Why can neural representations become smaller when neuronal tuning functions narrow?

In the context of perceptual learning, the narrowing of neuronal tuning functions can lead to a reduction in the size of a neural representation because it effectively increases the distance between neurons along the trained stimulus dimension.

When tuning functions sharpen, individual neurons become sensitive to a more restricted range of stimulus attributes. This process results in less overlap in the responses of neighboring neurons. Consequently, because the neurons are now more selective and distinct from one another, fewer total neurons are required to respond to a given stimulus, leading to a shrinkage in the overall neural representation.1