Active dendrites enable robust spiking computations despite timing jitter

Area of Science:

• Neuroscience
• Computational Neuroscience
• Biophysics

Background:

• Dendritic action potentials feature prolonged plateaus (tens of milliseconds), significantly longer than axonal spikes.
• The computational function of these slow dendritic events contrasts with the rapid information processing demands of nervous systems.

Purpose of the Study:

• To investigate the computational role of long-lived dendritic potentials in integrating asynchronous neural inputs.
• To develop a physiologically grounded model capturing dendritic nonlinearities and their computational implications.

Main Methods:

• Development of a simplified, physiologically grounded model of dendritic plateau potentials.
• Utilizing the model to analyze neuronal responses to asynchronous input spikes.
• Simulating a neural network performing an association/discrimination task with sparse spiking.

Main Results:

• The model demonstrates that long dendritic plateau potentials provide a resettable memory, enabling reliable neuronal spiking with asynchronous inputs.
• The model supports non-trivial computations in a network, even with timing jitter in sparse spiking.
• Dendritic potentials facilitate rapid, reliable decisions using a low number of spikes.

Conclusions:

• The timescale of dendritic potentials is crucial for reliable integration of asynchronous signals, supporting efficient neural computation.
• The findings offer testable hypotheses for dendritic action potential roles in cortical function.
• The model presents a bio-inspired approach for neuromorphic computing in analog hardware.