Ling-Qi Zhang

The paper derives a scaling law for biological learning by linking inductive bias to learning speed and validates it using human and mouse datasets.

The paper shows lab mice reorganize existing behaviors into structured bouts to become efficient hunters in a dynamic environment.

This study shows tree shrews can recognize objects despite lacking a fovea, using computational models and behavioral tests to demonstrate transformation-tolerant object recognition.

The paper shows that the tilt illusion results from increased neural coding precision for stimuli similar to an oriented surround, using psychophysical and fMRI data from humans.

The paper shows that gradient descent learning in neural networks creates representations sensitive to common image features, mirroring efficient coding principles observed in animal sensory systems.

The paper shows that a Bayesian model constrained by efficient coding explains both human visual speed perception and neural tuning in area MT, with priors from behavior matching those from neural data.

This paper develops a Bayesian image reconstruction framework to model initial visual encoding, analyzing information loss and perceptual effects in human vision.

This study shows that individuals with ASD have reduced sensory encoding capacity and limited adaptability to changing visual stimuli compared to controls.