The Impact of Weibo Features on User’s Information Comprehension: The Mediating Role of Cognitive Load

Abstract

Social media, such as Microblogs, have become an important source for people to obtain information. However, we know little about how this would influence our comprehension over online information. Based on the cognitive load theory, this research explores whether and how two important features of Weibo, which are the feedback function and information fragmentation, would increase cognitive load and may in turn hinder users’ information comprehension in Weibo. A 2 (feedback or non-feedback) × 2 (strong-interference or weak-interference information) between-participants experimental design was conducted. Our results revealed that the Weibo feedback function and interference information exerted a negative impact over information comprehension via inducing increased cognitive load. Specifically, these results deepened our understanding regarding the impact of Weibo features on online information comprehension and suggest the mechanism by which this occurs. This finding has implications for how to minimize the potential cost of using Weibo and maximize the adaptive development of social media.

Keywords

Weibo’s Feature Feedback function Interference information Information comprehension Cognition load

Introduction

With the rapid development of social media, we have entered into an era characterized by information explosion and fragmentation. According to a recent report by CNNIC (China Internet Network Information Center, 2020),¹ there are approximately 550 million active users on Weibo. Being one of the top rank social media platform, Weibo is an important source for the young generation to obtain instant information. As argued by Kaplan & Haenlein, (2010), social media has changed the way people read and understand information due to some if its features. Empirically, research has implied that microblog’s feedback function may shorten the time in which users’ digest the information presented (Bergman et al., 2006), and encourage urge them to concern other parts besides micro-blogging messages (Boyd et al., 2010). “Repost” features also may disrupt the coherence of target information (Mayer et al., 1996), truncate the original message to make it fit to repost (Kaplan & Haenlein, 2011), and produce excessive information (Bergman et al., 2006). Despite the long-standing interest in the impacts of Weibo on information comprehension, empirical studies on whether and how this effect occurs, especially employing objective measures, are far fewer than one might expect. Hence, in this study, using the eye movement technology, we explored the impacts of Weibo’s two typical features (i.e., feedback function and information fragmentation) on users’ information comprehension. Additionally, based on the Cognitive Load Theory (CLT), which suggests that shortened reading time and enormous information may lead to increased extraneous cognitive load, resulting in poorer information comprehension(Sweller, 2010), we further examined whether the extraneous cognitive load would play a mediating role between Weibo’s features and users’ comprehension.

Weibo Features and Microblog Information Comprehension

Weibo, similar to Twitter, is an online social media platform that enables users to share, communicate, and interact in text, pictures, videos, or other forms of instant information (Kaplan & Haenlein, 2010). As a typical micro-blogging platform, Weibo has two inherent structural features for feedback function and information fragmentation(Jiang et al., 2016). The feedback function allows users to “repost,” “comment,” or “like” the online content, which means the communication is interactive. The feature of information fragmentation stems from the relatively short content and it spreads without restriction whenever or wherever (Kaplan & Haenlein, 2011), presenting an explosion of content quantity and a fragmentation of quality (Jiang et al., 2016; Boyd et al., 2010; Mayer et al., 1996).

Research has implied that the feedback function of Weibo as a link navigation can hinder readers’ understanding of information (Jiang et al., 2016). This might be caused by users’ unconscious attention to the feedback interface, which would increase demands on visual processing and then detract from comprehension (DeStefano & LeFevre, 2007). Another possibility is that the feedback links presented on social media may shorten the information processing time by prompting users to respond quickly towards the content (Bergman et al., 2006), and impede individuals from establishing internal connections between information (Mayer et al., 1996).

The fragmentation of microblog information results from the large amount of “trivial information” that can be presented. Microblog users may encounter difficulties in capturing the target information when a large amount of distracting information is presented, as the interference information may reduce the accessibility of target information and disrupt a reader’s attention persistence (Jiang et al., 2016). In this way, the disruption hypothesis (Mayer et al., 2008) suggests that irrelevant content may disrupt readers’ cognitive activity aimed at building a coherent mental model of the to-be-learned system. A large amount of interference information on Weibo would compete with the limited cognitive resources for target information. Therefore, users tend to experience difficulties in establishing logical connections among the presented information, which finally impacts their comprehension performance (Jiang et al., 2016).

The Mediating Role of Extraneous Cognitive Load

Features such as the feedback function and interference information on Weibo may increase a users’ cognitive load. According to the CLT, cognitive load refers to the total working memory resources required to carry out a learning task (Kirschner et al., 2018). There are three categories of cognitive load (Sweller, 1988; Paas & Sweller, 2003), including intrinsic cognitive load, extraneous cognitive load, and germane cognitive load.

Intrinsic cognitive load is concerned with the intrinsic complexity of information, that is, the number of interactive elements that must be treated at the same time (Sweller & Chandler, 1994; Sweller, 1994). Extraneous cognitive load is evoked by the features of the presented information and does not directly contribute to learning (Sweller et al., 2019; de Jong, 2010; Van Merriënboer & Sweller, 2005). Cognitive Load Theory’s basic premise is that the working memory capacity is limited (Sweller et al., 2019), and that intrinsic and extraneous cognitive load are additive (Sweller, 2010; Sweller et al., 2011). The more resources that must be devoted to dealing with extraneous cognitive load, the less there will be allocated to intrinsic cognitive load (Sweller et al., 2019). Germane cognitive load is defined as the cognitive load required to learn, which refers to the working memory resources that are spared to intrinsic cognitive load rather than extraneous cognitive load (Sweller et al., 2019). Sweller et al. (2019) have removed germane load from the additive equation of total load and, as a result of this, only intrinsic and extraneous cognitive load are distinguished as basic categories of cognitive load. Thus, cognitive overload is determined solely by the increasing of intrinsic or extraneous cognitive load.

When browsing Weibo information, users may need some cognitive resources to focus on multiple feedback function keys, which may increase the extraneous cognitive load (occupying more working memory resources that could otherwise have been allocated to intrinsic cognitive load), which may lead to inaccurate information comprehension (DeStefano & LeFevre, 2007). Meanwhile, Weibo users have to deal with a large number of hyperlinks that may be irrelevant to the target information, with the switching across such hyperlinks also potentially leading to extraneous cognition overload (Carr, 2010), thus further hindering the information comprehension. It may therefore be speculated that extraneous cognitive load may mediate the relationship between interference information, feedback function, and information comprehension.

Eye Movement Technology

Literature has examined the impact of cognitive load on information comprehension, mostly adopting self-report method to measure participants perceived cognitive load (Cain et al., 2016; Baumgartner et al., 2014; Cardoso-Leite et al., 2016). However, individuals may have self-recollection bias because of limited memories about the past experience and accurate reports may elicit negative perceptions (Poltavski et al., 2018; Vraga et al., 2019; Willard & Gramzow, 2008). On the other hand, people may not be able to reliably introspect about how distracted they were during a task as they may not be consciously aware that it is happening (Vraga et al., 2019; Willard & Gramzow, 2008). Therefore, an objective measurement on cognitive load may offer more accurate conclusions.

By presenting participants with reading materials in real life, eye movement technology makes the experiment closer to real-world settings and increases ecological validity (Yan & Bai, 2018). Eye movement technology also provides information that cannot be obtained from self-report, or the attention test, which is based on task response time (Scott & Hand, 2016). Moreover, with the advantage of high spatial accuracy and temporal sensitivity, eye movement technology can measure individuals’ real-time visual attention processes, which enhances the objectivity of measurement (Scott & Hand, 2016; Rayner, 1998).

In the study, we adopted time-related and space-related eye movement indicators to reflect participants’ extraneous cognitive load, as indicated by total fixation duration and total number of fixations, respectively. Total fixation duration refers to the total gaze time in the area of interest (AOI), and it reflects the duration information cognitive processing (Yan et al., 2013). In this regard, Jin (2005) compared eye movement patterns among students with different comprehension abilities and found that students with stronger comprehension ability need less cognitive resources and recorded shorter total fixation duration when reading. Total number of fixations also indicate the cognitive load of reading process, with a greater numbers of fixation times indicating heavier cognitive load (Yan & Bai, 2018).

Current Study and Hypotheses

Prior research has explored the effect and mechanism of how online information impact cognitive load and comprehension. However, these have predominantly utilized subjective self-reports to assess cognitive load (e.g., Skulmowski & Xu, 2022; Jiang et al., 2016). In this study, therefore, we combined self-report, a reading comprehension test, and eye movement technology to exam the impact of Weibo features on user’s information comprehension. Based on the theoretical consideration aforementioned, we put forward our theoretical model (see Figure 1) and the following hypotheses.

Figure 1.

The mediation model.

First, as the feedback function and interference information on Weibo may induce the explosion of content quantity and fragmentation of content quality, and this can hinder users’ comprehension, we put forward that:

H1: The feedback function and interference information on Weibo interface would lead to the decrease in accuracy of users’ comprehension of Weibo information.

Secondly, the impact of microblog’s features on information comprehension may not be direct. The increased the extraneous cognitive load resulted from feedback function and interference information may mediate them. Accordingly, we put forward out second and third hypothesis:

H2: The feedback function and interference information group would increase the cognitive load perceived by participants.

H3: Perceived cognitive load would play a mediating role in the influence of the two features of Weibo (feedback function and interference information) on the comprehension of Weibo information.

Ultimately, given that users tend to unconsciously pay attention to Weibo’s interface design (Wathen & Burkell, 2002), and the feedback functions such as “repost” require users to respond quickly, this may lead to extraneous cognitive load and less time reflected on the target information (Bergman et al., 2006). Meanwhile, interference information would compete with cognitive resources of target information. Therefore, we proposed the following hypotheses:

H4a: With the feedback function, participants would exhibit shorter fixation duration and fewer fixations on Weibo information, while exhibit longer fixation duration and more fixations on feedback function interface.

H4b: With the interference information, participants would have longer fixation duration and more fixations on interference microblog information.

Methods

Participants

G*power Version 3.1 (Erdfelder et al., 1996) was used to calculate the sample size required for this study, and the parameters were set as follows: effect size η_p² = 0.1, f = 0.33, α = 0.05, Power (1 – β) = 0.8, number of groups = 4. The total sample size was calculated to be 75. Taking into account the possibility of extreme data and referring to previous studies (e.g., Jiang et al., 2016), the current study recruited 80 participants. Recruitment information was presented online, and 80 undergraduate (36 males and 54 females) Weibo users participated, with an average age of 20.2 years (SD = 1.51). The participants had normal or corrected vision, no color blindness or color weakness, all right-handed, and no reported history of psychiatric disorder. Participants were invited to the laboratory after obtaining their informed consent. After the experiment, each participant was paid 15 yuan as monetary compensation.

Materials

Experimental equipment

Software E-prime 1.0 (Psychology Software Tools, Inc., Pittsburgh, PA, USA) was used to present experimental tasks. The stimulus materials were presented on a screen with a display size of 19.7 inches, a display resolution of 1024 × 768 pixels, and 85 Hz. A desktop eye movement system EyeLink 1000 (SR research Ltd., Mississauga, ON, Canada) was used to record the participants’ eye movement (right eye) with a sampling rate of 500 Hz.

Experimental materials

Content and interface of Weibo

In order to restore the usage scenario of Weibo as realistic as possible, the experimental materials were 50 messages selected from Sina Weibo app, all in Chinese. Among them, 40 pieces were target messages, the theme of which was “The Ph.D. of Zhejiang University has lost contact.” (1) In the strong interference information group, participants needed to read 50 pieces of messages, 40 pieces about “The loss of Ph.D. of Zhejiang University”; the remaining 10 pieces of messages were selected from Sina Weibo app which were irrelevant to the target topic. (2) In the weak interference information group,¹ participants also needed to read 50 pieces of messages, 40 of which were related to the theme of “The Ph.D. of Zhejiang University has lost contact” and the remaining 10 pieces of messages were “Sorry, this Weibo had been deleted. View help: Weblink.” Some meaningless characters were added to ensure the number of characters presented in these two groups are identical. (3) In the feedback function group, there would be a feedback function interface on the bottom of each Weibo message, including “Repost,” “Like,” and “Next”; (4) In the non-feedback function group, only “Next” function interface was presented at the bottom of each Weibo message. Weibo details in each group are shown in Figure 2.

Figure 2.

Content and interface of Weibo in each group.

Measurement

Reading comprehension accuracy Comprehension test is consisted eight choice questions (See Appendix A for reading comprehension test). These questions were evaluated by 10 college students prior to the experiment (they didn’t participate in the formal experiment) from 1 = very easy to 9 = very difficult. The difficulty evaluation result (M = 5.48, SD = 1.09) indicated that these items had a good degree of discrimination. The item of the scale, for example, include “Where was the body of missing Dr. Hou found?,” “What caused the suicide of Dr. Hou?” and “Which of the following does not conform to the information presented on the WeChat moment of Dr. Hou?.” The equation of information understanding accuracy is calculated as number of correct answers divided by the total number of questions.

Cognitive Load Questionnaire Test Cognitive Load Questionnaire Test (Tsang & Velazquez, 1996) by answering seven items with a 10-point scale (from 1 = the resource is not occupied at all, to 10 = the resource is completely occupied). The cognitive load was to measure the degree of occupancy in cognitive resources during participants’ reading process, and seven types of cognitive resources were included: response selection, execution, spatial coding, word coding, visual reception, auditory reception, and action manipulation. The score reflected the total cognitive load.

Weibo usage time The time participants on Weibo and their level of dependence on it were interviewed, as prior studies have indicated that these factors may influence their reading style (Zhao & Qin, 2017; Wu et al., 2017), they may have better performance in reading comprehension because they used to comprehend this fragmented information. Thus, we asked participants “When did you first use Weibo?” and “How much time do you spend on Weibo per day on average?,” then consider them as covariate.

Weibo Excessive Use Scale Weibo Excessive Use Scale (Hou et al., 2014) by answering twelve items with a 5-point scale (from 1 = rarely, to 5 = always). Participants’ scores on the scale were used as the index of the extent of participants’ using Weibo, and considered it as covariate.

Design and procedure

Design

A 2 (strong-interference information vs. weak-interference information) × 2 (feedback function vs. non-feedback function) between-subject design was adopted. The independent variables were feedback function and interference information, the dependent variable was accuracy of comprehension over Weibo information, and the mediator variable was cognitive load. During the experiment, the total fixation duration and the total number of fixations on each Weibo information and feedback function interface were recorded as the indicators of extraneous cognitive load.

Procedure

Eighty participants were randomly allocated to 4 experimental conditions (i.e., with feedback and strong-interference, with feedback and weak-interference, without feedback and with strong-interference, and without feedback and weak-interference), with 20 participants in each group. None of participants had ever read the information about “The Ph.D. of Zhejiang University lost contact.” The entire experiment was carried out in an eye-movement laboratory, and each participant was tested individually for about 20 minutes.

The experimental procedures were as follows: (1) Participants filled in basic information and completed the Weibo Excessive Use Scale; (2) Participants were conducted exercise so as to be familiar with experimental procedures; (3) Calibration. Participants were asked to keep distance between their eyes and the screen at 70 cm as stably as possible. Calibration and validation were run on nine points to optimize spatial tracking accuracy. Average errors were considered acceptable below 1.0^◦ and 0.3^◦ for calibration and validation, respectively; (4) During the formal experimental procedure, participants needed to read the designed microblogs information. In the feedback group, after participants had read each Weibo message, they needed to click the left mouse button to make a choice among “Like,” “Repost,” or “Next” in function interface. If participants chose “Like” or “Repost,” then the characters of “Like!” or “Reposted!” appeared on the screen correspondingly, followed by the next message. If they clicked “Next,” they would skip directly to the next message. Participants in the non-feedback group just needed to click “Next” on the function interface to skip to the next message after reading. The maximum presentation time of each Weibo message is 15000 ms, and 50 pieces of messages were presented randomly. Participants’ eye movement data were recorded during the whole reading process; (5) After completing the experimental task, participants took a comprehension test; (6) Then, participants completed the Cognitive Load Questionnaire Test.

Data analysis

The eye movement data was preprocessed by Data Viewer 2.2.1 and exported to Excel format. All data were analyzed using SPSS 26.0 and Amos 23.0. First, the two-way analysis of variance (ANOVA) for accuracy of information comprehension was conducted to explore the influence of interference information and feedback function on participants’ reading comprehension performance. Second, the two-way ANOVA for cognitive load and testing of the mediation effect of cognitive load were conducted to examine the psychological mechanism explaining the impact of Weibo features on information comprehension. The mediating effect test was conducted by Amos. Finally, the two-way multivariable analysis of variance (MANOVA) for attention on microblogs information (including 40 target information, 10 interference information, and all microblogs information) and feedback function interface were conducted, comparing participants’ extraneous cognitive load under four experimental conditions in the process of reading Weibo information. One participants’ data were excluded from eye movement analysis as head shaking occurred and 79 valid data were retained. All analysis were conducted after controlling for age, gender, onset time of using Weibo, the average time spent on Weibo per day, and the score of excessive Weibo use, all analysis was done.

Result

The accuracy of information comprehension

We conducted a two-way ANOVA of 2 (with feedback vs. non-feedback) × 2 (with strong-interference vs. weak-interference) with the accuracy of microblog information comprehension as the dependent variable. The results showed that the main effect of feedback was significant (F_(1,70) = 17.24, p < .001, η_p² = 0.20), indicating that the comprehension performance for the feedback group was significantly poorer than the non-feedback group (0.52 vs. 0.62). The results of the ANOVA, however, revealed no significant main effect for the interference conditions (F_(1,70) = 1.14, p = 0.290, η_p² = 0.02) and no significant interaction effect between feedback and interference (F_(1,76) = 0. 002, p = 0.968, η_p² < 0.001), see Figure 3. Thus, H1 partially was supported.

Figure 3.

Different experimental conditions and accuracy of information comprehension. Error bars indicate standard deviation.

The Mediating Role of Perceived Cognitive Load

Weibo Features and Perceived Cognitive Load

A two-way ANOVA by 2 (feedback vs. non-feedback) × 2 (strong-interference vs. weak-interference) was conducted with perceived cognitive load as the dependent variable. The results showed that the main effect of feedback condition was significant (F_(1,70) = 14.82, p < .001, η_p² = 0.17), indicating that the cognitive load score for the feedback group was significantly higher than the non-feedback group (37.81 vs. 32.08). The significant main effect of the interference condition was also found (F_(1,70) = 13.49, p < 0.001, η_p² = 0.16), indicating that the cognitive load score for the strong-interference group was significantly higher than the weak-interference group (36.68 vs. 32.58). However, the interaction effect between feedback and interference was not significant, F_(1,70) = 0.11, p = 0.741, η_p² = 0.002 (Figure 4). Thus, H2 was supported.

Figure 4.

Different experimental conditions and perceived cognitive load. Error bars indicate standard deviation.

The Mediating Effect of Perceived Cognitive Load

With the Weibo features as the independent variables, and microblog information comprehension as the dependent variable, two moderated mediating effect models were constructed, respectively, to test the mediation effect of perceived cognitive load, see Figure 5a and Figure 5b. The results revealed that feedback function and interference information were found to be positive and significant predictors of cognitive load (β = 0.41, SE = 0.10, CR = 3.96, p < 0.001; β = 0.33, SE = 0.11, CR = 3.08, p < 0.01, respectively), but only feedback function was found to be a negative and significant predictor of microblog information comprehension (β = −0.34, SE = 0.11, CR = −3.17, p < 0.01). The cognitive load could negatively predict microblog information comprehension in both two models (β = −0.29, SE = 0.10, CR = −2.85, p < 0.01). And then, no significant interaction effects of feedback and inference on cognitive load and information comprehension were observed in either of the two models (β = −0.044, SE = 0.10, CR = −0.43, p = 0.669; β = −0.02, SE = 0.10, CR = −0.21, p = 0.833, respectively).

Figure 5a.

The mediating role of cognitive load in the influence of feedback on Weibo information comprehension.

We conducted a bootstrapping analysis (5000 samples) to test the mediating effects of cognitive load. As predicted in H2, the results showed that the 95% confidence intervals for both mediating effects do not include 0, indicating a partial mediating effect of perceived cognitive load in the influence of feedback function on comprehension of Weibo information and a full mediating effect of cognitive load in the influence of interference information on comprehension of Weibo information. The mediating effect is shown in Table 1.

Table 1.

Testing the mediating effect of cognitive load.

		β, %	p	95%CI
		β, %	p	BootLL CI	BootUL CI
Feedback → cognitive Load→ microblog information comprehension	Direct effect	−0.34	<0.01	−0.54	−0.12
	Indirect effect	−0.12	<0.01	−0.24	−0.03
	The proportion of indirect effect	26.67
Interference → cognitive Load→ microblog information comprehension	Direct effect	0.001	0.95	−0.26	0.23
	Indirect effect	−0.10	<0.05	−0.20	−0.02

The eye movements on microblogs

In eye movement data analysis, we divided each microblog into two areas of interest: the information of microblogs and the functional interface on the bottom of microblogs. Researchers suggested that data of fixation duration less than 80 ms or longer than 1200 ms were extreme values and should be excluded (e.g., Yan & Bai, 2018). Therefore, we retained the data between 80 ms and 1200 ms into analysis.

The Eye Movements on Microblog Information

Using the total fixation duration and total number of fixations on each piece of information from 50 microblogs as the dependent variables, a two-way MANOVA by 2 (with feedback vs. non-feedback) × 2 (with strong-interference vs. weak-interference) was conducted. The results of the MANOVA for the eye movements on microblog information showed the overall significant effects of the feedback condition (Wilks’ Λ = 0.82, F_(2,69) = 7.42, p < 0.01, η_p² = 0.18) and the interference condition (Wilks’ Λ = 0.88, F_(2,69) = 4.75, p < 0.05, η_p² = 0.12). The significant main effects of the feedback condition for total fixation duration (F_(1,70) = 15.05, p < 0.001, η_p² = 0.18) and total number of fixations (F_(1,70) = 13.56, p < 0.001, η_p² = 0.16) were observed in the univariate testing, indicating that participants in the feedback group had shorter fixation duration and fewer fixations on each piece of microblog information than those in the non-feedback group (5.62 s vs. 7.10 s, 25.79 fixations vs. 31.29 fixations), supporting H4a. The main effects of the interference condition were also found both in total fixation duration (F_(1,70) = 9.08, p < 0.01, η_p² = 0.12) and total number of fixations (F_(1,70) = 9.54, p < 0.01, η_p² = 0.12), revealing that participants’ total fixation duration and total number of fixations on each piece of microblog information in the strong-interference group were significantly higher than that of the weak-interference group (6.93 s vs. 5.78 s, 30.89 fixations vs. 26.18 fixations). There was no significant over all interaction effect between the feedback and interference conditions for the eye movements on microblog information (Wilks’ Λ = 0.995, F_(2,69) = 0.16, p = 0.853, η_p² = 0.01). The total fixation duration and number fixations on microblog information are presented in Figure 6a and Figure 6b, respectively.

Figure 6a.

The total fixation duration on information and function interface of 50 microblogs. Error bars indicate standard deviation. 6b. The total number fixations on information and function interface of 50 microblogs. Error bars indicate standard deviation.

The Eye Movement on the Feedback Function Interface

A two-way MANOVA by 2 (with feedback vs. non-feedback) × 2 (with strong-interference vs. weak-interference) was performed, using the total fixation duration and total number of fixations on the feedback function interface of 50 microblogs as the dependent variables. The results of the MANOVA for the eye movements on function interface showed an overall effect of the feedback condition, Wilks’ Λ = 0.90, F_(2,69) = 3.67, p < 0.05, η_p² = 0.10. The significant main effect of the feedback condition was observed in the univariate testing for total fixation duration (F_(1,70) = 6.98, p < 0.01, η_p² = 0.09) and total number of fixations (F_(1,70) = 5.53, p < 0.05, η_p² = 0.07), indicating that the participants in the feedback group fixated function interface longer and more frequently than those in the non-feedback group (0.33 s vs. 0.22 s; 1.37 fixations vs. 0.96 fixations), as predicted in H4a. However, the overall effects of the interference condition and the overall interaction effects between the feedback and interference conditions for the eye movements on function interface were not observed in the multivariable testing (Wilks’ Λ = 0.999, F_(2,69) = 0.02, p = 0.977, η_p² = 0.001). The total fixation duration and number fixations on function interface are presented in Figure 6a and Figure 6b, respectively.

The Eye Movements on Target Microblog Information

Using the total fixation duration and total number of fixations on each piece of information from 40 target microblogs as the dependent variables, a two-way MANOVA by 2 (with feedback vs. non-feedback) × 2 (with strong-interference vs. weak-interference) were performed. The results of the MANOVA for the eye movements on target microblog information showed an overall effect of the feedback condition, Wilks’ Λ = 0.80, F_(2,69) = 8.90, p < 0.001, η_p² = 0.21. The significant main effects of the feedback condition were observed in the univariate testing for total fixation duration (F_(1,70) = 18.04, p < .001, η_p² = 0.21) and total number of fixations (F_(1,70) = 16.33, p < .001, η_p² = 0.19), indicating that the participants in the feedback group fixated function interface shorter and less frequently than those in the non-feedback group (5.91 s vs. 7.73 s, 27.20 fixations vs. 34.19 fixations). However, the overall effect of the interference condition (Wilks’ Λ = 0.98, F_(2,69) = 0.77, p = 0.47, η_p² = 0.02) and the overall interaction effect between the feedback and interference conditions (Wilks’ Λ = 0.996, F_(2,69) = 0.14, p = 0.869, η_p² = 0.004) for the eye movements on target microblog information were not observed in the multivariable testing. The total fixation duration and number fixations on 40 target microblog information are presented in Figure 7a and Figure 7b, respectively.

Figure 7a.

The total fixation duration on 40 target microblog information and 10 interference microblog information. Error bars indicate standard deviation. 7b. The total number fixations on 40 target microblog information and 10 interference microblog information. Error bars indicate standard deviation.

The Eye Movements on Interference Microblog Information

A two-factor MANOVA by 2 (with feedback vs. non-feedback) × 2 (with strong-interference vs. weak-interference) was performed, using the total fixation duration and total number of fixations on each piece of information from 10 interference microblogs as the dependent variables. The results of the MANOVA for the eye movements on interference microblog information showed no overall effect of the feedback condition (Wilks’ Λ = 0.97, F_(2,69) = 1.24, p = 0.296, η_p² = 0.04), and no significant interaction effect between the feedback and interference conditions either (Wilks’ Λ = 0.96, F_(2,69) = 1.63, p = 0.20, η_p² = 0.05). The overall effect of the interference condition for the eye movements, however, was observed in the multivariable testing, Wilks’ Λ = 0.15, F_(2,69) = 195.24 p < 0.001, η_p² = 0.85. The significant main effects of the interference condition in total fixation duration (F_(1,70) = 190.21, p < .001, η_p² = 0.73) and total number of fixations (F_(1,70) = 331.76, p < 0.001, η_p² = 0.73) indicated that the participants in the strong-interference group fixated interference microblog information longer and more frequently than those in the weak-interference group (6.57 s vs. 2.24 s, 29.08 fixations vs. 9.85 fixations). Thus, H4b was supported. The total fixation duration and number fixations on 10 interference microblog information are presented in Figure 7a and Figure 7b, respectively.

Discussion

This study provides empirical support regarding the influence of microblog features on readers’ information comprehension and the mechanism by which this occurs. We found that the feedback and interference design on micro-blogging may increase users’ cognitive load and, in turn, lead to less accurate information comprehension. Additionally, by using the combination of a comprehension test, self-report cognitive load, and eye movement analysis, we were able to yield a more robust conclusion. These findings have supported the Cognitive Load Theory and extended it into the field of online information reading comprehension. Practically, our findings provide implications for the adaptive development of social media and suggestions on minimizing their potential negative impact.

The Influence of Weibo features on Information Comprehension

Hypothesis 1 was partially supported, as we found that Weibo’s feedback function interface had a negative impact on the accuracy of microblogs information comprehension, while interference information had no such effect. Initially, our study aimed to examine the impact of two features of Weibo (feedback function and interference information) on reading processing of Weibo messages. The results showed that compared with the condition of non-feedback, the accuracy of participants’ understanding of information in the feedback group was significantly lower, suggesting that the “repost” or “like” function may reduce the accuracy of information comprehension.

However, our study did not find the negative impact of interference information on messages comprehension. Despite not supporting our hypothesis, this result was consistent with the previous study that also observed no significant effect of interference in the microblog context (Jiang et al., 2016). A possible explanation for this non-significant relation between interference and information comprehension could be that the fragmented information presented on Weibo (Boyd et al., 2010) encourages people to browse information randomly (Jiang et al., 2016) and perform superficial processing on both target and non-target information. This shallow processing for repetitive information on microblogs will not only lead to limited comprehension of target information but also weaken the negative impact of interference information. Another explanation may be that when people are facing massive information, they employ a cognitive strategy that ignore irrelevant information to minimize the interference by non-target information.

The Mediating Role of Perceived Cognitive Load

To understand the underlying mechanism by which Weibo features impact reading comprehension, we further compared the participants’ perceived cognitive load in four different groups, and then explored the mediating role of perceived cognitive load in the relation between microblogs features and understanding of Weibo information. As expected, the results showed that the cognitive load perceived by participants with feedback or interference was significantly higher than those without feedback or interference. This indicated that the feedback function interface and interference information on Weibo may occupy readers’ cognitive resources and increase their extraneous cognitive load. This finding was also supported by the results of our eye movement analysis, discussed in the next section.

Specifically, the results showed that perceived cognitive load partially mediated the relation between the feedback function and information comprehension on Weibo. This can be explained by CLT (Paas & Sweller, 2003), which suggests that Weibo’s feedback function interface would compete for cognitive resources with target information (Shenk, 1998). This competition over cognitive resources may increase users’ extraneous cognitive load as users spend more time on the feedback function interface, and reduce the cognitive processing time for target information, finally leading to a decrease in accuracy of information comprehension (Jiang et al., 2016). Other studies have obtained the similar finding, that the higher demand of hyperlinked text on an individual’s cognitive resources, the worse the individual’s reading performance (e.g., DeStefano & LeFevre, 2007). Moreover, the setting of feedback function keys on Weibo required users to make decisions immediately about whether to “repost” or “like,” potentially occupying their cognitive resources in working memory, and increasing extraneous cognitive load, resulting in poor reading performance by reducing their effective cognitive load allocated to reading tasks (Paas & Sweller, 2003).

Perceived cognitive load fully mediated between interference information and microblog information comprehension. This finding revealed that interference information did not directly affect understanding of Weibo messages, except when cognition overload was induced. Such a result may be explained two ways: One is that, in our study, we used only 10 pieces of interference information as experimental materials, which might be too few to cause participants’ cognition overload and therefore did not interfere with understanding target information. The research of Eppler and Mengis (2004) revealed that the relation between the quantity of interference information and accuracy of information comprehension appeared in an inverted U shape, which suggest that the participants’ information comprehension performance would decrease only when interference information reaches a certain amount. Second, when an individual used Weibo, whether viewing relevant or irrelevant information, people just browsed and skimmed, and did not perform deep processing of information. Thus, interference information would not affect comprehension of target information (Jiang et al., 2016). Hence, H2 and H3 were supported, that perceived cognitive load plays a mediating role in the relationships between the feedback function interface as well as interference information and comprehension performance for microblog information.

Eye-Movement on Features of Weibo

According to previous research, total fixation duration and total number of fixations were considered to be effective indicators of extraneous cognitive load (Yan & Bai, 2018). In the current study, the results of eye-movement further indicated that the feedback function and interference information of Weibo would increase extraneous cognitive load.

Analysis on eye movement data revealed that in the feedback group, the total fixation duration and the total number of fixations on Weibo information were significantly lower than those in the non-feedback group, while the total fixation duration and total number of fixations on the feedback function interface was significantly higher than those in the non-feedback group. This finding is in line with previous research which found that people often pay attention to webpage features and interface designs such as “repost” or “liking a post,” rather than the content of message (Wathen & Burkell, 2002). Our results indicate that the design of the feedback function on the Weibo interface has a negative effect on users’ understanding of target information, because it encourages users to respond quickly and take away the time required for processing and integrating target information.

In addition, eye-movement in the interference group revealed that the total fixation duration and the total number of fixations are not significantly different on target microblogs, while for the strong interference group, the total fixation duration and the total number of fixations on interference microblogs were significantly higher than those in the weak interference group. These findings suggest that on the one hand, interference microblogs increased extraneous cognitive load; however, on the other hand, it did not occupy cognitive resources which were allocated to target information. To some extent, this also has echoed the findings that the main effect of interference on information comprehension was not significant. A possible explanation for these results could be that we presented target information and interference information on different microblogs separately, while in a real microblogs context, the two types of information tend to appear on the same screen.

Limitations

The current study conducted an exploration on the relationship and mechanism of microblog features and information comprehension. Most of our hypotheses have been supported. Despite this, there were still some limitations. First of all, even though we applied a laboratory design which improved internal validity of this research, ecological validity was reduced at the same time. Although the experimental materials were from the Sina Weibo app, it was still different from a real setting. For example, we only examined the influence of “repost” and “like” as feedback functions on information comprehending performance, while taking no consideration for the “comment” feature; Second, only 10 irrelevant messages were used as interference information in our experiment, which was far less than when using Weibo in reality. This might be a possible reason for why the current study did not find a direct impact of interference information on understanding of Weibo information. Finally, a microblog information comprehension test was developed and, though the difficulty level was assessed, its reliability and validity need to be further evaluated.

Conclusions

To sum up, the current study could draw the following conclusions. The design of Weibo feedback function interface had direct and negative association with the accuracy of Weibo information comprehension, while no such direct link was found between interference information on Weibo and information comprehension. Both the Weibo feedback function interface and interference information increased users’ cognitive load, which further lead to poorer information comprehension.

Footnotes

Author Contributions

Xu, Ying Xu: Conceptualization, Formal analysis, Investigation, Methodology, Writing—original draft. Xie, Jia-Qiong Xie: Conceptualization, Formal analysis, Investigation, Methodology. Wang, Fu-Xing Wang: Conceptualization, Methodology. Wang, Jin-Liang Wang: Conceptualization, Funding acquisition, Resources, Supervision, Writing—review & editing. Rebecca L Monk: Writing—review & editing. James Gaskin: Writing—review & editing

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Chongqing Social Science Planning Fund [grant number 2021YC006].

ORCID iDs

James Gaskin

Jin-Liang Wang

Note

Appendix

Author Biographies

Ying Xu is a graduate psychology student at the Southwest University in China. Her research interest is online communication and individuals’ social and cognitive development. m15117767579@163.com

Jia-Qiong Xie is a Ph. D student at the School of Law in Chongqing University. Her research focuses on excessive social media use among adolescents. JiaQiongX@163.com

Fu-Xing Wang is a professor of developmental psychology in the School of Psychology at Central China Normal University and his research interest focuses on exploring the cognitive processes of reading using eye-tracking movement technology. fxwang@mail.ccnu.edu.cn

Rebecca L Monk is a reader of Department of Psychology in Edge Hill University and she focuses on exploring the impact of social and environmental contextual influences on substance use and related beliefs, with a particular focus on alcohol. monkre@edgehill.ac.uk

James Gaskin is a professor of Marriott School of Business in Brigham Young University and his research interests include human computer interaction (particularly augmented reality and wearables), human robot interaction, and many other things. james.eric.gaskin@gmail.com

Jin-Liang Wang is a professor of educational psychology in the School of Psychology at the Southwest University in China. His research investigates the antecedents of internet use as well as its influence on users’ social development and psychological well-being. wangjinliang09@gmail.com

References

Baumgartner

S. E.

Weeda

W. D.

van der Heijden

L. L.

Huizinga

(2014). The relationship between media multitasking and executive function in early adolescents. Journal of Early Adolescence, 34(8), 1120–1144. https://doi.org/10.1177/0272431614523133

Bergman

Beyth-Marom

Nachmias

(2006). The project fragmentation problem in personal information management. In: Proceedings of the SIGCHI conference on human factors in computing systems, Montréal, QC, Canada, April 22–27, 2006

Boyd

Golder

Lotan

(2010). Tweet, tweet, retweet: Conversational aspects of retweeting on twitter. In: Proceedings of the Annual Hawaii International Conference on System Sciences, Honolulu, HI, USA, 05-08 January 2010

Cain

M. S.

Leonard

J. A.

Gabrieli

J. D. E.

Finn

A. S.

(2016). Media multitasking in adolescence. Psychonomic Bulletin and Review, 23(6), 1932–1941. https://doi.org/10.3758/s13423-016-1036-3

Cardoso-Leite

Kludt

Vignola

W. J.

Green

C. S.

Bavelier

(2016). Technology consumption and cognitive control: Contrasting action video game experience with media multitasking. Attention, Perception, and Psychophysics, 78(1), 218–241. https://doi.org/10.3758/s13414-015-0988-0

Carr

(2010). The shallows: What the internet is doing to our brains. W. W. Norton & Company

Center C I N I (2020). The 46th China statistical report on Internet development. Center C I N I. http://www.cnnic.net.cn/hlwfzyj/hlwxzbg/hlwtjbg/202009/t20200929_71257.htm

de Jong

(2010). Cognitive load theory, educational research, and instructional design: Some food for thought. Instructional Science, 38(2), 105–134. https://doi.org/10.1007/s11251-009-9110-0

DeStefano

LeFevre

J. A.

(2007). Cognitive load in hypertext reading: A review. Computers in Human Behavior, 23(3), 1616–1641. https://doi.org/10.1016/j.chb.2005.08.012

10.

Eppler

M. J.

Mengis

(2004). The concept of information overload: A review of literature from organization science, accounting, marketing, MIS, and related disciplines. The Information Society, 20(5), 325–344. https://doi.org/10.1080/01972240490507974

11.

Erdfelder

Faul

Buchner

(1996). Gpower: A general power analysis program. Behavior Research Methods Instruments & Computers, 28(1), 1–11. https://doi.org/10.3758/bf03203630

12.

Hou

Huang

Liu

Zhang

Yang

Liu

Jin

Zhang

(2014). Is the excessive use of microblogs an internet addiction? Developing a scale for assessing the excessive use of microblogs in Chinese college students [Journal Article; research support, non-U.S. Gov't]. PLoS One, 9(11), Article e110960. https://doi.org/10.1371/journal.pone.0110960

13.

Jiang

T. L.

Hou

Y. B.

Wang

(2016). Does micro-blogging make us "shallow"? Sharing information online interferes with information comprehension. Computers in Human Behavior, 59(6), 210–214. https://doi.org/10.1016/j.chb.2016.02.008

14.

Jin

M. Z.

(2005). Comparative study on character of eye movements while reading for junior high school student with different reading comprehension. Journal of Zhejiang Normal University (Nat.Sci.), 28(4), 465–468

15.

Kaplan

A. M.

Haenlein

(2010). Users of the world, unite! the challenges and opportunities of social media. Business Horizons, 53(1), 59–68. https://doi.org/10.1016/j.bushor.2009.09.003

16.

Kaplan

A. M.

Haenlein

(2011). The early bird catches the news: Nine things you should know about micro-blogging. Business Horizons, 54(2), 105–113. https://doi.org/10.1016/j.bushor.2010.09.004

17.

Kirschner

P. A.

Sweller

Kirschner

Zambrano R

J. R.

(2018). From cognitive load theory to Collaborative cognitive load theory. International Journal of Computer-Supported Collaborative Learning, 13(2), 213–233. https://doi.org/10.1007/s11412-018-9277-y

18.

Mayer

R. E.

Bove

Bryman

Mars

Tapangco

(1996). When less is more: Meaningful learning from visual and verbal summaries of science textbook lessons. Journal of Educational Psychology, 88(1), 64–73. https://doi.org/10.1037/0022-0663.88.1.64

19.

Mayer

R. E.

Griffith

Jurkowitz

I. T. N.

Rothman

(2008). Increased interestingness of extraneous details in a multimedia science presentation leads to decreased learning. Journal of Experimental Psychology: Applied, 14(4), 329–339. https://doi.org/10.1037/a0013835

20.

MILLER

G. A.

(1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63(2), 81–97

21.

Paas

F. R. A.

Renkl

Sweller

2003). Cognitive load theory and instructional design: Recent developments. Educational Psychologist, 38(1), 1–4. https://doi.org/10.1207/s15326985ep3801_1

22.

Poltavski

Van Eck

Winger

W. A. T.

Honts

(2018). Using a polygraph system for evaluation of the social desirability response bias in self-report measures of aggression. Applied Psychophysiology and Biofeedback, 43(4), 309–318. https://doi.org/10.1007/s10484-018-9414-4

23.

Rayner

(1998). Eye movements in reading and information processing: 20 years of research. Psychological Bulletin, 124(3), 372–422. https://doi.org/10.1037/0033-2909.124.3.372

24.

Scott

G. G.

Hand

C. J.

(2016). Motivation determines Facebook viewing strategy: An eye movement analysis. Computers In Human Behavior, 56(1), 267–280. https://doi.org/10.1016/j.chb.2015.11.029

25.

Shenk

(1998). Data smog: Surviving the information glut. San Francisco: Harper

26.

Skulmowski

K. M.

(2022). Understanding cognitive load in Digital and online learning: A New Perspective on extraneous cognitive load. Educational Psychology Review, 34(1), 171–196. https://doi.org/10.1007/s10648-021-09624-7

27.

Sweller

(1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257–285. https://doi.org/10.1207/s15516709cog1202_4

28.

Sweller

(1994). Cognitive load theory, learning difficulty, and instructional design. Learning and Instruction, 4(4), 295–312. https://doi.org/10.1016/0959-4752(94)90003-5

29.

Sweller

(2010). Element interactivity and intrinsic, extraneous, and germane cognitive load. Educational Psychology Review, 22(2), 123–138. https://doi.org/10.1007/s10648-010-9128-5

30.

Sweller

Ayres

Kalyuga

(2011). Cognitive load theory. New York, NY: Springer

31.

Sweller

Chandler

(1994). Why some material is difficult to learn. Cognition and Instruction, 12(3), 185–233. https://doi.org/10.1207/s1532690xci1203_1

32.

Sweller

van Merriënboer

J. J. G.

Paas

(2019). Cognitive architecture and instructional design: 20 years later. Educational Psychology Review, 31(2), 261–292. https://doi.org/10.1007/s10648-019-09465-5

33.

Tsang

P. S.

Velazquez

V. L.

(1996). Diagnosticity and multidimensional subjective workload ratings. Ergonomics, 39(3), 358–381. https://doi.org/10.1080/00140139608964470

34.

Van Merriënboer

J. J. G.

Sweller

(2005). Cognitive load theory and complex learning: Recent developments and future directions. Educational Psychology Review, 17(2), 147–177

35.

Vraga

E. K.

Bode

Smithson

Troller-Renfree

(2019). Accidentally attentive: Comparing visual, close-ended, and open-ended measures of attention on social media. Computers in Human Behavior, 99(1), 235–244. https://doi.org/10.1016/j.chb.2019.05.017

36.

Wathen

C. N.

Burkell

(2002). Believe it or not: Factors influencing credibility on the Web. Journal of the American Society for Information Science and Technology, 53(2), 134–144. https://doi.org/10.1002/asi.10016

37.

Willard

Gramzow

R. H.

(2008). Exaggeration in memory: Systematic distortion of self-evaluative information under reduced accessibility. Journal of Experimental Social Psychology, 44(2), 246–259. https://doi.org/10.1016/j.jesp.2007.04.012

38.

M. L.

Wang

Zhao

Liang

(2017). Distraction or not? Investigating the relationship between mobile social network engagement and task performance. ■■■. (pp. 1–5)

39.

Yan

G. L.

Bai

X. J.

(2018). The foundation and application of eye movement analysis technique. Beijing Normal University Press

40.

Yan

G. L.

Xiong

L. P.

Zang

C. L.

L. L.

Cui

Bai

X. J.

(2013). Review of eye-movement measures in reading research. Advances in Psychological Science, 21(4), 589–605. https://doi.org/10.3724/sp.j.1042.2013.00589

41.

Zhao

Qin

(2017). Study on hypertext reading in students'autonomous learning. Journal of Language Teaching and Research, 8(1), 157–162. https://doi.org/10.17507/jltr.0801.19