Limits...
Hippocampal remapping is constrained by sparseness rather than capacity.

Kammerer A, Leibold C - PLoS Comput. Biol. (2014)

Bottom Line: We find that the spatial decoding acuity is much more resilient to multiple remappings than the sparseness of the place code.Since the hippocampal place code is sparse, we thus conclude that the projection from grid cells to the place cells is not using its full capacity to transfer space information.Both populations may encode different aspects of space.

View Article: PubMed Central - PubMed

Affiliation: Department Biologie II, Ludwig-Maximilians-Universität München, Planegg, Germany; Graduate School for Systemic Neurosciences, Ludwig-Maximilians-Universität München, Planegg, Germany.

ABSTRACT
Grid cells in the medial entorhinal cortex encode space with firing fields that are arranged on the nodes of spatial hexagonal lattices. Potential candidates to read out the space information of this grid code and to combine it with other sensory cues are hippocampal place cells. In this paper, we investigate a population of grid cells providing feed-forward input to place cells. The capacity of the underlying synaptic transformation is determined by both spatial acuity and the number of different spatial environments that can be represented. The codes for different environments arise from phase shifts of the periodical entorhinal cortex patterns that induce a global remapping of hippocampal place fields, i.e., a new random assignment of place fields for each environment. If only a single environment is encoded, the grid code can be read out at high acuity with only few place cells. A surplus in place cells can be used to store a space code for more environments via remapping. The number of stored environments can be increased even more efficiently by stronger recurrent inhibition and by partitioning the place cell population such that learning affects only a small fraction of them in each environment. We find that the spatial decoding acuity is much more resilient to multiple remappings than the sparseness of the place code. Since the hippocampal place code is sparse, we thus conclude that the projection from grid cells to the place cells is not using its full capacity to transfer space information. Both populations may encode different aspects of space.

Show MeSH

Related in: MedlinePlus

Partial learning.Effect of place cell number  and of the fraction  that are trained to encode one environment on the number of environments . (A) Population sparseness as function of environments  stored, for  place cells. Different colors represent different fractions for partial learning, see legend in B. The critical value  at which sparseness reaches a biologically realistic value of  is obtained by interpolation. (B) Critical values of  as function of place cell number  and partial learning fraction . Data can be fitted by simple logarithmic functions . (C) Exponent  and coefficient  of fit from B. (D) Root mean square errors (RMSE) at the critical  for the  and  in B. (E) and (F): Mean place field size (E) and number (F) at the critical  for the  and  in B. Averages are over proper place cells.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4256019&req=5

pcbi-1003986-g010: Partial learning.Effect of place cell number and of the fraction that are trained to encode one environment on the number of environments . (A) Population sparseness as function of environments stored, for place cells. Different colors represent different fractions for partial learning, see legend in B. The critical value at which sparseness reaches a biologically realistic value of is obtained by interpolation. (B) Critical values of as function of place cell number and partial learning fraction . Data can be fitted by simple logarithmic functions . (C) Exponent and coefficient of fit from B. (D) Root mean square errors (RMSE) at the critical for the and in B. (E) and (F): Mean place field size (E) and number (F) at the critical for the and in B. Averages are over proper place cells.

Mentions: If one constrains the parameter space to biologically realistic mean population sparseness values for the hippocampal place fields about to (Supporting Information of [24] and [25], see Discussion) our simulations of the standard parameter regime (Fig. 8) show that such a regular place code can only be observed for up to about ten environments. Also for increased E% value the number of sparsely encoded environments is only increased to several tens (Fig. 9). A major factor limiting the number of environments is that in our model the synapses to the place cells are updated in each remapping, i.e., the place cells experience maximal interference. One can considerably extend the number of remappings for a given sparseness if the synaptic changes from different remappings are distributed to varying subsets of place cells, thereby increasing the overall number of putative place cells (partial learning). This strategy is motivated by an experimental report showing that only a small subset of CA1 pyramidal cells shows intracellular determinants for being recruited as a place cell in a novel environment [26]. We illustrate the benefits of partial learning by a further set of simulations in which the synaptic weights to only a fraction of the place cells are updated in each individual remapping (partial learning; Fig. 10). Using mean population sparseness as a criterion for the breakdown of the place code, partial learning increases the number of possible remappings (Fig. 10A) to over a hundred. As a measure for capacity, one can define a critical number of environments at which the mean population sparseness exceeds a (biologically motivated) threshold value of (see Discussion). This critical only weakly increases with the number of place fields but strongly decreases with increasing fraction of partial learning (Fig. 10B, C).


Hippocampal remapping is constrained by sparseness rather than capacity.

Kammerer A, Leibold C - PLoS Comput. Biol. (2014)

Partial learning.Effect of place cell number  and of the fraction  that are trained to encode one environment on the number of environments . (A) Population sparseness as function of environments  stored, for  place cells. Different colors represent different fractions for partial learning, see legend in B. The critical value  at which sparseness reaches a biologically realistic value of  is obtained by interpolation. (B) Critical values of  as function of place cell number  and partial learning fraction . Data can be fitted by simple logarithmic functions . (C) Exponent  and coefficient  of fit from B. (D) Root mean square errors (RMSE) at the critical  for the  and  in B. (E) and (F): Mean place field size (E) and number (F) at the critical  for the  and  in B. Averages are over proper place cells.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4256019&req=5

pcbi-1003986-g010: Partial learning.Effect of place cell number and of the fraction that are trained to encode one environment on the number of environments . (A) Population sparseness as function of environments stored, for place cells. Different colors represent different fractions for partial learning, see legend in B. The critical value at which sparseness reaches a biologically realistic value of is obtained by interpolation. (B) Critical values of as function of place cell number and partial learning fraction . Data can be fitted by simple logarithmic functions . (C) Exponent and coefficient of fit from B. (D) Root mean square errors (RMSE) at the critical for the and in B. (E) and (F): Mean place field size (E) and number (F) at the critical for the and in B. Averages are over proper place cells.
Mentions: If one constrains the parameter space to biologically realistic mean population sparseness values for the hippocampal place fields about to (Supporting Information of [24] and [25], see Discussion) our simulations of the standard parameter regime (Fig. 8) show that such a regular place code can only be observed for up to about ten environments. Also for increased E% value the number of sparsely encoded environments is only increased to several tens (Fig. 9). A major factor limiting the number of environments is that in our model the synapses to the place cells are updated in each remapping, i.e., the place cells experience maximal interference. One can considerably extend the number of remappings for a given sparseness if the synaptic changes from different remappings are distributed to varying subsets of place cells, thereby increasing the overall number of putative place cells (partial learning). This strategy is motivated by an experimental report showing that only a small subset of CA1 pyramidal cells shows intracellular determinants for being recruited as a place cell in a novel environment [26]. We illustrate the benefits of partial learning by a further set of simulations in which the synaptic weights to only a fraction of the place cells are updated in each individual remapping (partial learning; Fig. 10). Using mean population sparseness as a criterion for the breakdown of the place code, partial learning increases the number of possible remappings (Fig. 10A) to over a hundred. As a measure for capacity, one can define a critical number of environments at which the mean population sparseness exceeds a (biologically motivated) threshold value of (see Discussion). This critical only weakly increases with the number of place fields but strongly decreases with increasing fraction of partial learning (Fig. 10B, C).

Bottom Line: We find that the spatial decoding acuity is much more resilient to multiple remappings than the sparseness of the place code.Since the hippocampal place code is sparse, we thus conclude that the projection from grid cells to the place cells is not using its full capacity to transfer space information.Both populations may encode different aspects of space.

View Article: PubMed Central - PubMed

Affiliation: Department Biologie II, Ludwig-Maximilians-Universität München, Planegg, Germany; Graduate School for Systemic Neurosciences, Ludwig-Maximilians-Universität München, Planegg, Germany.

ABSTRACT
Grid cells in the medial entorhinal cortex encode space with firing fields that are arranged on the nodes of spatial hexagonal lattices. Potential candidates to read out the space information of this grid code and to combine it with other sensory cues are hippocampal place cells. In this paper, we investigate a population of grid cells providing feed-forward input to place cells. The capacity of the underlying synaptic transformation is determined by both spatial acuity and the number of different spatial environments that can be represented. The codes for different environments arise from phase shifts of the periodical entorhinal cortex patterns that induce a global remapping of hippocampal place fields, i.e., a new random assignment of place fields for each environment. If only a single environment is encoded, the grid code can be read out at high acuity with only few place cells. A surplus in place cells can be used to store a space code for more environments via remapping. The number of stored environments can be increased even more efficiently by stronger recurrent inhibition and by partitioning the place cell population such that learning affects only a small fraction of them in each environment. We find that the spatial decoding acuity is much more resilient to multiple remappings than the sparseness of the place code. Since the hippocampal place code is sparse, we thus conclude that the projection from grid cells to the place cells is not using its full capacity to transfer space information. Both populations may encode different aspects of space.

Show MeSH
Related in: MedlinePlus