The biological preparedness model suggests that survival-related visual cues elicit physiological changes without awareness to enable us to respond to our environment. Previous studies have reported some evidence for this effect. In the current article, we argue that this evidence is subject to methodological confounds. These include the use of a universal masked presentation threshold, the employment of hit rates (HRs) to measure meta-awareness, and the assertion of overall guess-level target detection using nonsignificance. In the current report, we address these issues and test whether masked emotional faces can elicit changes in physiology. We present participants with subjectively adjusted masked angry, fearful, happy, and neutral faces using HRs and signal detection theory. We assess detection performance using a strict Bayesian criterion for meta-awareness. Our findings reveal that HR adjustments in the detection threshold allow higher skin conductance responses to happy, fearful, and angry faces, but that this effect could not be reported by the same participants when the adjustments were made using signal detection measures. Combined these findings suggest that very brief biologically relevant stimuli can elicit physiological changes but cast doubt to the extent that this effect can occur in response to truly unconscious emotional faces.