7/ We observed that the amplitude of somatosensory responses did change as a function of the Vertex Potential phase! Thus, a large VP may reflect the rapid alternation of periods of depressed and enhanced cortical excitability, as it occurs during #sleep slow waves.

7/ These ANNs also naturally separate into sub-networks specialised for avoidance and interception, mirroring the modularity of both the macaque brain and the egocentric value map that we proposed.

6/ Remarkably, these action value fields naturally assemble into an *egocentric map*: a modular, composable predictive model of the world around the body, which allows rapid adaptation to novel situations by recombining existing knowledge.

5/ To test this, we used #ReinforcementLearning and computational modelling. We trained agents to interact with the world — and their networks spontaneously developed body-part-centered fields, not because of spatial proximity, but *because of value*.

3/ For decades we've known of neurons responsive to objects near the body—aka peripersonal neurons—initially viewed as proximity detectors.

However, these neurons also respond to stimulus value, speed, and motor repertoire.

Why are these neurons so flexible? What are they for?