* first iteration models

* second iteration models

* project description


* first iteration models: introduce stimuli by clicking on 'neurons'

To view this content, you need to install Java from java.com


[top]


* project description:

Highly abstracted neural models are used to explore the effects of spatial distribution and density on the qualitative nature of the system's memory.

The 'neurons' have no intrinsic memory. This can be seen from the first simulation above.
A stimuli is introduced by clicking on the 'neuron'. [refreshing the browser will generate new distributions]

The 'neurons' possess only the most rudimentary behaviour - generate a pulse when a stimuli has been received.
This can be seen from the second simulation.

We see that memory can only emerge from interaction with at least one other.

The introduction of a third 'neuron' generally results in a decaying memory, resulting from positive feedback. Rigid, locked structures develop. These become insensitive to further stimuli and turn black. The rate at which this occurs is a function of spatial proximity and the number of introduced stimuli. The spatial distribution is only very rarely suitable to prevent the system from 'running-away'.

Large numbers of 'neurons' begin to hold localised memories.
In this last simulation, the propagation distance has been capped for clarity. In order to counter the effect of positive feedback, time-varying thresholds and fatigue need to be introduced to the 'neurons'. These mechanisms of negative feedback are explored in the second iteration models.

The simulations are necessary vehicles for exploring and representing the time-based, and spatially dependant, qualitative nature of multiple interacting elements. Despite their simplicity, complex patterns can result from their interactions.

This project will be taken to full-scale in the near future. The stimuli will be acoustic.

[top]


Projects > Current > Indefinite Memory

work-in-progress [2006 - ongoing]

last updated: 17/08/07