By Frank Amthor

As information processors, neurons receive information from other neurons, perform computations on that information, and send the output of those computations to other neurons.

Neurons receive information via synapses. Most neurons have synapses on their cell bodies, and the first neurons probably were limited to such. However, synapses have a minimum size so that a normal sized cell could only have a maximum of a few hundred synapses on its cell body.

In order to have more synapses, cell bodies could grow larger, but doing so costs too much (a cubic change in volume translates to a square change in area). Neurons solved this problem by extending roughly cylindrical processes called dendrites such that a linear increase in length and volume produced a squared increase in area. Dendrites allow neurons to receive tens of thousands of synapses.

A single neuron is an enormously complicated computing device. Here, very briefly, are some key things to know about synapses and dendrites:

  • Synapses are either electrical or chemical. Electrical synapses are simple and fast, but inefficient and inflexible. Chemical synapses are more powerful and flexible, but they’re slower than electrical synapses.
  • The dendritic tree is a place for synapses that input to the neuron. But dendritic trees do more than just provide space for synapses. They also allow synapses that are close to each other to interact nonlinearly. Synapses close to the cell body have larger effects on the current spreading to the cell body than synapses that are farther away. The dendrites’ branching structure is also important because it provides numerous computational subunits at a level below that of the neuron itself, allowing tens of thousands of time-varying presynaptic inputs to interact in dynamically complex ways to produce a final output.