Neural circuitry underlying non-sensory responses in sensory cortex

A central function of the brain is to use information from the senses to guide decisions, plans and actions. In mammals, a key role in this process is played by the cerebral cortex, which transforms raw sensory input into the sense of a unified object with a meaning and processes relationships between objects, actions and context. Elucidating how the cortex is organised is necessary to understand brain function and has been the subject of decades of research. For a long time, it was thought that different cortical regions had separate functions, so that sensory processing, decision making or motor control were carried out in distinct regions. But recent work has discovered that neurons across the cortex can respond to many aspects of behaviour: thus, parts of cortex previously thought to process purely sensory information actually contain neurons whose responses reflect the animal's decisions and movements, and their outcomes. These non-sensory responses imply that the cortex is set up differently than was thought. Therefore, to understand how the cortex works, we need to find out how these non-sensory responses are organised.

This project will tackle this key gap in knowledge by elucidating which types of neurons in sensory cortex can generate non-sensory responses and how their connections shape those responses. Specifically, we will test the hypotheses that non-sensory responses differ in the more common class of excitatory projection neurons as compared to rarer but crucial classes of inhibitory neurons; that non-sensory responses depend on inputs from other areas of the brain involved in making decisions and selecting actions; and that non-sensory neurons themselves also connect back to those areas in a distinct way.

Reaching these aims requires observing and manipulating neuronal activity in the cortex while an animal is using sensory input to guide a decision to act. Two powerful techniques, two-photon imaging and optogenetics enable the measurement and manipulation of activity at the level of single neurons, and we will combine these techniques with a touch-based decision-making task that we have developed for mice. Mice will be given a pattern of stimulation to their whiskers and will lick for a reward if they recognise the pattern; meanwhile, we will observe and perturb neuronal activity in targeted sets of neurons within the part of sensory cortex that receives the most direct input from the whiskers. These state-of-the-art methods are already available and tested in our lab. By elucidating the organisation of non-sensory responses in this way, we will contribute new insight into how the cerebral cortex underpins brain function.

The cerebral cortex is involved in sensory processing, decision making and motor control, and a classical view holds that these functions are implemented in separate cortical regions. However, recent work from multiple groups including ours, in both humans and other animals, has found that the activity of many neurons in sensory cortical areas can reflect decisions, upcoming movements and even the ensuing outcomes. Such neurons may be part of a network broadcasting decisions and actions across the brain; in some cases, they may directly bias the animal's decision to act. These non-sensory responses are thus important to our understanding of cortex function and are now well-established experimentally, but little is known about how they are organised. This project will address this gap by examining how non-sensory responses vary across classes of neurons in local circuits of sensory cortex, and how those neurons are connected to the rest of the brain. Using the somatosensory "barrel" cortex of mice (S1bf) as an experimental model, we will test three hypotheses: (i) sensitivity to non-sensory variables differs by neuron type; (ii) non-sensory responses depend on top-down connections from frontal cortex, and this dependence changes as an animal learns to link stimuli with actions; (iii) non-sensory neurons themselves have distinct connections back to other brain areas involved in action selection. To do this, we will image activity in targeted neuronal populations of S1bf with two-photon microscopy and manipulate activity with optogenetics, while mice carry out a whisker-mediated sensory discrimination task. This novel task was developed in our lab, requires S1bf for successful performance, and its learning causes neurons with non-sensory responses to appear in S1bf. Thus, in conjunction with two-photon imaging and optogenetics -both ready in our lab- it provides a powerful model for elucidating the emergence and organisation of non-sensory responses in sensory cortex.