The difference in functional connectivity during human-human interaction and human-robot interaction.
Summary
The developments in artificial intelligence are leading us towards a new scientific frontier of socially engaged robots. This poses new questions regarding the impact of unfamiliar agents that alter our social environment. In order to optimise the communication with these robots, the question that has been answered in the current study is whether the social interaction between humans is distinct from the social interaction with robots. In previous studies the neural embodiment of social features has been examined through the activation of the main social networks, i.e. the person perception- and the theory-of-mind network. Here, a novel approach was performed by extracting the functional connections between the hubs of these networks (fusiform face area; FFA & temporoparietal junction; TPJ). In order to do so, an existing fMRI dataset was analysed. For the experiment the participants (N = 22) alternately conversated with a robot and a person during scanning. Correlation analysis between the extracted time series revealed identical connectivity patterns for human-human and human-robot interaction (HHI & HRI) that are stable over time. Specifically, the connection between the FFA and TPJ was increased, while control regions exhibited decreased functional connectivity with the FFA during conversation. As a result, I propose a novel theory concerning a general interaction network that is connected during communication regardless of the type of conversational agent. To further explore this theory, future research should include a full connectome study to ascertain whether functional connectivity during human-robot interaction remains similar to human- human interaction.