BackgroundSubclinical psychotic symptoms are present in the general population. Furthermore, they are quite common in diagnostic categories beyond psychosis, such as BPD patients. MethodsWe want to assess the differences between 3 groups: BPD (n = 68), FEP (n = 83) and controls (n = 203) in an experimental paradigm measuring the presence of speech illusions in white noise. The Positive and Negative Syndrome Scale was administered in the patient group, the Structured Interview for Schizotypy-Revised, and the Community Assessment of Psychic Experiences in the control and BPD group. The white noise task was also analysed within a signal detection theory (SDT) framework. Logistic regression analyses and the general linear models were used to analyse the adjusted differences between groups. ResultsDifferences were more prevalent in signals that were perceived as affectively salient in patients groups (9.6% in FEP vs 5.9% in BPD and 1% in controls; OR: 10.7; 95%CI: 2.2–51.6, p = 0.003 in FEP; OR: 6.3; 95%CI: 1.1–35.0, p = 0.036 in BPD). Besides, we found a worse general performance and more false alarms in the task for FEP group using SDT framework. ConclusionsExperimental paradigms indexing the tendency to detect affectively salient signals in noise may be used to identify liability to psychosis in people with vulnerability. Its predictable value in other diagnostic categories and general population requires further research.