Content-addressable memory with spiking neurons

Time evolution and cooperative phenomena of networks with spiking model neu rons are analyzed with emphasis on systems that could be used as content-ad dressable memories. Stored memories are represented by distributed patterns of neural activity where single cells either fire periodically or remain q uiescent. Two distinct mechanisms to generate relaxation behavior toward su ch periodic solutions are investigated: delayed feedback and subthreshold o scillations, Using theoretical analysis and numerical simulations it is sho wn that in both cases model networks with integrate-and-fire neurons posses s storage capabilities similar to those of conventional associative neural networks. [S1063-651X(99)09202-8].