Future intelligent virtual co-robots, or cobots, will work as extensions of the human mind and body in creative cognitivetasks, such as design, invention, or creation of art. Because these tasks depend on emotional attitudes, the cobot needsto maintain a social-emotional contact with the user. This can be achieved based on a cognitive architecture, in whichthe current emotional state of the user is represented in a two-dimensional weak semantic map. Cobot action selectionis determined by this state, the action appraisal, and the currently active M-schema. Main results include a significantlyhigher quality of the outcome, compared to the control condition, without a semantic map. It is remarkable that one andthe same cognitive model proves useful in various domains, including creative assistants of a choreographer, a composer,a designer, and an insight problem solver. The work yields preliminary results that suggest many potential practicalapplications. This research is supported by the Russian Science Foundation Grant # 18-11-00336.

Emotional capabilities of interactive virtual agents grow every year, while most of them lack in emotional intelligence. Their abilities to recognize and express emotions may be worthless, if the agent cannot decide how to behave adequately in a given social context. The central part of the problem, the logic of socially emotional behavior, remains an unsolved challenge. Here this challenge is addressed within a virtual reality paradigm of a poster presenter, designed on the basis of the emotional Biologically Inspired Cognitive Architecture (eBICA). The framework of eBICA combines the formalisms of semantic maps, moral schemas, mental states and narratives in order to achieve believable, socially acceptable interactive behavior of the presenter bot. The paradigm involves establishment and maintenance of a stable socially emotional contact with mutual empathy and trust. Results of evaluation of the prototype by participants at two international virtual conferences speak in favor of the selected approach.

A person who needs psychological help has to overcome an internal barrier before deciding to see a psychotherapist. On the other hand, obtaining an anonymous online consultation or taking an online test is psychologically easy. At the same time, state-of-the-art solutions of this sort are limited. Here a prototype of an intelligent assistant is presented that should help to alleviate the problem. The prototype is an embodied actor with socially emotional capabilities. Implementation is based on cognitive modeling and machine learning. The prototype is capable of conducting an initial psychological screening during a natural conversation with the user. As a result, the system can determine psychological characteristics of user’s personality and give further recommendations. Systems of this sort could be used beyond psychological counseling and can be expected to have an impact on the healthcare system in general.

Deep learning (DL) makes it possible to automate the work of theorists, developers, and programmers in creating intelligent agents. At the same time, it has limitations related to the statistical nature of the method, which largely ignores available domain knowledge and requires large training datasets. These limitations become a barrier in the field of modeling and characterization of human social-emotional behavior. Biologically inspired cognitive architectures (BICA) are based on facts from cognitive and brain sciences and therefore can be successful here, although at the great cost of intellectual human labor. The challenge therefore is to integrate the two approaches, combining their strengths and compensating for their weaknesses. Here a particular form of such integration is presented, which involves a scaffolding of DL by BICA. Experimental paradigms include a virtual registrar, a virtual partner dance, a virtual clownery, and more. This work was supported by the Russian Science Foundation Grant #22-11-00213.

Modern intelligent tutoring systems, exploiting technological advances in augmented and virtual reality and large language models, offer fluent natural language interaction between a virtual character and a student complemented with a multimodal interface, including recognition and synthesis of affects and intentions expressed in speech tonality, facial expression, gaze, and body language. Being concerned with consumer satisfaction, developers of such systems often miss the educational needs. Here we present a Virtual Tutor that, using the above technologies, helps students to self-regulate during learning. This is made possible based on the self-regulated learning theory integrated into an emotional cognitive architecture. Virtual Tutor uses its emotional intelligence to model, guide, and motivate students to engage in self-regulation. It does it in parallel with performing the basic tutoring functions. Results of our preliminary study provide some evidence of support for Virtual Tutor. This work was supported by the Russian Science Foundation Grant #22-11-00213, https://rscf.ru/en/project/22-11-00213/.