Entorhinal cortex grid cells can map to hippocampal place cells by competitive learning

Area of Science:

• Neuroscience
• Computational Neuroscience
• Cognitive Science

Background:

• Grid cells in the dorsocaudal medial entorhinal cortex (dMEC) exhibit periodic firing fields across environments.
• Hippocampal cells, such as those in the dentate gyrus (DG) and CA3, typically display spatially localized place fields.
• A discrepancy exists between the distributed firing of grid cells and the localized firing of hippocampal place cells.

Purpose of the Study:

• To investigate the computational mechanisms underlying the transformation of grid cell inputs into hippocampal place cell representations.
• To demonstrate how a competitive network in the dentate gyrus can recode entorhinal cortex grid cell inputs.
• To explore the role of learning and short-term memory in shaping hippocampal spatial representations.

Main Methods:

• Utilized a computational model of the hippocampus and entorhinal cortex.
• Simulated a competitive network of dentate granule cells receiving input from entorhinal grid cells.
• Incorporated associative learning and short-term memory traces into the network model.

Main Results:

• The competitive network model successfully recoded grid cell inputs to generate sparse, orthogonal representations.
• The model demonstrated that this recoding process can account for the emergence of dentate place cells from entorhinal grid cells.
• Inclusion of short-term memory traces in associative learning contributed to the formation of broader hippocampal place fields.

Conclusions:

• A competitive network within the dentate gyrus provides a viable computational explanation for the mapping of entorhinal grid cells to dentate place cells.
• Learning within this competitive network is crucial for achieving this spatial transformation.
• Short-term memory mechanisms can further refine these representations to match observed hippocampal place field characteristics.