Spike-triggered average

A method for analyzing neural responses to stimuli

Overview

The spike-triggered average (STA) is a computational technique used in neuroscience to analyze the relationship between neural spikes and external stimuli. It is a method for determining the average stimulus that precedes a neuron's action potential, or "spike." This technique is particularly useful for understanding how neurons encode sensory information and for characterizing the receptive field properties of sensory neurons.

Methodology

The STA is calculated by averaging the stimuli that occur before each spike over a set of recorded spikes. This involves the following steps:

1. Recording Spikes and Stimuli: First, a neuron's spikes are recorded along with the corresponding stimuli over a period of time. The stimuli can be visual, auditory, or any other type of sensory input.
2. Aligning Stimuli to Spikes: For each spike, the stimulus that occurred immediately before the spike is extracted. This is typically done for a fixed time window preceding the spike.
3. Averaging: The extracted stimuli are averaged across all spikes to produce the spike-triggered average. This average represents the typical stimulus pattern that leads to a spike.

Illustration of the spike-triggered average process

Applications

The STA is widely used in the study of sensory systems. It helps in identifying the features of stimuli that are most effective in driving a neuron's response. For example, in the visual system, the STA can reveal the preferred orientation or spatial frequency of a visual neuron. In the auditory system, it can help identify the frequency or temporal patterns that a neuron is most responsive to.

Limitations

While the STA is a powerful tool, it has limitations. It assumes that the neuron's response is linearly related to the stimulus, which may not always be the case. Nonlinearities in neural processing can lead to inaccuracies in the STA. Additionally, the STA requires a large number of spikes to produce a reliable average, which can be a limitation in experiments with low firing rates.

Related pages

• Neural coding
• Receptive field
• Action potential
• Neuroscience