Stimulus-dependent oscillatory activity in the lateral geniculate body of the cat.

Abstract

Recent observations of synchronous oscillatory activity of cell assemblies in cat primary visual cortex [1] have led to a marked resurgence of interest in the functional role of periodic neuronal behavior in information processing [2]. In the current discussion, however, precortical oscillatory activities in the visual system appear to be neglected. Neuronal oscillatory activities in the gamma frequency range have been described in retinal ganglion cells and in cells of the lateral geniculate body [3]. Moreover, it has been demonstrated that neurons in cat lateral geniculate body (LGB) show appropriate average firing rates in response to scanned figures [4]. The present study investigates responses of three simultaneously recorded neurons in cat LGB, which were stimulated with a photograph of a white three-dimensional cylinder illuminated from the top (Fig. 1 a). The photograph was moved stepwise across the receptive field (RF) of three adjacent neurons in order to reconstruct a response profile for the object [5]. The average firing rate and the dominant period length of the oscillatory multiunit activity (MUA) of the three cells during the stepwise scanning were analyzed and mapped in scale with the mirror-inverted stimulus picture. The experiment was conducted on an adult unanaesthetized cat weighing about 2.5 kg two days after preparation. During preparation, the cat was anaesthetized with intravenous injection of ketalar; pressure points, injection sites, and cuts were additionally infiltrated with a local anaesthetic. Multiunit activity of three neurons (two offand one on-neuron) was recorded extracellularly from layer A of the left LGB with a tungsten microelectrode (tip diameter about 1 /zm). During the experiment, total immobilization was achieved by regular intravenous administration of flaxedil ( 5 1 0 mg kg1 h 1). The refractive power of the contralateral eye was corrected by an appropriate contact lens and the nictitating membrane was withdrawn by the application of neosynephrine. The ipsilateral eye was covered with an opaque occluder. The cat was artificially respirated and the heart rate was controlled. The experiment lasted about 90 min. At the end of the experiment the cat received a lethal dose of barbiturate. During the experiment, the cat's head was fixed facing a tangential screen at 1 m distance. The RF of the three neurons overlapped and had an eccentricity of 5.3 ° visual angle along the horizontal meridian. Extracellular MUA was measured and spike occurrence was detected with a Schmitt trigger. Data were collected with a sampling rate of 1 500 s~ and analyzed in histograms of 2 ms bin width. The photograph of the cylinder was projected on a tangent screen. The stimulus had a maximum intensity of 10.83 cd /m 2 and the constant background intensity was of 1.11 cd /m 2. The picture (total size about 1 3 ° x 10 °) could be moved stepwise horizontally and vertically following a grid of 10 x 10 equidistant points in scale with the whole stimulus, using two mirrors which were placed in the light beam. Beginning from the upper left point the stimulus was tachistoscopically presented 30 times for 500 ms (on-condition) interrupted by dark periods of 500 ms (offcondition) and was then moved stepwise to the lower right point of the grid. Thus, the whole stimulus was systematically scanned by the RF. For each of the 100 grid points of stimulus presentation a poststimulus time histogram (PSTH) of the MUA of the 30 repeated measures was calculated for the onand off-responses separately. Figures 1 b, c illustrate the PSTHs of one grid point for the offand on-condition. The corresponding autocorrelation functions (ACF), presented in Fig. 1 d, e, make the inherent periodicity more obvious. The analysis of periodicity is based on a method published previously [6]. For each histogram a dominant periodicity was estimated by ACF and fast Fourier transformation (FFT) with a maximum lag of 128 histogram bins. The estimations based on the analysis of the trend-reduced difference distribution of the original PSTHs. With the exception of one PSTH of the off-condition, all histograms could be classified as multimodal periodic. Figure 1 f, g shows maps of the normalized average firing rate of the offand on-response across the grid points