Tracking individual mice within a group makes it possible to quantify and model social behavior, and reveals that the behavior of the group goes beyond pairwise interactions, and is shaped by complex interactions between multiple individuals.

In baboon gut microbiota, most pairwise correlations in bacterial abundances are weak and negative, and bacterial correlation patterns are largely shared across hosts, rather than personalized to each hosts.

Auditory cortical columns contain small subsets of neurons with highly synchronous activity (cNEs) that create robust sequences of coordinated activity suitable for enhanced information processing and signal transmission beyond the capability of individual neurons.

Aberration correction using 3D microprinting in ultrathin microendoscopes allows two-photon imaging of large neuronal networks with homogeneously high spatial resolution and minimal invasiveness in the deep mouse brain.

Visual sensitivity to correlation patterns in rats matches that previously measured in humans, as well as predictions from efficient coding theory based on the statistics of natural images.

Low-frequency correlations among neurons in monkey visual area V4 impair the animal's ability to perform an attention-demanding task, suggesting a causal role of these fluctuations in perception.

A high-throughput analysis of several self-cleaving ribozymes reveals the effect of every possible mutation, and every possible pair of mutations, and how patterns in the relative activity data can be mapped to canonical and non-canonical structural elements.

Psychophysical measures of human sensitivity to visual patterns reveal that the brain preferentially processes those features of stimuli that are more variable in the natural world.

Volumetric lifespan change in subcortical structures follows fundamental principles of brain development, lifespan continuity, genetic organization, and age-invariant relationships to cognition.

An inference paradigm for extracting neuronal correlations from two-photon imaging data, without requiring intermediate spike deconvolution, provides significant performance gains over existing methods as demonstrated by theoretical analysis, simulation studies, and real-data applications.