A Dendro-Centric View of Cognition in the Behaving Brain

The neuron is no longer viewed as a simple point-like integrator but as a sophisticated computational device whose power resides in its dendritic arbor. This review charts the paradigm shift driven by this new perspective. We synthesize recent in vivo findings from behaving animals, where active dendritic processes, from local nonlinear spikes to compartment-specific plasticity, are revealed to be fundamental for intelligence, including perception, action, and memory across key brain regions. We then explore how these biological mechanisms are formalized in theoretical and circuit models that explain complex neural computations at the network level. Finally, we highlight the potential of these findings for artificial intelligence, arguing that dendritic computation offers a compelling source of inspiration for future learning algorithms and hardware systems. This review solidifies dendritic computation as a cornerstone of modern neuroscience, linking cellular mechanisms to the principles of intelligent systems.