Tankut Can

The paper introduces a random tree model to explain quantitative features of narrative recall, showing sublinear recall length growth and scale-invariant summary distributions.

This paper introduces FINDR, a deep learning method that infers low-dimensional nonlinear dynamics from neural data, outperforming existing methods in capturing neuron responses and visualizing flow fields.

The paper uses large language models to design memory experiments and analyze human recall and recognition of narratives, finding linear scaling with length and preserved recognition in scrambled narratives.