| |
Abstract:
We present a neural network model that reconciles the
functional hypothesis of thalamic involvement in perceptual
selection with current anatomical and physiological data. Although
early research treated the thalamus as a simple relay station, it
is now generally accepted that the thalamus plays a more important
functional role. The reticular nucleus of the thalamus (TRN) is
widely thought to contribute to perceptual selection by enhancing
the contrast between attended and unattended stimuli. However,
existing mechanistic models of thalamic attention rely on patterns
of connectivity that are not well supported by current
electrophysiological and anatomical evidence, which instead
suggests that the TRN functions to equalize, or govern, differences
in activity across stimuli. We show that a previously
underappreciated property of TRN neurons, their broad dendritic
arbors, can form the basis for surround inhibition in the dorsal
thalamus. This surround inhibition results in contrast enhancement,
which can then be amplified by the reciprocal connectivity between
dorsal thalamus and early sensory cortex. Our model demonstrates
that these properties enable the thalamocortical circuit to mediate
some aspects of perceptual selection.
|
