Ann M. Hermundstad

The paper derives a scaling law for biological learning by linking inductive bias to learning speed and validates it using human and mouse datasets.

This study shows mice learn sparse path representations to navigate complex environments efficiently.

The paper develops an algorithm inspired by mouse behavior to rapidly infer navigational goals through Bayesian inference and trajectory composition.

The paper develops an algorithm to transform symmetric attractor models into functionally equivalent heterogeneous models, showing hidden symmetries in fly connectomes support ring attractor dynamics.

This study develops a reinforcement learning model to understand how mice adaptively forage across multiple options, finding that dopamine in the nucleus accumbens core reflects a global learning rate.

The paper develops a computational theory for rapid behavioral inferences under time pressure, showing how adaptive belief tracking improves escape behavior in fruit flies.

This paper develops a model of optimal foraging strategies under energetic constraints and shows how these strategies can be approximated by simple neural circuits.

The paper analyzes how environmental dynamics affect the optimality of matching behavior in foraging scenarios.

The paper shows that small neural networks with as few as 4 neurons can accurately integrate angular velocity for spatial navigation, challenging the need for large ring attractor networks.

This study investigates how flies balance exploration and exploitation during local search, finding that satiety drives a shift from exploiting to exploring, while optogenetic activation does not.

This paper describes a neural circuit framework in Drosophila that enables rapid behavioral adaptation through head direction and goal representations during navigation.

The paper shows that dynamic contrast adaptation in auditory cortex predicts perception of sounds in noise.

The paper shows human perceptual decisions align with optimal Bayesian inference in dynamic environments, with decision bias depending on context reliability, environmental stability, and time since context change.

The paper shows that Drosophila maintains a stable head direction representation in naturalistic environments by tethering to global cues, even when local landmarks create potential conflicts.

The paper presents the first complete connectome of the Drosophila central complex, revealing network motifs that support flexible navigation and context-dependent action selection.

This study develops a theoretical framework to analyze adaptive sensory codes and shows that codes optimized for stationary environments can perform poorly when the environment changes, proposing codes that balance encoding accuracy and adaptation speed.

The paper tests the 'variance is salience' hypothesis by showing it accurately predicts visual sensitivity to second-order grayscale image statistics in texture-segregation tasks.

The paper discusses principles of adaptive encoding in sensory systems and derives schemes that balance encoding cost and inference error.

The paper proposes a theory of olfactory circuit architecture in Drosophila that uses disorder to process complex odors, tested with empirical data.