One fundamental aspect of human cognition is the ability to understand (perceive and predict) the intentions of others via their actions. The ability to adapt to social situations and successfully interact with other people requires a (fairly) accurate prediction of unfolding events. Here, we take social signals to refer to all (externally) observable variables that are conducive to providing information about the intentions of other agents during movements and interactions (for instance, the kinematics of movements or given by the context of the movement). Although many studies addressed human social interactions, the question of which subset of all externally observable variables carry the most information and/or are most commonly used as a source of social signals remains a topic of much research. Relevant work includes investigations of mirror neuron mechanisms and affordances but is not limited to those.
Since the topic deals with predictions of intentions, it lends itself well to efforts in computational modelling. Relevant work addresses (1) predictive explanations of human performance in intention understanding from observation, (2) accounts – including information theoretic approaches – of how social signals can be used in identification and prediction of intentions and (3) accounts of how such mechanisms can tie in with, for example, research on Theory of Mind (ToM).
A particular application for research in this area is in the field of robotics, particularly in human robot interaction (although the concept can be extended to interaction with other machines), for example involving service robotics, socially assistive robots (SAR) or robot assisted therapy (RAT). Such robots can benefit from algorithms that directly predict likely intentions of humans. These can be informed from human studies (for instance on the use of social signals) but, since robots have limited sensory and analytical abilities compared to humans, particular care needs to be given to how results on human strategies (or models thereof) can be translated onto robots. Equally, purely robotic strategies to achieve such predictions from the observation of humans (or indeed other robots) can be developed and compared to human approaches.
Conversely, designing robots whose social signals are particularly intuitive to humans is also relevant. Research in this direction can profit from an understanding of what social signals humans process when observing a scene.
Finally, a highly relevant but to date relatively unexplored area is that of endowing robots with a ToM tailored to the humans (or robots) they interact with. Here, a particular interest is which observable variables are most conducive to/can be used in the creation and/or improvement of such mechanisms.
The present topic is aimed at state of the art research which addresses research into social signals either (a) from a behavioural, psychological and neuroscientific perspective (b) through computational models (c) with robotic applications. Interdisciplinary papers covering two or more of these are especially encouraged. Papers should further the understanding of the use of social signals in human interactions and/or address applications in human-machine interaction. Papers on computational models in particular therefore should be careful to make their relevance to such applications clear.
One fundamental aspect of human cognition is the ability to understand (perceive and predict) the intentions of others via their actions. The ability to adapt to social situations and successfully interact with other people requires a (fairly) accurate prediction of unfolding events. Here, we take social signals to refer to all (externally) observable variables that are conducive to providing information about the intentions of other agents during movements and interactions (for instance, the kinematics of movements or given by the context of the movement). Although many studies addressed human social interactions, the question of which subset of all externally observable variables carry the most information and/or are most commonly used as a source of social signals remains a topic of much research. Relevant work includes investigations of mirror neuron mechanisms and affordances but is not limited to those.
Since the topic deals with predictions of intentions, it lends itself well to efforts in computational modelling. Relevant work addresses (1) predictive explanations of human performance in intention understanding from observation, (2) accounts – including information theoretic approaches – of how social signals can be used in identification and prediction of intentions and (3) accounts of how such mechanisms can tie in with, for example, research on Theory of Mind (ToM).
A particular application for research in this area is in the field of robotics, particularly in human robot interaction (although the concept can be extended to interaction with other machines), for example involving service robotics, socially assistive robots (SAR) or robot assisted therapy (RAT). Such robots can benefit from algorithms that directly predict likely intentions of humans. These can be informed from human studies (for instance on the use of social signals) but, since robots have limited sensory and analytical abilities compared to humans, particular care needs to be given to how results on human strategies (or models thereof) can be translated onto robots. Equally, purely robotic strategies to achieve such predictions from the observation of humans (or indeed other robots) can be developed and compared to human approaches.
Conversely, designing robots whose social signals are particularly intuitive to humans is also relevant. Research in this direction can profit from an understanding of what social signals humans process when observing a scene.
Finally, a highly relevant but to date relatively unexplored area is that of endowing robots with a ToM tailored to the humans (or robots) they interact with. Here, a particular interest is which observable variables are most conducive to/can be used in the creation and/or improvement of such mechanisms.
The present topic is aimed at state of the art research which addresses research into social signals either (a) from a behavioural, psychological and neuroscientific perspective (b) through computational models (c) with robotic applications. Interdisciplinary papers covering two or more of these are especially encouraged. Papers should further the understanding of the use of social signals in human interactions and/or address applications in human-machine interaction. Papers on computational models in particular therefore should be careful to make their relevance to such applications clear.
