Toward Adaptive Trust Management for Human-automation Teaming Using an Instance-based Learning Cognitive model

Abstract

Trust in automation is seen as a core factor affecting human-automation teaming. Inappropriate calibration of trust in automation can damage the performance and safety goals of the collaborative team. It is essential to develop automation that can correctly calibrate human trust in it. Herein, based on the view that trust comes from interaction, we use an instance-based learning cognitive model to obtain the cognitive process involved in the interaction between dispatchers and automated Decision Support Systems(DSSs) in the Fully Automatic Operation (FAO) circumstances, and obtain from the model an internal estimate of the calibration state of human trust. We consider integrating the model into automation so that it can judge the hidden calibration status of the human teammate's trust, and respond to the trust dynamics in an online and adaptive way. We discuss our results and the potential of the instance-based computational cognitive process model to improve human-automation teaming. Our model has great potential to avoid the sluggish effect caused by dispatchers failing to obtain effective decision support in time in the FAO circumstances, especially when dealing with emergencies under high time pressure.

Keywords: Trust in automation, Human-automation teaming, Cognitive model

DOI: 10.54941/ahfe1001931