Figure 1
From: Spiking neural network model of reinforcement learning in the honeybee implemented on the GPU
Network diagram for the hypothesized model of reinforcement learning in the honeybee olfactory system. Excitatory connections are shown in black, inhibitory connections in blue and learning synapses in red. Grey arrows represent the abstractions modeled by implicit mechanisms. The model consists of the antennal lobe, lateral horn interneurons, mushroom body Kenyon cells and lobe neurons, and an octopaminergic/dopaminergic pathway for reinforcement, classically considered to be the VUMmx1 neuron. A conditioned stimulus is paired with an unconditioned stimulus (sugar to the antenna) to elicit the behavioral response (proboscis extension) in the training phase, which can then be rewarded by letting the bees drink. The association with reward facilitates plasticity in the synapses between the Kenyon cells and lobe neurons and between lobe neurons and pre-motor neurons. The size of weight changes is determined by an eligibility trace as a function of the delay between stimulus and reward.