Fig. 8: Modeling multiple unspecific fields and remapping experiments.
From: Place cell maps slowly develop via competitive learning and conjunctive coding in the dentate gyrus
a Model architecture. From the previous model design, a group of 450 spatially modulated non-grid/LV cells is added to the EC cell population, the number of Grid cells and LV cells is reduced to 150 to achieve the EC ratio of Grid cells and non-grid cells reported23, and a subgroup of 300 GCs under weak inhibition is added to the GC population. The firing fields of non-grid/LV cells were generated by the multiplication of noise and Gaussian signals to achieve lower spatial information23. The color scale is the same as that used in (b). b Transformation of spatial representations across iterations. Color-coded representation of rate maps for active GCs under high (blue) and low (green) inhibition, across iterations. In gray, the emergence of non-grid/LV cells across iterations. c Upper, proportion of GCs (irrespective of type) with a single field (black) and multiple fields (gray), across iterations. Lower, proportion of LV, periodic and unspecific GCs, among multiple-field GCs, across iterations. Note the similarity with experimental trends (Fig. 3c). d The rate map of EC LV cells (upper), grid cells (middle) and non-grid/LV cells (lower) used to model the original (left), reordered (middle) and new (right) belts. To simulate the reordered belt, the firing fields of EC LV cells are moved to the new location of the landmarks, and 40% of non-grid/LV firing fields are randomly regenerated. To simulate the new belt, EC LV cells are randomly assigned to new landmarks, and all non-grid/LV firing fields are randomly regenerated. The color scale is the same as that used in (b). e Fraction of GCs with single (black) and multiple (gray) place fields in each belt, across iterations (n = 3300 cells). Note the similarity with experimental trends (Fig. 5b). f The distribution of the number of fields per cell (using the belt with the largest number of fields) for the groups of GCs that are active in 1 (black), 2 (green), and 3 (purple) belts, across iterations. Note the similarity with experimental trends (Fig. 5d).
