Network: Comput. Neural Syst. 9 (1998) 235–264. Printed in the UK PII: S0954-898X(98)85200-7 A model of cortical associative memory based on a horizontal network of connected columns Erik Frans´ en and Anders Lansner† SANS – Studies of Artificial Neural Systems, Department of Numerical Analysis and Computing Science, Royal Institute of Technology, S-100 44 Stockholm, Sweden Received 18 June 1997, in final form 21 November 1997 Abstract. An attractor network model of cortical associative memory functions has been constructed and simulated. By replacing the single cell as the functional unit by multiple cells in cortical columns connected by long-range fibres, the model is improved in terms of correspondence with cortical connectivity. The connectivity is improved, since the original dense and symmetric connectivity of a standard recurrent network becomes sparse and asymmetric at the cell-to-cell level. Our simulations show that this kind of network, with model neurons of the Hodgkin–Huxley type arranged in columns, can operate as an associative memory in much the same way as previous models having simpler connectivity. The network shows attractor-like behaviour and performs the standard assembly operations despite differences in the dynamics introduced by the more detailed cell model and network structure. Furthermore, the model has become sufficiently detailed to allow evaluation against electrophysiological and anatomical observations. For instance, cell activities comply with experimental findings and reaction times are within biological and psychological ranges. By introducing a scaling model we demonstrate that a network approaching experimentally reported neuron numbers and synaptic distributions also could work like the model studied here. 1. Introduction 1.1. Background In this paper we present a model of the general auto-associative memory properties that the neocortex is presumed to support. One class of models used to describe this cortical property is the recurrent artificial neural network (ANN) model. In our model the functional units corresponding to the units in an ANN are represented by lamina II/III cells in cortical columns. This brings the connectivity closer to that of the cortex, since it becomes sparse, especially for the long-range connections, and also asymmetric at the level of single cells. The cell model is relatively detailed, with several compartments and ion channel dynamics of Hodgkin–Huxley type. The synapses are of several different types and have different cellular placement. The recurrent attractor ANN model (Anderson et al 1977, Hopfield 1982, Cohen and Grossberg 1983, Ackley et al 1985, Lansner and Ekeberg 1985, Smolensky 1986, Lansner and Ekeberg 1989, Amit et al 1990) shares phenomenology with the associative memory function of the cortex as revealed by psychological memory experiments. For instance, in priming experiments, if subjects are first shown a list of words and then asked to fill † E-mail: ala@sans.kth.se. Author to whom correspondence should be addressed. 0954-898X/98/020235+30$19.50 c  1998 IOP Publishing Ltd 235