The neocortex evolved after vertebrates were able to perform complex perceptual tasks. Its added value is almost certainly in providing mammals the capability to optimally function beyond genetic programming, optimization that emerges from rapid and long-term time scale dynamics in this structure and with its interconnected networks. A canonical example of rapid time-scale neocortical dynamics occurs with behavioral allocation of attention, following which the sensitivity and specificity of neocortical neurons changes substantially, in ways that predict enhanced information processing.
We study neocortical dynamics, with the goal of understanding their meaning for information processing. Crucial to understanding the meaning of dynamics, for example the expression of a neural oscillation during attention, is to understand underlying mechanisms, as these provide the real world 'embodied' constraints as to the potential functionality of the system.
An example of the progression between mechanism and meaning in our laboratory are recent studies of neocortical 'gamma' oscillations. Gamma oscillations are rhythmic activity patterns in the range from ~30-80 Hz that are increased in many neocortical areas during active processing, for example with the allocation of attention. The importance of this dynamics is a much fought over topic, with views ranging from the belief that these signals are key to consciousness perception, to the view that they are an ‘exhaust fume’ of computation, an epiphenomenal accident with no link to optimal sensory processing.
In initial mechanistic studies (Cardin et al., 2009 Nature; Cardin et al., 2010; Carlen et al., 2011), we used optogenetics to causally test in vivo the hypothesis that synchronization of a specific type of interneuron, fast-spiking cells, was key to the expression of gamma. We found that a highly naturalistic and specific expression of gamma was generated by selective drive of this ‘FS’ cell class, in agreement with a wealth of prior computational and correlative studies. We are now using this highly specific form of control to test the hypothesis that induced FS-gamma can enhance detection performance. We have found that endogenous expression of FS-gamma and specific emulation of FS-gamma by optogenetic drive both predict increased likelihood of correctly detecting a sensory input. These effects are interestingly unique to less salient stimuli—as with many effects in attention, there is no added benefit for the processing of innately salient, easier to perceive inputs (Siegle*, Pritchett* and Moore, under revision).
To test link between neocortical dynamics and perception, we leverage a variety of different methods with complementary strengths.
A major focus in the laboratory is 2-photon imaging in awake, behaving mice, using genetically encoded calcium signals to record activity across many neurons simultaneously, to study their role in sensory processing. We also combine neuronal imaging with tracking of non-neural elements such as blood vessels to study their dynamics.
A second major focus is the use of large chronic electrode implant arrays specifically adapted for use in mice. Led by Jakob Voigts, we developed the 'flexDrive,' an ultra-light (< 2g) implant capable of carrying up to 64 channels and 3 independently manipulated fiber optics (Voigts et al., 2013).
The study of dynamics in neural systems requires the ability to detect and perturb of enhance them in real time. A further innovation, made by Voigts and Josh Siegle, is their development of the Open Ephys electrophysiology system. This open source system is equivalent to commercial systems in performance, an order of magnitude cheaper (< $5K) and has already built a very active community for code sharing and hardware optimization, to allow rapid progress in the field. Perhaps most importantly, the system was built to optimize real time detection and intervention, a feature in which it exceeds almost all existing (and much more expensive) systems.
Optogenetics provides a unique tool for causal intervention in neocortical dynamics, and as indicated in our case study, we use this approach extensively in a wide variety of cellular targets to causally test the impact of their activation on local activity. Building on the recent explosion in development of optogenetic elements, we are also working on developing new causal control methods that are significantly less invasive.
Computational modeling of detailed neocortical networks is a major tool we employ for hypothesis generation and testing. Key to our use of this approach is a long-standing collaboration with Dr. Stephanie Jones. Collaborations with Dr. Jones are also focused on human neurophysiological testing of neocortical dynamics, for example in MEG. Our recent collaborative efforts include development of a low cost high performance open source EEG system with the Open Ephys architecture, to allow real time detection of human dynamics combined with causal intervention (e.g., brain stimulation).