– ACT-SENSE

The rhythmic sweep of rodent vibrissae during active touch is an ideal system to study the coding, circuitry, and dynamics of active sensing. We use this system to ask how an active sensory system can form a transient robust response that is invariant to details of the sensation. Here, the response in the input layer of cortex signifies the location of an object relative to the face, whose amplitude is invariant to the force and timing of vibrissa touch. The circuit for rhythmic motor control of the vibrissae is well delineated and part of an orofacial sensory system that is coordinated by the oscillator for inhalation. The rhythm is sufficiently precise so that whisking and touch are well described in terms of phase in the whisk cycle, as opposed to by kinetic parameters.
Our experiments gain from a breadth of anatomical, behavioral, electrophysiological and optical methods that include wavefront shaping for deep brain imaging. Our data is at the level of fields of single spines and boutons as well as single neurons, all in behaving mice. Preliminary results show that a major fraction of thalamocortical (TC) boutons encode phase in the whisk cycle so that the TC input to cortex is represented by a torus, with phase as the poloidal and neuron identity as the toroidal dimension. Excitatory cortical input (L4E) neurons inherit a phase preference for touch in the whisk cycle through a by mapping of the toroidal input, with disorder, across a cortical column. Aim 1 deals with thalamocortical transformation to the L4 network during free whisking. It involves the construction of theoretical models at various level of abstraction that are motivated by our preliminary data, yet are of interest in their own right. Ring architectures, here with an external rhythmic drive and asymmetric connectivity, are a predominant part of our theoretical effort to extend state-of-the-art continuous attractor models into the realm active sensing and sensorimotor control. Predictions from these models drive Aim 2 and Aim 3 Aim 2 focuses on the characterization and abstract understanding of the structure of thalamocortical input and predominantly subthreshold response of L4E neurons during rhythmic whisking. Both experiment and theory are focused on the balance between an excitatory TC input and the recurrent inhibitory input, mediated by L4 inhibitory interneurons, among with the L4E neurons.
Aim 3 focuses on the transformation of touch into spiking of L4E neurons. We conjecture that the excitatory connections function predominantly at the time of contact, unlike in the case of a standard ring model. In fact, preliminary results reveals that the excitatory synaptic component has a delayed onset whose time-lag appears to track the progression of phase in the whisk cycle.
Both experiment and theory will thus focus on the role of L4E connectivity to form a bump of activity that signifies touch in terms of phase and possibly auxiliary coordinates.
Intellectual merit: Our proposed study provides a unique scheme for neuronal computations during active sensation. It concerns the inherence of order in maps of sensory features between two layers, here phase from thalamus to cortex. It also concerns the interplay between excitatory inputs and inhibitory interactions to maintain a network at the threshold of activation, here L4 in the absence of touch. A third concerns transient recurrent activity to stabilize a response, here synaptic timing tied to preferred phase during rhythmic whisking. This work extends the notion of ring attractors and invariance to transient sensory responses.