Behavioral and facial thermal responses were recorded in twelve 3- to 4-month-old infants during the Still-Face Paradigm (SFP). As in the usual procedure, infants were observed in a three-step, face-to-face interaction: a normal interaction episode (3 min); the “still-face” episode in which the mother became unresponsive and assumed a neutral expression (1 min); a reunion episode in which the mother resumed the interaction (3 min). A fourth step that consisted of a toy play episode (5 min) was added for our own research interest. We coded the behavioral responses through the Infant and Caregiver Engagement Phases system, and recorded facial skin temperature via thermal infrared (IR) imaging. Comparing still-face episode to play episode, the infants’ communicative engagement decreased, their engagement with the environment increased, and no differences emerged in self-regulatory and protest behaviors. We also found that facial skin temperature increased. For the behavioral results, infants recognized the interruption of the interactional reciprocity caused by the still-face presentation, without showing upset behaviors. According to autonomic results, the parasympathetic system was more active than the sympathetic, as usually happens in aroused but not distressed situations. With respect to the debate about the causal factor of the still-face effect, thermal data were consistent with behavioral data in showing this effect as related to the infants’ expectations of the nature of the social interactions being violated. Moreover, as these are associated to the infants’ subsequent interest in the environment, they indicate the thermal IR imaging as a reliable technique for the detection of physiological variations not only in the emotional system, as indicated by research to date, but also in the attention system. Using this technique for the first time during the SFP allowed us to record autonomic data in a more ecological manner than in previous studies.