Stimulus-Specific Adaptation Decreases the Coupling of Spikes to LFP Phase

Abstract

Stimulus repetition suppresses the neural activity in different sensory areas of the brain. This mechanism of so-called stimulus-specific adaptation (SSA) has been observed in both spiking activity and local field potential (LFP) responses. However, much remains to be known about the effect of SSA on the spike-LFP relation. In this study, we approached this issue by investigating the spike-phase coupling (SPC) in control and adapting paradigms. For the control paradigm, pure tones were presented in a random unbiased sequence. In the adapting paradigm, the same stimuli were presented in a random pattern but it was biased to an adapter stimulus. In fact, the adapter occupied 80% of the adapting sequence. During the tasks, LFP and multi-unit activity were recorded simultaneously from the primary auditory cortex of 15 anesthetized rats. To clarify the effect of adaptation on the relation between spike and LFP responses, the SPC of the adapter stimulus in these two paradigms was evaluated. Here, we employed phase locking value method for calculating the SPC. Our data show a strong coupling of spikes to LFP phase most prominently in beta band. This coupling was observed to decrease in the adapting condition compared to the control one. Importantly, we found that adaptation reduces spikes dominantly from the preferred phase of LFP in which spikes are more likely to be present there. As a result, the preferred phase of LFP may play a key role in coordinating neuronal spiking activity in neural adaptation mechanism. This finding is important for interpretation of the underlying neural mechanism of adaptation and also can be used in the context of the network and related connectivity models.

Keywords: local field potential (LFP); multi-unit activity (MUA); primary auditory cortex; spike-phase coupling (SPC); stimulus-specific adaptation.

FIGURE 1

Adaptation experiment and MUA response. (A) Timeline of the auditory task. Pure tone stimuli were presented randomly for 50 ms and with 300 ms inter-stimulus interval. Stimuli were pure tones with a particular frequency and level out of five selected frequencies (f1–f5) and four intensities (40–70 dB SPL). (B) Two sequences of stimuli that were utilized for investigating the adaptation effect. In the first sequence (control) pure tones were presented with an equal probability. Each frequency–intensity combination was presented for 30 times. In the adapting sequence, the same combinations were presented with this a difference that the middle frequency at the level of 60 dB SPL (as the adapter) was presented with the probability of 80% of the whole sequence. (C) LFP and MUA were recorded from the primary auditory cortex of the anesthetized rat. (D) Raster plot and peristimulus time histogram (PSTH) of frequency f3 in 60 dB SPL intensity in a sample recording site. (E) Comparing PSTH of control vs. adapting conditions in the population of recording sites (n = 98).

FIGURE 2

Selecting the optimal threshold. (A) The average number of spikes in each electrode during the analysis period (100 ms). The X-axis shows the site indices and the Y-axis shows the spike rate. (B) The optimal spike count threshold for computing SPC. The X-axis indicates different values for the spike count threshold and Y-axis indicates the total spike count considered for each threshold (T). The optimal threshold (T = 14 spk/bin) produced the maximum spike count from the neural data. The bin is the 100 ms window from stimulus onset.

FIGURE 3

Spike-phase coupling at different frequency bands. (A) Power induced in six frequency bands for control and adapting conditions. The average band power was normalized to the maximum power. Significant power reduction observed in all frequency bands except alpha band. Maximum power difference between control and adapting condition arose in delta band and the minimum was in alpha band. (B) SPC strength for control and adapting conditions averaged across neurons for different frequency ranges of delta (δ; 1–4 Hz), theta (𝜃; 5–8 Hz), alpha (α; 9–12 Hz), beta (β; 13–30 Hz), low-gamma (γ_L; 31–70 Hz), and high-gamma (γ_H; 70–120 Hz). SPC strength is shown to be significantly different for the two conditions within the beta band (p < 0.0001; Wilcoxon rank sum test). (C) Comparison of SPC between the f3 condition of adapting sequence and a condition with similar spike rate in the control condition. The SPC in the adapting condition reduces (right panel) while the spiking activity was similar (left panel). (D) Scatter plot of the SPC strength (control vs. adapting) within the beta band (13–30 Hz) for all recording sites. The “cross” sign marks the average SPC (control vs. adapting) of the population of the recording site. The histogram in the upper right shows the distribution of recording sites toward two objective conditions, namely control and adapting (p < 0.001; t-test, n = 98, mean = –0.029, std = 0.071). It illustrates the distribution of SPC difference (SPC_adapting-SPC_control) across all recording sites. Each data point corresponds to a recording site. (E) Schematic of division of the adapting trials into partitions with the same length (L) based on the number of test trials in the control sequence (L). (F) Time course of adaptation effect on SPC. For all adapter trials in the adapting sequence, the data were partitioned based on its trials’ number in the control sequence (L trials in each bin). SPC strength in the adapting sequence is shown to be lower than the control sequence as the desired window increases (exponentially descends).

FIGURE 4

The preferred phase of LFP. (A) The histogram of phases for the preferred LFP phase in control (cyan) and adapting (magenta) conditions within the beta band (13–30 Hz). The mean locking phases are 0.91 and 0.88 rad (165° and 158°) for control and adapting conditions, respectively. (B) Spike-triggered average (STA) of normalized LFP across sites for control (cyan) and adapting (magenta) conditions. STA curves show the spikes are coupled to the ∼160° of the phase of LFP for two conditions. The inset in the upper section illustrates the phase values in a cycle. Shade areas depict standard error of mean across all recording sites.

FIGURE 5

Declining in the number of spikes from the preferred phase of LFP during adaptation. (A) Mean standard error of spike counts at both the preferred and the anti-preferred LFP phase for the control vs. adapting conditions. All LFP phases are divided into two groups of preferred phase ± π and anti - preferred phase ± π where the spike counts in these two groups are compared. (B) Histogram representation of spike counts in the preferred phase and anti-preferred phase for control (cyan) and adapting (magenta) conditions, respectively. (C) Schematic illustration of spike suppression from the preferred phase of LFP in the adapting condition compared to the control condition.