Show Thoughts

Deneve et al.'s model uses divisive normalization.