Elsevier

Neuroscience Research

Volume 122, September 2017, Pages 51-63
Neuroscience Research

Technical note
Design of multielectrode arrays for uniform sampling of different orientations of tuned unit populations in the cat visual cortex

https://doi.org/10.1016/j.neures.2017.04.004Get rights and content

Highlights

  • Uniform samplings of differently tuned neurons are required for population coding.

  • Neurons were recorded by 4-tetrodes array and 8-microelectrodes array in Cat A17.

  • Tetrode array is better suited for uniform sampling of orientation tuned neurons.

  • Experimental results were confirmed by simulated recordings on optical imaging data.

  • Spatial heterogeneity of orientation map provides diversity of optimal orientations.

Abstract

For better reconstruction of stimulus orientation from a single trial activity of the neuron population in the visual cortex, we need uniform samplings of differently oriented tuned neurons. We recorded multiple neurons simultaneously by using either a four-tetrode array or an eight-microelectrode array, and examined what kinds of electrodes and layouts provided a more homogeneous distribution of the units’ optimal orientations. The unit population sampled by a four-tetrode array showed more homogeneous distribution than those sampled by an eight-microelectrode array. We confirmed this property by simulated recording sessions based on the optical imaging data of the orientation map.

Introduction

Cortical cells show diverse response properties to stimulus features (sensory cortex) or movement directions of the limbs (motor cortex). One of the main interests in neuroscience is this problem: How can the brain process information by integrating its activities over a population of neurons showing different response properties? A model of population coding was proposed and its validity was tested using experimental data (Motor cortex: Georgopoulos et al., 1986, Salinas and Abbott, 1994; Visual cortex: Gilbert and Wiesel, 1990, Vogel, 1990, Graf et al., 2011, Berens et al., 2012, Inferotemporal cortex: Gochin et al., 1994, Young and Yamane, 1992, Wilson and McNaughton, 1993). Early experimental studies were limited to the population of neurons recorded by multiple sequential sessions with the same stimuli or the same motor behavior. Recent advancement of multineuron recordings has enabled us to record the activities of multiple neurons simultaneously. This enabled us to test whether additional information is provided by the correlation in activities among multiple neurons. This information includes spike count coherence and spike timing synchrony.

Well-established theoretical models of population coding are the vector method (Georgopoulos et al., 1986, Tanabe, 2013), the optimal linear estimator (Salinas and Abbott, 1994, Tanabe, 2013) and the Bayesian estimator with a multidimensional Gaussian model (Gochin et al., 1994, Maynard et al., 1999). Population coding has attracted more attention recently by practical applications with a brain-machine interface (Rao, 2013). The main research interest in this subject is the reconstruction of directed arm movements from a single trial response of the neuron population in the motor cortex. A comparison of various decoding algorithms suggested that uniformly distributed preferred directions over the neuron population have a large impact on the decoded performance in the vector method, especially when the number of neurons was not high (Salinas and Abbott, 1994, Koyama et al., 2010). However, there was not much reporting on how the design of an electrode array influences the distribution of preferred directions over the recorded neuron population.

In this study, we recorded neuron populations in a local area the size of a single hypercolumn (1 mm2) in the cat visual cortex using two types of high-density electrode arrays: an array of four tetrodes and an array of eight microelectrodes. We performed a statistical comparison of the distribution of the units’ optimal orientations based on the circular variance (CV) between the populations recorded by the two types of arrays. Although the tetrode can isolate a larger number of single units than the microelectrode, the isolated units are located in close proximity and are expected to have similar orientation preferences. On the other hand, a thinner diameter of the microelectrode allows for denser penetrations of eight electrodes, and each electrode is expected to sample a unit having a different optimal orientation. Therefore, it is not straightforward to expect the neuron population sampled by certain array to have more uniformly distributed optimal orientations. Our experimental data suggested that the unit population sampled by the four-tetrode array showed a larger CV (more homogeneous distributions) than those sampled by the eight-microelectrode array. We also confirmed this statistical property by simulated (in virtu or in silico) recording sessions based on the optical imaging data of an orientation map. Owing to a larger sampling range and ability in isolating more numbers of units than the microelectrode, the tetrode has an advantage in getting a diversity of optimal orientations in the neuron population recorded near the region of a fine spatial structure (pinwheel). Preliminary report of the data has appeared in abstract form (Maruyama and Ito, 2013a).

Section snippets

Physiological preparation

The procedures for multineuron recording in anesthetized, paralyzed cats have been described previously in detail (Maruyama and Ito, 2013b). Briefly, acute experiments were performed on five adult male cats weighing between 3 and 5 kg (American short hair, Liberty Research, Inc., USA). The animal’s head was mounted on a stereotaxic frame, and a small craniotomy was made above the primary visual cortex (A10-P10, L0-5) in one hemisphere. During recording, anesthesia was maintained by intravenous

Uniformity of optimal orientations in experimental neuron populations

We totally isolated 517 single units from 48 recording sessions in the cat visual cortex (N = 339 from 25 sessions with the four-tetrode array, and N = 178 from 23 sessions with the eight-microelectrode array). The number of neurons recorded simultaneously by the tetrode array in a single session (mean 13.5 ± 4.0) was twice as large as that by the microelectrode array (mean 7.7 ± 2.3). We selected 466 units (tetrode array N = 308, microelectrode array N = 158), which increased their firing rate

Discussion

We recorded multiple neurons simultaneously in the cat visual cortex by using either a four-tetrode array or an eight-microelectrode array, and examined what kinds of electrodes and their layouts provide more homogeneous distribution of the units’ optimal orientations. Experimental data suggested that the unit population sampled by the four-tetrode array showed larger circular variances (more homogeneous distribution) than those by the eight-microelectrode array. We further examined the

Conflict of interest

The authors have declared that no conflict of interest exists.

Acknowledgements

We would like to thank Drs. Toshiki Tani and Shigeru Tanaka for providing us optical imaging data of the orientation map in cat visual cortex. We also thank Kenichi Ohki for technical advises on the identification of pinwheel centers. Finally we thank the anonymous referees for their instructive suggestions which helped us to understand our results more in detail. This research was supported in part by Grant-in-Aid for Scientific Research (C) (21500226, 24500286 and 16K01966), Grant-in-Aid for

References (27)

  • P.E. Maldonado et al.

    Heterogeneity in local distributions of orientation-selective neurons in the cat primary visual cortex

    Vis. Neurosci.

    (1996)
  • P.E. Maldonado et al.

    Orientation selectivity in pinwheel centers in cat striate cortex

    Science

    (1997)
  • Y. Maruyama et al.

    Design of multi-electrode array for homogeneous samplings of differently orientation tuned unit population in cat visual cortex

    Soc. Neurosci. Abstr.

    (2013)
  • Cited by (0)

    View full text