What Are Dendrites (neurons)?

Dendrites are the branched, tree-like extensions of a neuron that receive synaptic inputs from other neurons and integrate those signals before transmitting the result toward the cell body and axon. A single neuron in the cerebral cortex may bear thousands of synaptic contacts distributed across its dendritic arbor, making dendrites the primary site of information convergence in neural circuits. The term shares its etymology with the word for crystalline branching structures in materials science, both deriving from the Greek for tree, but neuronal dendrites are functionally distinct: they are active computational structures capable of amplifying, filtering, and transforming synaptic inputs rather than passively summing them.

Neuronal dendrites are studied in neuroscience, neurophysiology, and computational neuroscience, and they have increasingly attracted attention in electrical engineering and computer science as researchers seek to incorporate dendritic computation principles into artificial neural network architectures.

Morphology and Branching

Dendritic morphology varies considerably across neuron types and brain regions. Pyramidal neurons in the neocortex bear an apical dendrite extending toward the cortical surface and a set of shorter basal dendrites radiating from the cell body; this geometry creates spatially segregated compartments that respond differently to inputs depending on their location along the dendritic tree. Purkinje cells in the cerebellum have elaborately branched, planar arbors that receive on the order of 100,000 synaptic inputs. Dendritic spines, small protrusions that host the majority of excitatory synapses, are sites of calcium signaling and synaptic plasticity. The geometry of the dendritic tree determines the spatial and temporal filtering properties of the neuron: distal inputs are electrotonically attenuated relative to proximal ones, and branching patterns affect how inputs from different sources interact. Research published in the Annual Review of Neuroscience on dendritic computation provides a comprehensive account of how dendritic morphology shapes computational function.

Signal Integration and Nonlinear Computation

Dendrites are not passive cables. Voltage-gated ion channels distributed along the dendritic membrane enable active signal propagation, amplification, and local spiking events that are distinct from the output spike generated at the axon initial segment. NMDA receptor-mediated nonlinearities allow individual dendritic branches to function as local coincidence detectors, responding supralinearly when spatially clustered inputs arrive simultaneously. This branch-level computation substantially expands the computational repertoire of a single neuron beyond what a simple integrate-and-fire model would predict. Studies in the Journal of Neuroscience on dendritic mechanisms in vivo demonstrate that pyramidal neurons use dendritic nonlinearities to multiplex contextual signals arriving at apical dendrites with sensory inputs arriving at basal dendrites, enabling flexible, context-dependent computation.

Dendritic Plasticity and Learning

Dendritic structure and function are not fixed; they change in response to experience and activity. Synaptic long-term potentiation (LTP) and long-term depression (LTD) alter the strength of individual synapses by modifying AMPA receptor density at dendritic spines, providing the cellular substrate for learning and memory. Structural plasticity includes spine growth and retraction, changes in dendritic branching, and the formation or elimination of entire synaptic connections. These mechanisms collectively form the basis of Hebbian learning in neural circuits. Research has shown that dendritic mechanisms offer solutions to core problems in machine learning, including credit assignment across network layers and resistance to catastrophic forgetting, as reviewed in Nature Communications research on dendrites in artificial neural networks.

Applications

Dendrites in neuroscience have direct relevance to several research and engineering domains, including:

• Neuromorphic computing, where dendritic branching and nonlinearity inspire hardware and software architectures for energy-efficient neural computation
• Neuropharmacology, where dendritic ion channels are targets for drugs treating epilepsy, pain, and neuropsychiatric disorders
• Brain-machine interfaces, where understanding dendritic signal integration guides the interpretation of local field potentials
• Computational neuroscience, where biophysical dendritic models constrain theories of learning and network function
• Artificial intelligence, where dendritic computation principles improve parameter efficiency and robustness in deep neural networks