Effects of Split-Attention and Task Complexity on Individual and Collaborative Learning

Abstract

School tasks often include individual and collaborative activities supported by a wide variety of learning materials. These materials can elicit varied levels of attention and learning depending on the complexity (i.e., element interactivity level) and physical separation of the information elements in the study material. The aim of this study was to explore the potential effects of the element interactivity level (i.e., high vs. low) and split attention (i.e., integrated vs. separated information) on individual and collaborative learning. An experimental design was implemented with 192 high school learners, with 64 working individually and 128 in dyads. The results revealed that in tasks with high element interactivity and integrated information, individual students learned more than groups. However, separated information benefited groups more than individual learners. It is concluded that the benefits of individual and group learning are mediated by task element interactivity and the physical separation of information sources in the study material, and recommendations for education professionals are presented.

1. Introduction

Learning and problem-solving in groups have come to be understood as key skills of the 21st century, both for education and for the global economy [1]. In an educational context, learning to solve problems may include both individual and collaborative tasks supported by a wide variety of materials and domains [2,3]. Consequently, more research is needed to guide decision-making regarding which instructional combination is most effective.

From the perspective of cognitive load theory [4], the effectiveness of individual or group learning depends on the levels of cognitive load based on the number of information elements and their interrelation [5,6]. Likewise, learning materials can elicit varied levels of cognitive load depending on the degree of physical separation between information segments, i.e., split attention [7]. However, little is known about the potential effects of social learning conditions and split attention on problem-solving learning. This article reports on a field experimental study that examined the level of element interactivity (i.e., high versus low) and split attention (i.e., integrated versus separated information) and discusses the findings.

1.1. Cognitive Load and Element Interactivity

Cognitive load theory is an instructional approach based on the characteristics of human cognitive architecture [4,8]. The fundamental components of this architecture are working memory, long-term memory, and the interaction between the two types of memory. When learning new information, working memory is severely limited because it can only process a few pieces of information simultaneously, while long-term memory has unlimited storage capacity [9,10]. As the number of new information elements increases, working memory experiences a higher processing load [11,12]. The limitations of working memory disappear when specific information elements are stored in long-term memory [13,14].

Cognitive load refers to the intensity of cognitive activities and information elements that need to be processed in working memory over a period of time (i.e., about 20 s) [4,15,16,17]. The load varies depending on the number of interacting information elements of the learning task. An information element can be a concept or procedure that needs to be processed and learned [11,12]. The element interactivity level (i.e., level of complexity) depends on the number of element tasks and their interconnection, as well as the learner’s level of expertise. [11,14]. For example, a novice learning chemical elements must process pairs of elements, their letters, and their respective names (e.g., H for hydrogen). Learning 118 pairs of elements is very difficult and time-consuming, but it imposes low element interactivity because each element can be processed individually. However, learning algebraic tasks, such as 2(3 + x) = 0, solve for x, may feel overwhelming because the learner must process many interacting elements (e.g., applying basic mathematical operations, breaking parentheses, and rearranging coefficients and constants, among others). Solving this problem requires processing 16 symbols (i.e., 6 + 2x = 0, 6 + 2x = 0, x = −3) by applying mathematical operations (i.e., distributing the 2 and multiplying it, subtracting 6, dividing by 2) while considering that changing one element affects other elements of the task. In contrast, an expert learner who has accumulated a lot of knowledge about solving problems like these would see this problem as a single element (i.e., low element interactivity level) that requires one or two solution steps (e.g., multiply both terms by 2 and change the sign of 3).

There are two categories of cognitive load that are related to the effectiveness of teaching [12]. Intrinsic load refers to the processing of the information elements that must be learned. This load can be low or high (e.g., 2x = 6 or 2(3 + x) = 0, respectively). Extraneous load refers to the processing of information elements unrelated to the task, such as learning to solve a problem through generic cognitive strategies like means–end analysis or trial and error or with poorly designed instructional materials and environments, i.e., split attention [7]. This load should be minimized by manipulating the presentation of information, for example, by using worked or partially worked examples, so that learners focus their working memory resources on the information elements to be learned. Instruction may be more effective when the interacting information elements of the task are kept within the limits of working memory and elements that impose an extraneous cognitive load are minimized. If the level of interactive information elements in the task is low, reducing elements that induce extraneous cognitive load may not be necessary if it does not overload working memory.

1.2. Collaborative Cognitive Load

Collaborative learning is an instructional strategy that involves creating small groups of learners with the purpose of learning to solve academic problems together [2,6,18]. Collaboration has been studied through a variety of approaches [3,19,20], and cognitive and social mechanisms that facilitate or inhibit collaboration have been identified [21]. An emerging approach is to consider groups as information processing systems [22,23]. These systems consist of limited individual working memories that create an extended workspace to process the information elements of the learning task through transactional activities [5]. From this perspective, transactional activities, defined as the elements of communication and interindividual communication, are an information processing mechanism whose cognitive load may either promote or hinder learning [8,22].

Recent research on collaborative learning from the perspective of cognitive load suggests that groups are more efficient and effective when the interactivity of information elements is high [5,6,24,25,26,27]. In other words, group members may acquire more knowledge because the cognitive load imposed by the interactive elements of task information and transactional activities (i.e., communication and interindividual coordination) is shared among group members. However, this extended capacity and the collaborative cognitive load associated with transactional activities would be redundant when the level of element interactivity is low. In these conditions, it is better to learn individually [25,28].

1.3. Split-Attention Effect

A common source of extraneous cognitive load is split attention, which occurs when learners must integrate multiple relevant sources of information that are spatially separated [7,29,30]. Shifting attention from one information element to another requires holding previously learned information in working memory while searching for and processing information from another source [7]. Searching for and integrating multiple sources of physically separated information consume working memory resources that may be used to understand and store the information in long-term memory.

Accordingly, for learners to comprehend and learn information, it should be presented in an integrated, spatial, and temporal manner [7,29]. However, it is worth noting that the split-attention effect has been mostly investigated under individual learning conditions. Nevertheless, the effect of the collective capacity of working memory [5,31] may allow groups to optimally manage separate sources of information. In other words, group members may distribute different information elements among themselves and integrate them through appropriate transactional activities [18,32].

1.4. The Present Study

This study examined how the level of element interactivity (i.e., low (partially worked examples) versus high (conventional problems)) and the type of attention induced by the material (i.e., integrated or separated information) influence individual and group learning. It was expected that the extraneous cognitive load associated with split attention in low-element-interactivity tasks would not be relevant because the cognitive activities of integrating sources of information do not overload the working memory capacity [11,12]. However, low element interactivity is unfavorable for groups because the benefits of the extended capacity of collective working memory are not greater than the cognitive costs induced by transactional activities [5,26,28]. Consequently, in tasks with low information element interactivity, we expected that individual learners would outperform collaborative groups using materials with integrated (h1.2) or separated information (h1.2).

Unlike low-element-interactivity tasks, tasks with high element interactivity demand more cognitive resources, which imposes a high intrinsic cognitive load due to the limited capacity of working memory [11,12]. In these conditions, integrated information material may alleviate the cognitive load of individual learners. On the other hand, groups may leverage their collective working memory capacity to address the extraneous processing demand associated with split attention and the intrinsic processing demand related to the high element interactivity of the task [5,8]. Consequently, it was expected that materials with integrated information would benefit individual learners more than groups (h2.1). In contrast, materials inducing split attention would not affect collaborative groups due to their large working memory capacity. Consequently, we expected that groups would outperform individuals in tasks with a high level of task element interactivity and with material containing separated information (h2.2).

2. Materials and Methods

2.1. Participants

The sample size was calculated using G*Power with a medium effect size of η_p² = 0.06, error = 0.05, and power = 0.80, resulting in 126 samples. An additional 50% of learners were added due to the collaborative learning condition in pairs, resulting in 192 participants. Students were Ecuadorian from a public school in Guayaquil, and the tasks were part of their mathematics class. Public schools tend to have large classes, and mathematics allows for the manipulation of element interactivity. The average age of the learners, of whom 97 were male (51%) and 95 were female (49%), was 15.50 years (SD = 0.79). Learners were randomly assigned, and institution authorities were informed of and approved their participation. All learners were novices in the learning topic.

2.2. Design and Procedure

A factorial design with 2 factors was used: element interactivity (i.e., low vs. high), material type (i.e., separated information vs. integrated information), and social condition (i.e., individual vs. group). The dependent variable was academic performance in mathematics. The study was conducted in four stages: pretest of knowledge, learning stage, immediate test stage, and delayed test stage. Each stage was structured with four 40 min sessions, equivalent to four regular class periods. A teacher provided participants with guidance on how to use study methods and treatment fidelity to ensure proper application. They were instructed to use a digital clock application to display the number of minutes assigned to each activity. Additionally, they were informed that they would receive 10 academic compensation points for participating.

The four involved stages were executed as follows: In the first stage, there was a preliminary test, where all learners in the class were individually assessed on their knowledge of complex numbers. In the second stage, learners were assigned using a randomization scheme. Four groups were formed, each with 16 learners, in the following way: (1) individual learning with low-element-interactivity tasks and material with integrated information; (2) individual learning with low-element-interactivity tasks and material with separated information; (3) individual learning with high-element-interactivity tasks and material with integrated information; and (4) individual learning with high-element-interactivity tasks and material with separated information.

Another four groups were formed, each with 32 learners, who worked in pairs as described below: (1) collaborative learning in pairs with low-element-interactivity tasks and material with integrated information; (2) collaborative learning in pairs with low-element-interactivity tasks and material with separated information; (3) collaborative learning in pairs high-element-interactivity tasks and material with integrated information; (4) collaborative learning with high-element-interactivity tasks with separated information. Instructional guidance was provided according to the type of learning for each treatment. The third stage involved an immediate test after the learning stage, which included retention and transfer tests. Finally, the fourth stage was an 8-day-delayed test. Additionally, learners were informed that they could ask the teacher if they had any doubts.

2.3. Materials

The material was based on operations with complex numbers (i.e., real and imaginary), which are part of the content of mathematics in the national curriculum. Learners received instructional guidance and techniques for adding, subtracting, multiplying, and dividing complex numbers, as well as solved examples and practical problems to solve. Materials were designed considering three key sources of information: a detailed introduction to each operation (i.e., addition, subtraction, multiplication, and division), a worked example for each operation (i.e., detailed and numbered steps accompanied by its solution), and a practice problem for each operation. The level of interactivity of each element of the tasks was manipulated in the practice problem. A task with low element interactivity consisted of completing a partially worked example. For instance, Step 1 of the task below requires regrouping two pairs of information elements. But, the partially solved Step 1 only requires completing each group. Step 2 requires identifying, calculating, and representing imaginary numbers. However, the partially solved Step 2 indicates the imaginary number to be calculated and represented. These instructional procedures decreased the element interactivity level and its respective cognitive load.

Task 1: Complete the processes and solve the following sum with complex numbers.

$\left({\displaystyle \frac{2}{5}}\sqrt[ 3]{7}+\sqrt{-25}\right)+\left({\displaystyle \frac{3}{2}}\sqrt[ 3]{7}+\sqrt{-36}\right)=$

Step 1: Group the real part and the imaginary part.

$\left({\displaystyle \frac{2}{5}}\sqrt[3]{7}+ \right)+\left( +\sqrt{-36}\right)=$

Step 2: Represent the roots of the imaginary numbers using the imaginary unit.

$\begin{array}{c}\sqrt{-25}=\\ \sqrt{-36}=\end{array}$

Step 3: Add the real grouping and the imaginary grouping.

$\left( \sqrt[3]{7}+ \sqrt[ 3]{7}\right)+\left( +6i \right)=$

A task with high element interactivity consisted of solving a conventional problem. Conventional problems only included the step description, requiring learners to solve each step. An example of a conventional problem is as follows:

Task 1: Complete the processes and solve the following sum with complex numbers.

(1) $\left({\displaystyle \frac{2}{5}}\sqrt[ 3]{7}+\sqrt{-25}\right)+\left({\displaystyle \frac{3}{2}}\sqrt[ 3]{7}+\sqrt{-36}\right)=$

Step 1: Group the real part and the imaginary part.

Step 2: Represent the roots of the imaginary numbers using the imaginary unit.

Step 3: Add the real grouping and the imaginary grouping.

The integrated information material for the tasks with low element interactivity consisted of numbered step descriptions, followed by the respective steps’ solutions, and the worked example was followed by the respective practice problem for each operation (i.e., the practice problem for multiplication after the worked example for multiplication).

Separated information for tasks with low element interactivity comprised two sections. The first section presented a worked example for each operation, and each example had two blocks: one block contained the numbered step descriptions, and the second block contained the corresponding numbered solved steps. Learners needed to pair each step description with its corresponding solution. The second section consisted of four practice problems, one per operation. Learners had to complete or solve each practice problem by integrating it with the respective worked example provided in the first section. Separating blocks and sections of information imposed a higher extraneous load compared to the integrated information condition.

For tasks with high element interactivity, the solution steps were removed. The material used for both individual and group settings was identical but varied in element interactivity and information integration according to the condition. However, groups were instructed to interact to study and solve the problems.

2.4. Performance Measure

During the pretest stage, learners were assessed with a written mathematics test comprising four tasks related to the operations of addition, subtraction, multiplication, and division of complex numbers, a new topic not yet taught, which they had to solve. In the learning stage, learners received instructional guidance containing the topic of complex numbers with four worked examples and the respective procedure related to the operations of addition, subtraction, multiplication, and division, as well as four practice problems to solve according to their condition. The academic performance of learners who worked with high element interactivity was graded out of 13 points. Participants had to solve four tasks with a total of 13 steps. Participants were awarded 1 point for each step answered correctly, 0 points if the step was not completed or the answer was incorrect, and a proportional score if the step was partially resolved. For students who worked with low element interactivity (i.e., they had to solve partially worked examples), they could earn 6.5 points: 0.5 points for each step solved correctly, 0 points if the step was not completed or the answer was incorrect, and a proportion if the step was partially resolved. In the immediate assessment, participants received a booklet with four problems to solve one day after the learning stage. In the delayed test, participants received another booklet with four problems eight days after the learning stage.

In the pretest, immediate posttest, and delayed posttest, participants received a document with learning activities containing the procedures for the operations of addition, subtraction, multiplication, and division of complex numbers with the same level of complexity as the learning tasks, along with their respective assessment criteria. These tasks were performed independently by the participants. They were also instructed to complete all practice tasks provided to them in the learning stage.

3. Results

Analyses were conducted using SPSS version 26 and a 2 (element interactivity: high vs. low) × 2 (social condition: individual vs. group) × 2 (format: separated information vs. integrated information) analysis of variance (ANOVA). The dependent variable was academic performance, measured during the learning, immediate test, and delayed test stages. A significance level of 0.05 was used for all analyses. Partial eta squared (η_p²) was used as the effect size measure, with values of 0.01, 0.06, and 0.14 corresponding to small, medium, and large effects, respectively [33]. The levels of element interactivity in the learning stage were analyzed separately.

3.1. Learning

Descriptive results for the learning stage are shown in Table 1. Regarding tasks with low element interactivity, the ANOVA showed that the main effect of material was statistically significant in favor of material with integrated information (M = 6.01, SD = 0.47, F(1, 92) = 13.55, MSE = 0.50, p < 0.01, η_p² = 0.13) compared to separated information (M = 5.45, SD = 0.90). The main effect of the social condition was also significant (F(1, 92) = 5.05, MSE = 0.50, p = 0.03, η_p² = 0.05), suggesting that group performance (M = 5.84, SD = 0.79) was better than individual performance (M = 5.50, SD = 0.67). However, the interaction between main effects was not significant (p = 1.00).

Regarding performance on tasks with high element interactivity, the ANOVA showed that the main effects of the material and social condition were not significant (p = 0.59 and p = 0.31, respectively). However, the interaction between these effects was significant (F(1, 92) = 28.96, MSE = 3.43, p < 0.01, η_p² = 0.24). Bonferroni post hoc tests suggest that material with integrated information is more beneficial for the individual condition than for the group condition (p < 0.01, η_p² = 0.18) and that separated information favors the group condition more than the individual condition (p < 0.01, η_p² = 0.09).

3.2. Immediate Test

The ANOVA (Table 2) revealed that only the main effect of element interactivity was significant, suggesting that low element interactivity (M = 9.92, SD = 0.31) resulted in superior performance compared to high element interactivity (M = 7.5, SD = 0.31). The remaining main effects were not statistically significant.

Regarding the significant interactions between main effects (Table 2), Bonferroni post hoc tests for the interaction between element interactivity and social condition showed that for tasks with low element interactivity, learning individually (p < 0.01, η_p² = 0.20, M = 12.05, SD = 0.51) was more effective than learning in groups (M = 7.80, SD = 0.36). However, for tasks with high element interactivity, learning in groups (p < 0.01, η_p² = 0.08, M = 8.78, SD = 0.36) was better than learning individually (M = 6.22, SD = 0.51).

Bonferroni tests for the interaction between the material and social condition indicated that material with integrated information benefits individual learners more (p < 0.01, η_p² = 0.08, M = 9.83, SD = 0.51) than groups (M = 7.41, SD = 0.36). However, material with separated information produced similar results for both individual and group learning (p = 2.4).

The interaction between element interactivity, social condition, and material was significant. This suggests that tasks with low element interactivity, whether with integrated information (p < 0.01, η_p² = 0.11) or separated information (p < 0.01, η_p² = 0.11), benefit individual learners more than groups (Table 3). Tasks with high element interactivity and integrated information benefited both individual and group learners equally (p = 0.45). However, separated information proved to be better for group learners (p < 0.01, η_p² = 0.19).

3.3. Delayed Test

The ANOVA (Table 4) showed that the main effect of the social condition was significant and suggests that individual learning (M = 7.87, SD = 0.41) resulted in better performance than group learning (M = 6.06, SD = 0.29). The remaining main effects were not statistically significant.

Regarding statistically significant main-effect interactions (Table 4), Bonferroni tests for the interaction between element interactivity and material indicated that for low element interactivity, integrated information (M = 7.91, SD = 0.51) is more effective than separate information (p < 0.01, η_p² = 0.05, M = 5.70, SD = 0.51). However, for high element interactivity, both types of material are equally effective (p = 0.52).

Bonferroni tests on the interaction between the material and social condition indicated that integrated information material benefits individual learners more (p < 0.01, η_p² = 0.13, M = 9.27, SD = 0.58) than those learning in groups (M = 5.53, SD = 0.41). However, both learning conditions equally achieved similar results with separated information material (p = 0.87).

4. Discussion

Developing problem-solving skills is gaining popularity among those advocating for the existence of key knowledge for contemporary society [34,35]. Meanwhile, in classrooms, instructional decisions for problem-solving learning often combine collaborative and individual learning activities with the support of a wide variety of instructional materials. In this context, the decision between individual and group learning to promote the acquisition of problem-solving skills will be more accurate and effective if the characteristics of human cognition, the task, and the learning material are considered [36]. Consequently, the purpose of this study was to examine how the level of interactivity of task elements (i.e., low (partially worked examples) versus high (problem-solving)) and split attention (i.e., material with integrated and separated information) interact in individual and collaborative learning.

Our hypotheses regarding tasks with low element interactivity, such as partially worked examples, assumed that these induce low cognitive load, leaving space available to store task information in long-term memory and cope with other sources of cognitive load. Therefore, we expected that individual learners would exhibit superior performance compared to groups, whether they studied materials that induced split attention (h1.1) or those that did not (h1.2). Immediate and delayed tests provide evidence supporting both hypotheses. These results are consistent with previous studies on the split-attention effect [37,38,39] and low-element-interactivity tasks (i.e., worked examples) [26,28]. This finding suggests that the attentional resources invested during the individual learning stage may have been sufficient to cope with the intrinsic cognitive demand of the learning task and the extraneous load associated with separated information [7]. The low element interactivity induced by partially solved tasks probably induced lower activity in working memory or cognitive load. This perhaps allowed individual learners to integrate the separated information from the material without exceeding the total capacity of working memory.

On the other hand, collaborative groups had to invest additional working memory resources in transactional activities [5,32]. Activities such as sharing problem information, performing shared calculations, or discussing the result of a task step [32,40] likely further increased working memory activity. Prior research about resource interdependence suggests that learners can combine their resources to achieve their goals [41,42,43]. Group members may have effectively shared their working memory resources to meet the cognitive demand of both the interactions and mental operations involved in integrating separate sources of information. Regarding the latter, it is plausible that group members may have coped with the cognitive costs of split attention by sharing elements of the material that were physically separated. In other words, by sharing task information, the negative effect of split attention led to increased collaborative interactions. However, transactional activities may have consumed a significant amount of attentional resources to the point that they interfered with the storage of information in long-term memory [25]. Apparently, as previous research suggests [5,24,26,28], the advantages of being able to distribute the cognitive load of a task among group members and having a larger reservoir of collective memory resources are not justified in tasks with low element interactivity. Transactional activities in undemanding learning tasks impose cognitive costs that seem to inhibit information acquisition [32].

The lack of evidence for these hypotheses during the learning phase might be explained by factors associated with the low element interactivity. Although individual learners performed better than groups in the subsequent tests, their similar performance to groups during the learning tasks may be due to their better information acquisition [6] being equivalent to the group performance associated with the extended working memory capacity [5]. It appears that the groups leveraged their interactions to complete the tasks, although this did not translate into better retention. The greater knowledge acquired by individual learners and the higher cognitive capacity of the groups may explain why the split-attention material did not affect performance during the learning phase, as we predicted in our hypothesis.

Regarding our hypotheses about the effects of tasks with a high level of element interactivity (i.e., problem-solving), we expected them to consume substantial mental resources to the point of interfering with the storage of task information in long-term memory and the handling of extraneous cognitive load. Therefore, it was hypothesized that tasks with high element interactivity with materials containing integrated information would benefit individual learners more than groups because groups have extra-high cognitive costs associated with transactional activities (h2.1). Only the results of the learning stage provided evidence for this hypothesis. Apparently, during the learning process, materials with integrated information allowed individual learners to invest more working memory resources in acquiring information compared to groups. Problem-solving required a series of steps following a worked example. The worked example with integrated information may have reduced the number of working memory activities, leaving more mental resources to solve the learning tasks [44,45]. However, the quality of the acquired mental representations was not enough to solve problems on subsequent tests. It is plausible that, although the worked example with integrated information may have reduced the cognitive load during learning, solving each of the steps following the example was still cognitively demanding. As a result, working memory resources were focused on problem-solving, leaving few attentional resources available to create lasting learning [46].

Unlike individual learners, collaborative groups had to invest more working memory resources in the learning stage due to transactional activities. These activities may include distributing the elements of the task or discussing how to complete partial solutions, imposing additional information processing on working memory [5,32]. Although groups may have been able to handle the collaborative load due to their extended working memory capacity, previous research [32] provides some evidence suggesting that some interactions may have interfered with information retention processes (e.g., task-unrelated conversations), leading to a lower performance. Nonetheless, group work resulted in learning equivalent to that of individual learners in the subsequent test due to operational transactive discussions, such as sharing information elements and calculations or explaining solutions [47,48]. These interactions may have resulted in enduring task elaborations similar to those of individual learners.

Regarding tasks with high element interactivity with materials that induce split attention, according to the collective working memory effect [5,31], it was estimated that collaborative groups would leverage their large memory resources to meet the demands of task information, as well as transactional activities and distributed information. Consequently, we expected that groups would perform better than individuals (h2.2). The findings from both the learning stage and the immediate test support this hypothesis. Apparently, groups more effectively managed the attentional load associated with the task, interactions, and material through a distributive advantage and collective working memory [23]. Because of this, the cognitive load associated with split attention was due to the physical separation not only between the step descriptions but also between the worked example and the practice problem. It is plausible that group members might have decided to process the high cognitive demand of separated information by productive coordination interactions (i.e., mutual cognitive interdependence principle) [8,32,49]. Previous studies suggest that group members implicitly or explicitly agree to process separated sources of information and integrate them with productive interactions such as verbally shared calculations [32,40,42]. Furthermore, it is reasonable to suggest that groups agree to process separated information by having one member review the information from the material at hand while another integrates that information with what they have in another section of the material. In this way, the larger working memory capacity, along with more productive transactional activities, likely induced the creation of durable mental representations in long-term memory.

5. Conclusions

In the educational context, learning to solve problems can encompass both individual and collaborative tasks across a wide variety of materials and domains. This study provides a contribution to the current evidence indicating that the outcomes of individual and group learning are mediated by the characteristics of the task and the learning material. Tasks with a low level of element interactivity (e.g., partially worked examples), presented either integrated or spatially separated (inducing or not inducing split attention, respectively), result in similar performance, whether learned individually or in groups. However, tasks with a high level of element interactivity (i.e., conventional problems), whose information is integrated, seem to benefit individual learning processes, but the test results are similar to those of collaborative groups. Nevertheless, if the information in the material is physically separated, inducing split attention, groups learn more due to their greater collective working memory capacity. These findings were observed only in immediate tests; in tests delayed by 8 days, the performance was similar.

This study has limitations and implications for future research. One limitation was that we did not conduct an analysis to explore the transactional activities that may explain the results of the subsequent tests. Future studies should confirm our findings and include analyses, for example, of the types of interactions deployed to process material with separated information. Split attention is likely to evoke more regulatory management interactions [32,50], as is the case with individual learning [51].

Our study was conducted with pairs of learners in order to keep interactions focused on the task [52]. However, in classrooms, there may be groups with varying numbers of members. A future line of research may estimate the limits of the distributive advantage and collective working memory of groups in high-element-interactivity tasks considering the number of members per group [8]. An increase in the number of members would increase both the group’s processing capacity and interactions, but this may reduce responsibility and effort in learning the task [53]. Further, it is surprising that the evidence from the 8-day test did not support our hypotheses. Nevertheless, this opens up an opportunity for future research to explore the cognitive load factors associated with group interactions that can enhance long-term retention, such as collaborative retrieval practice [54] and prior knowledge [6]. Additionally, further studies will be valuable in confirming our hypotheses.

Finally, the results of this study have implications for education professionals. Collaborative learning is a strategy that can be used to foster the acquisition of problem-solving skills in several curriculum areas, such as mathematics, computer science, or writing. However, grouping students to work together and solve problems requires a significant amount of time, effort, and resources and does not always result in learning gains compared to individual learning. This study suggests that acquiring domain-specific problem-solving skills should carefully consider the complexity of the task and the spatial distribution of the information in the study material. Individual or collaborative learning can be effective when the task information is not very complex, regardless of whether it induces split attention or not. However, it is important to consider the spatial distribution of the information when tasks are highly complex. If the material requires integrating many sources of physically separated information, it is more effective to promote problem-solving learning with collaborative groups.

Author Contributions

Conceptualization, J.Z.R.; methodology, J.Z.R. and J.G.; formal analysis, J.Z.R.; writing—original draft preparation, J.Z.R. and J.G.; writing—review and editing, J.Z.R. and J.G.; supervision, J.Z.R. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

This study was conducted in accordance with the Declaration of Helsinki and approved by the Universidad Del Pacífico (form code GP-FR-T-01 and date of approval 11 March 2023).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

Data are available on request from the authors.

Conflicts of Interest

The authors declare no conflicts of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Tables

Table 1

Means and standard deviations for performance in the learning stage.

	Integrated Information		Separated Information
	M	SD	M	SD
Low element interactivity
Individual	5.78	0.55	5.22	0.68
Group	6.12	0.38	5.56	0.98
High element interactivity
Individual	12.00	1.21	9.63	2.80
Group	9.44	1.72	11.38	1.64

Table 2

Two-way ANOVA for immediate test.

Source	MS	F (1184)	p	ηp2
Element interactivity (E)	250.26	29.77	<0.01	0.14
Material (M)	1.50	0.18	0.67	0.01
Social condition (S)	30.38	3.61	0.06	0.02
E × M	0.01	0.01	0.97	<0.01
E × S	495.04	58.88	<0.01	0.24
M × S	106.26	12.64	<0.01	0.06
E × M × S	117.04	13.92	<0.01	0.07
Error	8.41

Table 3

Means and standard deviations for immediate test performance.

	Integrated Information		Separated Information
	M	SD	M	SD
Low element interactivity
Individual	11.91	1.13	12.19	1.57
Group	7.73	2.55	7.86	3.93
High element interactivity
Individual	7.75	4.41	4.69	3.98
Group	7.08	2.64	10.48	1.43

Table 4

Two-way ANOVA for delayed test.

Source	MS	F (1184)	p	ηp2
Element interactivity (E)	4.27	0.39	0.53	<0.01
Material (M)	32.38	2.97	0.09	0.02
Social condition (S)	139.56	12.78	<0.01	0.07
E × M	75.70	6.94	0.01	0.04
E × S	9.85	0.90	0.34	<0.01
M × S	158.24	14.50	<0.01	0.07
E × M × S	0.71	0.07	0.80	<0.01
Error	10.92

References

1. Graesser, A.C.; Fiore, S.M.; Greiff, S.; Andrews-Todd, J.; Foltz, P.W.; Hesse, F.W. Advancing the science of collaborative problem solving. Psychol. Sci. Public Interest; 2018; 19, pp. 59-92. [DOI: https://dx.doi.org/10.1177/1529100618808244] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30497346]

2. Slavin, R.E. Cooperative learning in elementary schools. Education 3–13; 2014; 43, pp. 5-14. [DOI: https://dx.doi.org/10.1080/03004279.2015.963370]

3. Hmelo-Silver, C.; Chinn, C.A. Collaborative learning. Handbook of Educational Psychology; Corno, L.; Anderman, E.M. Routledge: New York, NY, USA, 2015; pp. 349-363.

4. Sweller, J.; Ayres, P.; Kalyuga, S. Cognitive Load Theory; Springer: New York, NY, USA, 2011; [DOI: https://dx.doi.org/10.1007/978-1-4419-8126-4]

5. Kirschner, F.; Paas, F.; Kirschner, P.A. Task complexity as a driver for collaborative learning efficiency: The collective working-memory effect. Appl. Cogn. Psychol.; 2011; 25, pp. 615-624. [DOI: https://dx.doi.org/10.1002/acp.1730]

6. Zambrano, J.; Kirschner, F.; Sweller, J.; Kirschner, P.A. Effects of prior knowledge on collaborative and individual learning. Learn. Instr.; 2019; 63, 101214. [DOI: https://dx.doi.org/10.1016/j.learninstruc.2019.05.011]

7. Ayres, P.; Sweller, J. The split-attention principle in multimedia learning. The Cambridge Handbook of Multimedia Learning; Mayer, R.E.; Fiorella, L. Cambridge University Press: New York, NY, USA, 2022; pp. 199-211. [DOI: https://dx.doi.org/10.1017/9781108894333.020]

8. Kirschner, P.A.; Sweller, J.; Kirschner, F.; Zambrano, J. From cognitive load theory to collaborative cognitive load theory. Int. J. Comput.-Support. Collab. Learn; 2018; 13, pp. 213-233. [DOI: https://dx.doi.org/10.1007/s11412-018-9277-y]

9. Baddeley, A.D. Working memory: Theories, models, and controversies. Annu. Rev. Psychol.; 2011; 63, pp. 1-29. [DOI: https://dx.doi.org/10.1146/annurev-psych-120710-100422]

10. Gilchrist, A.L.; Cowan, N. Can the focus of attention accommodate multiple, separate items?. J. Exp. Psychol. Learn. Mem. Cogn.; 2011; 37, 1484. [DOI: https://dx.doi.org/10.1037/a0024352]

11. Chen, O.; Paas, F.; Sweller, J. A cognitive load theory approach to defining and measuring task complexity through element interactivity. Educ. Psychol. Rev.; 2023; 35, 63. [DOI: https://dx.doi.org/10.1007/s10648-023-09782-w]

12. Sweller, J. Element interactivity and intrinsic, extraneous, and germane cognitive load. Educ. Psychol. Rev.; 2010; 22, pp. 123-138. [DOI: https://dx.doi.org/10.1007/s10648-010-9128-5]

13. Chen, O.; Kalyuga, S.; Sweller, J. The expertise reversal effect is a variant of the more general element interactivity effect. Educ. Psychol. Rev.; 2016; 29, pp. 393-405. [DOI: https://dx.doi.org/10.1007/s10648-016-9359-1]

14. Tricot, A.; Sweller, J. Domain-specific knowledge and why teaching generic skills does not work. Educ. Psychol. Rev.; 2014; 26, pp. 265-283. [DOI: https://dx.doi.org/10.1007/s10648-013-9243-1]

15. Kalyuga, S. Knowledge elaboration: A cognitive load perspective. Learn. Instr.; 2009; 19, pp. 402-410. [DOI: https://dx.doi.org/10.1016/j.learninstruc.2009.02.003]

16. Sweller, J.; Van Merriënboer, J.J.G.; Paas, F. Cognitive architecture and instructional design: 20 years later. Educ. Psychol. Rev.; 2019; 31, pp. 261-292. [DOI: https://dx.doi.org/10.1007/s10648-019-09465-5]

17. Peterson, L.R.; Peterson, M.J. Short-term retention of individual verbal items. J. Exp. Psychol.; 1959; 58, pp. 193-198. [DOI: https://dx.doi.org/10.1037/h0049234] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/14432252]

18. Webb, N.M. Task-related verbal interaction and mathematics learning in small groups. J. Res. Math. Educ.; 1991; 22, pp. 366-389. [DOI: https://dx.doi.org/10.2307/749186]

19. Webb, N.M.; Burnheimer, E. Cooperative and collaborative learning. International Encyclopedia of Education; Elsevier: Amsterdam, The Netherlands, 2023; pp. 593-599. [DOI: https://dx.doi.org/10.1016/b978-0-12-818630-5.14070-9]

20. Crook, C. CSsCL: The performance of collaborative learning. Int. J. Comput.-Support. Collab. Learn; 2022; 17, pp. 169-183. [DOI: https://dx.doi.org/10.1007/s11412-022-09364-y]

21. Nokes-Malach, T.; Richey, J.E.; Gadgil, S. When is it better to learn together? Insights from research on collaborative learning. Educ. Psychol. Rev.; 2015; 27, pp. 645-656. [DOI: https://dx.doi.org/10.1007/s10648-015-9312-8]

22. Tindale, R.S.; Kameda, T. ‘Social sharedness’ as a unifying theme for information processing in groups. Group Process. Intergroup Relat.; 2000; 3, pp. 123-140. [DOI: https://dx.doi.org/10.1177/1368430200003002002]

23. Hinsz, V.B.; Tindale, R.S.; Vollrath, D.A. The emerging conceptualization of groups as information processors. Psychol. Bull.; 1997; 121, pp. 43-64. [DOI: https://dx.doi.org/10.1037/0033-2909.121.1.43]

24. Zambrano, J.; Kirschner, F.; Sweller, J.; Kirschner, P.A. Effects of group experience and information distribution on collaborative learning. Instr. Sci.; 2019; 47, pp. 531-550. [DOI: https://dx.doi.org/10.1007/s11251-019-09495-0]

25. Kirschner, F.; Paas, F.; Kirschner, P.A.; Janssen, J. Differential effects of problem-solving demands on individual and collaborative learning outcomes. Learn. Instr.; 2011; 21, pp. 587-599. [DOI: https://dx.doi.org/10.1016/j.learninstruc.2011.01.001]

26. Retnowati, E.; Ayres, P.; Sweller, J. Can collaborative learning improve the effectiveness of worked examples in learning mathematics?. J. Educ. Psychol.; 2016; 109, pp. 666-679. [DOI: https://dx.doi.org/10.1037/edu0000167]

27. Zambrano, J.; Kirschner, F.; Kirschner, P.A. The effect of the prior collaborative experience on the effectiveness and efficiency of collaborative learning. Rethinking Learning in the Digital Age: Making the Learning Sciences Count, Proceedings of the 13th International Conference of the Learning Sciences (ICLS), London, UK, 23–27 June 2018; Kay, J.; Luckin, R. International Society of the Learning Sciences: London, UK, 2018; pp. 112-119.

28. Retnowati, E.; Ayres, P.; Sweller, J. Worked example effects in individual and group work settings. Educ. Psychol.; 2010; 30, pp. 349-367. [DOI: https://dx.doi.org/10.1080/01443411003659960]

29. Sweller, J.; Ayres, P.; Kalyuga, S. The split-attention effect. Cognitive Load Theory; Sweller, J.; Ayres, P.; Kalyuga, S. Springer: London, UK, 2011; pp. 111-128. [DOI: https://dx.doi.org/10.1007/978-1-4419-8126-4_9]

30. De Koning, B.B.; Rop, G.; Paas, F. Effects of spatial distance on the effectiveness of mental and physical integration strategies in learning from split-attention examples. Comput. Hum. Behav.; 2020; 110, 106379. [DOI: https://dx.doi.org/10.1016/j.chb.2020.106379]

31. Kirschner, F.; Paas, F.; Kirschner, P.A. A cognitive load approach to collaborative learning: United brains for complex tasks. Educ. Psychol. Rev.; 2009; 21, pp. 31-42. [DOI: https://dx.doi.org/10.1007/s10648-008-9095-2]

32. Zambrano, J.; Kirschner, F.; Sweller, J.; Kirschner, P.A. Effect of task-based group experience on collaborative learning: Exploring the transaction activities. Br. J. Educ. Psychol.; 2023; 93, pp. 879-902. [DOI: https://dx.doi.org/10.1111/bjep.12603]

33. Cohen, J. Statistical Power Analysis for the Behavioral Sciences; 2nd ed. L. Erlbaum Associates: Hillsdale, NJ, USA, 1988; p. xxi, 567p

34. Han, A.; Krieger, F.; Greiff, S. Assessment of collaborative problem-solving: Past achievements and current challenges. International Encyclopedia of Education; Elsevier: Amsterdam, The Netherlands, 2023; pp. 234-244. [DOI: https://dx.doi.org/10.1016/b978-0-12-818630-5.09041-2]

35. Karaca-Atik, A.; Meeuwisse, M.; Gorgievski, M.; Smeets, G. Uncovering important 21st-century skills for sustainable career development of social sciences graduates: A systematic review. Educ. Res. Rev.; 2023; 39, 100528. [DOI: https://dx.doi.org/10.1016/j.edurev.2023.100528]

36. Zhang, L.; Kirschner, P.A.; Cobern, W.W.; Sweller, J. There Is an Evidence Crisis in Science Educational Policy. Educ. Psychol. Rev.; 2022; 34, pp. 1157-1176. [DOI: https://dx.doi.org/10.1007/s10648-021-09646-1]

37. Schroeder, N.L.; Cenkci, A.T. Spatial contiguity and spatial split-attention effects in multimedia learning environments: A meta-analysis. Educ. Psychol. Rev.; 2018; 30, pp. 679-701. [DOI: https://dx.doi.org/10.1007/s10648-018-9435-9]

38. Noetel, M.; Griffith, S.; Delaney, O.; Harris, N.R.; Sanders, T.; Parker, P.; del Pozo Cruz, B.; Lonsdale, C. Multimedia design for learning: An overview of reviews with meta-meta-analysis. Rev. Educ. Res.; 2021; 90, pp. 6-23. [DOI: https://dx.doi.org/10.3102/00346543211052329]

39. Chandler, P.; Sweller, J. Cognitive load while learning to use a computer program. Appl. Cogn. Psychol.; 1996; 10, pp. 151-170. [DOI: https://dx.doi.org/10.1002/(SICI)1099-0720(199604)10:2<151::AID-ACP380>3.0.CO;2-U]

40. Webb, N.M.; Troper, J.D.; Fall, R. Constructive activity and learning in collaborative small groups. J. Educ. Psychol.; 1995; 87, 406. [DOI: https://dx.doi.org/10.1037/0022-0663.87.3.406]

41. Johnson, D.W. Social Interdependence: Interrelationships among Theory, Research, and Practice. Am. Psychol.; 2003; 58, pp. 934-945. [DOI: https://dx.doi.org/10.1037/0003-066X.58.11.934] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/14609388]

42. Wong, R.M.; Adesope, O.O. Promoting collaboration through resource interdependence to enhance collaborative concept mapping. Soc. Psychol. Educ.; 2023; 27, pp. 89-113. [DOI: https://dx.doi.org/10.1007/s11218-023-09830-x]

43. Butera, F.; Buchs, C. Social interdependence and the promotion of cooperative learning. Social Psychology in Action; Sassenberg, K.; Vliek, M.L.W. Springer International Publishing: Cham, Switzerland, 2019; pp. 111-127. [DOI: https://dx.doi.org/10.1007/978-3-030-13788-5_8]

44. Renkl, A.; Atkinson, R.K.; Plass, J.L.; Moreno, R.; Brunken, R. Learning from worked-out examples and problem solving. Cognitive Load Theory; Cambridge University Press: Cambridge, UK, 2010; pp. 91-108. [DOI: https://dx.doi.org/10.1017/cbo9780511844744.007]

45. Chen, O.; Retnowati, E.; Chan, B.K.Y.; Kalyuga, S. The effect of worked examples on learning solution steps and knowledge transfer. Educ. Psychol.; 2023; 43, pp. 914-928. [DOI: https://dx.doi.org/10.1080/01443410.2023.2273762]

46. Soderstrom, N.C.; Bjork, R.A. Learning versus performance: An integrative review. Perspect. Psychol. Sci.; 2015; 10, pp. 176-199. [DOI: https://dx.doi.org/10.1177/1745691615569000]

47. Webb, N.M. Student interaction and learning in small groups. Learning to Cooperate, Cooperating to Learn; Slavin, R.E.; Sharan, S.; Kagan, S.; Hertz-Lazarowitz, R.; Webb, C.; Schmuck, R. Plenum Press: New York, NY, USA, 1985; pp. 147-172. [DOI: https://dx.doi.org/10.1007/978-1-4899-3650-9_6]

48. Jurkowski, S.; Mundelsee, L.; Hänze, M. Strengthening collaborative learning in secondary school: Development and evaluation of a lesson-integrated training approach for transactive communication. Learn. Instr.; 2024; 92, 101934. [DOI: https://dx.doi.org/10.1016/j.learninstruc.2024.101934]

49. Janssen, J.; Kirschner, P.A. Applying collaborative cognitive load theory to computer-supported collaborative learning: Towards a research agenda. Educ. Technol. Res. Dev.; 2020; 68, pp. 783-805. [DOI: https://dx.doi.org/10.1007/s11423-019-09729-5]

50. Malmberg, J.; Järvelä, S.; Järvenoja, H. Capturing temporal and sequential patterns of self-, co-, and socially shared regulation in the context of collaborative learning. Contemp. Educ. Psychol.; 2017; 49, pp. 160-174. [DOI: https://dx.doi.org/10.1016/j.cedpsych.2017.01.009]

51. Eitel, A.; Endres, T.; Renkl, A. Self-management as a bridge between cognitive load and self-regulated learning: The illustrative case of seductive details. Educ. Psychol. Rev.; 2020; 32, pp. 1073-1087. [DOI: https://dx.doi.org/10.1007/s10648-020-09559-5]

52. Zhan, Z.; He, G.; Li, T.; He, L.; Xiang, S. Effect of groups size on students’ learning achievement, motivation, cognitive load, collaborative problem-solving quality, and in-class interaction in an introductory AI course. J. Comput. Assist. Learn.; 2022; 38, pp. 1807-1818. [DOI: https://dx.doi.org/10.1111/jcal.12722]

53. Latané, B.; Williams, K.; Harkins, S. Many hands make light the work: The causes and consequences of social loafing. J. Personal. Soc. Psychol.; 1979; 37, pp. 822-832. [DOI: https://dx.doi.org/10.1037/0022-3514.37.6.822]

54. Wissman, K.T.; Rawson, K.A. Why does collaborative retrieval improve memory? Enhanced relational and item-specific processing. J. Mem. Lang.; 2015; 84, pp. 75-87. [DOI: https://dx.doi.org/10.1016/j.jml.2015.05.003]