Spatial Concept Learning: A Spiking Neural Network Implementation in Virtual and Physical Robots

Figure 5

Dynamical neural representations of the virtual simulation, showing the ability of the SNN to adapt its behavior according to given rewards. Graphics A–K represent neural spike events and L–O consist of synaptic STDP coefficients. The final test includes a novel location of the stimulus. In the top row of the figure, only few images from the whole set are shown for clarity, the arrow indicating the perceived stimulus.