Exploring the Cognitive Mechanisms Behind the Adoption of Algorithmic Advice

Algorithm-based decision support systems are increasingly employed across various decision environments. This article investigates how algorithmic support influences the human decision-making process from a neuro-information systems (neuro-IS) perspective. The goal of this study is twofold: first, to elucidate the cognitive mechanisms behind the adoption of algorithmic advice, and second, to draw managerial insights from these cognitive mechanisms. Utilizing electroencephalography to measure human brain activity during decision-making with and without algorithmic support, the study finds that the presence of algorithmic support significantly influences a series of decision-making processes. Specifically, it impacts information conflict processing (observed in beta-band activity at the early stage of decision-making), decision-making itself (observed in theta-band activity at the late stage of decision-making), and reward prediction error processing (observed in theta-band activity at the outcome evaluation stage). Furthermore, there is a significant correlation between the changes in subjects' neural activity due to the introduction of algorithmic support and their levels of compliance with the algorithmic advice during decision-making. Additionally, the results show that neural measures outperform self-reported measures in explaining and predicting algorithm compliance. This study highlights the roles of cognitive and emotional factors in shaping algorithm compliance and yields practical implications for designing and promoting the usage of algorithmic support to achieve higher compliance levels. Overall, by observing the relationship between neural processes and behavioral changes, this study leverages the complementary value of neurophysiological methods to offer new managerial insights into the design of IS, thereby contributing to the development of the neuro-IS discipline.