Single neuron binding properties and the magical number 7

When we observe a scene, we can almost instantlyrecognize a familiar object or can quickly distinguish among objects differingby apparently minor details. Individual neurons in the medialtemporal lobe of humans have been shown to be crucial for the recognitionprocess, and they are selectively activated by different views ofknown individuals or objects. However, how single neurons couldimplement such a sparse and explicit code is unknown and almostimpossible to investigate experimentally. Hippocampal CA1 pyramidalneurons could be instrumental in this process. Here, in an extensive seriesof simulations with realistic morphologies and active properties, wedemonstrate how n radial (oblique) dendrites of these neurons may beused to bind n inputs to generate an output signal. The results suggest apossible neural code as the most effective n-ple of dendrites that can beused for short-term memory recollection of persons, objects, or places.Our analysis predicts a straightforward physiological explanation for theobserved puzzling limit of about 7 short-term memory items that can bestored by humans. VVC 2008 Wiley-Liss, Inc.