uncertain stimulus and the speed of information processing in perceptual decision making: more uncertainty, or noise in the stimulus leads to slower processing. This link highlights the temporal nature of information processing and provides meaning to variation in the time needed to respond in perceptual decision making tasks. This, in turn, allows us to use response timing to inform our analyses of information processing in the brain.
I have previously shown that more classical, drift diffusion models of perceptual decision making are a special case of Bayesian inference [Bitzer2014]. I have also demonstrated that attractor models of perceptual decision making can be embedded in a Bayesian framework to extend the more classical models, for example, to situations in which the stimulus may switch unexpectedly [Bitzer2015]. I have, further, argued that the brain adapts its uncertainty about a stimulus to the noise in the stimulus even on the fast (hundreds of ms) timescale of perceptual decisions [Bitzer2015a]. In work with Hame Park [Park2016] we show that transient stimulus features influence perceptual decisions and, especially, that they can enrich possible inferences about the underlying information processing mechanisms. In the future I seek to further broaden the set of situations and stimuli to which we can apply Bayesian inference ideas in perceptual decision making and am looking forward to the new insights into information processing in the brain that this will provide.