Collaborative problem solving (CPS)
Collaborative problem solving is defined as “
the capacity of an individual to effectively engage in a process whereby two or more agents attempt to solve a problem by sharing the understanding and effort required to come to a solution and pooling their knowledge, skills, and efforts to reach that solution” (OECD,
2017, p. 26). (OECD,
2017) developed the PISA survey, which aims to investigate whether a student has acquired the key knowledge and skills for full participation in modern societies near the end of his or her compulsory education. Overall, twelve CPS skills are assessed in the PISA 2015 survey (Herborn et al.,
2020). Three social related skills are found to be important in each of the CPS phases, including (1) establishing and maintaining a shared understanding, (2) taking appropriate action to solve the problem, and (3) establishing and maintaining team organization. These skills can help the student better solve a problem collaboratively. Andrews-Todd and Forsyth (
2020) proposed a method to evaluate CPS skills based on interactive content and analyzed the performance of collaborative groups with different combinations of CPS skills. The authors found that groups with at least one student with high CPS skills showed significantly better learning performance. This is consistent with findings in areas such as cooperative learning that show that beneficial cooperative behaviors exhibited by team members contribute to team success (Barron,
2003).
To better organize a CPS activity, it is necessary to understand students’ collaboration processes in CPS learning activities. Scholars have tried to define different phases of the whole CPS process and to map the required skills in individual phases. For example, (OECD,
2017) defined the CPS process, as starting with (A) exploring and understanding and then moving on to (B) representing and formulating, (C) planning and executing, and (D) monitoring and reflecting. Hayes (
2013) categorized the problem-solving process into six phases: finding the problem, representing the problem, planning the solution, carrying out the plan, evaluating the solution, and consolidating gains. Although these definitions have minor differences, the whole process consists of two major phases: problem-understanding and solution development (Kwon et al.,
2019). The quality of a solution is strongly influenced by the problem-understanding phase (Simon & Hayes,
1976) which has been referred to as ‘‘
a cognitive structure corresponding to a problem, constructed by a solver based on his domain-related knowledge and its organization’’ (Chi, Feltovich, et al.,
1981). Then students work together to develop corresponding solutions based on the collaborative cognitive structure in the solution development phase. Therefore, group dynamics (i.e., how students interact with each other) is the critical element during the process (Chi, Glaser, et al.,
1981). Research efforts have been launched to understand the influencing factors, the quality of learning outcomes, and the collaboration patterns during these two phases. For example, Chi, Glaser, et al. (
1981) found that there are considerable differences between novices and experts in problem-solving. Novices will stick to the problem definition or problem-understanding as they work on a solution, whereas experts will move forward toward solution development. Kwon et al. (
2019) found that solution-oriented students gained more domain knowledge than problem-oriented students. The authors believe that students’ focus more on the problem-solving process rather than on the problem-understanding process is more conducive to the improvement of academic performance. Zheng et al. (
2020) coded the online collaboration behaviors of students in their study, used the Apriori algorithm to find the high-frequency jump relationship between CPS behaviors, and analyzed the collaboration patterns of students with different academic performances. Their results show that, at the problem-solving stage, the group with high scores repeatedly modified and improved the solution, while the group with low scores seldom modified the possible solution after it was proposed.
Literature shows that most CPS studies have obtained data through questionnaires or observations. CPS is rooted in the social constructivist view of learning, which asserts that in-depth learning occurs when students engage in building a shared understanding of a problem through social interactions (Jermann & Dillenbourg,
2008; Pear & Crone-Todd,
2002). More analytic results culled not merely from perceptual data, are needed to understand the details of the individual CPS phases, especially from the aspect of group members’ dynamics and the mutual effects of two phases on the quality of CPS outcomes.
Physiological synchrony (PS) and brain-to-brain synchrony (BS)
The development of emerging technologies opens new possibilities in collecting and analyzing students’ behaviors and interactions without interfering in the learning process (Chanel & Muhl,
2015). Physiological data, such as EDA, heart rate, gesture, body pose, and EEG, reflect the personal physical and/or psychological states of a person (Cukurova et al.,
2020; Sharma & Giannakos,
2020). Such data have been adapted to make up for some of the gaps in perceptional data analysis (Ashwin & Guddeti,
2020; Dikker et al.,
2017; Noroozi et al.,
2020). PS is one of the analytic approaches used to obtain insights from physiological data. Studies for years in psychophysiology indicated that human cognition cannot be separated from the body (Critchley et al.,
2013). This connection is bidirectional, many of the mental states are reflected in the body’s physiological signals (Pecchinenda,
1996). On the other hand, the physiology of the body influences human consciousness and cognition (Critchley & Garfinkel,
2018). PS refers to the interdependence of, or the associated activity between, the physiological signals of collaborating individuals. It is an unintentional and spontaneous phenomenon that can be indexed through measures of the human autonomic nervous system (Palumbo et al.,
2017). PS appears when there are the same attention objects or when there is effective interaction, and the phenomenon is that the physiological indicators rise or fall simultaneously. Studies have shown that PS can be used to measure whether the interaction is effective or whether students are focused on the same item (Stuldreher et al.,
2020a,
2020b). It was found that students who shared their reflected views also showed higher physiological synchrony (Haataja et al.,
2018). As CPS is rooted in the social constructivist view of learning, which asserts that in-depth learning occurs when students engage in building a shared understanding of a problem through social interactions (Jermann & Dillenbourg,
2008; Pear & Crone-Todd,
2002). Thus the PS in the CPS process is mainly influenced by the interaction effectivity, and the relationship between PS and learning between students and teachers is worth studying (Davidesco,
2020; Nam et al.,
2020). Dindar, Järvelä, et al. (
2020) recorded students’ EDA in CPS and analyzed the relationship between PS and metacognitive experiences. The PS was calculated through a Multidimensional Recurrence Quantification Analysis (MdRQA). The results show a positive relationship between continuous PS episodes and groups’ collective mental effort. Dindar, Malmberg, et al. (
2020) investigated the interplay of temporal changes in self-regulated learning processes (i.e., behavioral, cognitive, motivational, and emotional) and their relationship with academic achievement in computer-supported collaborative learning. The PS of the dyads in the collaborating groups was determined by calculating a single session index. The results show that PS among the collaborating students was found to be related to cognitive regulation. Sobocinski et al. (
2021) collected heart rate data and videos of students during collaboration. The authors combined video observation and PS as a possible indicator to identify monitoring and adaptation events. The studies have shown that PS is an effective indicator to reflect the process of collaborative learning.
Scalp-recorded electric potentials or electroencephalograms (EEGs) are the most popular instruments to collect a participant’s brain wave signals. The signals provide estimates of synaptic action at large scales that are closely related to behavior and cognition. Thus, EEG has been recognized as a genuine “window on the mind” (Nunez & Srinivasan,
2006). The original EEG records electric potentials and can be further divided into specific ranges through the frequency, namely the delta (1–4 Hz), theta (4–7 Hz), alpha (8–13 Hz), beta (13–30 Hz), and gamma (> 30 Hz) bands (Alarcao & Fonseca,
2019). Different wavebands of the EEG reflect different types of activity in the brain (Alarcao & Fonseca,
2019). The literature shows that the delta band is related to signal detection or the unconscious mind (Alarcao & Fonseca,
2019). The theta band is positively correlated with working memory load or cognitive load (Muthukrishnan et al.,
2020). The alpha band is related to cognitive load and mediation (Chen & Wang,
2018; Yang et al.,
2019). The beta band is related to attention and decision-making (Chen & Wang,
2018; Yang et al.,
2019). The gamma band has been demonstrated in a wide range of brain processes, including multisensory and sensorimotor integration, attention, memory formation, and perceptual binding (Chand et al.,
2016; Min et al.,
2016). The relationships between the EEG bands and brain activities are shown in Table
1.
Table 1
Relationships between EEG bands and brain activities
Delta (\(\updelta\)) | The unconscious mind, signal detection |
Theta (\(\uptheta\)) | Positively correlated with cognitive load |
Alpha (\(\mathrm{\alpha }\)) | Negatively correlated with cognitive load, related with mediation |
Beta (\(\upbeta\)) | Related with attention and decision making |
Gamma (\(\upgamma\)) | Related with multisensory and sensorimotor integration, memory formation, and perceptual binding |
BS is a type of PS. It refers to the synchronization of brain activity between two or more people (Nam et al.,
2020). Compared with PS reflected by EDA and heart rate data, BS can reflect students' cognitive states more accurately (Stuldreher et al.,
2020a,
2020b). Dikker et al. (
2017) studied the relationship between BS and the self-reported engagement of twelve students in a traditional classroom. BS was computed using the method of total interdependence (Wen et al.,
2012). The authors found that students with a higher level of BS had higher levels of engagement and social dynamics during the lecture. Bevilacqua et al. (
2019) came to a similar conclusion in their study. The authors calculated the level of BS between students and the teacher and studied the relationship between the level of BS and self-reported engagement level of twelve students in an offline lecture. The results show that students with a higher level of BS with the teacher had higher levels of perceived engagement and closeness. Davidesco et al. (
2019) studied the relationship between BS and academic performance. The authors calculated the BS between students and between students and teachers in a traditional classroom. The results show that students with high performance had higher BS with teachers and that the BS between students was more pronounced when they learned what they got wrong on the pretest and right on the posttest. Due to the limitations of devices, the sample sizes of the above studies were between twelve and thirty-six. These studies have shown that BS is an indicator that reflects academic performance and the learning process. The above studies showed that BS can provide more insights and make up some gaps in studies that rely on perceptional data only. However, the studies discussed the relationship between BS and the learning process in a traditional classroom without collaboration. More research efforts should be expended, to understand the CPS process, and to study the unique findings that can be extracted from BS. The literature shows that most CPS studies analyze the process through PS (Dindar, Järvelä, et al.,
2020; Dindar, Malmberg, et al.,
2020; Sobocinski et al.,
2021). As CPS is a process of building a shared understanding (Jermann & Dillenbourg,
2008; Pear & Crone-Todd,
2002), BS can better serve the research, since it can reflect students' cognitive state more accurately (Stuldreher et al.,
2020a,
2020b). Most of the studies have a limit on analyzing the original EEG signals (Bevilacqua et al.,
2019; Davidesco et al.,
2019; Dikker et al.,
2017). Since different wavebands of the EEG reflect different types of activity in the brain (Alarcao & Fonseca,
2019), analyzing the BS through different wavebands, rather than through the original signal, will help reveal the CPS in more detail.