The More, the Better? Exploring the Effects of Modal and Codal Redundancy on Learning and Cognitive Load: An Experimental Study

Abstract

This study explores how receiving identical information from different sources affects learning and cognitive load, focusing on two types of redundancy: modal redundancy, where redundant information comes from two visual sources (images and written text), and codal redundancy, where redundant information comes from two sources of different modalities which utilize the same symbol system (spoken and written text). Using a 2 × 2 between-subjects design involving modal (yes/no) and codal (yes/no) redundancy, 158 participants completed twenty learning tasks, consisting of ten construction and ten recall tasks. Additionally, they rated their cognitive load by indicating their perceived task difficulty and mental load. Overall, results indicate positive main effects of modal redundancy and negative effects of codal redundancy on learning and cognitive load. Furthermore, significant interaction effects suggest that modal redundancy may counterbalance the negative effects of codal redundancy, implying a compensatory mechanism in cognitive processing for construction tasks. These results highlight the importance of considering both modal and codal redundancy and their interaction in instructional design.

1. Introduction

Multimedia learning materials play an essential role in daily teaching practices, including the integration of text, images, and visualizations (e.g., images or animations) to build learning environments. This integration aligns with the principles of multimedia learning, which advocate for the incorporation of both verbal (written or spoken text) and visual components (graphics, animations, and images) within instructional materials [1]. Empirical studies have consistently underscored the advantages of multimedia learning environments, demonstrating that combining written text with visualizations enhances learning compared to text-only presentations, a phenomenon known as the multimedia principle [2]. Similarly, the modality effect [3] suggests that presenting text in spoken rather than written form, in addition to visualizations, can further enhance learning. However, while research on multimedia learning offers numerous advantages, it also introduces challenges related to cognitive load, particularly regarding the redundancy effect [4]. The redundancy effect refers to the negative impact when two or more sources provide redundant information simultaneously, potentially overloading the cognitive capacity of working memory [5]. However, empirical evidence demonstrates a diverse range of redundancy effects [6], from enhancements in learning [7] to drawbacks [8] or no effects [9]. This variability arises from the multiple facets of redundancy. One aspect of redundancy, as proposed by Cognitive Load Theory (CLT) [5], involves the contentual overlap within learning materials, exemplified by instances where written text restates information depicted in diagrams [10]. Results indicate that presenting identical information in different formats can increase cognitive load and lead to poorer learning performance [11]. Moreover, the Cognitive Theory of Multimedia Learning (CTML) [1] highlights another facet of redundancy related to separate processing channels for verbal and visual information. According to CTML, an ineffective combination of verbal and visual information can overwhelm working memory capacity, thereby hindering learning due to the effort required to process and compare multiple streams of information. For instance, if individuals learn more effectively from visualizations paired with spoken text rather than from visualizations, spoken text, and redundant written text. In such cases, the visual channel can become overloaded as learners must switch between visualizations and written text and expend additional mental effort trying to reconcile spoken and written text. [2].

While prior research has primarily focused on redundancy concerning the overlap between spoken and written text, recent studies have introduced distinctions between different types of redundancy that account for the presentation of redundant information across various working memory channels [12]. This study aims to expand on these findings by exploring the concept of working memory channel redundancy, as referred to by Trypke et al. [6]. An experimental study was conducted to analyze the impact of two types of redundancy on learning and cognitive load.

2. Why Should We Distinguish Different Types of Redundancy?

Distinguishing between different types of redundancy is essential due to its multifaceted nature, which impacts learning and cognitive load in various ways. Researchers such as Albers et al. [12], Bohec and Jamet [13], Mayer [2], and Trypke et al. [6] have highlighted the importance of understanding these distinctions.

First, considering the nature of redundant information is crucial. Learners may encounter unnecessary or irrelevant information (unrelated to the learning content) within instructional materials, leading to a higher extraneous cognitive load and impeding learning. This perspective aligns with CLT, suggesting that irrelevant content increases cognitive load and detracts from the learning process [5]. This observation also corresponds with the findings from the CTML, particularly the coherence effect [14] and the seductive details effect [15], where irrelevant information diverts attention from the essential learning content.

Additionally, the presence of identical information (the focus of this study) can have varying impacts on learning. For example, adding written text that restates information depicted in diagrams can lead to negative effects, as observed by many researchers [16,17,18]. However, some studies indicate the positive effects of providing identical information through multiple sources, as it can guide learners’ attention to relevant aspects of the material [19,20,21]. This principle is known as the signaling or cueing principle [22]. Nevertheless, these studies often do not account for the degree of contentual overlap, but recent empirical evidence [12] illustrates that even presenting identical information (100% overlap) can enhance learning outcomes and reduce cognitive load.

The second crucial aspect of redundancy is the presentation mode of information, which refers to the use of multiple modalities. Based on Albers et al. [12] and Trypke et al. [6], we distinguish between modal and codal redundancy.

2.1. Modal Redundancy

Modal redundancy involves scenarios where images and written text are presented and contain identical information. In this scenario, cognitive processing in working memory channels is influenced as images and written text “compete for limited cognitive resources in the visual channel” [23] (p. 124). This may lead to an overload of the visual channel as learners may struggle to differentiate between the images and accompanying written text or process them simultaneously.

2.2. Codal Redundancy

Codal redundancy occurs when learners receive narration (spoken text) and identical written text. In this scenario, the narration competes with the written text for cognitive processing resources, particularly impacting the verbal processing channel in working memory. This phenomenon aligns with Mayer’s concept of verbal redundancy [2,24].

Distinguishing between the nature and presentation mode of redundant information is crucial for several reasons. First, this distinction allows for a more precise understanding of how different types of redundancy affect learning and cognitive load. Prior research has often treated redundancy as a uniform concept, failing to account for the nuanced ways in which different forms of redundant information can interact with working memory channels. By considering both content and the modality, instructional designers can create more effective learning materials that optimize cognitive processing and enhance learning efficiency. Moreover, understanding these distinctions can lead to the development of instructional strategies that mitigate the negative effects of redundancy while leveraging its potential benefits. For instance, educators can design multimedia presentations that strategically use redundancy to highlight critical information without overloading working memory, as demonstrated by the verbal redundancy effect [24].

2.3. Purpose of the Study

The specific research gap this study aims to fill involves the nuanced understanding of how different types of redundancy (modal and codal) affect learning and cognitive load when the provided content is identical. While previous research has examined the general effects of redundancy on learning, there has been limited exploration into how these effects differ based on the type of redundancy and the specific cognitive processes involved. This study seeks to clarify these distinctions by systematically investigating the interplay between modal and codal redundancy and their respective impacts on learning and cognitive load while the provided content is identical among all groups. Therefore, we formulated the following research question: How do modal and codal redundancy affect learning and cognitive load when the provided content is identical?

3. Hypotheses

This experimental study examines whether different types of redundancy affect learning and cognitive load differently. To investigate the research question, we formulated the following hypotheses:

Hypothesis 1.

Presenting identical information through images accompanied by narration (no modal or codal redundancy) results in superior learning, along with reduced cognitive load compared to narration and written text (codal redundancy) or images and written text (modal redundancy). This superiority is linked to the distribution of information across different modalities, alleviating cognitive processing systems in line with the modality effect [1,3].

Hypothesis 2.

Presenting identical information through narration and written text (codal redundancy) leads to lower learning and higher cognitive load compared to the other experimental groups. This simultaneous engagement of both sensory channels requires working memory to coordinate information from both sources, leading to potential inferences between the narration and written text, as illustrated by Diao and Sweller [25].

Hypothesis 3.

Presenting identical information through images, narration, and written text (modal and codal redundancy) will enhance learning and decrease cognitive load compared to groups with only one type of redundancy (group 2 codal redundancy and group 3 modal redundancy). In this scenario, we anticipate that redundancy can enhance learning, as the transient nature of narrations may be complemented by additional images and written text, affording learners’ multiple retrieval cues and facilitating information processing [2,22].

4. Method

4.1. Participants and Design

The online data collection was administered via the digital survey tool Unipark, with participants recruited through the subject pool Prolific (Prolific is an online platform that connects researchers with participants for survey and study recruitment. https://www.prolific.com (accessed on 11 March 2024)). To qualify for participation, participants had to be fluent in German and had access to either a tablet, laptop, or desktop computer with an audio device (headphones). A total of 158 participants (68.4% male; 31.73 years (Min. 18; Max. 66), SD = 11.63) (n = 1 (0.6%) divers; n = 2 (1.3%) not mentioned) participated in the experiment. The study utilizes a 2 × 2 factorial between-subjects design, with modal redundancy (yes vs. no) and codal redundancy (yes vs. no). As outlined in

Table 1. Assignment of participants to groups.

Redundancy		Presentation	N = 158
Modal	Codal
No	No	Images and narration (IN)	n = 43 (27.2%)
No	Yes	Narration and written text (NT)	n = 35 (22.2%)
Yes	No	Images and written text (IT)	n = 41 (25.9%)
Yes	Yes	Images, narration, and written text (INT)	n = 39 (24.7%)

, participants were randomly assigned to one of four groups resulting from the factorial design. Utilizing G*Power 3.1.9.7 [26], we determined the necessary sample size, which indicated that a minimum of 128 participants was required, assuming a medium effect size (f = 0.25), a Type I error of 0.05, and a Type II error of 0.20.

4.2. Learning Material (Rhombus Logic Tasks)

Since the expertise reversal effect [27] demonstrated that additional redundant information (e.g., additional written text to a diagram) initially benefits novice learners, it becomes detrimental as they become more familiar with the task; we aimed to exclude the effects of prior knowledge and used an adapted form of system-paced rhombus logic tasks from Albers et al. [12]. Participants were presented with rhombus shapes containing numbers and positions. Each task consisted of a row of three rhombuses with a position (top, down, left, right) and a number changing according to a logical pattern.

For the construction tasks, we tested learners’ ability to understand and apply logical patterns and use inductive reasoning to construct the fourth rhombus. For the corresponding recall tasks, learners had to reconstruct one randomly chosen rhombus from the original three. Therefore, the recall tasks assessed memory retrieval, particularly the ability to remember and reconstruct information. All groups received the same tasks, but they were presented differently depending on the types of redundancy (Figure 1). For modal redundancy, identical information was presented via images and written text, while for codal redundancy, it was presented via narration and written text. The group with modal and codal redundancy received identical information via images, written text, and narration, whereas the group with neither modal nor codal redundancy via images and narration. Overall, participants had to complete the construction and corresponding recall tasks in a randomized order.

4.3. Tests and Scoring

4.3.1. Learning Outcomes

Learning outcomes are operationally defined as the participants’ performance on the rhombus logic tasks. The rhombus logic tasks represent a formal learning activity, as they are designed with a clear structure, requiring participants to follow specific instructions to complete them. The tasks are standardized, measuring participants’ performance through accuracy, and aim to test specific cognitive abilities such as logical reasoning and memory retrieval. Participants could attain up to two points for each of the ten construction tasks and two points for each of the ten recall tasks, based on the correctness of the number and position (for each task, participants had to write the number in a designated field and select one of the four positions (top, down, left, right) from a drop-down menu). Consequently, each task had a potential maximum score of four points. Therefore, the total possible score for all ten construction tasks was 20 points, and similarly, the total possible score for all ten recall tasks was also 20 points. The overall learning score was calculated by adding the total scores from the construction and the recall tasks, leading to a maximum possible overall score of 40 points.

4.3.2. Perceived Task Difficulty and Mental Load

To measure cognitive load after each construction task and after each recall task, participants were asked to rate their perceived task difficulty and their experienced mental load. Both are related concepts but represent different aspects of cognitive processing. Perceived task difficulty reflects learners’ subjective evaluation of how challenging they perceived the task to be. In contrast, the task-related experienced mental load indicates the amount of cognitive resources required to perform a task [28,29].

To measure participants’ perceived task difficulty, they had to rate how easy or difficult it was to solve the task. The items were oriented on Bratfisch et al. [30] and Kalyuga et al. [17]. For the construction tasks, the item (in German) was “How easy or difficult was it for you to construct the fourth rhombus?”. For the recall tasks, the item (in German) was “How easy or difficult was it for you to recall the [first, second, or third] rhombus?”. Ratings were made using a nine-point Likert scale ranging from 1 = very, very easy to 9 = very, very difficult. To measure the mental load, we implemented a facet of the NASA Task load index (TLX) from Hart and Staveland [31]. To investigate the experienced mental load, we asked the questions (in German) “How mentally demanding was it to construct the fourth rhombus?” (for construction tasks) and “How mentally demanding was it to recall the [first, second, or third] rhombus?” (for recall tasks). Ratings were made using a nine-point Likert scale ranging from 1 = very, very little to 9 = very, very much.

The perceived task difficulty score and the mental load score were each computed using the mean score of the ratings for each participant. Finally, three scores were calculated: one construction, one recall, and one overall mean score of both construction and recall scores. All three scores could range from 1 to 9.

4.4. Procedure

After accessing the study via the subject pool, Prolific participants gave their informed consent to participate. They then completed an audio device check and answered questions on their demographics (gender, age, hearing impairment, mother tongue (n = 140 (88.6%) German as mother tongue; n = 18 (11.4%) other language. All participants indicated to be fluent in German, and analyses revealed no significant impact of mother tongue on the dependent variables), and the existence of left–right confusion). Afterward, they had to pass an attention check (type the color “green” into the text field below). At this point, they were randomly assigned to one of four groups according to the experimental design of the study (Table 1). Before participants entered the tasks, they watched an instructional video (2 min) explaining the procedure and providing an example task. After that, participants had to solve twenty tasks following a standardized procedure: After participants pressed the “Start” button, they viewed a video displaying each rhombus for four seconds. Afterward, they engaged in a distraction task, during which they were shown five animals or countries in different modalities (narration, written text, or images), followed by a question like “Which animal is smallest?”. The purpose of the distraction tasks was to occupy participants’ working memory. They had to type the answer into a text field. In the following, participants received instructions to construct the fourth rhombus (“What should the 4th rhombus in the logical series look like?”) and rated their perceived task difficulty and mental load on a nine-point Likert scale. For the recall tasks, participants were asked “What did the [first, second, or third] rhombus in the logical series look like?” and rated their perceived task difficulty and mental load. After that, they received the prompt that they had to click “next” to go to the next task. Once participants had completed the tasks, they had to pass an additional attention check. At the end of the survey, they were asked questions regarding cheating and technical issues and should indicate if they tried to ignore some information in the learning material. The survey duration was approximately 25–30 min, and participants received their monetary reward afterward (6 euros).

5. Results

5.1. Experimental Factors (Modal and Codal Redundancy)

The results were computed using SPSS 28. The internal consistency was satisfying, with Cronbach’s alpha >0.80 for construction and recall tasks. To ensure group comparability, we examined the distributions of gender and age. A chi-square test revealed no significant differences in gender distribution among the four groups, χ²(9, N = 158) = 13.705, p = 0.133. Similarly, ANOVA results indicated no significant differences in age distribution across the groups, F(3, 154) = 0.494, p > 0.05.

A 2 × 2 MANOVA with modal and codal redundancy as between-subject variables was performed on the construction and recall tasks and on perceived task difficulty and mental load for construction and recall tasks. For effect sizes, ηp² is reported where values of 0.01, 0.06, and 0.14 represent small, medium, and large effect sizes [32]. The results from Levene’s test for homogeneity of variance indicated that homogeneity of variance was met for all dependent variables except the recall tasks (to account for the violation of homogeneity of variances, we additionally conducted a Welch–ANOVA for the recall tasks). Assumption of normality (measured with a Kolmogorov–Smirnov test) was applied for perceived task difficulty and mental load but was violated for the test scores (construction and recall). However, since MANOVAs are considered reasonably robust against violations of normality [33], we refrained from presenting non-parametrical methods in this case.

5.2. Learning

To investigate how modal and codal redundancy affect learning, we examined the learning scores for construction and recall tasks (

Table 2. Learning Scores for all groups (N = 158).

	Construction Score		Recall Score		Overall Learning Score *
Groups	M	SD	M	SD	M	SD
IN (n = 43)	12.98	4.13	11.21	5.26	24.18	8.57
NT (n = 35)	8.74	5.05	8.80	6.09	17.54	10.66
IT (n = 41)	11.95	4.43	10.61	4.79	22.56	8.24
INT (n = 39)	13.00	3.45	10.79	4.51	23.79	6.88

* Minimum Score is 0; Maximum Score is 20 for Construction and Recall Score, 40 for Overall Learning score. Redundancy (modal/codal): IN (no/no), NT (no/yes), IT (yes/no), INT (yes/yes).

and Figure 2). For construction tasks, modal redundancy had a positive effect, F(1,154) = 5.592, p < 0.05, ηp2 = 0.035, enhancing learning outcomes, whereas codal redundancy had a negative effect, F(1,154) = 5.432, p < 0.05, ηp2 = 0.034. Additionally, a significant interaction between modal and codal redundancy was observed, F(1,154) = 14.943, p < 0.001, ηp² = 0.088, indicating that modal redundancy mitigates the negative effects of codal redundancy. No significant effects were found for recall tasks (this non-significant finding was confirmed by a Welch–ANOVA, which was conducted to account for the violation of homogeneity of variances, F(3, 83.696) = 1.211, p = 0.311).

5.3. Perceived Task Difficulty

To investigate how modal and codal redundancy affect perceived task difficulty, we examined the ratings for construction and recall tasks (

Table 3. Perceived Task Difficulty Scores for all groups (N = 158).

	Difficulty Construction		Difficulty Recall		Overall Difficulty *
Groups	M	SD	M	SD	M	SD
IN (n = 43)	5.62	1.68	5.66	1.81	5.64	1.65
NT (n = 35)	7.24	1.25	6.82	1.62	7.03	1.39
IT (n = 41)	5.59	1.68	5.70	1.61	5.65	1.54
INT (n = 39)	6.36	1.34	6.30	1.30	6.33	1.22

* Minimum score is 1; Maximum score is 9. Redundancy (modal/codal): IN (no/no), NT (no/yes), IT (yes/no), INT (yes/yes).

and Figure 3). Results indicate that codal redundancy significantly increased perceived task difficulty for both construction, F(1,154) = 24.349, p < 0.001, ηp2 = 0.137, and recall tasks, F(1,154) = 11.974, p < 0.001, ηp2 = 0.072. In contrast, modal redundancy did not significantly affect perceived task difficulty.

5.4. Mental Load

To investigate how modal and codal redundancy affect metal load, we examined the ratings for construction and recall tasks (

Table 4. Mental Load Scores for all groups (N = 158).

	Mental Load Construction		Mental Load Recall		Overall Mental Load *
Groups	M	SD	M	SD	M	SD
IN (n = 43)	5.58	1.71	5.57	1.84	5.58	1.69
NT (n = 35)	7.24	1.26	6.76	1.67	7.00	1.43
IT (n = 41)	5.61	1.67	5.66	1.59	5.64	1.54
INT (n = 39)	6.17	1.42	6.17	1.48	6.17	1.40

* Minimum Score is 1; Maximum Score is 9. Redundancy (modal/codal): IN (no/no), NT (no/yes), IT (yes/no), INT (yes/yes).

and Figure 4). For the construction tasks, modal redundancy reduced mental load, F(1,154) = 4.501, p < 0.05, ηp2 = 0.028, and codal redundancy increased it, F(1,154) = 20.283, p < 0.001, ηp2 = 0.116. Additionally, the interaction between modal and codal redundancy was also significant, F(1,154) = 5.049, p < 0.05, ηp2 = 0.032, suggesting that the presence of modal redundancy can alleviate the increased mental load caused by codal redundancy. However, no significant effects were found for the recall tasks.

6. Discussion

The results of this study contribute to our understanding of redundancy in multimedia learning, building upon existing research grounded in Cognitive Load Theory (CLT) and Cognitive Theory of Multimedia Learning (CTML). Our investigation explored how modal and codal redundancy impact learning outcomes and cognitive load.

The first hypothesis posited that presenting identical information through images accompanied by narration (no modal or codal redundancy) would result in superior learning outcomes and reduced cognitive load compared to the groups involving either codal or modal redundancy. Our results partially support this hypothesis. Specifically, the group with images and narration demonstrated higher scores in the construction tasks and reported lower cognitive load than the group with narration and written text (codal redundancy). However, contrary to expectations, no significant superiority was observed in learning and cognitive load between the group with images and narration and the group with images and written text (modal redundancy). This finding contrasts with prior research by Mayer and Moreno [34], who observed a modality effect where animation with narration was more effective than animation with written text when understanding the process of lightning formation and the car’s braking system. One possible explanation for our findings is that the complexity of the learning tasks in our study may not have been sufficiently robust to highlight distinctions in learning outcomes and cognitive load. Unlike the specific scenarios investigated by Mayer and Moreno [34], our study focused on a setting where both narration and written text provided comparable support for cognitive processing. Therefore, it is conceivable that both groups in our study derived benefits from dual coding [35], which involves integrating visual and verbal information to enhance encoding and retrieval processes rather than overloading the verbal channel (codal redundancy). Given that both groups needed to integrate visual (images) and verbal (text or narration) information, it is plausible that the cognitive demands associated with integrating these types of information were sufficiently similar [36], resulting in comparable cognitive load and performance in construction tasks between the groups.

The second hypothesis posited that presenting identical information through narration and written text (codal redundancy) would result in lower learning outcomes and higher cognitive load compared to the other experimental groups. Our findings confirm this hypothesis. Codal redundancy was indeed associated with decreased learning outcomes and higher perceived task difficulty and mental load. This suggests that the concurrent engagement required to process information from both narration and written text increased cognitive load and impeded learning. This underscores that codal redundancy can overload the verbal processing channel, consistent with the verbal redundancy effect [2,37].

The results support Hypothesis 3, which posited that presenting identical information through images, narration, and written text (modal and codal redundancy) would enhance learning outcomes and decrease cognitive load compared to groups with only one type of redundancy (group 2 codal redundancy and group 3 modal redundancy). The significant interaction observed between modal and codal redundancy on the construction tasks suggests that the combined use of both types of redundancy enhanced learning. Furthermore, this interaction appeared to mitigate the detrimental effects typically associated with codal redundancy alone, as indicated by the overall higher scores for construction tasks. This compensatory mechanism implies that while codal redundancy alone may increase cognitive load by requiring simultaneous processing of verbal and written information, the inclusion of modal redundancy serves to balance and distribute cognitive load more effectively across multiple channels. Consequently, learners benefit from improved learning and reduced cognitive load when information is presented through a combination of images, narration, and written text. These findings align with the reversed redundancy effect illustrated by many studies [23,38,39,40].

6.1. Answering the Research Question

How do modal and codal redundancy affect learning and cognitive load when the provided content is identical?

For learning, modal redundancy seems to have a positive effect on learning outcomes (construction tasks). Participants who received information in both visual and textual formats performed better on construction tasks compared to those who received the information through narration and written text (codal redundancy). The significant interaction between modal and codal redundancy shows that when both types of redundancy were present, the negative effects of codal redundancy were mitigated by the positive effects of modal redundancy, suggesting a compensatory mechanism. This mechanism can be explained by the findings from Mayer and Johnson [23], indicating that “redundant on-screen phrases may be particularly helpful for directing the learner’s attention with static illustrations” [23] (p. 385).

For cognitive load, modal redundancy did not significantly affect perceived task difficulty, indicating that participants did not find tasks more or less difficult when information was presented through images and written text. For mental load, modal redundancy had a significant effect on reducing mental load during construction tasks. We assume that the distribution of information across image and written text eased cognitive processing, leading to lower reported mental load. In contrast, codal redundancy significantly increased perceived task difficulty and mental load for construction tasks. Participants reported higher difficulty levels and placed a higher demand on the required cognitive resources to solve the tasks when they had to process both written and spoken information simultaneously. The significant interaction between modal and codal redundancy on mental load further demonstrates a compensatory mechanism similar to that observed for construction scores. When both types of redundancy were present, the negative impacts of codal redundancy were alleviated by the positive effects of modal redundancy.

But why do the effects only apply to the construction and not the recall tasks? When we refer to Chen et al. [41], we can assume that the construction tasks inherently involve a higher intrinsic cognitive load due to the need to understand the logical patterns to create the fourth rhombus (higher knowledge complexity). Since tasks with higher intrinsic cognitive load are more likely to be affected by additional extraneous cognitive load [4], changes in the manner of information presentation (which affect extraneous cognitive load) have a pronounced impact. In contrast, the recall tasks involve retrieving information from memory, typically resulting in a lower intrinsic cognitive load (lower knowledge complexity). Consequently, changes in how information is presented have a lesser effect on learning, perceived difficulty, and mental load for the recall tasks.

6.2. Limitations

While this study provides valuable insights into the effects of modal and codal redundancy, several limitations must be acknowledged. The study’s high internal validity is a significant strength, owing to its controlled experimental design and meticulous management of variables such as redundancy type, presentation mode, and task difficulty. This control minimizes confounding factors, ensuring that observed differences in learning and cognitive load can be attributed to the manipulated variables rather than extraneous influences. Additionally, the use of system-paced tasks ensured standardized administration across all participants, reducing variability arising from individual pacing preferences. However, these strengths are counterbalanced by the study’s limitations, particularly its low external validity. The tasks employed in this research were abstract and not representative of the complex, real-world learning scenarios encountered in educational settings. Authentic learning environments typically involve diverse materials that demand varied cognitive processes, which may not align with the simplified tasks used here. Consequently, the generalizability of findings to practical educational contexts remains uncertain. Another limitation concerns participant recruitment via platforms like Prolific, offering access to a broad participant pool but potentially introduces variability in engagement levels and task completion conditions. Despite efforts to mitigate these issues with attention checks, some participants may not have fully engaged or adhered to instructions consistently, which could affect data reliability. Furthermore, assessing participants’ processing of learning materials was constrained by reliance on self-reporting, which may not always accurately reflect actual engagement or selective attention (we asked participants to report if they attempted to ignore some redundant information, but these self-reported data may not be entirely reliable). Future studies could benefit from integrating objective measures such as eye-tracking to provide more nuanced insights into how learners interact with and process the content. Such technologies can pinpoint where and for how long attention is directed, offering deeper insights into cognitive processing. Additionally, incorporating within-subjects designs and considering individual differences in cognitive abilities, such as working memory capacity, could enhance understanding of how learners benefit differently from various presentation formats. This approach would enable researchers to explore whether cognitive factors moderate the effects of redundancy on learning outcomes. Regarding the measurement of cognitive load, this study relied on perceived task difficulty and mental load. Since redundancy can increase extraneous cognitive load, employing more precise measures such as those proposed by Klepsch et al. [42] would be advantageous. Additionally, assessing active mental effort, as suggested by [28], could provide further insights into cognitive engagement. However, subjective measures are inherently prone to various biases, including social desirability and inaccuracies in self-perception [43], which can potentially undermine the validity and reliability of findings. To enhance the robustness of future studies, it is recommended to include complementary objective measures of cognitive load. Objective measures offer the advantage of providing quantifiable data that is less susceptible to individual biases. For instance, physiological indicators such as heart rate or pupil dilation can offer additional information by reflecting autonomic nervous system responses to cognitive demands [44]. However, expanding the number of dependent variables poses the risk of inducing a fatigue effect among participants, which should be carefully considered in future research designs.

7. Conclusions

The aim of this study was to investigate the effects of different types of redundancy—modal (images and written text) and codal (narration and written text)—on learning and cognitive load when the provided content is identical. An experimental study was conducted with participants completing tasks under varying redundancy conditions. The main findings indicate that modal redundancy enhances learning outcomes and reduces cognitive load, while codal redundancy increases cognitive load and hinders learning. When both types of redundancy are present, the positive effects of modal redundancy can mitigate the negative impacts of codal redundancy.

The key takeaway for educators and instructional designers is the importance of strategically combining visual and verbal information and the notion that repeated information, when presented strategically, is helpful to learning. Therefore, it is essential to consider the types of redundancy and that a balanced combination of both types of redundancy can optimize learning outcomes and cognitive load by leveraging the strengths of each modality. However, researchers and educators must be aware of the degree of contentual overlap.