Hybrid brain network models predict neurophysiological processes that link structural and functional empirical data across scales and modalities in order to better understand neural information processing and its relation to brain function.

Combining biologically-plausible neural network models of vision with traditional decision-making models enables a detailed characterization of how the visual system extracts representations that guide decisions from raw sensory inputs.

A neural network showed better prediction of upcoming states when it was selective in when it encoded and retrieved episodic memories, thereby explaining why humans show this selectivity in studies of naturalistic memory.

Biologically plausible changes in the excitabilities of single neurons may suffice to selectively modulate sequential network dynamics, without modifying of recurrent connectivity.

For brain imaging to be useful despite its limitations in measuring neural activity, the neural code must be smooth both in a traditional sense and functionally.

Two-photon Ca²⁺ imaging and computational modeling reveal major developmental trajectories of spontaneous activity in developing CA1 and identify important roles of network bi-stability and synaptic input characteristics for hippocampal burstiness before eye opening.

A comprehensive, data-driven and interpretable nonlinear computational modeling framework based on deep neural networks uncovers different nonlinear transformations of speech signal in the human auditory cortex.

The age-dependent shift of prefrontal excitation-inhibition (E-I) ratio toward inhibition causes sparser and decorrelated activity, while its impairment might relate to neurodevelopmental disorders.