AMS Professorship award for Professor Randy Bruno, University of Oxford

An AMS Professorship would substantially accelerate my embedding into the UK and enhance productivity.

First, its rapid turn-around would allow me to hire new research staff immediately to continue my research program in the UK (a technician and 3-year postdoc), rather than waiting until I secure other UK-based awards requiring longer review cycles. Second, it would accelerate my original slow, phased move plan.

With 3 US-based students defending and completing experiments as temporary postdocs in 2022, I have shipped only a quarter of my lab equipment and am delaying the rest for one mass shipment when the last person completes at the end of the year.

An AMS Professorship would allow me to bring 2 of these individuals for ~6 months each to Oxford for short postdocs and move my equipment 6-12 months earlier.

Their physical presence would also facilitate the transfer of experimental know-how to the new technician and 3-year postdoc and promote UK science to foreign scientists.

Specifically, the new staff would test the idea that the upper cortical layers exist to create a high-dimensional representation of task context to enhance an otherwise low-dimensional deep layer representation.

High-dimensional representations have computational advantages, distributing information across multiple neurons in a format easily readout by downstream brain structures and allowing flexible sensory-motor associations. Low-dimensional representations are better at generalizing to previously unseen stimuli and conditions.

For instance, in driving, using lines on the road to direct steering movements to keep a car left requires only a low-dimensional sensory-motor representation.

Recognizing that one is in the USA or the UK on the basis of arbitrary environmental cues requires a high-dimensional representation, and this can switch the operations of a low-dimensional one (for identical stimuli, drive right not left).

Hallucinations in schizophrenia and many autistic traits are potentially more complex examples that would correspond to disrupted high-dimensional representations.

By crossing transgenic mouse cre lines with a reporter mouse expressing a cre-conditional light-sensitive inhibitory opsin, we have previously silenced specific layers in behaving mice.

We will combine layer-selective optogenetic inhibition and array recording in mice trained to discriminate different objects (concave versus convex shapes, rough vs smooth textures, etc).

Our theory predicts that upper cortical layers, under direct long-range control from frontal cortical areas encoding the task, will continue to exhibit task-dependent representation during deep layer silencing.

Conversely, silencing upper layers should destroy any information about the task in deep layer neural activity if inherited from the upper layers.

Generalization will be assessed from how well decoder analysis of the neural activity can distinguish novel shapes with less/more curvature and will be analyzed both with and without layer-specific silencing.

Establishing distinct computational and behavioral contributions of the layers would be a milestone in understanding neocortex. It would establish a framework for future dissection of the multiple cell types within layers.

As each of the cell types are biophysically different and have different complements of neuromodulatory receptors, one may more rationally target different circuit elements with pharmacological agents in treating neurological and psychiatric disorders.