Available online at www.sciencedirect.com
ScienceDirect Challenges in the quantification and interpretation of spike-LFP relationships Supratim Ray Brain signals often show fluctuations in particular frequency bands, which are highly conserved across species and are associated with specific behavioural states. Such rhythmic patterns can be captured in the local field potential (LFP), which is obtained by low-pass filtering the extracellular signal recorded from microelectrodes. However, LFP also captures other neural processes that are associated with spikes, such as synaptic events preceding a spike, low-frequency component of the action potential (‘‘spike bleed-through’’) and spike afterhyperpolarization, which pose difficulties in the estimation of the amplitude and phase of the rhythm with respect to spikes. Here we discuss these issues and different techniques that have been used to dissociate the rhythm from other neural events in the LFP. Addresses Centre for Neuroscience, Indian Institute of Science, Bangalore 560012, India Corresponding author: Ray, Supratim ([email protected])
Current Opinion in Neurobiology 2015, 31:111–118 This review comes from a themed issue on Brain rhythms and dynamic coordination Edited by Gyo¨rgy Buzsa´ki and Walter Freeman For a complete overview see the Issue and the Editorial Available online 3rd October 2014 http://dx.doi.org/10.1016/j.conb.2014.09.004 0959-4388/# 2014 Elsevier Ltd. All rights reserved.
the excitability of these neurons varies within one period of the rhythm [10–12], which can be used to synchronize the spiking of neurons that process information about different features of a stimulus (BBS), generate temporal windows of communication between two assemblies (CTC), or encode information in the position of the spike relative to the rhythm (PC). Some of the hypotheses can be tested by recording spikes from multiple neurons and checking for the presence of rhythmicity and phase lag in the cross-correlation function computed using spike trains. However, if the phase locking of the spike with the rhythm is weak (which is typically the case, at least for gamma), rhythmicity is often difficult to detect using spike trains alone ([13], Supplementary Figure 3 of [14]). Local field potential (LFP), which is obtained by low-pass filtering the raw signal (typically below 300–500 Hz) recorded from a microelectrode and thought to mainly reflect the summed transmembrane currents flowing through the neurons within a local region around the microelectrode tip [15,16,17,18], is therefore often used to capture the rhythm. This review focuses on the problems faced by the researcher who, from the LFP, tries to accurately estimate the phase (and amplitude) of the intracellular periodic fluctuations in a neuron (generated, for example, by an inhibitory network) relative to its spikes. This is a challenging problem because LFP also captures other synaptic and non-synaptic events related to spiking [17,18] (summarized in Figure 1), which overlap with the rhythm in both spectral and temporal domains.
LFP events associated with spikes A distinctive feature of brain signals is the presence of oscillations at different frequencies that are highly conserved across species [1] and are coupled to specific behavioural states and stimulus features [2]. Understanding the relationship between oscillations and underlying neuronal processes is an important goal in neuroscience [3], and over the last two decades, several ideas have been put forward regarding a potential role of rhythms in cognitive processing, such as binding-by-synchrony (BBS) [4], communication-through-coherence (CTC) [5,6] and phase coding (PC) [7–9] hypotheses. These are based on the observation that several rhythms (such as gamma, which has a centre frequency between 30 and 80 Hz) are generated through inhibitory networks that produce periodic fluctuations in the intracellular potential of the target post-synaptic neurons such that www.sciencedirect.com
Excitatory inputs arriving from an upstream area generate EPSPs in the post-synaptic cell. The effect of these EPSPs on the field potential can be estimated by computing the mean target LFP response following spikes in the input region. For example, V1 LFP triggered on the spiking of an LGN neuron shows a transient synaptic response of 5 ms duration (Figure 2a; [19,20]). If the target neuron also fires, the extracellular action potential (EAP) shows a transient response whose shape varies depending on the position of the microelectrode (Figure 2b, [17,18,21,22,23,24]). Near the soma where the EAP amplitude is largest, it has a prominent negative dip followed by a slower positive upswing. EAP has power at all frequencies, including low frequencies, and is present even after low pass filtering (spike ‘bleedthrough’; Figure 2c; [25]; shape of the bleed-through depends on the low-pass filter settings). Taken together, Current Opinion in Neurobiology 2015, 31:111–118
112 Brain rhythms and dynamic coordination
Figure 1
(a)
(b)
Inputs from thalamus
Local Field Potential (LFP) events around a spike
10 0 –10
Rhythmic inhibition synaptic events (unfiltered) synaptic events (0–300 Hz)
Microelectrode tip
Extracellular Action Potential (EAP) EAP below 300 Hz: “bleed-through” Post-spike upswing
Rhythmic inhibition from an inhibitory network
LFP: sum of the components above
10
Goal: to extract the phase of the rhythmic inhibition (blue trace on top) from the LFP
0 –10 –50
0
50
Time (ms) Current Opinion in Neurobiology
Summary of LFP events locked to a spike. (a) A pyramidal neuron (grey) receives thalamic inputs (red) and rhythmic inhibition from an interneuron network (blue). A researcher tests the hypothesis that the rhythmic inhibition biases the timing of spikes produced by the pyramidal cell by inserting a microelectrode (tip shown as a black dot) and measuring the extracellular potential. Action potentials produced by the pyramidal cell and the rhythmic intracellular fluctuations are studied by filtering the raw signal in high (>1000 Hz) and low (
ScienceDirect Challenges in the quantification and interpretation of spike-LFP relationships Supratim Ray Brain signals often show fluctuations in particular frequency bands, which are highly conserved across species and are associated with specific behavioural states. Such rhythmic patterns can be captured in the local field potential (LFP), which is obtained by low-pass filtering the extracellular signal recorded from microelectrodes. However, LFP also captures other neural processes that are associated with spikes, such as synaptic events preceding a spike, low-frequency component of the action potential (‘‘spike bleed-through’’) and spike afterhyperpolarization, which pose difficulties in the estimation of the amplitude and phase of the rhythm with respect to spikes. Here we discuss these issues and different techniques that have been used to dissociate the rhythm from other neural events in the LFP. Addresses Centre for Neuroscience, Indian Institute of Science, Bangalore 560012, India Corresponding author: Ray, Supratim ([email protected])
Current Opinion in Neurobiology 2015, 31:111–118 This review comes from a themed issue on Brain rhythms and dynamic coordination Edited by Gyo¨rgy Buzsa´ki and Walter Freeman For a complete overview see the Issue and the Editorial Available online 3rd October 2014 http://dx.doi.org/10.1016/j.conb.2014.09.004 0959-4388/# 2014 Elsevier Ltd. All rights reserved.
the excitability of these neurons varies within one period of the rhythm [10–12], which can be used to synchronize the spiking of neurons that process information about different features of a stimulus (BBS), generate temporal windows of communication between two assemblies (CTC), or encode information in the position of the spike relative to the rhythm (PC). Some of the hypotheses can be tested by recording spikes from multiple neurons and checking for the presence of rhythmicity and phase lag in the cross-correlation function computed using spike trains. However, if the phase locking of the spike with the rhythm is weak (which is typically the case, at least for gamma), rhythmicity is often difficult to detect using spike trains alone ([13], Supplementary Figure 3 of [14]). Local field potential (LFP), which is obtained by low-pass filtering the raw signal (typically below 300–500 Hz) recorded from a microelectrode and thought to mainly reflect the summed transmembrane currents flowing through the neurons within a local region around the microelectrode tip [15,16,17,18], is therefore often used to capture the rhythm. This review focuses on the problems faced by the researcher who, from the LFP, tries to accurately estimate the phase (and amplitude) of the intracellular periodic fluctuations in a neuron (generated, for example, by an inhibitory network) relative to its spikes. This is a challenging problem because LFP also captures other synaptic and non-synaptic events related to spiking [17,18] (summarized in Figure 1), which overlap with the rhythm in both spectral and temporal domains.
LFP events associated with spikes A distinctive feature of brain signals is the presence of oscillations at different frequencies that are highly conserved across species [1] and are coupled to specific behavioural states and stimulus features [2]. Understanding the relationship between oscillations and underlying neuronal processes is an important goal in neuroscience [3], and over the last two decades, several ideas have been put forward regarding a potential role of rhythms in cognitive processing, such as binding-by-synchrony (BBS) [4], communication-through-coherence (CTC) [5,6] and phase coding (PC) [7–9] hypotheses. These are based on the observation that several rhythms (such as gamma, which has a centre frequency between 30 and 80 Hz) are generated through inhibitory networks that produce periodic fluctuations in the intracellular potential of the target post-synaptic neurons such that www.sciencedirect.com
Excitatory inputs arriving from an upstream area generate EPSPs in the post-synaptic cell. The effect of these EPSPs on the field potential can be estimated by computing the mean target LFP response following spikes in the input region. For example, V1 LFP triggered on the spiking of an LGN neuron shows a transient synaptic response of 5 ms duration (Figure 2a; [19,20]). If the target neuron also fires, the extracellular action potential (EAP) shows a transient response whose shape varies depending on the position of the microelectrode (Figure 2b, [17,18,21,22,23,24]). Near the soma where the EAP amplitude is largest, it has a prominent negative dip followed by a slower positive upswing. EAP has power at all frequencies, including low frequencies, and is present even after low pass filtering (spike ‘bleedthrough’; Figure 2c; [25]; shape of the bleed-through depends on the low-pass filter settings). Taken together, Current Opinion in Neurobiology 2015, 31:111–118
112 Brain rhythms and dynamic coordination
Figure 1
(a)
(b)
Inputs from thalamus
Local Field Potential (LFP) events around a spike
10 0 –10
Rhythmic inhibition synaptic events (unfiltered) synaptic events (0–300 Hz)
Microelectrode tip
Extracellular Action Potential (EAP) EAP below 300 Hz: “bleed-through” Post-spike upswing
Rhythmic inhibition from an inhibitory network
LFP: sum of the components above
10
Goal: to extract the phase of the rhythmic inhibition (blue trace on top) from the LFP
0 –10 –50
0
50
Time (ms) Current Opinion in Neurobiology
Summary of LFP events locked to a spike. (a) A pyramidal neuron (grey) receives thalamic inputs (red) and rhythmic inhibition from an interneuron network (blue). A researcher tests the hypothesis that the rhythmic inhibition biases the timing of spikes produced by the pyramidal cell by inserting a microelectrode (tip shown as a black dot) and measuring the extracellular potential. Action potentials produced by the pyramidal cell and the rhythmic intracellular fluctuations are studied by filtering the raw signal in high (>1000 Hz) and low (
