Michael J. Tarr

The paper validates a diffusion-based method called BrainDiVE for generating stimuli that activate specific brain regions, showing enhanced category selectivity and revealing functional gradients in visual cortex.

The paper uses a texture statistics model to predict fMRI responses across human visual cortex, showing increasing complexity of texture features from early to higher visual areas.

The paper shows that low-level visual feature tuning in visual cortex correlates with semantic informativeness, supporting the natural scene statistics hypothesis.

This study investigates cortical selectivity to real-world object size in natural images, finding consistent organization across occipitotemporal cortex and preserved spatial patterns despite controlling for visual features.

The paper identifies two food-selective regions in the ventral visual cortex using fMRI and natural images, confirming food selectivity across multiple datasets and analysis methods.

This paper uses explainable AI to analyze brain responses to natural images, revealing distinct spatial feature selectivity in different visual regions.

GLMsingle improves single-trial fMRI response estimates using custom HRFs, noise regressors, and ridge regression, enhancing reliability and decoding performance across datasets.

The paper shows that using natural language supervision with large, diverse datasets improves models' ability to predict human high-level visual cortex responses compared to image-only or text-only models.

The paper uses CNNs to simulate visual category learning with blurred inputs, finding that gradual blur reduction improves basic and subordinate-level categorization, especially with grayscale images.

The paper investigates how functional context influences scene processing in specific brain regions, finding that some regions are more sensitive to top-down perceptual expectations than others.

The paper uses computer vision tasks to predict brain activity and maps task-specific visual features to brain regions, revealing task-specific architecture in the human visual system.

The paper examines spatio-temporal neural dynamics in the human visual cortex using neural network features, finding non-feedforward processes influence early visual cortex activity.

This study compares high-density EEG (SND) with standard density (ND) in capturing spatiotemporal aspects of early visual processing, finding SND provides more accurate and detailed neural information.