Fig. 8
Distribution of cell types and the influence of network size on the model. a The Pie chart shows the distribution of spatial cells and non-spatial cells from LAHN network with 50 neurons. Cells are classified as spatial cells based on their spatial information (SI) index. Values are averaged over 20 trials (i.e., training the LAHN 20 times). Distribution shows that most of the neurons in the network after training code for one or the other spatial variable. b Clustering of different spatial cell types into place cells, border cells, grid cells, and plane cells in the descriptor space (grid score, border score, and plane index). c Pie chart shows the distribution of different spatial cell types from LAHN network with 50 neurons. Distribution shows that out of the 95% of spatial cells formed, 32.43% emerge as place cells, 23.97% emerge as grid cells, 28.1% form border cells, and 15.5% form plane cells. d Graph (mean ± s.d.) shows the influence of LAHN size on the distribution of spatial cells. The logarithmic trend line (red curve) shows that the percentage of average spatial cell distribution increases as the size of the network increases. The black arrow indicates the network size from which the calculations in a–c are drawn. e Shaded plot of the cumulative variance graph of the principal components. Intersection of the black dotted lines shows the coordinate points at which the cumulative variance reaches 99% at 30th principal component. f–i Variation of spatial cell descriptors such as spatial information index, border score, grid score, and plane index, respectively, with respect to the principal component (PC) index. (For e.g. PC index 1 represents the first principal component and so on and so forth.) The dotted line in each figure represents the threshold value of the respective spatial descriptor. In f, the black dotted line indicates the spatial information index threshold and the blue dotted line indicates the threshold value above which the cell qualifies as a place cell
