Thinking About Brain Cell Assemblies

Abstract

Neurobiologists from Sherrington to Hebb (1 ) have struggled with the problem of how perception, cognition, and memory are coded by assemblies of neurons within the cerebral cortex of the brain. We can now simultaneously monitor the activity of hundreds of neighboring cells (2) and have identified functional properties of ensembles of neurons in the cortex and hippocampus (3-7). In this issue of Science, Wilson and McNauehton ( 8 ) reDort one of the . . . first of these studies on the hippocampus, a phylogenetically old part of the brain that is a final stage of information processing for some kinds of memory. They are able to monitor the activity of neural ensembles and show that, when it comes to functional coding, many neurons are better than one. These kinds of studies-plus insights from computational neuroscience-are beginning to show how the hierarchical organization of the cerebral cortex can be coupled with distributed coding. We may soon understand the nature of representation in the thinking parts of the brain. A comprehensive understanding of representation by neural ensembles will require reconciliation of two extreme views of information coding in the cortex. One view espouses a systematic organization com~osed of a hierarchv of filters or that encode simple stimulus features and com~lex events bv the activitv of single neuron' (9). A cont;asting view'espouses a fully distributed representation that encodes each item by distinct spatiotemporal activity patterns of homogeneous arrays of neurons (1 0). These opposing views may in fact be reconcilable; in the brain, both are employed to different extents at successive stages of information processing (see fieure). ~viYdence in support of the idea that the brain codes information with feature detectors followed closely the development of the extracellular microelectrode in the late 1950s. This technology unexpectedly revealed that cortical firing patterns could readily be related to recognizable dimensions of stimuli and actions ( I I). Many such observations, perhaps most widely promoted by Hubel and Wiesel's studies on the visual cortex, revealed an organized topo-