Features of the local field potential (LFP) in the superior colliculus during target selection

Abstract

Purpose : Single unit activity (SUA) which discriminates targets from distractors has been characterized in the superior colliculus (SC); however, it is unclear whether, when, and by what degree neuronal populations exhibit target selectivity in the SC. The purpose of our experiment was to address this knowledge gap through recording both SUA and local field potential (LFP) signals. We aimed to determine whether target selective activity was present within the LFP, and if so, how it differs from that described in SUA.

Methods : Broadband (0.5 Hz – 8 kHz) microelectrode signals were sampled at 40 kHz from the intermediate layers of the SC in two male rhesus monkeys (Macaca mulatta) while each performed a salience-based visual search task. SUA and LFP’s were extracted, respectively, with high- (250 Hz cutoff) and low- (300 Hz cutoff) pass zero-phase Butterworth filters. We used two time-domain metrics to quantify selectivity for both signals: selection time and selection magnitude. Each was derived from ROC analyses comparing neuronal activity between target-in and target-out trials at each millisecond after stimulus onset.

Results : Selective activity was found in both SUA and the LFP. Selection times derived from the LFP and SUA were not significantly different (mean_LFP: 167.29 +/- 4.79, mean_SUA: 189.09 +/- 6.17, p = 0.051); however, were highly correlated across sites (rho = 0.59, p < 0.0001). On the other hand, the selection magnitude was significantly higher in the LFP as compared to SUA (mean_LFP: 0.92 +/- 0.01, mean_SUA: 0.69 +/- 0.01, p < 0.0001) and was not correlated across sites (rho = 0.20, p > 0.05).

Conclusions : The relationship between selective activity in SUA and LFP’s sheds light on the underlying computational architecture of the SC. Specifically, the presence of selectivity and similarity in selection times between the two measures suggests that population-level extracellular voltage fluctuations maintain the spatial and temporal resolution found in single units. One explanation for this preservation is a clustering of functionally-similar subunits within the SC: indeed, the greater selection magnitude found in the LFP may reflect the summation of selection signals from such neighborhoods. Thus, the LFP may serve as a robust indicator of underlying single unit computations and therefore be used to link micro- and macro-level measurements of neuronal activity.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.