Eliciting caregiving behavior in dyadic human-robot attachment-like interactions

Abstract

Based on research in developmental robotics and psychology findings in attachment theory in young infants, we designed an arousal-based model controlling the behaviour of a Sony AIBO robot during the exploration of a children play mat. When the robot experiences too many new perceptions, the increase of arousal triggers calls for attention from its human caregiver. The caregiver can choose to either calm the robot down by providing it with comfort, or to leave the robot coping with the situation on its own . When the arousal of the robot has decreased, the robot moves on to further explore the play mat. We present here the results of two experiments using this arousal-driven control architecture. In the first setting, we show that such a robotic architecture allows the human caregiver to influence greatly the learning outcomes of the exploration episode, with some similarities to a primary caregiver during early childhood. In a second experiment, we tested how human adults behaved in a similar setup with two different robots: one needy, often demanding attention, and one more independent, requesting far less care or assistance. Our results show that human adults recognise each profile of the robot for what they have been designed, and behave accordingly to what would be expected, caring more for the needy robot than the other. Additionally, the subjects exhibited a preference and more positive affect whilst interacting and rating the robot we designed as needy. This experiment leads us to the conclusion that our architecture and setup succeeded in eliciting positive and caregiving behaviour from adults of different age groups and technological background. Finally, the consistency and reactivity of the robot during this dyadic interaction appeared crucial for the enjoyment and engagement of the human partner.In the remainder of this paper, we present early research trying to bridge the field of developmental robotics with attachment theory, which besides a thought experiment [Kaplan 2001] and a fairly remote use of the theory [Arkin 1998], has largely remained an unexplored area of research. The main goal is to address how a robotic platform could use the properties of this bond in order to thrive from it, as children most often manage to do. To that end, we present a body of related work that we took inspiration from, then describe what properties of attachment we used and why they are relevant to the design of robotic architectures. We finally present the results of two experiments that bring together our findings, and assess to what extent we improved the state of the art of the field concerned with developing robots and improving human-robot interactions.

Making New “New AI” Friends: Designing a Social Robot for Diabetic Children from an Embodied AI Perspective

“…3.2.2 were all very much in line with, and could be interpreted in the light of, our Embodied AI approach to autonomous robots [16,17], and our research on motivationally autonomous robots [15,18,23,35] and on modeling the development of attachment [20,29]. We thus decided to design a motivationally autonomous affective robot companion-named "Robin" after "Robot Infant"-to implement the "robot actor" and "insulin companion" scenarios that were given as part of the needs of the medical staff (Sect.…”

Modelling emotional attachment: An integrative framework for architectures and scenarios

“…So it is more like a precocial species than altricial human infants. Hiolle and Canamero and co-workers [9] also constructed robots that explored their surroundings but which curtailed this exploration and returned to the proximity of their carer when they experienced too much 'arousal' resulting from too much stimulation from novel perceptual input. So these simulations demonstrated systems learning about objects but not learning about carer effectiveness in how to support effective exploration.…”

“…There are three kinds of emotional interrupt distinguished: [5] represents alarms from changes in the reactive layer, such as fear or wariness; [6] represents alarms from predicting the future, such as apprehension a carer may leave again after returning from a previous separation; and [7] represents tertiary emotions which are not alarms but perturbations to normal meta-management, deliberative and reactive processes, perhaps triggered by repeated references to an attachment figure being activated for deliberation. In symmetry with the perception column, there are also three levels of action [8,9,10], which deal with comparable information transfers, like movements, linguistic actions, and sharing of attachment models. Processes which transfer information in the three layers differ in their representational requirements: in the reactive layer information processes may lack compositional semantics, in the deliberative layer this is required, and in metamanagement layer information is represented with meta-semantics.…”

Long-Term Cohabitation with a Social Robot: A Case Study of the Influence of Human Attachment Patterns

“…A social robot possessing human characteristics and operating as a home assistant may therefore be a subject of human attachment. Bowlby's theory has already been considered in the context of social robotics [37], however, the conducted research was related to its implementation in the behavior of robots [21,22,25]. It is believed that the emotional equipment and robot behavior should depend on who the user is.…”