In everyday life, people continuously interact with each other to achieve goals or to simplyexchange states of mind. How people react to and interact with the surrounding world is a productof evolution: the success of our species is also due to our social intellect, allowing us to live ingroups and share skills and purposes. In other words, our brain has evolved not only in term ofcognitive but also of social processing.From one side, social neuroscience has recently stressed the investigation of human interactions toreveal brain areas involved in these complex processes. Interestingly enough, in the past, socialsignals during interactions have been studied mainly by social sciences, highlighting characteristicgestures (such as hand shaking) but neglecting the contribution that the motor system may give tohuman interplays. Actually, the motor system is a fundamental part of the brain networks involvedin social cognition, as motor predictive mechanisms may contribute to the anticipation of whatothers are going to do next and regulate our own reactions, a principal function of social cognition.On the other side, social interactions are nowadays accessible to automatic analysis throughcomputer science methods, namely, computer vision and pattern recognition, the main disciplinesused for automatic scene understanding. Observation activities have never been as ubiquitous astoday and they keep increasing in terms of both amount and scope. Furthermore, the involvedtechnologies progress at a significant pace (some sensors exceed now human capabilities) and, asthey are cheap and easily available, have an increasingly large diffusion. This does not happen bychance: automatic tools make the observation objective and rigorous while safer and more2comprehensive, so that public and private environments can be monitored 24 hours a day fromdifferent points of view with limited human intervention.To date, however, neuroscientific findings about social interactions have been rarely shared withcomputer vision, as these disciplines are traditionally far from each other. The goal of this paper isto survey the methods for understanding human behavior in social interactions from both acomputational and a neuroscientific perspective, showing how they can gain large mutual benefitswhen these issues are tackled in an unified manner.In particular, methods for real-time gesture recognition, algorithms for the analysis of the bodypostures and the extraction of proxemics cues are only few examples of features that may help theonline registration and characterization of interactions under a genuine neuroscience perspective. Inthis way, the video modality could be finally considered in the analysis, whereas the audio channelhas been traditionally the most considered information source by neuroscientists so far.Similarly, understanding the processes underlying human behavior in social interactions startingfrom motor gestures is extremely important to design computer systems able to model specificsituations and events in an principled way. This can be faced by capturing novel features (specificpostures, subtle gestures) which have a precise meaning as consequences of activations of welldefined parts of the brain network.In conclusion, in this work, we will show not only that a joint approach involving neurosciencesand computational sciences can tackle in more rigorous way the studies on human social behavior,but will also disclose new perspectives and open up fresh research issues in both domains.

Automatic human interaction understanding: lessons from a multidisciplinary approach

CRISTANI, Marco;MURINO, Vittorio
2012-01-01

Abstract

In everyday life, people continuously interact with each other to achieve goals or to simplyexchange states of mind. How people react to and interact with the surrounding world is a productof evolution: the success of our species is also due to our social intellect, allowing us to live ingroups and share skills and purposes. In other words, our brain has evolved not only in term ofcognitive but also of social processing.From one side, social neuroscience has recently stressed the investigation of human interactions toreveal brain areas involved in these complex processes. Interestingly enough, in the past, socialsignals during interactions have been studied mainly by social sciences, highlighting characteristicgestures (such as hand shaking) but neglecting the contribution that the motor system may give tohuman interplays. Actually, the motor system is a fundamental part of the brain networks involvedin social cognition, as motor predictive mechanisms may contribute to the anticipation of whatothers are going to do next and regulate our own reactions, a principal function of social cognition.On the other side, social interactions are nowadays accessible to automatic analysis throughcomputer science methods, namely, computer vision and pattern recognition, the main disciplinesused for automatic scene understanding. Observation activities have never been as ubiquitous astoday and they keep increasing in terms of both amount and scope. Furthermore, the involvedtechnologies progress at a significant pace (some sensors exceed now human capabilities) and, asthey are cheap and easily available, have an increasingly large diffusion. This does not happen bychance: automatic tools make the observation objective and rigorous while safer and more2comprehensive, so that public and private environments can be monitored 24 hours a day fromdifferent points of view with limited human intervention.To date, however, neuroscientific findings about social interactions have been rarely shared withcomputer vision, as these disciplines are traditionally far from each other. The goal of this paper isto survey the methods for understanding human behavior in social interactions from both acomputational and a neuroscientific perspective, showing how they can gain large mutual benefitswhen these issues are tackled in an unified manner.In particular, methods for real-time gesture recognition, algorithms for the analysis of the bodypostures and the extraction of proxemics cues are only few examples of features that may help theonline registration and characterization of interactions under a genuine neuroscience perspective. Inthis way, the video modality could be finally considered in the analysis, whereas the audio channelhas been traditionally the most considered information source by neuroscientists so far.Similarly, understanding the processes underlying human behavior in social interactions startingfrom motor gestures is extremely important to design computer systems able to model specificsituations and events in an principled way. This can be faced by capturing novel features (specificpostures, subtle gestures) which have a precise meaning as consequences of activations of welldefined parts of the brain network.In conclusion, in this work, we will show not only that a joint approach involving neurosciencesand computational sciences can tackle in more rigorous way the studies on human social behavior,but will also disclose new perspectives and open up fresh research issues in both domains.
2012
Social Signal Processing; Neuroscience; Pattern Recognition
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11562/470791
Citazioni
  • ???jsp.display-item.citation.pmc??? 1
  • Scopus 3
  • ???jsp.display-item.citation.isi??? ND
social impact