Bicyclists remain among the most vulnerable road users, particularly at urban intersections where visibility limitations and complex maneuvers increase the risk of severe crashes. Connected vehicle (CV) technologies, such as Collision Warning Systems (CWS), have the potential to enhance situational awareness and reduce the likelihood of such events. Yet, little is known about their impacts in mixed connectivity environments, where only some road users are equipped with these systems, especially when bicyclists are involved. This study leverages a novel dual-simulator framework that enables real-time interaction between a human-controlled driver and a human-controlled bicyclist within a high-fidelity Digital Twin of Delaware Avenue in Newark, Delaware, USA. The simulation environment integrates CARLA for immersive 3D visualization and PTV Vissim for background traffic operations, providing both naturalistic and controlled conditions. The experimental design simulated scenarios in which a driver executes a left turn while a bicyclist, traveling straight in a dedicated bike lane, remains within the driver’s blind spot. The analysis first examines whether designed interactions can be classified as conflicts by combining a kinematic surrogate safety indicator, Time-to-Collision (TTC), with physiological stress responses (heart rate). Building on this analysis, the study evaluates the role of CWS across four connectivity scenarios, focusing on how connectivity impacts users’ perception, anticipation, and decision-making. Finally, the study analyzes whether repeated exposure to similar events across different intersections produces behavioral adaptation, impacting how participants respond over time. This research advances conflict analysis methodologies in VR and offers new insights into how connectivity technologies influence bicyclist and driver abilities, and their behaviors in multimodal traffic environments.