Video of Compound Affect (VoCA) is a collection of 156 colour facial expression videos, posed by one subject. Each expression is presented in frontal, forty-five degree profile, and ninety-degree profile angles under natural lighting conditions. These expressions form a catalogue of compounds associated with the emotion labels happiness, sadness, anger, disgust, fear, surprise, and contempt. The videos are organised by compound type. Single upper, single lower (SUSL) compounds depict the actions associated with one emotion label in the upper face and another in the lower face. Neutral upper, single lower (NUSL) compounds depict neutrality in the upper face and the actions associated with one emotion label in the lower face. Single upper, neutral lower (SUNL) compounds depict neutrality in the lower face and the actions associated with one emotion label in the upper face. Neutral upper, dual lower (NUDL) compounds depict neutrality in the upper face and the actions associated with two emotion labels in the lower face. Dual upper, neutral lower (DUNL) compounds depict neutrality in the lower face and the actions associated with two emotion labels in the upper face. Single upper, dual lower (SUDL) compounds depict the actions associated with one emotion label in the upper face and two in the lower face. Dual upper, single lower (DUSL) compounds depict the actions associated with two emotion labels in the upper face and one in the lower face. Dual upper, dual lower (DUDL) compounds depict the actions associated with two emotion labels in the upper face and two in the lower face. Contempt (C) compounds depict (1) the actions associated with one emotion label in the upper face with another in the lower face modified by a unilateral action associated with contempt, (2) the actions associated with one emotion label in the upper face and a unilateral action associated with contempt in the lower face, and (3) neutrality in the upper face and the actions associated with one emotion label in the lower face modified by a unilateral action associated with contempt. Such arrangements are particularly suited to studies of face processing and emotion labelling in clinical populations. They also suit studies on partial facial neutrality and emotion labelling. Broader potential applications include testing and training FACS-based technologies, guiding the construction of multi-subject sets, proving concept and feasibility for complex controlled poses, and assisting creative industry projects.

I recorded the videos across two sessions, on consecutive days, in August 2021. I made reasonable efforts to hold consistent camera; camera angle; room location; head, body, and eye position; clothing; and lighting. I selected emotion labels by reviewing several prominent resources and extracting consensus codes. I describe this process in Table 1 of the metadata document. I became Facial Action Coding System (FACS; Ekman et al. 2002) certified in 2015. It was not possible to involve other coders due to time and cost limitations; accordingly, where use of the material relies upon the FACS code metafields in version 1 of the VoCA supplementary documentation, I advise users to seek input from at least one other FACS certified coder. I’d also be grateful of the chance to add their scores to future versions with permission, as this information is vital for the dataset’s continuous improvement.