10.Information Coding

Information coding

time to first spike

In a $\mathrm{SRM_0}$ model, N presynaptic neurons target on the postsynaptic neuron. At $t^\mathrm{pre}$, n presynaptic spikes arrive. $u_i(t)=nw\epsilon(t-t^\mathrm{pre})$ $u_i(t)$ is membran potential, $w$ is the constant value of all synaptic weights. When $u_i(t)$ cross threshold $\theta$, a spike is generated. So the time difference (time to fist spike) codes the strength of inputs.

phase coding

Reference is necessary: periodic background signal Phase codes constant input strength?

correlation coding

sharing same inputs or connected.

noise in neuron activities

facts: in vivo recording revealed noise phenomena in in-tact neuron fast moving stimuli: no noise slowly moving stimuli: changing

noise sources

1. thermal noise
2. number of channels detected
3. multi-inputs from background

Statistics of spike trains

To reflect noisy neuron activity, input-dependent renewal process is required.