Neurons linking the brain region that controls movement to the region involved in auditory control have been found to suppress auditory responses when mice move, but the reason for this inhibition is unclear. See Article p.189
The key to human cognition lies in the neocortex, a modular brain structure that is unique to mammals. Within each neocortical module, small ensembles of neurons are wired together in stereotyped patterns. Subsets of these neurons send long-range axonal projections to other modules to create systems of circuits that transform the activity of single neurons into complex behaviours such as perception, cognition and motor control. Understanding how different neocortical regions — including the motor, visual and auditory cortices — coordinate their activity is a central challenge in systems neuroscience. In this issue, Schneider et al.1 (page 189) describe a technically sophisticated set of experiments that unravels the mechanisms by which the motor cortex exerts control over the auditory cortex during locomotion.
Locomotion facilitates visual responses in the visual cortex2 but, conversely, Schneider and colleagues observed that it suppresses sound-evoked responses in the auditory cortex. This observation is intriguing because these responses are also suppressed when an animal vocalizes3 or engages in an auditory task4, behavioural states that require careful auditory processing. What is the mechanism by which locomotion suppresses neuronal responses in the auditory cortex?
Neuronal firing rates are determined by the balance between signals that promote and inhibit firing, so, in principle, firing can be suppressed by either a decrease in excitatory signals or increased inhibition. To distinguish between these possibilities, Schneider and co-workers performed the challenging feat of making intracellular-activity recordings from neurons in the auditory cortex of mice running on a treadmill. These experiments revealed that decreased auditory responses during locomotion are the result of an increase in inhibition. Cortical inhibition arises almost entirely from local inhibitory interneurons that make only short-range connections with nearby neurons, so the interneurons are probably driven by long-range excitatory inputs that transmit signals into the auditory cortex. But which long-range inputs are responsible?
The authors hypothesized that long-range inputs arrive from the motor cortex. To test this, they labelled the subset of motor-cortex neurons that sends axonal projections to the auditory cortex (motor–auditory neurons) with a protein that fluoresces when activated, and monitored the neurons during locomotion. They found that the activity of motor–auditory neurons is increased before and throughout movement, indicating that they could be responsible for auditory-cortex suppression (Fig. 1). The researchers therefore set out to demonstrate that activation of motor–auditory neurons was not just correlated with suppression, but was also causally involved.
To establish causality, Schneider et al. infected motor–auditory neurons with a virus that enabled them to express channelrhodopsin-2 protein. Expression of channelrhodopsin-2 (which is originally derived from algae5) allows neurons to be activated in response to light. Selective stimulation of the axon terminals of motor–auditory neurons with light resulted in a suppression of the auditory cortex that was indistinguishable from that elicited by locomotion, supporting a causal role for this direct projection. However, this experiment alone was inconclusive, because excitation of motor–auditory axons may travel backwards along the motor projection, exciting other targets of the motor neurons and so indirectly affecting auditory responses. To rule out the possibility that suppression was indirect, the authors repeated the experiments while pharmacologically blocking activity in the motor cortex, and achieved the same result.
Finally, Schneider and colleagues inhibited motor-cortex neurons during locomotion, which disabled motor inputs to the auditory cortex. The authors found that in the absence of motor-cortex activity, locomotion was not accompanied by auditory suppression. Thus, the motor-to-auditory cortex projection is both necessary and sufficient for locomotion to suppress auditory responses.
Why should the auditory cortex be suppressed during locomotion? One might imagine that decreased activity in the auditory cortex implies reduced auditory sensitivity. However, behavioural conditions that require enhanced auditory processing typically suppress responses in the auditory cortex3,4, raising the possibility that suppressed responsiveness serves to increase sensitivity. Such a seemingly paradoxical increase in sensitivity in the face of a general decrease in auditory cortical activity would occur if a privileged subset of cortical outputs were spared the effect of feedback suppression. In much the same way that shushing a noisy audience makes it possible to hear the seminar speaker, so feedback suppression may act to 'shush' all but the most important outputs from the auditory cortex.
The current results might be best considered in the framework of active sensation — that is, how animals separate self-induced sensory inputs from externally induced ones6. Movement and locomotion generate various types of self-induced sensation (for example, the movement of an object on your retina as you move your head), and so sensory inputs consist of both externally derived and self-induced sensations. Our perception separates these two sources of sensation to provide us with a movement-independent representation of the environment. To achieve this separation, a copy of the motor command might be used to indicate to the sensory cortices that movement is occurring. This copy could then be used to subtract the self-induced motor signal from the externally generated signal. The present results provide a detailed description of a circuit that may be involved in just such a computation.