Using single-neuron recordings in the human brain to inform cognitive models of error monitoring

我们如何监督我们的行动？当我们拨错电话号码，下错电梯，脱口而出一些不恰当的话，或者发送一封我们真的不想发送的电子邮件时，我们很快就会注意到，即使没有人告诉我们我们犯了一个错误。这些行为都是我们每天监控自己表现的例子，这对我们从错误中学习至关重要。虽然外部反馈在监控我们的行为方面也发挥着重要作用，但值得注意的是，注意到我们犯了错误往往是仅仅通过自我监控。该研究项目利用难得的机会记录单个人类脑细胞，以破译大脑如何实现这一非凡的壮举。确定大脑如何使我们能够发现自己的错误具有广泛的意义，因为它可以使我们开发新的培训和教学策略，以帮助那些未能发现自己的错误或相反，对自己的错误过于敏感的人。此外，它还将提供有关错误监测的大脑机制的基础知识，这可能有助于开发一种非侵入性生物标志物，使我们能够评估训练策略的成功或失败，并能够创建具有更好性能监测能力的人工智能系统。该研究团队参与了几项针对代表性不足的群体的外展教育活动，旨在为初中和高中水平提供丰富课程，向学生传授认知神经科学的最新进展，并为高中和本科生带来神经科学STEM研究经验的机会。为了了解大脑如何使我们能够检测自己的错误，这项建议旨在测试三个主要的误差监控理论的预测。虽然这些理论是心理学和认知神经科学中大量研究的核心，但我们目前并不知道这些理论中的哪一个（如果有的话）实际上反映了我们的大脑如何实现这种关键的表现监控能力。受他们最近发现的在人类大脑内侧前额叶皮层（mPFC）中发出错误信号的单个神经元的启发，研究小组将检查单个mPFC神经元的反应（在神经外科手术之前监测癫痫患者以去除癫痫脑组织，他们自愿参加这项研究）。参与者将执行认知任务，以测试三种现存理论的不同预测，这可能会解决哪种理论是正确的。沿着来自mPFC的单个神经元记录，研究人员还将同时记录头皮和颅内EEG。这些串联记录技术将是特别有用的理解错误相关负波（ERN），这是一个经常研究的头皮EEG成分的神经基础。结合新的定量方法，研究人员希望能够从机制上了解人类大脑如何监控错误。研究人员还计划开始开发一个准确的计算模型，以了解人类mPFC如何监控错误，并结合本提案中的研究中获得的见解。该奖项反映了NSF的法定使命，并且通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

How do we monitor our actions? We notice quickly when we dial a wrong number, get off the elevator at the wrong floor, blurt out something inappropriate, or send an email that we really didn’t mean to send, even when no one tells us that we made a mistake. These behaviors are everyday examples of monitoring our own performance, which is critical for us to learn from our mistakes. Although external feedback can also play an important role in monitoring our actions, remarkably, noticing that we made an error often occurs from self-monitoring alone. This research project utilizes rare opportunities to record individual human brain cells to decipher how the brain achieves this extraordinary feat. Determining how the brain enables us to detect our own errors is of broad significance because it may enable us to develop new training and teaching strategies to help those who fail to detect their own errors or, conversely, are overly sensitive to their own mistakes. Also, it will provide fundamental knowledge about the brain mechanisms underlying error-monitoring that may facilitate the development of a non-invasive biomarker that might allow us to assess the success or failure of training strategies and also enable the creation of artificial intelligence systems with better abilities at performance monitoring. The research team is involved in several outreach educational activities to under-represented groups aimed at providing enrichment classes at the middle and high school level, to teach students about exciting recent advances in cognitive neuroscience, and also bringing opportunities for STEM research experiences in neuroscience for high school and undergraduate students.To understand how the brain enables us to detect our own errors, this proposal aims to test the predictions of three prominent theories of error monitoring. While these theories are at the core of a large number of studies in psychology and cognitive neuroscience, we do not know at present which (if any) of these theories actually reflects how our brain enables this critical ability of performance monitoring. Motivated by their recent discovery of single neurons which signal error in the medial prefrontal cortex (mPFC) of the human brain, the research team will examine the response of individual mPFC neurons (in patients being monitored for epilepsy prior to neurosurgery to remove epileptic brain tissue, who have volunteered to participate in this study). Participants will perform cognitive tasks to test the differing predictions of the three extant theories which may resolve which theory is correct. Along with the single neuron recordings from the mPFC, researchers will also simultaneously record scalp and intercranial EEG. These tandem recording techniques will be especially useful for understanding the neural basis of the error related negativity (ERN), which is a frequently studied scalp EEG component. Combined with novel quantitative methods, the researchers expect to gain a mechanistic understanding of how the human brain monitors errors. The researchers also plan to begin development of an accurate computational model of how the human mPFC monitors errors incorporating insights gained from the research in this proposal.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.