Timezone: »

A Framework for Testing Identifiability of Bayesian Models of Perception
Luigi Acerbi · Wei Ji Ma · Sethu Vijayakumar

Thu Dec 11 11:00 AM -- 03:00 PM (PST) @ Level 2, room 210D #None

Bayesian observer models are very effective in describing human performance in perceptual tasks, so much so that they are trusted to faithfully recover hidden mental representations of priors, likelihoods, or loss functions from the data. However, the intrinsic degeneracy of the Bayesian framework, as multiple combinations of elements can yield empirically indistinguishable results, prompts the question of model identifiability. We propose a novel framework for a systematic testing of the identifiability of a significant class of Bayesian observer models, with practical applications for improving experimental design. We examine the theoretical identifiability of the inferred internal representations in two case studies. First, we show which experimental designs work better to remove the underlying degeneracy in a time interval estimation task. Second, we find that the reconstructed representations in a speed perception task under a slow-speed prior are fairly robust.

Author Information

Luigi Acerbi (University of Geneva)

I am a computational neuroscientist and machine learning researcher. I recently joined Alex Pouget‘s lab in Geneva, Switzerland. From 2014 to 2017, I was a postdoc in Wei Ji Ma‘s lab at NYU, NY, USA. I obtained my PhD at the Doctoral Training Centre for computational neuroscience based in Edinburgh, UK, under the supervision of Sethu Vijayakumar and Daniel Wolpert. I spent several months of my PhD at the Computational and Biological Learning Lab in Cambridge, UK. My research in computational neuroscience focuses on how the brain combines different sources of perceptual information when there are multiple possible underlying explanations (causal inference). I also study how the brain represents and computes with uncertainty. I explore these question with mathematical modelling, computational analysis and human psychophysics. In my applied machine learning research, I have been developing methods for model fitting and approximate inference, building upon Bayesian optimization, variational inference and Monte Carlo techniques.

Wei Ji Ma (New York University)
Sethu Vijayakumar (University of Edinburgh)

More from the Same Authors