Biomedical subjects

Drazen Domijan

Publications and source records attributed to Drazen Domijan.

5 recordsLinked to original sources

PubMed2004

A neural model of quantity discrimination.

A neural network model is proposed with the ability to extract abstract numerical representation from visual input. It simulates properties of a number detection system which is hypothesized to underlie simple language-independent numerical abilities. The network has three layers where the first layer computes the sum of the nearest neighbour inputs. The first layer is also augmented with multiplicative gating and gradient tonic activation which prevents interference. The second layer implements local lateral inhibition which enables a single node to represent a single object. The third layer exhibits number-tuning similar to recently described responses of neurons in the prefrontal cortex. Computer simulations showed that network response does not depend on visual attributes like the object's size, position or shape. The model is based on several biophysical mechanisms such as multiplicative interaction on dendrites, independent processing on different dendritic branches and disinhibition by glutamate spill-over on kainate receptors on inhibitory axons.

Animals↗

PubMed2004

Recurrent network with large representational capacity.

A recurrent network is proposed with the ability to bind image features into a unified surface representation within a single layer and without capacity limitations or border effects. A group of cells belonging to the same object or surface is labeled with the same activity amplitude, while cells in different groups are kept segregated due to lateral inhibition. Labeling is achieved by activity spreading through local excitatory connections. In order to prevent uncontrolled spreading, a separate network computes the intensity difference between neighboring locations and signals the presence of the surface boundary, which constrains local excitation. The quality of surface representation is not compromised due to the self-excitation. The model is also applied on gray-level images. In order to remove small, noisy regions, a feedforward network is proposed that computes the size of surfaces. Size estimation is based on the difference of dendritic inhibition in lateral excitatory and inhibitory pathways, which allows the network to selectively integrate signals only from cells with the same activity amplitude. When the output of the size estimation network is combined with the recurrent network, good segmentation results are obtained. Both networks are based on biophysically realistic mechanisms such as dendritic inhibition and multiplicative integration among different dendritic branches.

Action Potentials↗

PubMed2003

A mathematical model of persistent neural activity in human prefrontal cortex for visual feature binding.

A two-stage model of sustained neural activity in the prefrontal cortex is proposed in order to simulate feature binding and capacity limits in visual working memory. In the first stage, object features are stored in parallel network layers without explicit conjunctions. A second stage binds features into integrated objects consistent with the recent proposal of Wheeler and Treisman (J. Exp. Psychol. Gen. 131 (2002) 48). Model neurons have extended dendrites which are capable of active non-linear integration. Computer simulation illustrates model ability to segregate feature values of the different objects into cells with different activity amplitude and to maintain segregated feature representations for a limited number of objects. Depending on the task demands, features are retrieved in a second stage and form a unified object representation.

Models, Theoretical↗

PubMed2003

A neural model for visual selection of grouped spatial arrays.

Psychophysical and electrophysiological studies indicate that visual attention operates on early retinotopic maps and selects spatially grouped arrays of locations which correspond to objects or perceptual groups. A neural model is proposed which is able to select an array of locations labelled by the same activity level and suppress all other regions that are not in the focus of attention. The model uses self-recurrent dendritic inhibition to compute the maximum activity level in the input and suppress a feedforward flow of activity from input to selection layer at unattended locations. Computer simulations also illustrate the model's ability to detect abrupt onsets of new objects in the visual scene, perform visual search and track moving objects.

Animals↗

PubMed2003

Neural mechanism for noise exclusion in spatial cueing.

Spatial cueing in an orientation discrimination task with targets embedded in high or low external noise indicates noise exclusion as a primary mechanism for attentional modulation. To implement noise exclusion in a neural network, a new mechanism is proposed based on a dendritic computation of difference between self-inhibition and lateral inhibition signals. A computer simulation illustrates that the model exhibits a strong cueing effect for high noise input and no effect when the noiseless input is presented, as is consistent with behavioral signatures of noise exclusion. It is argued that the model could also exhibit object-based selection if uniform activity distribution is assumed for all cells representing the object.

Attention↗