The Reality and Limitations of Implicit Control based on Passive Brain-Computer Interfacing

拟议的项目涉及一种新的人机交互(HCI)形式，我们称之为隐式控制。来自生理计算、自适应自动化和被动脑-计算机接口(PBCI)等领域的思想和方法的统一允许解决基本的人机界面问题：用户和机器之间仍然非常不对称和受限的通信形式，其中用户必须将他们的想法和意图转换成特定的小命令序列(例如按钮按下、点击等)。有了pBCI，与任务相关的信息可以直接从用户正在进行的大脑活动中收集，而不需要他们付出任何额外的努力。使用这些信息来实时构建和调整用户模型，系统可以适当地响应其用户的当前认知和情感状态。因此，有效地，用户有可能在不知道这样做的情况下控制系统--隐含地，通过他们的隐蔽状态。我们已经使用一种光标控制形式演示了这一概念证明：参与者隐含地将光标指向目标。认识到这种技术的潜力--在机器可以实际理解其用户的情况下，自然人机界面的潜力--我们现在建议在我们的概念功能证明的基础上，进一步研究这一概念的潜力及其局限性。将对四个基本方面进行调查。1)隐式光标控制的生产潜力。可以使用这项技术做出选择决定吗？参与者将把光标指向从多个选项中选择的一个目标。2)该技术对用户的不同解释起作用的潜力。还能检测到多个备选方案中哪一个是最不受欢迎的吗？3)最初的演示器是与事件相关的，但我们有强烈的迹象表明，对用户状态的连续解释是可能的，从而可能实现更快、连续的光标控制。这件事将得到调查。4)道德方面的考虑发挥了重要作用，因为使用这类系统的用户可能不知道正在进行的确切交流。还将调查这项技术的这些和其他伦理方面，以及潜在的用户接受度。因此，拟议的项目涵盖了与这一新兴技术有关的重要主题，并将初步概述其未来可能的用途和应用、可能要处理的问题以及设计者和用户在伦理方面的指导方针。

The proposed project concerns a new form of human-computer interaction (HCI) which we call implicit control. A unification of ideas and methods from the fields of, among others, physiological computing, adaptive automation, and passive brain-computer interfacing (pBCI) allows a fundamental HCI problem to be addressed: the still very asymmetrical and restrictive form of communication between a user and a machine, where the user must translate their ideas and intentions into a specific sequence of small commands (e.g. button presses, clicks, etc.). With pBCI, task-relevant information can be gathered from a user's ongoing brain activity directly, without requiring any additional effort from them. Using this information to build and adapt a user model in real time, a system can respond appropriately to the current cognitive and affective states of its user. Effectively, it is thus possible that a user controls a system without being aware of doing so--implicitly, through their covert states. We have demonstrated this proof of concept using a form of cursor control: implicitly, participants guided a cursor towards a target. Recognising the potential of such technology--the potential of natural HCI where the machine, as it were, actually understand its user--we now propose to further investigate the concept's potential and its limitations, building up on our functional proof of concept. Four essential aspects will be investigated. 1) The productive potential of implicit cursor control. Can selection decisions be made using this technology? Participants will steer the cursor towards one self-selected target out of multiple alternatives. 2) The potential of the technology to work with different interpretations on the side of the user. Can it also be detected which of multiple alternatives is least desirable? 3) The original demonstrator is event-related, but we have strong indications that a continuous interpretation of user state, and thus faster, continuous cursor control is possible. This will be investigated. 4) Ethical considerations play an important role as users using such systems may not be aware of the exact communication taking place. These and other ethical aspects of the technology, as well as potential user acceptance, will be investigated as well. The proposed project thus covers essential topics of interest related to this emerging technology, and will result in an initial overview of its possible future uses and applications, possible issues to be dealt with, and guidelines for designers and users with respect to ethical considerations.