The effect of argumentation on cognitive functions at human -machine interaction

Attention has been focused on building argumentation based XAI which reinforces mutual explainability. Nevertheless, there are not many researches that explore the cognitive compatibility in HMI through the user perspective. Therefore, the aim of the study was to investigate some effects of argumentation on cognitive aspects such as decision making, problem solving and cognitive load in a XAI context, namely, Machine Coaching. By using a 93 factorial design (N=258), we saw that example based explanations mostly affect the confidence level of participants while prototype based, influence the behavior of participants. A pattern of people with high level of expertise preference of Near miss and people with lower level of expertise preference of Far miss, was visible as well and needs further investigation. Argumentation promotes not only consistency of cognitive behavior but also, higher performance. In general, our findings suggest: 1) integration of psychometrics such as trust, comprehension and persuasiveness combined with self-report tools, 2) personalized XAI which distinguishes users based on cognitive skills and background knowledge, 3) examination of Cognitive load while using EEG, and 4) further empirical studies on this matter.