Neural Computing with Oscillations

点击翻译按钮获取中文摘要

Understanding how intelligent behaviour emerges from a network of neurons has been the holy grail of psychology and neuroscience for decades. Scientists are gradually peeling back the layers, exposing the neurological mechanisms behind simple behaviours, one by one. Oscillations are ubiquitous in the brain. Understanding their role is fundamental in our understanding of how the brain works in general. This project proposes to investigate the role of oscillations in neural computing, with a focus on navigation and localization, an animal's perception of where it is in space. A number of biological observations of the entorhinal cortex (EC) and hippocampus (HC) suggest that they are involved in navigation and localization. Recent computational models have proposed a variety of mechanisms for how oscillations in these regions contribute to this encoding. However, these models address the observations in a somewhat piecemeal and disconnected fashion, and lack an overarching, simple framework. Through our experience in medical image processing, we have seen how the Fourier transform and its properties offer a theoretical framework for understanding and manipulating oscillations. We plan to apply our knowledge of signal processing and propose a Fourier framework for these navigation processes. Moreover, we will investigate what self-organizing principles might be involved in forming these oscillating networks. A principled study of oscillations in this system might translate to other systems and other oscillations. Our work could pave the way to a better understanding of oscillation-related neurological disorders such as epilepsy, Parkinson's, Alzheimer's, schizophrenia, and possibly ADHD and bipolar disorder. A deeper understanding of how oscillations are leveraged in the animal brain could also lead to breakthroughs in the use of neural models for intelligent robotics and brain-machine interfaces.