Most invasive Brain Computer Interfaces (iBCIs) use spike and Local Field Potentials (LFPs) from the motor or parietal cortices to decode movement intentions. It has been debated whether harvesting signals from other brain areas that encode global cognitive variables, such as the allocation of attention and eye movement goals in a variety of spatial reference frames, may improve the outcome of iBCIs. Here, we explore the ability of LFP signals, sampled from the lateral prefrontal cortex (LPFC) of macaque monkeys, to encode eye-movement intention during the pre-movement fixation period of a delayed saccade task. We use spectral dimensionality reduction to examine the spatiotemporal properties of the extracted non-rhythmic broadband activity and explore its usefulness in decoding saccade goals. The dynamics of the broadband signal in low spatial dimensions across the pre-movement fixation period uncovered saccade target separation; its discriminative potential was confirmed using support vector machine classifications. These findings reveal that broadband LFP from the LPFC can be used to decode intended saccade target location during pre-movement periods. We further provide a general workflow that can be implemented in iBCIs and it is relatively robust to the loss of spikes in individual electrodes.

中文翻译：

---

使用光谱、空间和时间降维从灵长类动物前额皮质局部场电位解码眼跳意图

大多数侵入性脑机接口 (iBCI) 使用来自运动或顶叶皮层的尖峰和局部场电位 (LFP) 来解码运动意图。从其他大脑区域获取编码全局认知变量的信号（例如在各种空间参考框架中的注意力分配和眼球运动目标）是否可以改善 iBCI 的结果一直存在争议。在这里，我们探索了从猕猴外侧前额叶皮层 (LPFC) 采样的 LFP 信号在延迟扫视任务的运动前固定期间编码眼球运动意图的能力。我们使用频谱降维来检查提取的非节奏宽带活动的时空特性，并探索其在解码扫视目标中的有用性。在运动前固定期间低空间维度的宽带信号的动态揭示了扫视目标的分离；使用支持向量机分类确认了它的判别潜力。这些发现表明，来自 LPFC 的宽带 LFP 可用于在运动前期间解码预期的扫视目标位置。我们进一步提供了一个可以在 iBCI 中实施的通用工作流程，它对于单个电极中尖峰的丢失相对稳健。