Task Matters: Investigating Human Questioning Behavior in Different Household Service for Learning by Asking Robots

TL;DR

The paper addresses how task structure influences human questioning in Learning by Asking (LBA) by contrasting a Goal-Oriented task (refrigerator organization) with a Process-Oriented task (cocktail mixing) through a human-human study of 28 participants. Questions are coded using a three-dimensional K-C-Q framework to reveal content, cognitive processes, and expression strategies, uncovering task-dependent patterns in both question types and sequencing. The findings show that Goal-Oriented tasks elicit early user-preference questions, while Process-Oriented tasks drive ongoing, parallel exploration of procedures and preferences, with overall questioning remaining sparse yet highly targeted. These results offer practical guidance for designing task-sensitive, LBA-enabled robots that adapt their questioning strategies to user needs and task structure, potentially improving personalization and collaboration in home environments.

Abstract

Learning by Asking (LBA) enables robots to identify knowledge gaps during task execution and acquire the missing information by asking targeted questions. However, different tasks often require different types of questions, and how to adapt questioning strategies accordingly remains underexplored. This paper investigates human questioning behavior in two representative household service tasks: a Goal-Oriented task (refrigerator organization) and a Process-Oriented task (cocktail mixing). Through a human-human study involving 28 participants, we analyze the questions asked using a structured framework that encodes each question along three dimensions: acquired knowledge, cognitive process, and question form. Our results reveal that participants adapt both question types and their temporal ordering based on task structure. Goal-Oriented tasks elicited early inquiries about user preferences, while Process-Oriented tasks led to ongoing, parallel questioning of procedural steps and preferences. These findings offer actionable insights for developing task-sensitive questioning strategies in LBA-enabled robots for more effective and personalized human-robot collaboration.

Figures (5)

• Figure 1: Illustration of Learning by Asking in two household service tasks. Robots ask different types of questions depending on the nature of the task: Goal-Oriented (e.g., organizing a refrigerator, left) vs. Process-Oriented(e.g., mixing drinks, right).
• Figure 2: Participants in the human-human study performing two representative household tasks: refrigerator organization (left) and cocktail mixing (right).
• Figure 3: 3D visualization of question type combinations in $(K, C, Q)$ space. Each point represents a unique combination used during task execution. Color encode frequency.
• Figure 4: Distributions of K-type, C-type, and Q-type questions across T1 and T2.
• Figure 5: Distribution of questioning strategies across task types.