An attractor network model of cortical associative memory functions ha
s been constructed and simulated. By replacing the single cell as the
functional unit by multiple cells in cortical columns connected by lon
g-range fibres, the model is improved in terms of correspondence with
cortical connectivity. The connectivity is improved, since the origina
l dense and symmetric connectivity of a standard recurrent network bec
omes sparse and asymmetric at the cell-to-cell level. Our simulations
show that this kind of network, with model neurons of the Hodgkin-Huxl
ey type arranged in columns, can operate as an associative memory in m
uch the same way as previous models having simpler connectivity. The n
etwork shows attractor-like behaviour and performs the standard assemb
ly operations despite differences in the dynamics introduced by the mo
re detailed cell model and network structure. Furthermore, the model h
as become sufficiently detailed to allow evaluation against electrophy
siological and anatomical observations. For instance, cell activities
comply with experimental findings and reaction times are within biolog
ical and psychological ranges. By introducing a scaling model we demon
strate that a network approaching experimentally reported neuron numbe
rs and synaptic distributions also could work like the model studied h
ere.