Lowe Robert, Almér Alexander, Lindblad Gustaf, Gander Pierre, Michael John, Vesper Cordula
Ice Lab, Applied IT, University of GothenburgGothenburg, Sweden; Interaction Lab, School of Informatics, University of SkövdeSkövde, Sweden.
Ice Lab, Applied IT, University of Gothenburg Gothenburg, Sweden.
Front Comput Neurosci. 2016 Aug 23;10:88. doi: 10.3389/fncom.2016.00088. eCollection 2016.
Joint Action is typically described as social interaction that requires coordination among two or more co-actors in order to achieve a common goal. In this article, we put forward a hypothesis for the existence of a neural-computational mechanism of affective valuation that may be critically exploited in Joint Action. Such a mechanism would serve to facilitate coordination between co-actors permitting a reduction of required information. Our hypothesized affective mechanism provides a value function based implementation of Associative Two-Process (ATP) theory that entails the classification of external stimuli according to outcome expectancies. This approach has been used to describe animal and human action that concerns differential outcome expectancies. Until now it has not been applied to social interaction. We describe our Affective ATP model as applied to social learning consistent with an "extended common currency" perspective in the social neuroscience literature. We contrast this to an alternative mechanism that provides an example implementation of the so-called social-specific value perspective. In brief, our Social-Affective ATP mechanism builds upon established formalisms for reinforcement learning (temporal difference learning models) nuanced to accommodate expectations (consistent with ATP theory) and extended to integrate non-social and social cues for use in Joint Action.
联合行动通常被描述为一种社会互动，它要求两个或更多的共同行动者之间进行协调，以实现一个共同目标。在本文中，我们提出了一个关于情感评估神经计算机制存在的假设，这种机制可能在联合行动中被关键地利用。这样一种机制将有助于促进共同行动者之间的协调，从而减少所需的信息。我们假设的情感机制提供了一种基于价值函数的联想双过程（ATP）理论实现，该理论根据结果预期对外部刺激进行分类。这种方法已被用于描述涉及不同结果预期的动物和人类行为。到目前为止，它尚未应用于社会互动。我们将应用于社会学习的情感ATP模型描述为与社会神经科学文献中的“扩展通用货币”观点一致。我们将此与一种替代机制进行对比，该机制提供了所谓社会特定价值观点的一个示例实现。简而言之，我们的社会情感ATP机制建立在强化学习的既定形式主义（时间差分学习模型）之上，并进行了细微调整以适应预期（与ATP理论一致），并扩展到整合非社会和社会线索以用于联合行动。
