A descriptive model to characterize a physiological property called receptive field, which is fundamental in deciphering how particular neurons encode the incoming visual stimulus, was built to accurately predict the direction of stimulus motion based on recorded directional selective retinal ganglion cells (DSRGCs) neural response. Briefly, building the model, space-time inseparability of the receptive fields of ON-OFF DSRGCs together with additional static non-linearity provided the principal key. The main steps we used were: analysis of the extracellularly recorded data from ON-OFF DSRGCs stimulated with white noise stimulus; calculation of spike triggered average in response to white noise stimulus and extracting the kernel; finding the optimal kernel by adjusting the estimated firing rate to the recorded firing rate; analysis of the extracellularly recorded data using new stimulus, consisting in drifting grating bars moving 45 degrees apart over the RFs of the ON-OFF DSRGCs; construction of the linear model; adding static non-linearity (i.e. spike threshold and saturation); comparing predicted data with recorded data. Similar methodology was used to test different scientific hypothesis , . The optimal kernel needed to describe direction selectivity of ON-OFF DSRGCs was first extracted using white noise stimulus and produces an estimate of the firing rate that is a linear filter of the stimulus to which a nonlinear function was added. This is a function of the linear filter value instantaneously evaluated at the time of the rate estimation. Half wave rectification and saturation limit negative or unrealistically large values of the function. Static nonlinearity is then used to introduce firing thresholds into estimates of neural responses of ON-OFF DSRGCs. The direction of stimulus motion was accurately predicted for different preferred directions of the recorded cells by our model, as exemplified in Figure 1. However, the degree of direction selectivity (DSi) calculated from the tuning curves  was not 100% predicted. This slight underestimation of the direction selectivity degree (see Figure 1.B) suggests that for complex tasks, such as direction of stimulus motion, higher order non-linearity should be taken into account to improve the model performance. Intrinsic properties of ON-OFF DSRGCs like bursting or paired spiking activity can presumably improve the model accuracy. These results are in accordance with other scientific work related to different types of neurons and to different stimuli , .
Department of Computer Science, Technical University Munich, Garching, Germany
Faculty of Environmental Science & Engineering, Babeş-Bolyai University, Cluj-Napoca, Romania
Pillow JW, Paninski L, Uzzell VJ, Simoncelli EP, Chichilnisky EJ: Prediction and decoding of retinal ganglion cell responses with a probabilistic spiking model. The Journal of Neuroscience. 2005, 25: 11003-11013. 10.1523/JNEUROSCI.3305-05.2005.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.