Algorithm interruption

Interruptions are pervasive in digital work, yet most research assumes their source is irrelevant. This study challenges that assumption by investigating how algorithmic versus human interruptions differentially affect autonomy, attentional residue, performance, and strain. Drawing on self-determination and attentional residue theories, we theorize that algorithmic interruptions are uniquely autonomy-thwarting and residue-amplifying, particularly when opaque and imposed during non-routine tasks. We tested this model across two complementary studies: a 15-day field experiment in China using digital logs and daily surveys (N ≈ 300 employees, 30 teams) and a cross-cultural laboratory experiment in China and the UK (N = 480). Multilevel path analyses showed that algorithmic interruptions reduced autonomy by .3 SDs, which in turn elevated residue and decreased performance accuracy by 10–12%. Transparency dashboards and alignment with routine tasks buffered these effects, while cultural tightness shaped their magnitude—effects were stronger in the UK (looser) than China (tighter). The findings extend interruption research by showing that source and design matter, refine autonomy theory by recognizing algorithms as autonomy-thwarting actors, and extend attentional residue theory by highlighting design opacity as an amplifier. Practically, we recommend transparency features, snooze options, and culturally sensitive deployment of algorithmic tools to support psychologically sustainable digital workplaces