THE NEURAL BASIS OF INTERNAL REPRESENTATION OF PLACE

点击翻译按钮获取中文摘要

The objective of this project is to apply a new technique for the isolation of spike trains from several single neurons in behaving rats to the problem of how internal representations of familiar places are maintained upon removal of their initiating cues (i.e. "how does the organism continue to know where it is when the lights go out?"). There are two problems here. The first is to demonstrate beyond question that experimental animals (rats) actually do make use of internal representations to solve spatial problems. The second is to discover the neural basis of this ability. This proposal is based upon O'Keefe's and Nadel's hypothesis that the hippocampal formation acts as a cognitive mapping system. Apart from the lesion literature leading to this notion, its main support is the fact that hippocampal unit activity separates into two clear classes: 'place' cells which carry position and direction information, and 'theta' cells which fire during translational movements. These two types of information could form the elements of a cognitive mapping system. The extension of this theory presented here requires that a) specific stable patterns of neural discharge be set up whenever the animal finds itself in a familiar place, b) these patterns should exhibit hysteresis in that the place specific information required to maintain them should be much less than that required to initiate them (in the limit, all external information should be removable,) and c) learned correspondences between transitions among these neural states and the motor sequences leading to them should result in the ability of the system to reactivate an appropriate series of such place specific states on the basis of the corresponding motor sequences alone (given that some initial state has been set up by externally provided place information). The new recording method to be applied to this problem is based on the principle that cells which are a unique ratio of distances from two closely spaced electrodes will generate spikes with unique ratio of amplitudes on the corresponding recording channels. Evidence is presented that a method based on this principle effectively solves the perennial problems of single unit isolation in the hippocampus where cells are densely packed and may exhibit considerable intrinsic variation in spike amplitude. In addition, it removes some of the selection bias towards large cells, and permits analysis of interactions among spike trains recorded simultaneously from several neurons.