Investigating the behavioural and cerebral bases of empathy using human-computer interactions

同理心是在某种程度上理解和分享他人感受的能力。这种复杂的现象已经从神经科学和神经生理学的角度进行了多年的研究。即使这项关于同理心的大脑基础的研究在过去十年中取得了长足的进步，这一领域也遇到了一些挑战。这些挑战之一是使用控制良好但不是很吸引人的涉及情感或痛苦的实验设计，因此不能很好地代表实际的互动。这个项目的主要目标是通过观察人体和大脑在移情互动过程中的反应来继续研究移情。这一次，我们将通过人机交互将其提升到另一个水平，机器(计算机生成的角色，也称为化身或虚拟代理)可以扮演受难者的角色，也可以扮演照顾者的角色。这个项目的第一个目标是研究参与者对(1)非语言行为，(2)不协调行为和(3)组合情感状态(当时不止一种情感)的表达的神经生理反应，所有这些都是由虚拟主体呈现的。我们将使用不同的痛苦和情感背景来研究患者或照顾者的观点。对于第二个目标，我们将记录与虚拟代理人互动的大组参与者的行为和生理反应，并使用先进的统计方法(机器学习)来确定哪个(“标记物”)最能预测共情反应。然后，第三个目标将是创建和测试标记预测组合(算法)的使用，以从计算机生成的角色引入实时自适应回溯。这个移情的虚拟代理将被与预编程的虚拟代理进行比较。最后，第四个目标是通过使用先前开发的相同算法来改善人与人之间的互动，以将信息发送到计算机屏幕等常见设备(增强的同理心)。这将允许人们一起互动，以获得他们通常无法获得的关于另一个人的情感状态的线索。这一研究方案将为来自不同研究领域的学生和人员提供独特的和具有启发性的多学科培训机会。该项目将有助于更好地理解人类在移情互动过程中的大脑和生理反应，同时也将对虚拟代理人和计算机界面的开发产生影响，这些虚拟代理人和计算机界面考虑到用户的情绪状态。最后，它将为设计最先进的技术手段来改善人与人之间的互动打开大门。

Empathy is the ability to understand and share to some extent the feeling of another person. This complex phenomenon has been studied from a neuroscience and neurophysiological perspective for a number of years. Even if this research on the cerebral bases of empathy has advanced considerably in the last decade, this field has met a number of challenges. One of these challenges has been the use of well-controlled but not very engaging experimental designs involving emotions or pain, thus not very representative of actual interactions. The main goal of this project is to continue the study of empathy by looking at the responses of the human body and brain during empathic interactions. This time, we will bring it to another level by using human-machine interactions, where the machine (computer-generated characters, also named avatars or virtual agents) can play either the role of the person suffering or of the care taker. The first objective of this project is to study the participants' neurophysiological responses to (1) non-verbal behaviours, (2) incongruent behaviours and (3) the expression of combined affective states (more than one emotion at the time), all presented by virtual agents. We will use various contexts of pain and emotions to study both the perspectives of the sufferer or of the carer. For the second objective, we will record the behaviour and physiological responses of large groups of participants interacting with virtual agents and use advanced statistical methods (machine learning) to determine which one ("markers") are the most predictive of an empathic reaction. Then, the third objective will be to create and test the use of the predictive combination of markers (algorithms) to introduce real-time adaptive retroactions from the computer-generated characters. This empathic virtual agent will be compared to pre-programmed virtual agents. Finally, the fourth objective is to improve human-human interactions by using the same algorithms previously developed to send information to a common device such as a computer screen (augmented empathy). This will allow people interacting together to obtain cues, which are typically not available to them, about the affective state of the other person. This research programme will provide unique and stimulating multidisciplinary training opportunities to students and personnel from different fields of research. The project will contribute to a better understanding of human's cerebral and physiological responses during empathic interactions, while also having an impact on the development of virtual agents and computer interfaces that take into account the emotional state of the use. Finally, it will open the door to designing state-of-the-art technological means to improve human-human interaction.