Neuronal spatial learning

作者:Dorian Aur, Mandar S. Jog

摘要

Neurons are electrically active structures determined by the evolution of ion-specific pumps and channels that allow the transfer of charges under the influence of electric fields and concentration gradients. Extensive studies of spike timing of neurons and the relationship to learning exist. However, the properties of spatial activations during action potential in the context of learning have to our knowledge not been consistently studied. We examined spatial propagation of electrical signal for many consecutive spikes using recorded information from tetrodes in freely behaving rats before and during rewarded T-maze learning tasks. Analyzing spatial spike propagation in expert medium spiny neurons with the charge movement model we show that electrical flow has directionality which becomes organized with behavioral learning. This implies that neurons within a network may behave as “weak learners” attending to preferred spatial directions in the probably approximately correct sense. Importantly, the organization of spatial electrical activity within the neuronal network could be interpreted as representing a change in spatial activation of neuronal ensemble termed “strong learning.” Together, the subtle yet critical modulations of electrical flow directivity during weak and strong learning represent the dynamics of what happens in the neuronal network during acquisition of a behavioral task.

论文关键词:action potential, computation, information theory, machine learning, tetrode recordings, weak learning

论文评审过程:

论文官网地址:https://doi.org/10.1007/s11063-006-9029-2