For perceptual inference, human observers do not estimate sensory uncertainty instantaneously from the current sensory signals alone, but by combining past and current sensory inputs consistent with a Bayesian learner.
New modelling, statistics, and experiments show that cellular populations of mitochondrial DNA (mtDNA) evolve during development according to solvable stochastic dynamics involving binomial partitioning and random turnover, facilitating a predictive and quantitative theory of the mtDNA bottleneck.
Deep neural networks can be trained to automatically find mechanistic models which quantitatively agree with experimental data, providing new opportunities for building and visualizing interpretable models of neural dynamics.
In this ideal example of pharmacogenomics, individuals with a common variant in a gene encoding for an inflammatory lipid mediator benefit selectively from standard-of-care anti-inflammatory treatment used for tuberculous meningitis.
A novel method and software provides researchers with the capability to rapidly, flexibly, and robustly perform Bayesian parameter estimation of theoretically meaningful models in cognitive neuroscience that were heretofore intractable.