Eero P. Simoncelli

This paper investigates how visual feature representation changes with eccentricity by comparing human perception and models of luminance and spectral energy.

This study shows fixational eye movements improve retinal signal precision by enhancing spike timing and cell interactions in macaque retinas.

This study examines neuronal and behavioral responses to naturalistic texture images in macaque monkeys, finding that V2 neurons better predict texture perception than V1 neurons.

The paper develops a model to explain how V1 and V2 neurons in macaques respond to visual stimuli by combining afferent signals, capturing 43% and 38% of variance respectively.

The paper investigates how neurons in macaque V4 respond to textures and objects, finding that object selectivity emerges in two phases with late-emerging responses sensitive to image degradation.

The paper shows that targeted gain modulation in V1 neurons supports task-adaptive sensory decisions by enabling rapid downstream readout of task-relevant information.

The paper introduces a method to compare neural models by measuring their perceptual discriminability using metric tensors, demonstrating its effectiveness with color and autoencoder models.

The paper shows that perceptual learning improves discrimination accuracy but distorts appearance estimates, suggesting learning enhances category distinctions.

The paper develops a method to reconstruct natural images from retinal ganglion cell activity using deep denoiser priors, achieving high perceptual similarity with fewer parameters than previous approaches.

The paper proposes a unified framework that links perceptual sensitivity and intensity judgments by modeling them as properties of an internal representation, showing compatibility with Webers law and Stevens power law.

The paper maps spatial frequency preferences in human primary visual cortex using fMRI, finding a relationship between preferred spatial frequency and eccentricity, with small effects from stimulus orientation.

The paper investigates neural signatures of visual recognition memory in rhesus monkeys and finds that sensory referenced suppression (SRS), not repetition suppression, better explains behavioral patterns.

This study develops a stochastic recurrent neural network model with reward-modulated Hebbian plasticity to learn efficient, task-specific sensory representations in two tasks: stimulus estimation and discrimination.

The paper introduces a method to infer nonlinear spatial subunits in primate retina using spike-triggered clustering, showing improved response prediction with subunit models in both natural and controlled stimuli.

The paper investigates motion selectivity in macaque MT neurons using single-unit recordings and a novel model, finding that velocity-separable models better describe responses to complex stimuli like plaids.

This paper introduces a decoding method that uses stochastic modulation to identify task-relevant neurons and achieves performance close to an optimal decoder in simulations.