Teacher-learner interaction quantifies scaffolding behaviour in imitation learning

Shuntaro Okazaki,Rieko Osu,Yoshihiro Muraoka

doi:10.1038/s41598-019-44049-x

Abstract

Teachers often believe that they take into account learners’ ongoing learning progress in their teaching. Can behavioural data support this belief? To address this question, we investigated the interactive behavioural coordination between teachers and learners during imitation learning to solve a puzzle. The teacher manually demonstrated the puzzle solution to a learner who immediately imitated and learned it. Manual movements of teachers and learners were analysed using a bivariate autoregressive model. To identify bidirectional information exchange and information shared between the two agents, we calculated causality and noise covariance from the model. Information transfer observed from teacher to learner in the lateral component of their motion indicated imitation of the spatial information of the puzzle solution. Information transfer from learner to teacher in the vertical component of their motion indicated the monitoring process through which teachers adjust their timing of demonstration to the learner’s progress. The shared information in the lateral component increased as learning progressed, indicating the knowledge was shared between the two agents. Our findings demonstrated that the teacher interactively engaged in and contingently supported (i.e. scaffolded) imitation. We thus provide a behavioural signature of the teacher’s intention to promote learning indispensable for understanding the nature of teaching.

Highlights

Teaching is a fundamental human activity to transfer and accumulate knowledge and skills among people and their descendants, fostering their culture[1,2]
The puzzle duration of learners during imitation phase (IP) gradually decreased with durations of the second trial and later all being significantly shorter than the first trial (2: p = 0.006, R = 0.88; 3: p = 0.001, R = 0.88; 4: p = 0.001, R = 0.85; 5: p = 0.004, R = 0.81; 6: p = 0.009, R = 0.88; 7: p = 0.001, R = 0.88; 8: p = 0.001, R = 0.88; 9: p = 0.001, R = 0.83; 10: p = 0.001, R = 0.88)
The performance of learners in the IP significantly improved from the second trial, while that of teachers in the teacher phase (TP) improved from the third

Summary

Introduction

Teaching is a fundamental human activity to transfer and accumulate knowledge and skills among people and their descendants, fostering their culture[1,2]. Some quantitative studies have demonstrated non-verbal scaffolding such as exaggerated and slowed demonstration, or physical guidance when a human adult teaches a specific action to an infant or a robot by imitation[19,20,21] It is unclear whether these behaviours for infants and robots fulfilled the three characteristics of scaffolding because it might be a unidirectional support based on teachers’ a priori information about the learners (age or appearance). Kostrubiec and colleagues have recently reported empirical evidence of bidirectional information flow between a computer-driven virtual teacher and human learner using a sophisticated experimental setup[22] It remains elusive how a human teacher interacts with a learner to promote her/his learning because the virtual teacher adjusts its behaviour in a designated manner. The noise covariance of this model was evaluated to detect the simultaneous component of their movement that was not explained by the bidirectional causality but identified as commonly shared driving information

Methods

Results

Conclusion