Detection of Code Juggling with Spike Count Data Analysis

Fluctuating activities may manifest in a single neuron’s activation patterns when preserving information from multiple stimuli. Considering two simultaneously presented stimuli for simplicity, we introduce a code juggling paradigm which relates the code associated with a composite signal to the elementary codes associated with each constituent stimulus. We model four potential biologically meaningful modes of activation patterns using Poisson mixtures with unknown mixing distributions and construct a hypothesis testing between these competing modes. Under a Bayesian paradigm, the testing requires an integration of an infinite-dimensional nonparametric mixing distribution for the marginal likelihood estimation, bringing a practicability challenge. We circumvent this computing problem by adopting a pseudo-Bayesian framework of Predictive Recursion Marginal Likelihood (PRML) and design a PRML-based test. Numerical experiments suggest our PRML test has a decent detection rate of code juggling. We re-analyze four datasets in previous studies, confirming the existence of code juggling in multiple brain areas and explore how coding patterns associate to object identification and attention.