Embodied Language Learning and Cognitive Bootstrapping: Methods and Design Principles

Abstract

Co-development of action, conceptualization and social interaction mutually scaffold and support each other within a virtuous feedback cycle in the development of human language in children. Within this framework, the purpose of this article is to bring together diverse but complementary accounts of research methods that jointly contribute to our understanding of cognitive development and in particular, language acquisition in robots. Thus, we include research pertaining to developmental robotics, cognitive science, psychology, linguistics and neuroscience, as well as practical computer science and engineering. The different studies are not at this stage all connected into a cohesive whole; rather, they are presented to illuminate the need for multiple different approaches that complement each other in the pursuit of understanding cognitive development in robots. Extensive experiments involving the humanoid robot iCub are reported, while human learning relevant to developmental robotics has also contributed useful results.Disparate approaches are brought together via common underlying design principles. Without claiming to model human language acquisition directly, we are nonetheless inspired by analogous development in humans and consequently, our investigations include the parallel codevelopment of action, conceptualization and social interaction. Though these different approaches need to ultimately be integrated into a coherent, unified body of knowledge, progress is currently also being made by pursuing individual methods.

Predicting Perceived Age: Both Language Ability and Appearance are Important

“…Moreover, though we do not design our robots for deliberate affective grounding (i.e., the coordination effect of building common understanding of what behaviors can be exhibited, and how beahvior is interpreted emotionally) as in Jung ( 2017), we hypothesize that how our robots behave effects how they are perceived. Kiela et al (2015) compared tutoring sequences in parent-child and human-robot interactions with varying verbal and demonstrative behaviors, and Lyon et al (2016) brought together several areas of research relating to language acquisition in robotics. We differ from this previous work in that we do not explcitely tell our participants to interact with the robots as they would a child, effectively removing the assumption that participants will treat robots in an age-appropriate way.…”

A Robot Is Not Worth Another: Exploring Children’s Mental State Attribution to Different Humanoid Robots

“…A recent review of the literature by Marchetti et al (2018) showed that different physical characteristics of HSRs may significantly affect the quality of interaction between humans and robots at different ages. The construction of robots that integrate and expand the specific biological abilities of our species led to two different directions in robotic development based on different, though related, theoretical perspectives: developmental cybernetics (DC; Itakura, 2008 ; Itakura et al, 2008 ; Moriguchi et al, 2011 ; Kannegiesser et al, 2015 ; Okanda et al, 2018 ; Di Dio et al, 2019 ; Wang et al, 2020 ; Manzi et al, 2020a ) and developmental robotics (DR; De La Cruz et al, 2014 ; Cangelosi and Schlesinger, 2015 , 2018 ; Lyon et al, 2016 ; Morse and Cangelosi, 2017 ; Vinanzi et al, 2019 ; Zhong et al, 2019 ; Di Dio et al, 2020a , b ). The first perspective (DC) consists of creating a human-like system, by simulating human psychological processes and prosthetic functions in the robot (enhancing the function and lifestyle of persons) to observe people’s behavioral response toward the robot.…”

Crossmodal Language Grounding in an Embodied Neurocognitive Model

“…A research question related to representation learning is natural language acquisition since representations for language production and language perception in the human brain seem to form embodied and cross-modally integrated (Cangelosi and Schlesinger, 2015 ; Heinrich and Wermter, 2018 ). The data set is therefore particularly suited for research on the grounding of language in sensorimotor perception because the recording diligently followed the developmental robot approach (Lyon et al, 2016 ). Mechanisms for representation formation and bidirectional hierarchical composition and decomposition can get tested in the biologically plausible setting.…”