Single Trial Reconstruction of Acoustic Stimuli from Multielectrode Cortical Recordings of Awake Ferrets

感觉错觉表明，我们对世界的感知是多么强烈地受到大脑结构的影响。虽然人类可以报告他们的感知，但在动物实验中，我们必须依靠其他场所。在这个项目中，我们的目标是重建声刺激的频谱图基于多电极记录从一个神经元群在清醒的雪貂听觉皮层。模仿大脑的实时任务，我们将开发在单次试验中操作的重建技术，并尊重时间因果关系。为了解释神经元之间相关性的可能作用，我们将采用多线性建模框架进行刺激重建，提供有限范围的二阶相互作用。接下来，我们想要应用这些方法来估计动物在行为设置中对模糊刺激（例如流类型）的感知，从而验证重建。一般来说，解释神经元活动的能力将是一系列假肢应用的关键。该项目将由Shamma教授（马里兰大学）主持，他的团队在将计算分析与行为神经生理学相结合方面具有丰富的经验。

Sensoryose illusions demonstrate how strongly our perception of the world can be biased by the structure of the brain. While humans can report their percept, we have to rely on other venues in the case of animal experiments. In this project we aim to reconstruct the spectrograms of acoustic stimuli based on multielectrode recordings from a neuronal population in the auditory cortex of awake ferrets. Mimicking the brain’s real-time task, we will develop reconstruction techniques that operate on single trials and respect temporal causality. To account for the possible role of correlations between neurons, we will adapt the multilinear modeling framework for stimulus reconstruction, providing a limited range of second order interactions. Next, we want to apply these methods to estimate the animals percept of an ambiguous stimulus (e.g. streaming type) in a behavioral setting, allowing to verify the reconstructions. Generally, the ability to interpret the neuronal activity will be key for a range of prosthetic applications. This project will be hosted by Prof. Shamma (Univ. of Maryland), wh group has substantial experience in combining computational analysis with behavioral neurophysiology.