Example Of Polychronization
There has been recent experimental evidence that neurons can transmit information through precisely timed spike patterns. Patterns can be found in the precise spike timings of a group of neurons, which form a functional neuron group. Polychronization  can be defined as the ability to produce ‘time-locked’ but not synchronous firing patterns with milli-second precision. A network consisting of a group of neurons recurrently connected, with axonal delays and a synaptic efficacy tuning technique called Spike-Timing-Dependent Plasticity
(STDP) can be shown to display polychronization.
A simple example of polychronization is depicted below.
Figure 6.1: Polychronization
The figures above depict polychronous activity among …show more content…
The network consists of 80% excitatory and 20% inhibitory neurons.
STDP: Synaptic weights in the network evolve according to the STDP learning rule. In an input is presented multiple times,
STDP causes the corresponding polychronous group to grow stronger. 6.2 polychronous spikes
Even in a small group of 1000 neurons, we can find more than 5000 polychronous groups. These are not present initially, but as STDP acts on the network, groups are identified and weights are tuned so that frequently active groups grow stronger. During long periods of simulation, it is common to find polychronous groups appearing and disappearing, as STDP works on the network. The total number of polychronous groups at any time is however found to be much larger than the number of neurons or synapses.
Often, multiple polychronous groups can share more than 1 neuron.
This is evident from the fact that we discover more number of polychronous groups than the number of neurons in the network. Sometimes the same subgroup can behave as two different polychronous groups due to different firing times and spike order. There is no am6.3 significance of polychronous activity …show more content…
Even when no external input stimuli are presented to the network, polychronous groups emerge. This shows that the network is capable of generating random memories even without prior experiences.
However, when certain common input stimuli are repeatedly presented to the network, and STDP is allowed to run on the network, we can observe the presence of subgroups that are activated every time the corresponding stimuli is presented to the network. Another interesting property is that different polychronous groups are built corresponding to different inputs even when all the inputs are presented to the same set of input neurons.
As an example, 2 different inputs are presented to a network as different sets of spike timings. Initially, there are no distinct groups, but as STDP is allowed to run on the network, groups begin to develop.
When the network settles down, we observe that the presence of the above stimuli will give rise to the same set of polychronous groups.
This can be thought of as the development of memory within a network based on previous experiences. We can also see that, as different inputs are introduced and the network is made to learn these