A computational model of the retina

M.J. Mulder, M.N. Steijaert, N.H.L. Kuijpers, H.M.M. Eikelder, ten, P.A.J. Hilbers

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

3 Downloads (Pure)


Although the retina is one of the best understood parts of the central nervous system, manymysteries of this complex tissue have not yet been unraveled. We have used computationalmodeling to comprise knowledge from literature and gain more knowledge in the interactionsbetween neurons on two levels. Here we present two models: the network model, which describesthe whole neuronal network of the retina and the ephaptic model, which deals with aspecific feedback mechanism between neurons (cones and horizontal cells) in the first layers ofthe retina.

In the network model we have included all known information on the pathways involved in colorvision in the primate retina. The behavior and output of this model depends on the spatial,temporal as well as the spectral information of the visual input. The model contains 20 typesof neurons, representing 3 types of cones (photoreceptors), 2 types of horizontal cells, 8 types ofbipolar cells and 7 types of ganglion cells. For each type of neuron there is a layer with cells ofthat type placed on a hexagonal grid. Each individual neuron is described by differential equationsbased on an electrical network of a capacitor in parallel with a number of resistors. Someof these resistors represent synaptic connections, for which the resistance can be controlled byadjacent neurons in the same layer as well as neurons in other layers.

The ephaptic model provides a more detailed description of the interaction between cones andhorizontal cells. For long it has been known that cone type photoreceptors convert light stimuliinto a chemical signal that is received by horizontal cells and bipolar cells. There is also a feedbacksignal from horizontal cells to cones. With our ephaptic model we study the hypothesisthat this feedback is caused by an ephaptic phenomenon, i.e., by the local modulation of extracellular ion concentrations. The differential equations describing the behavior of neurons in thismodel resemble those in the network model, but contain much more physiological detail, as wellas a description of the extra cellular potential in a specific region of the synapse.

Simulations with the network model show qualitative agreement with biological phenomena.Examples are differences in spatial resolution of the various pathways and the change in networkbehavior if one type of cone is missing (color blindness). However, the model lacks physiologicaldetail of the ephaptic interaction between cones and horizontal cells. This lack of physiologicaldetail can be filled in with our model of ephaptic feedback. The ephaptic model fits well withexperimental data and shows that ephaptic interaction can indeed explain feedback from horizontalcells to cones.

Our network model can be used to study the whole retina as if it were an image processingunit, whereas the ephaptic model can be used to study the specific interaction between conesand horizontal cells. The combination of the two models yields a model that can explain bothphysiological and perceptual phenomena.

Original languageEnglish
Title of host publicationProceedings of the First Dutch Conference on Bio-Medical Engineering (BME 2007) 18-19 January 2007, Egmond aan zee, The Netherlands
Publication statusPublished - 2007
Event1st Dutch Bio-Medical Engineering Conference (BME 2007), January 18-19, 2007, Egmond aan Zee, The Netherlands - Hotel Zuiderduin, Egmond aan Zee, Netherlands
Duration: 18 Jan 200719 Jan 2007


Conference1st Dutch Bio-Medical Engineering Conference (BME 2007), January 18-19, 2007, Egmond aan Zee, The Netherlands
Abbreviated titleBME 2007
CityEgmond aan Zee


Dive into the research topics of 'A computational model of the retina'. Together they form a unique fingerprint.

Cite this