Dmitriy Aronov

This paper introduces a framework for multi-view animal pose estimation that improves accuracy and uncertainty estimation with limited labeled data, validated across multiple species.

This paper introduces a recurrent network model of the hippocampus that generates barcodes to bind memory elements, reducing interference and enabling flexible retrieval.

The paper shows that place cells in chickadees activate both when the bird is in a location and when it gazes at it from a distance, linking place coding to gaze through head saccades.

The paper identifies unique neural activity patterns in chickadees during caching events, suggesting they represent episodic memories.

The paper maps input patterns into the hippocampal formation of black-capped chickadees, revealing topographic organization along both the transverse and long axes, with similarities to mammalian brains.

The paper identifies an entorhinal-like structure in chickadees that exhibits similar anatomical and physiological features to the mammalian entorhinal cortex, suggesting conserved functions in hippocampal processing.

This paper shows that a recurrent neural network can compute the successor representation when synaptic weights match transition probabilities, and demonstrates this with hippocampal data from random foraging.

This study shows chickadees use flexible memory strategies for caching and retrieving food in a lab setting, combining memory-independent behaviors with precise spatial memory.

The paper identifies spatial representations and sharp wave ripples in the hippocampus of a food-caching bird, suggesting conserved mechanisms across vertebrates.

The paper identifies a new cell class in the mouse medial entorhinal cortex that encodes visual cue information to correct path integration errors during navigation.

An unsupervised algorithm with temporal dynamics recovers hippocampal activity manifolds in rats from neural activity alone, recasting trial-to-trial variability as movement along an extra manifold dimension.