Thilakarathne Dilhan J, Treur Jan
Agent Systems Research Group, Department of Computer Science, VU University Amsterdam, De Boelelaan 1081, 1081 HV, Amsterdam, The Netherlands.
Brain Inform. 2015 Jun;2(2):77-106. doi: 10.1007/s40708-015-0013-3. Epub 2015 Mar 20.
This paper presents a computational cognitive model for action awareness focusing on action preparation and performance by considering its cognitive effects and affects from both prior and retrospective form relative to the action execution. How action selection and execution contribute to the awareness or vice versa is a research question, and from the findings of brain imaging and recording techniques more information has become available on this. Some evidence leads to a hypothesis that awareness of action selection is not directly causing the action execution (or behaviour) but comes afterwards as an effect of unconscious processes of action preparation. In contrast, another hypothesis claims that both predictive and inferential processes related to the action preparation and execution may contribute to the conscious awareness of the action, and furthermore, this awareness of an action is a dynamic combination of both prior awareness (through predictive motor control processes) and retrospective awareness (through inferential sense-making processes) relative to the action execution. The proposed model integrates the findings of both conscious and unconscious explanations for both action awareness and ownership and acts as a generic computational cognitive model to explain agent behaviour through the interplay between conscious and unconscious processes. Validation of the proposed model is achieved through simulations on suitable scenarios which are covered with actions that are prepared without being conscious at any point in time, and also with the actions that agent develops prior awareness and/or retrospective awareness. Having selected an interrelated set of scenarios, a systematic approach is used to find a suitable but generic parameter value set which is used throughout all the simulations that highlights the strength of the design of this cognitive model.
本文提出了一种针对动作意识的计算认知模型，该模型通过考虑其认知效应以及相对于动作执行的先验和回顾形式的影响，专注于动作准备和执行。动作选择和执行如何促进意识，反之亦然，这是一个研究问题，并且从脑成像和记录技术的研究结果中，关于这方面已有更多信息。一些证据导致一种假设，即动作选择的意识并非直接导致动作执行（或行为），而是在动作准备的无意识过程之后产生的结果。相比之下，另一种假设认为，与动作准备和执行相关的预测和推理过程可能有助于动作的意识，而且，这种动作意识是相对于动作执行的先验意识（通过预测性运动控制过程）和回顾意识（通过推理性意义构建过程）的动态组合。所提出的模型整合了关于动作意识和归属的有意识和无意识解释的研究结果，并作为一个通用的计算认知模型，通过有意识和无意识过程之间的相互作用来解释智能体行为。通过在合适的场景上进行模拟来验证所提出的模型，这些场景涵盖了在任何时间点都无需意识准备的动作，以及智能体产生先验意识和/或回顾意识的动作。在选择了一组相互关联的场景后，采用系统方法来找到一个合适但通用的参数值集，该参数值集用于所有模拟中，突出了这个认知模型设计的优势。
