Social affective touch is an important aspect of close relationships in humans. It has been also observed in many non-human primate species. Despite the high relevance of behaviours like embraces for personal wellbeing and mental health, they remain vastly under-investigated in behavioural neuroscience. This may be because human neuroscience often relies on a limited repertoire of behavioural measurements such as error rates and reaction time measurements. These are, however, insufficient to capture the multidimensional complexity of highly interactive dyadic behaviours like embraces. Based on recent advances in computational ethology in animal models, the rapidly emerging field of human computational ethology utilizes an accessible repertoire of machine learning methods to track and quantify complex natural behaviours. In this short communication, we highlight how such techniques can be utilized to investigate social touch and which preliminary conditions, motor aspects and higher-level interactions need to be considered. Ultimately, integration of digital ethology with mobile neuroscience techniques such as ultraportable EEG systems will allow for an ecologically valid investigation of social touch in humans that will advance affective behavioural neuroscience.