About this Research Topic
We have an ever-expanding array of tools that alter neural activity, but do we know how to use them? Control goes beyond stimulation by specifying a desired outcome, defining a quantitative measure of error between the desired and achieved outcomes, and providing a foundation of theoretical and practical tools for designing stimulation that minimizes that error. Control theory as an engineering discipline developed over the last century primarily in the context of low dimensional mechanical or electronic systems. Direct application of this theory to the brain confronts a distinct set of assumptions appropriate to biological systems that renders many control problems ill posed or intractable.
Currently successful neural engineering applications exemplify the limitations we see. The brain appears to function in an exciting but challenging area between discrete and continuous systems, somewhere between a network of neurons and a slab of tissue. Most theories of sensory or cognitive function are described at a circuit or network level, in which the identity of individual neurons plays a key role. Successful stimulation applications in such problems are limited primarily to cases in which there is a fortuitous spatial arrangement of identities, such as the tonotopic labeled lines in the ear underlying cochlear implants. Alternatively, some highly successful neuroengineering applications, notably deep brain stimulation, work at a level of large scale synchronization. Although neural identity may play a critical role in the success of DBS, the problem is typically defined at a level of low-dimensional, continuous fields, without concern for neural identity. We suggest advances in neurocontrol will need to address the space between single neuron tuning and regional synchronization.
In this Research Topic, we welcome submissions that highlight limits to applying traditional control theory in neural applications; provide promising alternative formulations and methods for controllability, observability and/or model validation; employ control objectives that are cognitive or behavioral, or otherwise at a scale above network biophysics; connect current frontiers in the theory of brain function (e.g. coding in sensory and motor systems, or hippocampal function) to control objectives for scientific investigation or neuroengineering translation; contrast the robustness, complexity and specificity of control at peripheral vs. central stimulation targets (for example, cochlear vs. brainstem or cortical implants); and/or argue for new research directions, contrasted with existing literature, that are likely to substantially advance neurocontrol capabilities. It is not our goal to include hardware issues per se, but we encourage submissions that suggest alternative formulations of control problems motivated by the peculiar constraints imposed by available or foreseeable neurotechnology. We envision a broad range of possible submissions having experimental and/or theoretical content, with a unifying focus on working towards a coherent but general methodology for neurocontrol, as distinct from stimulation in the sense defined above.