Publications

While escape is often viewed in classical ethology as an action that is released upon presentation of specific stimuli, successful and adaptive escape behaviour relies on integrating information from sensory systems, stored knowledge, and internal states. This article reviews recent research showing how elements from multiple information streams are integrated to generate flexible escape behaviour.

Theoretical work in collaboration with Mate Lengyel developing a new method using a hierarchical cascade of linear-nonlinear subunits to model the somatic response of neurons receiving complex spatiotemporal synaptic input patterns. In L2/3 pyramidal neurons, linear input integration with a single global dendritic nonlinearity achieved above 90% prediction accuracy.

This paper describes an ethological paradigm to evaluate the influence of learned spatial knowledge on defensive behaviors in mice. The behavioral assay is based on the Barnes maze spatial memory assay, and tests how mice navigate to a shelter during escape responses to imminent threats in novel environments, and how they adapt to acute changes in the environment.

Work showing that the decision to initiate instinctive escape is well described by a theoretical model where evidence of threat is accumulated towards an escape threshold, and that neurons in the medial superior colliculus (mSC) integrate threat evidence, while dorsal periaqueductal gray (dPAG) neurons to compute escape initiation. The synaptic connection between the mSC and the dPAG is weak and unreliable due to low release probability, and provides a synaptic-level realization of the threat evidence thresholding operation.

This paper describes the results of behavioural experiments showing that learned knowledge about the spatial environment controls escape behaviour. Mice form a memory of shelter location within seconds, and this memory guides not only the execution of escape, but also changes the defensive strategy from escape to freezing, demonstrating that instinctive defensive actions are surprisingly flexible and not simple stimulus-reactions.

Neurons are well suited for computations on millisecond timescales, but some neuronal circuits set behavioral states over long time periods, such as those involved in energy homeostasis. This work, done in collaboration with Scott Sternson, shows that multiple types of hypothalamic neurons are near-perfect synaptic integrators that summate inputs over extended timescales, due to the voltage-gated sodium channel Nav1.7.