DuckSoup: An open-source experimental platform to manipulate participants’ facial and vocal attributes during social interactions

Social interaction research, like the study of attraction or friendship between humans in natural situations, is incredibly difficult to conduct.

Researchers typically cannot control the variables under study, and there is a myriad of multimodal factors potentially influencing our feelings and decisions. To address this problem, we have created an experimental platform called DuckSoup to study human social interactions.

DuckSoup is a videoconference application, just as Zoom or Skype, but where researchers can manipulate the video- and audio feed in real time, and control participants’ facial and vocal attributes during unscripted interactions (for example, increase smiling or frowning or change the emotional tone of voice).

The processing done by DuckSoup operates with very little latency, giving researchers the possibility to efficiently collect large cross-cultural and cross-sectional data sets. The current proposal has two main purposes. Firstly, we will further develop and enhance DuckSoup and release it to the public as an open source platform.

Secondly, we will use the plattform to study the role of emotional display in attraction and cooperation, using ecologically-valid scenarios like romantic dating.

This will give us an unprecedented ability to go beyond previous correlational findings, and directly infer causation about social interaction processes.