© 1999 Macmillan Magazines Ltd
To the extent that the models can be related
to brain systems, they can be tested and,
eventually, the chasm that once separated the
mind and the brain can be bridged.
One of the most surprising and puzzling
recent discoveries in brain science has been
the degree to which the cerebral cortex, the
most highly evolved part of the human brain,
remains malleable. The surface of the body is
mapped onto the surface of the cortex in
such a way that nearby points on the body
map onto nearby neurons in the cortex.
When a sensory nerve that innervates the
fingers of a monkey is cut, the map reorga-
nizes and the cortical area once dedicated to
that patch of the body surface is, over time,
reassigned to neighbouring body surfaces.
This process involves both cortical and sub-
cortical mechanisms for neural plasticity.
Conversely, when a monkey is asked to use its
fingers repeatedly over a long period, the
area devoted to those fingers enlarges. Some-
thing similar happens in the somatosensory
cortex of human Braille readers.
Neural-network models of cortical maps
with Hebbian synaptic plasticity on the sen-
sory inputs can reproduce the changes that
occur during loss of neuronal input. Shortly
after limb amputation in humans, vivid
phantom sensations can occur, often accom-
panied by intractable pain that is referred to
the missing limb. Curiously, reports of phan-
tom limbs are rare following spinal injury
leading to paraplegia. Why should these two
ways of cutting off neuronal input to the cor-
tex have such different perceptual conse-
One possibility, explored by Spitzer in a
cortical model, assumes that there is noise in
the stump of the severed nerve. In the model,
the noise excites cortical neurons and leads
to cortical reorganization but, in the case of
spinal injury where it is assumed that such
noise is absent or reduced, the reorganiza-
tion of the map in the model is attenuated.
The advantage of the model is that plausible
hypotheses can be generated and explored in