Extracellular electrophysiology#
Understanding how the brain produces cognition, emotion, and behavior requires that we measure the signals that neurons are sending to one another. One powerful tool to do this is extracellular electrophysiology.
The basic premise: listening to the concert from the outside#
Extracellular electrophysiology is analogous to eavesdropping on the electrical conversations of neurons from outside their cell membranes. Unlike its counterpart, intracellular electrophysiology, which provides detailed insights from within a single neuron, placing electrodes in the extracellular space allows better experimental access to intact brains, and in some configurations allows us to record the collective activity of multiple neurons in their natural ensemble. The primary readouts of this technique are extracellular action potentials or ‘spikes’ and local field potentials (LFPs), each offering distinct insights into neuronal dynamics.
Spikes and LFPs: the dual storytellers#
spikes: These are the digital currency of the brain. Each spike signifies the action potential of a neuron, which propagates down that neuron’s axon and causes neurotransmitters to be released onto downstream neurons. In vertebrates, the majority of synaptic release is triggered by all-or-none action potentials, and as a result recording the action potentials from a neuron to a first approximation describes that neuron’s contribution to the ongoing computations occurring in the brain. With advancements in multi-electrode arrays and silicon probes, we can now concurrently monitor the spiking activity of hundreds, if not thousands, of neurons spanning many brain regions.
local field potentials (LFP): These oscillatory signals are slower and reflect the summed activity of many neurons, encapsulating both their synaptic potentials and the inputs they receive. The power and phase of LFPs across different frequency bands provide a measure of network synchrony, connectivity, and information flow, providing additional information to understand brain dynamics.
Techniques and technologies: from glass pipettes to silicon probes#
The tools which neuroscientists use to measure extracellular voltages has evolved with the field. From glass pipettes to tetrodes, multi-electrode arrays, and now silicon probes, the granularity and scale of recordings possible with these techniques has markedly grown.
The silicon revolution: silicon probes in electrophysiology#
Silicon probes are microfabricated devices that pack many electrodes together at very high spatial densities. For the reasons below, silicon probes have dramatically improved our capacity to observe neural dynamics:
High-density recording: The sheer density of electrodes on a silicon probe allows the measurement of an individual action potential with multiple electrodes, or channels. It also allows multiple regions of the brain to measured at the same time, as a single shank of silicon can have electrodes patterned along its length.
Enhanced spike sorting: Observing individual action potentials on several electrodes inherently improves spike sorting—the process of attributing spikes to specific neurons. With traditional methods, overlapping spikes from neighboring neurons were challenging to disentangle. Silicon probes, due to their spatial configuration, offer higher-resolution data, aiding more accurate spike discrimination and sorting.
Minimally invasive: Their microscale size ensures reduced tissue damage for the same number of electrodes, permitting longer recording sessions with maintained tissue integrity.
Spike sorting: isolating discrete action potentials from continuous voltage recordings#
While extracellular recordings allow us to eavesdrop on these neuronal conversations, the signals we obtain are typically a mixture of multiple neurons’ activities intermingled with noise. Disentangling this mixture signal to identify the action potentials of individual neurons is where the essential technique of spike sorting comes into play.
What is spike sorting?#
Spike sorting is a computational process used to identify and categorize the action potentials—or ‘spikes’—of individual neurons from the continuous voltage recordings obtained during extracellular electrophysiology experiments. Given that multiple nearby neurons can contribute to the recorded signals, it’s not always straightforward to determine which spike belongs to which neuron. This problem is an example of blind source separation, where the objective is to identify the contribution of distinct sources with unknown properties to the mixed signal. Spike sorting aims to clarify this ambiguity.
The nitty-gritty of spike sorting#
Detection: The first step involves identifying significant voltage deflections amidst background noise. Typically, this involves setting a threshold value, and any signal exceeding this threshold is considered a potential spike.
Extraction: Once spikes are detected, they are extracted—often as short snippets of the continuous voltage trace to be analyzed further.
Feature Extraction: To differentiate spikes from different neurons, the waveform of each spike is transformed into a set of features that describe its shape. Techniques like principal component analysis (PCA) are often employed to reduce the dimensionality of the data while retaining the most distinguishing characteristics of different spike waveforms.
Clustering: Based on the extracted features, spikes are grouped into clusters, with each cluster ideally representing spikes from a single neuron. Various algorithms, ranging from k-means clustering to more sophisticated probabilistic models, are used to achieve this segregation. Kilosort is an algorithm specifically tailored to perform spike sorting with data collected from silicon probes.
Validation: Post clustering, it’s essential to validate the results, ensuring that each cluster indeed represents a distinct neuron and not an artifact or overlapping spikes from multiple neurons. This often involves examining the inter-spike intervals and the refractory period, as neurons typically have a brief period after firing during which they cannot fire again.
Why spike sorting matters#
Spike sorting is crucial for several reasons:
Neuronal Identity: It allows neuroscientists to attribute recorded spikes to individual neurons, providing insights into the firing patterns and roles of specific cells in neural circuits.
Network Analysis: By understanding which neuron emits which spike, researchers can map out interactions and connectivity patterns within neuronal networks.
In summary, spike sorting is an indispensable tool in modern neuroscience, bridging the gap between raw, continuous voltage recordings and the discrete, individualistic firing patterns of neurons.