Abstract
The neurons of many basal ganglia nuclei, including the external and internal globus pallidus (GPe and GPi, respectively) and the substantia nigra pars reticulata (SNr) are characterized by their high-frequency (50-100 spikes/s) tonic discharge (HFD). However, the high firing rate of GPe neurons is interrupted by long pauses. We studied the extracellularly recorded spiking activity of 212 well-isolated HFD GPe and 52 GPi/SNr neurons from five monkeys during different states of behavioral activity. An algorithm that maximizes the surprise function was used to detect pauses and pauser cells ("pausers"). Only 6% of the GPi/SNr neurons versus as many as 56% of the GPe neurons were classified as pausers. The GPe average pause duration equals 0.62 s. The interpause intervals follow a Poissonian distribution with a frequency of 13 pauses/minute. No linear relationship was found between pause parameters (duration or frequency) and the firing rate of the cell. Pauses were preceded by various changes in firing rate but not dominantly by a decrease. The average amplitude and duration of the spike waveform was modulated only after the pause but not before it. Pauses of pairs of cells that were recorded simultaneously were not correlated. The probability of GPe cells to pause spontaneously was extremely variable among monkeys (30-90%) and inversely related to the degree of the monkey's motor activity. These findings suggest that spontaneous GPe pauses are related to low-arousal periods and are generated by a process that is independent of the discharge properties of the cells.
Introduction
The external segment of the globus pallidus (GPe) is a central component of the basal ganglia, traditionally known for its high-frequency tonic discharge (HFD). While GPe neurons typically fire at rates of 50–100 spikes per second, this activity is frequently punctuated by long periods of silence, or "pauses." These pauses are a hallmark of GPe activity, yet their statistical properties and underlying mechanisms have historically been poorly understood. This study aims to provide a quantitative definition of these pauses, determine whether they arise from intrinsic cellular properties or external network inputs, and explore their relationship with behavioral arousal.
Methodology: The Pause Surprise Algorithm
A critical challenge in studying pauses is their objective detection. Traditional methods often rely on simple inter-spike interval (ISI) thresholds (e.g., any gap >300 ms). However, such methods are sensitive to "spoiler" spikes—isolated spikes that occur during a period of relative silence, which would technically break a single long pause into two shorter ones.
To address this, the researchers employed a "pause surprise" algorithm. This method maximizes a Poisson surprise function to identify segments where the probability of observing so few spikes is statistically improbable given the neuron's average firing rate. This approach allows for the inclusion of occasional spikes within a pause if the overall density remains significantly lower than the baseline. To ensure data quality, the team also utilized a zero-to-one "isolation score" to objectively classify well-isolated single units.
Characterization of GPe Pauses
The study revealed that pauses are highly specific to the GPe. While 56% of GPe neurons were classified as "pausers," only 6% of neurons in the GPi and SNr exhibited similar behavior. The average GPe pause lasts 0.62 seconds, with an average frequency of 13 pauses per minute.
Statistical analysis of the interpause intervals (IPIs) showed they follow a Poisson distribution. This indicates that the timing of pauses is random; the occurrence of one pause does not predict when the next will happen. Furthermore, no linear relationship was found between a cell's baseline firing rate and its pause parameters, suggesting that pausing is not simply a function of high-frequency firing.
Mechanisms and Behavioral Context
A key finding of this research is that pauses are likely driven by network activity rather than intrinsic cellular fatigue. If pauses were caused by intrinsic mechanisms like ion channel inactivation, one would expect a decrease in firing rate or a change in spike shape immediately preceding the pause. Instead, the firing rate remains stable until the pause onset, and spike waveforms only change after the pause, likely reflecting recovery from sodium channel inactivation during the silent period.
Additionally, cross-correlation analysis of simultaneously recorded neurons showed that pauses are not synchronized across the population. This rules out global inhibitory signals or changes in the extracellular medium as primary causes. Instead, the probability of pausing was found to be inversely related to the animal's motor activity. Monkeys in active, high-arousal states showed fewer pauser cells (30%), whereas those in idle, low-arousal states showed significantly more (up to 90%).
Conclusion
These findings suggest that GPe pauses are a signature of low-arousal or "offline" states, generated by independent network inputs. While the exact source of these inputs remains to be identified—potential candidates include inhibitory striatal inputs or a reduction in excitatory drive from the subthalamic nucleus—the random, asynchronous nature of these events points toward a complex network-level regulation of basal ganglia activity during quiet wakefulness.
Summary updated