2.6.4. PyGeNN Implementation of Pavlovian Conditioning Experiment

To perform the Pavlovian conditioning experiment described by Izhikevich (2007) using this model, we chose 100 random groups of 50 neurons (each representing stimuli S₁…S₁₀₀) from amongst the two neural populations. Stimuli are presented to the network in a random order, separated by intervals sampled from U(100, 300)ms. The neurons associated with an active stimulus are stimulated for a single 1 ms simulation timestep with a current of 40.0 nA, in addition to the random background current of U(−6.5, 6.5)nA, delivered to each neuron via I_exti throughout the simulation. S₁ is arbitrarily chosen as the Conditioned Stimuli (CS) and, whenever this stimuli is presented, a reward in the form of an increase in dopamine is delivered by setting DA(t) = 0.5 after a delay sampled from U(0, 1000)ms. This delay period is large enough to allow a few irrelevant stimuli to be presented which act as distractors. The simplest way to implement this stimulation regime is to add a current source to the excitatory and inhibitory neuron populations which adds the uniformly-distributed input current to an externally-controllable per-neuron current. In PyGeNN, the following model can be defined to do just that:

where the n parameter sets the magnitude of the background noise, the $(injectCurrent, I) function injects a current of InA into the neuron and $(gennrand_uniform) samples from U(0, 1) using the “XORWOW” pseudo-random number generator provided by cuRAND (NVIDIA Corporation, 2019). Once a current source population using this model has been instantiated and a memory view to iExt obtained in the manner described in section 2.3, in timesteps when stimulus injection is required, current can be injected into the list of neurons contained in stimuli_input_set with:

The same approach can then be used to zero the current afterwards.