(J-393) PERIODIC AND APERIODIC POWER SPECTRUM ANALYSIS OF LOCAL FIELD POTENTIALS IN THE MONKEY SUPERIOR COLLICULUS

The ability to direct our visual axis to stimuli of interest is crucial for engaging with the environment around us. The superior colliculus (SC) is a midbrain structure that plays a fundamental role in both the processing of visual stimuli and generating eye movements. Neuronal layers within the SC are arranged by function, with superficial layers being primarily sensory and deeper layers being primarily motoric. Neurons between these superficial and deep layers show a mix of sensory and motoric responses. Areas of the superior colliculus are also organized by receptive fields and respond to stimulus present in certain areas of the visual field.

Local Field Potentials, or LFPs, are aggregate voltage signals generated by an ensemble of neurons located within a volume around the recording electrode site. These signals can be categorized by their power within specific frequency bands: delta (0-4 Hz), theta (4-8 Hz), beta (10-30 Hz), and gamma (30-70 Hz). Traditionally, these frequency bands are thought to contain key information on the coordination of activity of groups of neurons.

All electrophysiological signals also feature an aperiodic decay property as well. While in the past this component has been either ignored or thought of as something to correct for, Gao et al. claims that the rate of decay of this aperiodic component is correlated with the balance between synaptic inhibition and excitation [1]. We therefore sought to quantify the periodic and aperiodic features of the LFP spectrogram for visual targets presented either inside or outside the neuron’s response field.

Data for this study was gathered from two adult male rhesus monkeys (M. mulatta). The subjects were trained to initially fixate on a target presented in the center of their vision. Trials were performed where another stimulus was presented in the receptive field (RF1) or 180 degrees away to be classified as out of the receptive field (RF0), and the data was compared. Analysis of neuronal response was centered around the appearance of a new visual stimulus (target onset).

We applied multitaper spectral analysis in MATLAB to transform the time-series LFP data into the frequency domain. This transformed data revealed the power at each frequency over the course of each trial. The individual power spectrums from each time window were then decomposed into its aperiodic and periodic components over time, using Custom MATLAB code and a curve fitting algorithm written by Donoghue et al. [2]. This code fits each power spectrum in the time series along a double exponential decay function and then modeled each deviation from this decay as gaussian peaks. From there, the periodic peaks and the aperiodic decay exponent were plotted against time.

Spectrogram analysis shows a distinct reaction to stimulus when comparing RF1 to RF0. Just after the presentation of the stimulus, a large increase in power is shown in low frequencies in RF1, while RF0 shows no reaction to stimulus. Surprisingly, the periodic peaks against time in RF1 and RF0 were remarkably similar, while the aperiodic decay exponent shows a sharp double peaked response to stimulus in RF1 not seen in RF0.

Following Gao et al., the large peaks in the decay exponent in response to the visual stimulus are to be expected, as an increase in this decay exponent is correlated with an increase in excitation. However, the lack of any difference in the periodic content is very surprising, as this suggests that the neural response to the visual stimulus can be explained broadly by the aperiodic components of the power spectra and the periodic components hold little information.

Funding was provided by the Swanson School of Engineering and the Office of the Provost at the University of Pittsburgh.

1. Richard Gao, Erik J. Peterson, Bradley Voytek, Inferring synaptic excitation/inhibition balance from field potentials, NeuroImage, Volume 158, 2017, Pages 70-78, ISSN 1053-8119, https://doi.org/10.1016/j.neuroimage.2017.06.078.

2. Donoghue, T., Haller, M., Peterson, E.J. et al. Parameterizing neural power spectra into periodic and aperiodic components. Nat Neurosci 23, 1655–1665 (2020). https://doi.org/10.1038/s41593-020-00744-x