Shape_input captures the signal from the presenting objects and connects to all the neurons in shape_register via all_to_all mapping. Shape_register has the firing-threshold increment feature enabled, so higher neurons activate progressively only when they receive sufficient PSP. MP accumulation is enabled to encode how many objects appear within a given time window. For instance, if two objects are shown, shape_input spikes twice, delivering enough PSP to activate the first and second (lower-threshold) neurons in shape_register, but not enough to reach neurons with higher thresholds.

Because the circuit requires a single-voxel output, shape_register feeds into shape_count via an excitatory block-to-block mapping and an inhibitory shape_count_inh mapping (Figure 1). Every active neuron in shape_register excites its corresponding neuron in shape_count, while shape_count_inh suppresses all lower neurons. As a result, only the highest activated neuron remains active after inhibition. For example, if three neurons are activated, only the third neuron in shape_count fires, producing a single-voxel output that encodes the number of objects present.