Whole-brain computational modelling, incorporating novel ML-based parameter estimation techniques, reveals how transcranial magnetic stimulation-evoked brain responses are driven at earlier timepoints by local echoes of the external stimulus, and at later timepoints by large-scale network reverberation across the connectome.

A proposed model of optimal inference under cognitive costs accounts for human sequential effects, including subtle patterns of attractive and repulsive influence of past observations, in a binary prediction task across a wide range of stimulus conditions.

A shift in fitness optimum of a polygenic trait rapidly introduces small frequency differences between alleles with effects aligned with and opposing the shift, which gradually translate into small differences in fixation probability.

A new Bayesian algorithm for tracking subjects’ choice strategies on every trial reveals when subjects learn and what they tried while doing so, providing strong evidence that reward- and loss-driven exploration change independently.

A bulk barcoded quantitative trait locus approach increases the power and resolution of genotype-phenotype mapping in yeast, revealing that the genetic architecture of 18 complex traits is highly polygenic, and is characterized by widespread epistatic interactions and pleiotropic effects.

Network symmetry represents a new vantage point for dissecting complex information processing characteristics in multisite modification, and the breaking of symmetry can confer ordering of modification and absolute concentration robustness.

Mice can use self-motion cues to learn a cognitive map thereby permitting them to take shortcut trajectories in an open maze without requiring prior experience of such routes.