Neural Correlates of Trust During an Automated System Monitoring Task: Preliminary Results of an Effective Connectivity Study

Abstract

As autonomous systems become more prevalent and their inner workings become more opaque, we increasingly rely on trust to guide our interactions with them especially in complex or rapidly evolving situations. When our expectations of what automation is capable of do not match reality, the consequences can be sub-optimal to say the least. The degree to which our trust reflects actual capability is known as trust calibration. One of the approaches to studying this is neuroergonomics. By understanding the neural mechanisms involved in human-machine trust, we can design systems which promote trust calibration and possibly measure trust in real time. Our study used the Multi Attribute Task Battery to investigate neural correlates of trust in automation. We used EEG to record brain activity of participants as they watched four algorithms of varying reliability perform the SYSMON subtask on the MATB. Subjects reported their subjective trust level after each round. We subsequently conducted an effective connectivity analysis and identified the cingulate cortex as a node, and its asymmetry ratio and incoming information flow as possible indices of trust calibration. We hope our study will inform future work involving decision-making and real-time cognitive state detection.

Neural Correlates of Trust in Automation: Considerations and Generalizability Between Technology Domains

“…They also report that this activation negatively correlates with human trust; an increase in activation suggests lower trust. Validation of this experiment was perfomed by Sanders et al (2019), who also observed increased activation of the region around FCz during unreliable conditions. All EEG studies that manipulated reliability report similar findings, where at least one of the locations Fz, F1, F2, FC1, FCz, FC2, C1, Cz, C2 are identified with identical activation direction (increased activity for lower trust).…”

Heading Toward Trusted ATCO-AI Systems: A Literature Review

“…Contemporary technology advances in automation, artificial intelligence (AI), and computational algorithms have created great benefits for human operators, as well as increases in adverse or unexpected consequences [1], [2], [3]. To ensure safe and seamless human-machine interaction (HMI), trust between humans and machines (in the form of AI or automated tools) has become a widely discussed topic over the past three decades [4], [5], [6].…”

“…More specific and operational definitions of human-machine trust that has been widely accepted in the current human factors literature concern the perception of trust as: (i) "the extent to which a user is willing to act on the basis of the information, recommendations, actions, and decisions of a computer-based tool or decision aid" (p. 11 in [7], adapted from p. 1 in [9])) or (ii) "the attitude that an agent will help achieve an individual's goals in a situation characterized by uncertainty and vulnerability" (p. 6 in [10]). Here, it is worth noting that according to the second operational definition, trust is characterized as decision-making heuristic that machine operators use in situations that are too uncertain or changing rapidly [3], [2].…”

“…Together with these two popular operational definitions of trust, the concept of calibrated trust must also be mentioned. Calibrated trust refers to the condition in which the anticipated or perceived trust of a machine, automation or AI matches the experienced trust associated with the actual use of the tool [10], [1], [3]. Seen in this light, trust becomes an evolving process that gets updated with increased information of the operating machine or system [1].…”

“…A phenomenon common to all these studies pertains to the use of experimental design or tasks that involved the modulation of reliability (or dependability) offered by an automated tool (Table II, column 2). Reliability refers to how consistent an automation can perform in providing accurate information and was largely manipulated through the programming and presentation of automated agents or advisories that can provide information with different probabilities of accuracy [35], [1], [36], [3] or risk-taking tendencies [37], [38].…”