Selective visual attention in a neurocomputational model of phase oscillators

Abstract

In order to understand the dynamic property of covert selective visual attention, which is different from the proposed mechanism of the spotlight metaphor, a two-layered network of phase oscillators was developed. The first layer is related to the hippocampus and controls attention focus formation. The second layer is related to the visual cortex, and each cortical oscillator in it simulates an assembly of cells coding for a particular stimulus in the sense of feature binding. Selective visual attention is interpreted as the result of the emergent synchronization of hippocampus oscillators and a part of cortical oscillators. Numerical experiments are presented to illustrate attention focus formation and attention shifting from one set of stimuli to another. From a neurocomputational point of view, our results demonstrate that attention is an emergent property of the dynamical cell assemblies responding to the whole visual field.

Neural network model of selective visual attention using Hodgkin?Huxley equation

“…, 40 in the visual cortical layer are entrained to a state with a frequency of about 100 in the hippocampal layer, but the other neurons are not. This result for the frequency is in qualitative agreement with that obtained for the phase oscillatory neurons by Wu and Guo (1999).…”

“…The value of the frequency of each neuron in the hippocampal layer becomes almost constant, and that of each neuron in the visual cortical layer also becomes constant. This result for the frequency is in qualitative agreement with that obtained for the phase oscillatory neurons by Wu and Guo (1999). …”

“…We notice that the neurons in the hippocampal layer synchronize with each other not only for the frequency but also for the firing time, which is in contrast to the results obtained by Wu and Guo (1999). We interpret this as meaning that only the visual information related to the focused information by the neurons in the hippocampal layer can be obtained in the visual cortex in order to ensure that the necessary visual information and other visual information are removed.…”

“…Thus the results obtained by our model are different from those obtained by the model proposed by Wu and Guo (1999). However, the time dependence of the frequency for the neurons described by the Hodgkin-Huxley equation is in qualitative agreement with that of the frequency for the neurons described by the phase oscillator.…”

“…: +81-22-2175844) by a bottom-up mechanism that separates objects from the background for both space and time and in part by a top-down mechanism that selects the relevant objects to current behavior, (3) the objects are recognized as a result of the competition among neuronal assemblages because the objects are reconstructed by combinations of many different features across the visual cortex in parallel, and (4) the selective visual attention is obtained much like an emergent property of competitive interactions that work across the visual cortex in parallel. Wu and Guo (1999) proposed a mathematical model of the selective visual attention and verified part of the assumption proposed by Desimone and Duncan (1995). The model is based on a two-layered neural network that consists of a layer of the hippocampal formation and a layer of the visual cortex by using neurons described by a phase oscillator.…”