An Exploration of Internal States in Collaborative Problem Solving

Sifatul Anindho, Videep Venkatesha, Mariah Bradford, Anne M. Cleary, Nathaniel Blanchard · 2025

Read An Exploration of Internal States in Collaborative Problem Solving by Sifatul Anindho, Videep Venkatesha, Mariah Bradford, Anne M. Cleary, Nathaniel Blanchard (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-02229.txt.

Introduction

Sifatul Anindho and four co-authors explore the inner emotional states people experience while solving problems together in teams. Their paper came out in 2025. Collaborative problem solving mixes thinking, social interaction, and emotion, and it is common in schools and workplaces.

Teams of four first completed a new group task. Right afterward, each person sat alone and watched the video of their team, narrating what they had felt at each moment. The team analyzed that language for single words, word pairs, and emotion labels. Distinct patterns emerged, including characteristic phrases for different feelings and links between related emotion words. The results give a window into how emotion moves through group work.

Catalog record

CREATORSifatul Anindho, Videep Venkatesha, Mariah Bradford, Anne M. Cleary, Nathaniel Blanchard
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE17 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-02229

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

An Exploration of Internal States in Collaborative Problem Solving Sifatul Anindho, Videep Venkatesha, Mariah Bradford, Anne M. Cleary, Nathaniel Blanchard Abstract Collaborative problem solving (CPS) is a complex cognitive, social, and emotional process that is increasingly prevalent in educational and professional settings. This study investigates the emotional states of individuals during CPS using a mixed-methods approach. Teams of four first completed a novel CPS task. Immediately after, each individual was placed in an isolated room where they reviewed the video of their group performing the task and self-reported their internal experiences throughout the task. We performed a linguistic analysis of these internal monologues, providing insights into the range of emotions individuals experience during CPS. Our analysis showed distinct patterns in language use, including characteristic unigrams and bigrams, key words and phrases, emotion labels, and semantic similarity between emotion-related words. 1 Introduction Collaborative problem solving (CPS) occurs when small teams are required to work together to accomplish a task/goal [ 7 ] . Recently, collaboration has been emphasized as an essential skill that is ubiquitous in industry settings. Several recent works have explored the feasibility of automatically parsing these team dynamics [ 5 , 30 , 40 , 39 , 25 , 41 ] . However, little work has explored individual internal states, e.g., the cognitive and emotional processes that underlie teamwork during team activities. Here, we explore exactly this, specifically focusing on individual reflections of their internal states. We hypothesized that combining team task recordings with self-reported individual reflections after the task provides insights into the internal processes that coincide with an individual’s actions during the collaborative task. This raises important research questions: What vernacular do individuals use when describing their internal monologues during CPS tasks? What kinds of internal states are the most/least prevalent during collaborative tasks? How do these internal states link to the actions employed during the learning task? We examined individuals’ internal states as they perform a novel collaborative task. After the task was complete, we collected a spoken retrospective from individuals as they watched a video of their team progressing through the task — specifically, they self-report what their internal states were moment-to-moment across the task as they watch the video. The collaborative task was a Lego building task in which each participant was given partial information about the Lego structure (see Section 3.1 ). Our analysis focuses on identifying patterns in emotion words and themes to help us better understand the emotional experiences that underlie CPS [ 13 ] . Our findings on the distribution of these states also lays the groundwork for future data collections to target distinct internal states. 2 Related Work CPS has been extensively researched across various domains, including education, psychology, and computer science. To understand the intricacies of team dynamics, researchers have employed multiple methods, such as observational studies, surveys, and physiological measures. For instance, Sun et al. [ 37 ] developed a framework for assessing CPS, while Stahl et al. [ 36 ] examined the roles of individuals in small-group collaboration. Dindar et al. [ 8 ] developed and implemented a new method for detecting shared physiological arousal events (SPAE) during CPS using skin conductance response data. VanderHoeven et al. [ 40 ] explored the importance of multimodal features to modeling CPS tasks and outlined various tool sets that can be leveraged to support each of the individual features and their integration. Research on CPS has traditionally focused on externally observable behaviors, with less emphasis on the internal states of individuals engaged in these activities. Literature has also shown results on the influence of emotional experiences during learning and problem solving. Graesser et al. [ 17 ] highlighted the importance of emotions in shaping learning outcomes, demonstrating that emotional states, such as frustration and engagement, play a crucial role in learning [ 10 ] . Tyng et al. [ 38 ] introduced a basic evolutionary approach to emotion to understand the effects of emotion on learning and memory and the functional roles played by various regions of the brain and their mutual interactions in relation to emotional processing. Self-reports have been used to assess mental states in various contexts, including learning strategies and approaches. According to Azevedo [ 1 ] , self-reports, in addition to classroom discourse, are the only proven approach that can be used to measure the cognitive, meta cognitive, affective, and motivational constructs of student engagement [ 15 ] . Beheshitha et al. [ 4 ] made use of existing self-report instruments to measure students’ goal orientations and to interpret and triangulate findings obtained through the use of trace data [ 14 ] . Self-report measures have been used to study internal experiences, but primarily with solo activities, leaving a gap in our understanding of internal states during collaborative tasks. Our study addresses this gap by exploring the internal monologues of individuals during CPS tasks. 3 Methods 3.1 Data Collection Participants This study recruited a total of 32 participants, organized into 8 groups of 4 individuals each. Participants were required to be at least 18 years old, fluent in English and affiliated with Colorado State University as students or staff members. Recruitment efforts were focused within the Computer Science department of Colorado State University, and potential participants were informed of the study through verbal communication. Table 1 presents the demographics of our participants. The age of participants ranged from 20 to 32 years. A total of 21 participants identified English as their first language, while others reported first languages including Telugu, Tajik, Hindi, Kannada and Persian. Providing demographic information was voluntary and recorded anonymously. Participants were recruited upon request, and all personal data was de-identified. Table 1: Demographics of Participants Gender Male 21 Female 10 Nonbinary 1 Native Language English 21 Telugu 7 Tajik 1 Hindi 1 Kannada 1 Persian 1 Age 18-24 19 25-31 12 32+ 1 Figure 1: Experimental setup for the collaborative Lego building exercise. To maintain the integrity of group dynamics and task outcomes, participants were allowed to participate in the study multiple times, provided that they were assigned to a different group each time, with at least 3 new members of the group [ 24 ] . This enabled researchers to collect a more comprehensive data set while minimizing potential biases arising from repeated interactions among familiar group members. Experiment Setup This study used a collaborative laboratory task paradigm [ 2 ] . The participants were organized into teams of 4 and completed a two-part situated task involving a Lego building exercise. The task setup (see Figure 1 ) consisted of several key components: • Lego blocks and a building base plate for constructing the Lego structure. • A round table team members collaborated around. • Azure Kinect Cameras that captured high-quality video and depth information of the task. • A GoPro Camera provided wide-field, high-resolution video for playback during self-report collection. • Skullcandy Hesh Evo headphones used for audio playback. The built-in microphone was used to record participant self-reports. The setup ensured that the participants’ faces and hands were clearly visible to the Azure cameras. The Azure cameras were calibrated to ensure clear visuals of the participants during video recording. Experiment Procedure Before the experiment began, participants reviewed and

Read the complete text