Thomas Hannagan

Recurrent neural networks as neuro-computational models of human speech recognition

This study shows that LSTM RNNs trained on speech data predict human neural responses to speech, suggesting RNNs model cortical speech recognition processes.

Reset Networks: Emergent Topography by Composition of Convolutional Neural Networks

The paper introduces a Reset model composed of convolutional neural networks that self-organize into clusters and exhibit topographic organization similar to parietal cortex number representation.

Simulating the emergence of the Visual Word Form Area: Recycling a convolutional neural network for reading

The paper simulates the emergence of the visual word form area by training a convolutional neural network to recognize written words, showing it can replicate key properties of the VWFA.

A random-matrix theory of the number sense

This paper proposes a random-matrix theory to explain the emergence of number sense in the brain without learning, linking neural activity patterns to numerical representations.