Title: Attentional effects in a model of two reciprocally connected visual areas Authors: Andreas Knoblauch and Günther Palm Abstract: Even when creating a biologically realistic model for an apparently very simple cognitive task like seeking a certain object in the visual field, one is confronted with severe problems concerning binding of distributed representations. In this work we present simulation results from a model of two reciprocally coupled visual cortical areas. One area is a peripheral visual area where local object features are represented, the other is a more central visual area where whole objects are recognized. In our model, correct binding is achieved by the simultaneous switching of the activation state of corresponding neuron groups. We relate our simulations to neurophysiological findings concerning attention and biased competition, and demonstrate how these findings can be explained very naturally by assuming different kinds of bindings between neuron groups in different areas as produced by our model. While the binding is fluctuating in the absence of attention, it becomes static by the attentional bias. This leads us to several predictions for neurophysiological experiments. Even when creating a biologically realistic model for an apparently very simple cognitive task like seeking a certain object in the visual field, one is confronted with severe problems concerning binding of distributed representations. In this work we present simulation results from a model of two reciprocally coupled visual cortical areas. One area is a peripheral visual area where local object features are represented, the other is a more central visual area where whole objects are recognized. In our model, correct binding is achieved by the simultaneous switching of the activation state of corresponding neuron groups. We relate our simulations to neurophysiological findings concerning attention and biased competition, and demonstrate how these findings can be explained very naturally by assuming different kinds of bindings between neuron groups in different areas as produced by our model. While the binding is fluctuating in the absence of attention, it becomes static by the attentional bias. This leads us to several predictions for neurophysiological experiments. Poster-Paper-Submission (2 pages) for the ICCM meeting 2003 in Bamberg (Germany) Title and author information: Title: Attentional effects in a model of two reciprocally connected visual areas Authors: (i) Andreas Knoblauch Department of Neural Information Processing University of Ulm Oberer Eselsberg O27 D-89069 Ulm Germany Tel.: +49 – 731 – 5
[1]
L. Shastri,et al.
From simple associations to systematic reasoning: A connectionist representation of rules, variables and dynamic bindings using temporal synchrony
,
1993,
Behavioral and Brain Sciences.
[2]
M. Carandini,et al.
Stimulus dependence of two-state fluctuations of membrane potential in cat visual cortex
,
2000,
Nature Neuroscience.
[3]
Andreas Knoblauch,et al.
Pattern separation and synchronization in spiking associative memories and visual areas
,
2001,
Neural Networks.
[4]
A. Grinvald,et al.
Linking spontaneous activity of single cortical neurons and the underlying functional architecture.
,
1999,
Science.
[5]
W Singer,et al.
Visual feature integration and the temporal correlation hypothesis.
,
1995,
Annual review of neuroscience.
[6]
Günther Palm,et al.
Scene segmentation by spike synchronization in reciprocally connected visual areas. II. Global assemblies and synchronization on larger space and time scales
,
2002,
Biological Cybernetics.
[7]
R. Desimone,et al.
The Role of Neural Mechanisms of Attention in Solving the Binding Problem
,
1999,
Neuron.