This paper proposes a conceptual model derived through current ongoing research that incorporates the potential relationship that mental health may have on trust and reliance calibration in automated systems (AS). Understanding the variables involved in the human-AS interaction allows system designers to better achieve trust calibration and avoid AS misuse and disuse. However, most of the research area is saturated with understanding how external and internal (both to the human) short-term cognitive symptoms mediate this critical relationship. Therefore, the present paper extends human-AS trust literature with common mental disorders (CMDs) as outlined by the Adult Psychiatric Morbidity Survey (APMS) and incorporates them into existing models within the engineering and psychology subject areas to begin to understand what this relationship may look like. It is hoped that this paper will expand the scope of human factors (specifically human-AS trust) to include mental disorders.

This paper explores whether generalised anxiety disorder (GAD) and depression have any effect on an individual's explicit general propensity to trust automated systems (trust that is unspecific to any one automated system) and whether those that do have these disorders have an implicit bias towards automated systems over other humans. The human-automated system literature to date has discovered that individual differences in humans, such as self-confidence, mood, and personality types, can influence the human-automated system relationship through human trust and reliance attitudes and behaviour. However, whether suffering from a mental disorder influences an individual's attitudes towards automated systems generally is yet to be explored. In this study, 184 UK university students responded to online experiments between December 2019 -January 2020 and were subjected to the cultural trust instrument survey and the implicit association test in a between-subjects design to measure their general propensity to trust and implicit association towards automated systems respectively. A two-way ANOVA was performed to evaluate GAD × depression interaction effects on the dependent variables. Results suggest there was a significant interaction between GAD and depression regarding propensity to trust automated systems but they have little to no influential effect on mean implicit association test scores. Furthermore, those without depression showed a significantly higher trust score when they also had GAD. It can be concluded that GAD and depression have potential critical influence over human-automated system trust, thus creating potential issues with misuse and disuse and must be accounted for in automated system design.