Political and economic challenges, amplified by social media algorithms, are increasingly polarising how people receive and evaluate information. This study examines how political views and immediate emotional reactions shape the perceived trustworthiness of information. We conducted a preregistered laboratory experiment in which participants evaluated politically charged statements while their emotional responses were assessed in real time. Facial expressions were recorded to identify seven discrete emotions, while galvanic skin response and heart rate were measured to capture physiological arousal and reaction intensity. Consistent with expectations and prior research, we observed a clear political confirmation bias: ideologically congruent statements were rated as more trustworthy, with the strongest effect among the most conservative participants. Increases in anger and sadness were positively associated with trust ratings across the sample, whereas joy was linked to reduced confirmation bias. Among conservatives, multiple emotions predicted overall trust and the extent of confirmation bias. Machine-learning models incorporating physiological data likewise identified anger and joy as the most informative emotional predictors; however, physiological measures did not improve predictive performance beyond the baseline model, contrary to our expectation. By integrating behavioural and real-time physiological measures, the findings highlight how emotional and cognitive processes interact to shape trust in political information.