How does the brain generate and update representations of the world?

Our ability to perceive and respond to our environment requires the brain to build representations of the outside world. How does the brain do this?

A key function of the brain is to extract features of the sensory environment (e.g., through vision, audition, olfaction) and use them to build more complex abstract representations (e.g., a sense of space). Research groups at the SWC focus on understanding the computations that give rise to these representations and how these computations map onto the underlying neural circuits.

In many instances, sensory and abstract representations of the outside world are used to inform motor actions. SWC studies how neocortex, basal ganglia, midbrain, cerebellum, brainstem and spinal circuits coordinate sensory-motor computations to obtain general insights into how sensory information is transformed into meaningful actions.

For this purpose, SWC researchers record from hundreds of neurons using multi-channel electrodes (e.g. Neuropixels probes) or two-photon microscopy to reveal the computations contributing to sensory representations and sensorimotor transformations. To investigate the underlying circuitry, we make use of whole-cell recordings, anatomical tracing based on in-house viral tools and advanced microscopy platforms which enable us to dissect the circuits at the level of cell types, projection pathways and synapses. This effort requires dedicated teams of experimentalists, engineers, and theorists, to tackle the wider problem of how the brain generates representations and executes actions relating to the outside world.

How does the brain make decisions and select actions?

Making a decision involves combining ongoing sensory information with prior knowledge before committing to an action. What are the neural mechanisms underlying these stages of a decision?

Theories on decision-making suggest the brain carries out several computations before committing to a choice. Evidence must be sampled, integrated and maintained in working memory before comparing it against a decision threshold, a process informed by prior knowledge. This process is also dependent on certainty of the information, motivational state and social and environmental context.

Research at the SWC focuses at the cellular and circuit levels across multiple brain structures to uncover how these factors influence choice. Key questions are addressed using a variety of decision-making paradigms in rodents, ranging from instinctive behavioural assays to complex tasks that require learning of abstract or changing rules. The underlying neural processes are elucidated by measuring and manipulating activity in identified regions and circuits, applying computational methods to analyse complex datasets, and constructing models that explain decision making. This work will help us generate a theoretical framework about how the brain’s circuits evaluate sensory information to make appropriate choices and how such processes are modified in diseased brain states.

How does the brain learn and remember?

Memories are vital to cognition. The brain is constantly integrating new information with prior knowledge allowing us to update our internal models of the world and thereby optimise our actions. But how does the brain learn and remember?

Researchers at the SWC study the fundamental principles by which the nervous system collects and stores information in the process of learning. We focus on ethologically relevant forms of memory across different time scales, from short term, working memory, to longer lasting memories such as those involved in spatial navigation and social interactions. We also study how the brain infers meaningful statistical patterns in the environment and how these abstract relations are formed, represented and stored. 

To do this we develop customised behavioural tasks, often in virtual reality (VR) environments, and make use of a range of recording technologies including tetrode and high density neuropixels electrophysiological probes as well as two photon imaging.

Circuit and cellular mechanisms of memory are interrogated by combining experimental and computational techniques in order to determine how specific forms of learning lead to changes in the excitability of individual neurons and in the strength of synaptic connections.

Much of this work is done in close collaboration with the Gatsby Computational Neuroscience Unit with whom we build models and test theories on how the brain stores and recalls information.