Andrea Ivan Costantino

Low-Dimensional and Optimised Representations of High-Level Information in the Expert Brain

This study identifies three principles of expert brain representation using chess as a model, showing shifts in content, structure, and location of neural representations.

Impact of a transient neonatal visual deprivation on the development of the ventral occipito-temporal cortex in humans

This study shows that early visual deprivation in humans impairs the early visual cortex but leaves the categorical coding in the ventral occipito-temporal cortex largely unaffected.

Recurrent issues with deep neural networks of visual recognition

This study compares recurrent and feedforward DNNs in modeling human visual recognition behavior, finding that model size, not recurrence, better aligns with human-like confusion patterns.

Partial information transfer from peripheral visual streams to foveal visual streams is mediated through local primary visual circuits

The paper investigates feedback from peripheral to foveal visual cortex using fMRI and finds limited information transfer with local V1 circuits.

A computational understanding of zoomorphic perception in the human brain

This study investigates the human bias toward perceiving zoomorphic objects as animal-like and shows that this bias can be induced in deep neural networks with specific training.