The information encoded by single neurons has been extensively studied at both the sensory and motor levels. Neuronal circuits that adaptively transform sensory signals into motor signals have been a topic of great interest, but experimentally, such transformations have almost always been subjected to only pair-wise analyses. In contrast, the work proposed here is a study of adaptive information processing by networks of sensorimotor neurons. I propose initially to develop mathematical and computational tools to explore the role of network connectivity in information processing. This will be accomplished both with a large-scale simulated network of spiking neurons, and by using arrays of electrodes implanted chronically in the sensorimotor cortex of behaving rats. With these analytical tools, I will induce and then characterize connectivity changes in both the simulated and in vivo networks. Numerous methods have been developed for measuring functional connectivity amongst neurons. In particular, a class of point process models has been shown to perform quite well in a number of preparations. I propose to extend these models by evaluating them with a Bayesian framework. This adds two important features to the connectivity model. First, the addition of priors allows preexisting knowledge about the nervous system to be incorporated into the model in a principled manner. Second, this model framework generates not just estimates of the connectivity, but confidence intervals on that estimate as well. This is crucial to allow us to say that the measured changes in connectivity are of statistical significance. In the rat, I will induce predictable changes in the strength of individual connections within the rat sensorimotor cortical network using tetanic and spike-timing dependent potentiation protocols. This will demonstrate controlled, predictable changes in the functional connection between pairs and potentially ensembles of neurons. This work has relevance for, and benefits from, the development of the brain-machine interfaces (BMIs) that directly connect to the central nervous system. Either as a source of forward motor commands, or sensory feedback, a BMI involves a small number of neurons: recording and stimulating from tens to hundreds of neurons.
| Nazarpour, Kianoush; Ethier, Christian; Paninski, Liam et al. (2012) EMG prediction from motor cortical recordings via a nonnegative point-process filter. IEEE Trans Biomed Eng 59:1829-38 |
| Rebesco, James M; Miller, Lee E (2011) Enhanced detection threshold for in vivo cortical stimulation produced by Hebbian conditioning. J Neural Eng 8:016011 |
| Rebesco, James M; Miller, Lee E (2011) Stimulus-driven changes in sensorimotor behavior and neuronal functional connectivity application to brain-machine interfaces and neurorehabilitation. Prog Brain Res 192:83-102 |
| Rebesco, James M; Miller, Lee E (2010) Altering function in cortical networks by short-latency, paired stimulation. Conf Proc IEEE Eng Med Biol Soc 2010:1674-7 |
