In dynamic (social) environments, an affective state of “stress” can be adaptive and promote agent wellbeing, but maladaptive if not appropriately regulated. The presence of (and interactions with) affect-based social support has been hypothesised to provide mechanisms to regulate stress (the “social buffering” hypothesis), though the precise, underlying mechanisms are still unclear. However, the hormone oxytocin has been implicated in mediating these effects in at least two ways: by improving social appraisals and reducing the short-term release of stress hormones (i.e., cortisol), and adapting an agent’s long-term stress tolerance. These effects likely facilitate an agent’s long-term adaptive ability by grounding their physiological and behavioural adaptation in the (affective) social environment, though these effects also appear to be context-dependent. In this paper, we investigate whether two of the hypothesised hormonal mechanisms that underpin the “social buffering” phenomenon affect the long-term wellbeing of (artificial) social agents who share affective social bonds, across numerous social and physical environmental contexts. Building on previous findings, we hypothesise that “social buffering” effects can improve the long-term wellbeing of agents who share affective social bonds in dynamic environments, through regular prosocial interactions with social bond partners. We model some of the effects associated with oxytocin and cortisol that underpin these hypothesised mechanisms in our biologically-inspired, socially-adaptive agent model, and conduct our investigation in a small society of artificial agents whose goal is to survive in challenging environments. Our results find that, while stress can be adaptive and regulated through affective social support, long-term behavioural and physiological adaptation is determined by the contextual perception of affective social bonds, which is influenced by early-stage interactions between affective social bond partners as well as the degree of the physical and social challenges. We also show how these low-level effects associated with oxytocin and cortisol can be used as “biomarkers” of social support and environmental stress. For socially-situated artificial agents, we suggest that these “social buffering” mechanisms can adapt the (adaptive) stress mechanisms, but that the long-term efficacy of this adaptation is related to the temporal dynamics of social interactions and the contextual perception of the affective social and physical environments.
During communication, humans express their emotional states using various modalities (e.g., facial expressions and gestures), and they estimate the emotional states of others by paying attention to multimodal signals. To ensure that a communication robot with limited resources can pay attention to such multimodal signals, the main challenge involves selecting the most effective modalities among those expressed. In this study, we propose an active perception method that involves selecting the most informative modalities using a criterion based on energy minimization. This energy-based model can learn the probability of the network state using energy values, whereby a lower energy value represents a higher probability of the state. A multimodal deep belief network, which is an energy-based model, was employed to represent the relationships between the emotional states and multimodal sensory signals. Compared to other active perception methods, the proposed approach demonstrated improved accuracy using limited information in several contexts associated with affective human–robot interaction. We present the differences and advantages of our method compared to other methods through mathematical formulations using, for example, information gain as a criterion. Further, we evaluate performance of our method, as pertains to active inference, which is based on the free energy principle. Consequently, we establish that our method demonstrated superior performance in tasks associated with mutually correlated multimodal information.
The affective motion of humans conveys messages that other humans perceive and understand without conventional linguistic processing. This ability to classify human movement into meaningful gestures or segments plays also a critical role in creating social interaction between humans and robots. In the research presented here, grasping and social gesture recognition by humans and four machine learning techniques (k-Nearest Neighbor, Locality-Sensitive Hashing Forest, Random Forest and Support Vector Machine) is assessed by using human classification data as a reference for evaluating the classification performance of machine learning techniques for thirty hand/arm gestures. The gestures are rated according to the extent of grasping motion on one task and the extent to which the same gestures are perceived as social according to another task. The results indicate that humans clearly rate differently according to the two different tasks. The machine learning techniques provide a similar classification of the actions according to grasping kinematics and social quality. Furthermore, there is a strong association between gesture kinematics and judgments of grasping and the social quality of the hand/arm gestures. Our results support previous research on intention-from-movement understanding that demonstrates the reliance on kinematic information for perceiving the social aspects and intentions in different grasping actions as well as communicative point-light actions.