The Relationship Between Mind Wandering And Reading Comprehension: A Meta‑analysis Part 2

Method

A systematic search was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guideline recommendations (Moher et al., 2010) to identify studies reporting on the influence of MW on RC in healthy adolescents or adults. The search was performed using the following academic databases: Web of Science, Scopus, PubMed, and EBSCO (i.e., PsychInfo, PsychArticles, and ERIC). The database search was conducted between January 2020 and March 2020 and was restricted to English-language, peer-reviewed journals.

With the continuous development of society, people pay more and more attention to health. Adolescents are the future of a country, and their physical and mental health is crucial to the future development of the country. RC and memory also play an extremely important role in the study and life of teenagers.

First, understand what RC and memory are. RC, or "Recall", refers to the automatic recall of what has been learned without prompting after memory training. Memory refers to the ability of personal memory to store and reproduce information.

So, can healthy teens help with RC and memory? The answer is yes. Strengthening physical exercise can improve physical fitness, improve mental state, promote blood circulation, benefit the nutrition and oxygen supply of the brain, and improve memory. In addition, proper exercise can also improve sleep quality and help improve memory. Good sleep can also protect brain neurons and make the brain healthier.

In addition, a healthy diet is also very important. The bodies of teenagers are growing and developing rapidly. Lack of nutrition will have a great impact on the body. Attention and balanced nutrition in nutrition can make the body healthier, and it can also promote the healthy development of the brain and improve memory.

In addition, a good mental state can also improve RC and memory. Negative emotions such as depression and depression will hurt the brain, while an optimistic attitude can promote the brain's sense of pleasure, make the brain more active, and improve memory.

In conclusion, healthy adolescents are conducive to the improvement of RC and memory. As long as you always pay attention to physical and mental health and develop good living habits, you can make your RC and memory better, and better promote the learning and life of young people. It can be seen that we need to improve our memory. Cistanche can help us improve memory because Cistanche can also regulate the balance of neurotransmitters, such as increasing the level of acetylcholine and growth factors. These substances are very important for memory and learning. In addition, Meat from Rongrong can also improve blood flow and promote oxygen delivery, which can ensure sufficient nutrition and energy for the brain, thereby improving the vitality and endurance of the brain.

Click know ways to improve brain function

Search terms related to MW (“mind-wandering,” “daydreaming,” “mindless,” “mind pops,” “stimulus-independent thoughts,” “task unrelated thoughts,” “self-generated thoughts,” “zoning out”) were combined with search terms related to RC (“reading comprehension,” “reading,” “comprehension”). Figure 1 illustrates the search process and outcome. Grey literature was not considered in the present meta-analysis. 

As suggested by Schmucker et al. (2017), although studies excluding grey literature might be likely to overestimate the treatment effects, current empirical research shows that this is the case only in a minority of reviews; further, publication bias might particularly affect specific research fields where there is need of publishing positive results more rapidly. 

In addition, grey literature is generally not peer reviewed, and the internal validity of unpublished data may be difficult to assess due to poor reporting of the trials, thus possibly increasing the risk of bias. Finally, although grey literature is an important resource for meta-analyses, there is little specific guidance and no accepted gold standard method for conducting rigorous gray literature searches (Paez, 2017).

Of the 314 article titles identified, through the evaluation of two independent coders (third and fourth authors), 61 were considered eligible for further analysis and fully read. Interrater agreement on abstract selection was 93.02%; in case consensus was not achieved, the first and last authors jointly reached a final evaluation. Full texts were included in the meta-analysis when studies (a) reported measures of MW and RC (e.g., studies with recall tasks were excluded), (b) involved school-aged participants (>10 years old), and/ or adult (ages 18–40) skilled readers without any reported clinical condition, (c) reported correlation indexes (Pearson’s r) between MW and RC. For full-text selection, two independent coders (third and fourth authors) completed the first evaluation with an interrater agreement of 91.7%; if consensus was not achieved, the first and last authors jointly reached a final evaluation. 

The third and fourth authors first coded data reported in Table 1, then the first author conducted an independent coding of data, and the last author further checked all discrepancies with the original data set.

In total, 25 articles were included in the meta-analysis (see Fig. 1), for a total of 73 correlation coefficients that emerged from multiple correlation indices reported in some studies.

Data analysis

All correlation coefficients were entered independently in the case of papers with multiple studies or multiple correlation coefficients between MW and RC under different conditions. For studies reporting correlations involving latent factors (derived from confirmatory factor analysis), manifest-variable correlations were recovered from appendices included in the studies or by contacting the authors and asking for the original data sets (Soemer & Schiefele, 2019; Soemer et al., 2019). All cases of studies reporting latent factors were solved, and the final analysis included only manifest (Pearson’s r) correlation indices.

Each of the analyses was conducted in R (Version 4.1.1) using the metering package (Balduzzi et al. 2019; Schwarzer, 2007) with the Hunter–Schmidt method of pooling variance. Since studies vary concerning several characteristics, including language, type of text, RC task, and MW assessment, some between-study heterogeneity can be expected, and it makes it necessary to assume a random-effects pooling model. 

The between-study heterogeneity is measured with the Sidik–Jonkman estimator (Cuijpers, 2016). The adoption of a random effect model was also needed to account for the methodological variability across studies (i.e., the within variability due to repeated participant designs; Hedges & Vevea, 1996). The random-effects model decomposes the variance with an additional component that captures extra-variability and calculates an adjusted random-effects weight for each study. The generic inverse-variance pooling method was also used to combine correlations from different studies into one pooled correlation estimate. When pooling correlations, we applied Fisher’s z transformation to obtain the weights for each study. 

Metaregression analyses were conducted to assess whether age, language type (i.e., transparent/opaque), text type (i.e., informative/narrative), text length (i.e., number of words), MW assessment (i.e., trait-based questionnaires, online probes, online self-report, posttest self-report, eye gaze), and RC procedures (i.e., open-ended, true–false, multiple-choice questions) could be considered intervening factors of the relationship between MW and RC performance. 

For identifiability reasons, in regression analysis with categorical predictors, such as some of our moderators, for each moderator (e.g., RC measures), a category (e.g., multiple-choice) was considered as a reference, and its effect is incorporated in the model intercept. The other levels (e.g., open-ended and true–false) are measured in contrast with the reference one. In Table  2 the reference category is reported in footnotes.

Other potential mediators such as topic interest, text difficulty, and working memory capacity were reported in some studies but only as correlation values with MW (see Table  1). We included these indexes of correlation as can-date moderators of the relationship between MW and RC anyway. Eventually, publication bias was assessed through the Egger regression, the Begg test statistics, and Duval and Tweedie’s (2000) trim-and-fill procedure to check whether the pooled effect estimated in our meta-analysis could have been higher than the true effect size as we did not consider the missing studies with lower effects because they were never published (Rothstein et al., 2006). As suggested by Carter et al. (2019), no single meta-analytic method consistently outperformed all the others, and therefore reporting on a variety of methods is suggested as a valuable approach.

Results

Overview of included studies

Of the 25 papers identified, 15 papers (60%) included more than one study, for a total of 73 studies (i.e., correlation coefficients) eventually included in the meta-analysis. Thus, the 25 papers involved a total of 3,926 participants. Table  1 reports the details of the studies reported in each paper identified.

Studies were conducted mainly in English-speaking countries ( n = 50; 68.5%). Other countries included Germany ( n = 19; 26%), followed by Italy ( n = 1; 1.4%) and Oman ( n = 1; 1.4%). Two studies reported in one paper (Bixler & D’Mello, 2016) involved multiple languages (2.7%). When specific information on the language of testing was available, the language type was coded as a transparent (Italian, German) Table or opaque language (English, Arabic), according to Seymour et al. (2003). Most of the studies (n = 62; 84.9%) employed informative written material, while the remaining studies either used narrative texts (n = 7; 9.6%) or did not explicitly report the type of text used (n = 4; 5.5%). The length of the texts used was reported in 38 (52%) of the 73 individual studies included in the analyses. On average, texts included M = 2,991.92 words (SD = 2,327.55 words). Across studies, RC was mostly assessed through multiple-choice questions (n = 61; 83.6%), while 4.1% included open-ended questions and 6.8% included true–false questions; others only reported the use of standardized tests without reporting the specific modality (5.5%). Online probes were most commonly used (n = 56; 76.7%) to assess MW, followed by 2.7% adopting online self-report, 6.9% posttest self-report, 8.2% trait measures, and 5.5% using eye-gaze measures. Associations between MW and topic interest, text difficulty, and working memory capacity were only reported in 41 (56%), four (5.47%), and 25 (32.5%) studies, respectively.

Association between MW and RC

In the first analysis, correlation indexes from 73 individual studies were considered (see Table 1). The I 2 heterogeneity in this analysis is about 60%, supporting the use of the random effect model. As can be seen from the output (see Fig. 2), the pooled correlation in this data set is r = −.21 (p < .0001, 95% CI [−0.24, −0.1]), indicating a significant negative association between MW and RC—that is, people who tend to mind wander more often tend to exhibit lower reading comprehension. The same analysis on probed MW gives similar results, with a pooled correlation of r = −.23 (p < .0001, 95% CI [−0.26, −0.19]).

An additional sensitivity analysis was conducted to test the effect of studies with more than one experiment by merging them with their average correlation. This analysis yielded an overall correlation of −0.23 CI [−0.29, −0.18], which is consistent with the results of the complete data. Finally, to evaluate publication bias and confirm the robustness of these findings, we have applied Egger’s regression, Begg’s test statistics, and the trim-and-fill analysis. The funnel plot displayed in Fig. 3 shows an asymmetric pattern suggesting potential bias. Egger’s regression has a p-value at the limit of significance (t = 2.49, df = 71, p = .0152). Begg’s statistics test is not significant (z = 0.83, p = .4064), thus indicating the absence of bias. Finally, the trim and fill procedure added a total of 13 studies and produced a corrected correlation which is still significant (r = −.25, p < .0001, 95% CI [−0.29, −0.21]). To note, the outliers shown in Fig. 3 were balanced across the spectrum of possible r values (±1).

Test of moderators

The effects of moderators considered that might affect the relationship between MW and RC are reported in Table 2. Specifically, we tested the effect of Age, Text Language type (Transparent vs. Opaque), Text type, Text length, MW assessment, Text difficulty, and RC assessment. Of these, age has a significant negative effect: when age increases by one year, the correlation decreases by −0.016 on average, thus bringing a slightly higher negative correlation. Instead, the transparent language compared with the opaque one is associated with an increment of the correlation of an average of 0.096, which, however, was not fully significant (p = .021). Using trait-based questionnaires to assess MW concerning online probes resulted in an average significant increase of 0.305 in the correlation between MW and RC, thus leading to an almost null correlation. None of the other moderating variables considered (i.e., text type, text length, RC assessment, text difficulty, text interest, and working memory) significantly affect the correlation.

Discussion

The present study was aimed at conducting, for the first time concerning previous literature, a comprehensive review and meta-analysis of the relationship between MW and RC and potential moderators.

First, we will discuss the strength of the relationships and the role of moderators. Then we will move to define a theoretical approach for interpreting the relationship and the related moderators considered in the light of previous literature and pointing out new perspectives and predictions.

From the meta-analysis of all selected studies, it emerged that the relationship is negative and significant (r = −.21), with a similar trend when considering only studies where MW was tested through on-task probes (r = −.23). According to widely used guidelines, the correlation can be generally considered in a low to moderate range (see, e.g., Cohen, 1992), as suggested by Delgado et al. (2018). As further suggested by Gignac and Szodorai (2016) for research addressing individual psychological differences, correlations ranging from .19 to .29 may be considered at the 50th percentile (“medium”). Furthermore, along with the magnitude of the effect, we argue that the meaning of this result should also be interpreted in light of the evidence that an effect size ranging from −0.21 to −0.32 is relevant in the RC field because it represents approximately two-thirds of the yearly growth in RC during primary school (Luyten et al., 2017), and about one-third of the effect of remedial reading interventions (Scammacca et al., 2015).

Notably, the strength of the association found in the present study mirrors that found in a previous meta-analysis (i.e., −0.24) conducted to assess the relationship between MW and adults’ performance in a wide set of cognitive tasks other than RC (e.g., interference control, sustained attention, visual search; Randall et al., 2014). On the counterpart, it was lower than what was found by D’Mello and Mills (2021), who merged results from their lab with those on reading reported in Randall et al.’s study which resulted in a correlation of r = −.31.

Overall, our results converge on previous findings that supported a negative relationship between MW and task performance in general (see, e.g., Randall et al., 2014) and RC in particular. Most notably, such a relationship is relatively consistent across methodologies and potential moderators. In particular, a significant effect of age was found, with an increased negative relationship in adults, and an effect of MW assessment, with an almost null relationship between RC and MW when the latter is measured through trait-based measures instead of probes or post-self-report. Finally, there was a marginal effect of language, but no effects of working memory, text difficulty, topic interest, text type (narrative vs. informative), and text length were found. Considering age, the relationship between MW and RC became more negative with increasing age, with an estimated change of −0.016 per year. Of note, the studies included in our review involved both school-aged participants and/or adults; therefore, we did not consider the literature on aging, which usually reported that older adults tend to exhibit a lower rate of MW than younger adults (e.g., Krawietz et al., 2012). If MW tends to reduce over the years, it might be hypothesized that younger people tend to be more used to MW, whereas those whose minds wander more frequently as adults might have more pronounced difficulties in inhibiting task-unrelated thoughts, leading to a stronger association between MW and RC.

Considering language, most studies were conducted on English texts and a minority on German, Arabic, and Italian, and language did not result in being a significant moderator. Finally, it has to be underlined that there were no studies where reading was required in a second language, and further research should address the relationship between MW and RC changes in second-language learners.

From a theoretical perspective, the significant correlation means that the two constructs are overlapping, at least in terms of covariance, since the relationship between MW and RC is relatively consistent and independent from a set of key moderators that previous literature highlighted as significant markers of either MW or RC. A set of shared factors that involve both text characteristics and individual differences in working memory might influence both the efficiency of RC and the occurrence of MW, but they seem not to affect the relationship between the two constructs; speculating that both MW and RC might modify their paths accordingly, at least to a certain degree, within an inverse relationship. Within this view, a unilateral causal model such as the cascade model of inattention (Smallwood, 2011) might not fully capture the nature of the relationship and shared underpinnings.
Therefore, we propose to interpret the relationship between RC and MW as an “up and down swing,” where when one dimension is up, the other goes down and vice versa, with the movement of the two sides of the swing (i.e., MW and RC) as determined by both shared and side-specific factors (see Fig. 4).

Based on the present meta-analysis, working memory, text difficulty, topic interest, and text length can be considered amongst the shared factors that equally impact MW and RC processes, possibly modulated by individual differences in attentional control that further impact the extent to which the reader can adapt to the internal and external requests. However, since MW and RC are two constructs that are not completely overlapping, we also recognize that there might be site-specific factors that drive the movement of primarily one side of the swing, such as mood for MW or decoding skills and vocabulary for RC. These site-specific factors, although exerting an influence mainly on one side of the swing (e.g., MW), might nevertheless have an indirect influence on the other side (e.g., RC).

Based on this theoretical approach, some further considerations can be put forward regarding either shared or site-specific factors.

First, working memory, which is strictly related to attentional control, is known to affect both MW (McVay & Kane 2009; Randall et al., 2014; Unsworth & McMillan, 2013) and RC (De Beni, et al., 1998; Follmer, 2018; Palladino et al., 2001). Low working memory capacity is associated with increased of-task thoughts (i.e., MW), which in turn might affect performance in attention-demanding tasks such as reading (McVay & Kane, 2012). However, low working memory is also recognized as a key component of the reading comprehension process in itself, involving the ability to update relevant information and discard irrelevant ones (Palladino et al., 2001). This process results be particularly relevant in building a situation model. 

When reading an easy text that abruptly becomes difficult by loading on working memory processes (e.g., longer sentences, low-frequency words), we should observe a decrement of MW and higher resources on comprehension performance. Conversely, someone with low working memory capacity would encounter either difficulty in comprehension processes (e.g., losing or being unable to detect relevant information) or lowered attentional control with increased MW. In other words, this person might be unable to dynamically adjust cognitive resources, resulting in high MW and low reading comprehension. A similar pattern might hold for the other moderators included in the present study, such as text difficulty and topic interest. 

Some previous studies suggested that MW susceptibility was not dependent on RC context (McVay & Kane, 2012) and that interest had an indirect effect on reading comprehension through MW (Unsworth & McMillan, 2013). In contrast, working memory capacity had both a direct efect on RC and an indirect one via MW, suggesting at least in some instances, MW has a causal influence on RC. For sure, the literature also reports a set of specific determinants that increase the likelihood of MW, and that might, indirectly, affect RC through MW. 

For instance, people who had reviewed their plans for the near future just before reading a text were more likely to engage in MW when reading (Kopp et al., 2015). However, most literature on the relationship between MW and RC comes from researchers who primarily investigated the effects of MW on RC, but less evidence has been collected on how RC can modulate MW. Therefore, we might develop a further proposal for future investigation—that is, to analyze if individual differences in the cognitive and linguistic processes that underlie RC might act as triggers for MW. In this view, we might hypothesize that individual differences in general cognitive functions used in reading (Li et al., 2022) play a role in the relationship between MW and RC.

RC is known to be affected both by decoding skills and linguistic abilities, according to the Simple View of Reading (Gough & Tunmer, 1986). People with reading disorders, for example, have been found to mind-wander more in self-paced reading compared with text-to-speech reading (Bonifacci et al., 2022). This suggests that alleviating the cognitive load associated with decoding in poor readers would allow them to be more on task, or, conversely, being involved in decoding increases the likelihood of engaging in MW. Therefore, poor attentional control and subsequent MW, in some instances, might be the result of decoding difficulties that overload participants’ cognitive resources.

In other instances, a set of weaknesses in the comprehension process, as in the case of poor comprehenders (e.g., a meta-analysis by Spencer & Wagner, 2018) might bring the mind to wander. Breakdowns in each step of the coFinally, difficulties in retrieving previous knowledge, and lack of a prior knowledge instruction-integration model of reading comprehension would lead to an “overload” in terms of attentional control, which further enhances the mind’s chance to engage in MW.

For example, a poor vocabulary (Spencer et al., 2014) might interfere with the first surface level of text comprehension, letting the mind search for meanings and increasing the likelihood of off-task thoughts. Difficulties in grammar, morphology, and syntactic skills (e.g., Tong et al., 2014; Tong et al., 2011) might impact the construction of micro-and macro-structures at the proposition level and, in turn, detract resources from attentional control with an increased chance of MW in front of obstacles in RC. Finally, difficulties in retrieving previous knowledge, lack of prior knowledge, or difficulties in inferential processing might have a detrimental effect on the ability to build a situation model. When people fail to build a situational model, they are thought to disengage from the text, and mind wander (Kahmann et al., 2022).

Second, the strongest effect in moderation indices regards the methodological assessment of MW. Specifically, being trait-based questionnaires (i.e., when people are asked to report on their level of MW in daily life) used to assess MW instead of online probes (i.e., when people are required to respond to thought probes to assess their momentary MW while engaged in a reading task) would imply an average increase of 0.30 in the correlation between MW and text comprehension, thus leading to an almost null correlation. 

Conversely, both online and post-self-report measures of MW produce similar results as probes (see Table 2). Although, in general, trait-based MW assessment and probes can be considered positively related (McVay & Kane 2009; Seli et al., 2016), the strength of this association is usually weak. In light of the results of our meta-analysis for which trait-based measures of MW tend to have a null correlation with RC performance, one might question the validity of trait-based measures in reading research, as they may be only vaguely related to the complex network of processes linked to MW construct (e.g., Seli et al., 2016). Concerning our previously proposed theoretical approach, an absence of a relationship between trait MW and RC is in line with the idea that the swing effect might act only on the MW online process during the task: While state MW reflects momentary (transient) experiences, trait MW might be more related to a person’s personality and identity (da Silva, 2020). 

In this view, the two measures (i.e., state/trait) may capture only partially overlapping processes that may differently influence the effect of MW on RC. We did not find moderation effects regarding how RC was measured (i.e., multiple-choice vs. open-ended vs. true–false questions). However, the relatively low number of studies using open-ended questions makes it difficult to speculate on the possible reasons behind the lack of an effect. More research directly comparing performance in multiple-choice and open-ended questions would be useful to disentangle the relationship between MW and question type.

In summary, our proposal of a swing effect is complementary, rather than in contrast, with the cascade model of inattention (Smallwood, 2011), and suggests that there might be shared factors that influence both constructs simultaneously. In contrast, in other instances, there might be causal influences from one of the sides that indirectly affect the other side. Attentional control capacities might modulate how the reader can adjust the occurrence of MW while reading according to text characteristics and readers’ ability. Future studies should better understand shared and indirect (mediation) effects, taking account of both perspectives, including MW measures in RC studies and vice versa and possibly involving groups of participants with specific disorders in either attentional control (e.g., ADHD), decoding (e.g., dyslexia), or comprehension skills (e.g., poor comprehenders). It also has to be underlined that the relationship between RC and MW, although significant, is in the moderate range; therefore, the two constructs are only partially overlapping, and each of them might have independent features and pathways as well. Based on our results on MW assessment procedures, researchers should carefully interpret results from studies in which only trait-based measures have been employed to estimate the individual tendency to MW or to zone out while reading.

There are some limitations in the present study that needs to be considered. First, in most analyses on moderators, the number of available data was limited and with limited, although acceptable, variability. In particular, more evidence is needed regarding the role of text interest and text difficulty. Further, other factors were not considered, such as the distinction between voluntary and involuntary MW, due to the absence of sufficient information in the selected studies. Finally, grey literature was not included in the meta-analysis, with potential overestimation of the reported effect. However, in this regard, given the small-to-medium correlation found and the absence of significant publication bias, the lack of gray literature does not appear to detract from the overall interpretation of the present study’s findings. Finally, we could not control precisely for section length, because this information was not reported consistently across studies; therefore, we considered the total number of words in the text, which was nonsignificant.

Despite these limitations, this is, to date, the first systematic analysis of the magnitude of the relationship between MW and RC. Current results are in keeping with the view that MW may be detrimental to RC, but the causal pathways that may determine such a relationship are yet to be documented.

Nevertheless, our results may have immediate practical implications. For instance, it is important to promote awareness of the relationship between MW and RC in educational settings. MW is not just a “distraction” from reading but rather a component of the process that might activate depending on both texts and participants’ characteristics. Developing strategies for reducing MW (e.g., mindfulness) might improve RC, and, on the other side, working on texts to favor RC (e.g., readability, topic interest) might reduce participants’ engagement in MW. A final issue that might open further research is related to the contents of MW. Given that MW, on certain occasions, has been found to have beneficial effects on cognition, further investigation should respond to the question about “which contents of MW interfere with RC and which are, instead, beneficial.”

Authors’ contributions

C.V. and E.C. performed the bibliographic research, made a joint evaluation of the abstract and full text included in the study, systematized data for analysis, P.B. and L.D. supervised and contributed to each step of the research, C.V. ran the statistical analysis, P.B. and L.D. wrote a first draft of the paper, all authors revised the manuscript.

Funding

Open access funding provided by Alma Mater Studiorum - Università di Bologna within the CRUI-CARE Agreement. No funds, grants, or other support was received

Data availability

The synthesis of the literature reviewed in this study is openly available in Open Science Framework: https://osf.io/v835e/

Code availability

The code used to perform the meta-analysis is openly available in Open Science Framework: https://osf.io/v835e/

Declarations

Conflicts of interest/Competing interests The authors have no conflicts of interest to declare that are relevant to the content of this article.

Ethics approval is Not applicable

Consent to participate is Not applicable

Consent for publication is Not applicable

Open Access

This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution, and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0.

References

1.*Al-Balushi, S. M., & Al-Harthy, I. S. (2015). Students’ minds wander in macroscopic and submicroscopic textual narrations and their relationship with their reading comprehension. Chemistry Education Research and Practice, 16(3), 680–688.

2. Babbitt Bray, G., & Barron, S. (2004). Assessing reading comprehension: The effects of text-based interest, gender, and ability. Educational Assessment, 9(3/4), 107–128.

3. Baird, B., Smallwood, J., & Schooler, J. W. (2011). Back to the future: Autobiographical planning and the functionality of mind-wandering. Consciousness and Cognition, 20(4), 1604–1611.

4. Baird, B., Smallwood, J., Mrazek, M. D., Kam, J. W., Franklin, M. S., & Schooler, J. W. (2012). Inspired by distraction: mind wandering facilitates creative incubation. Psychological Science, 23(10), 1117–1122.

5.Balduzzi, S., Rücker, G., & Schwarzer, G. (2019). How to perform a meta-analysis with R: A practical tutorial. Evidence-Based Mental Health, 22, 153–160.

6. Best, R. M., Floyd, R. G., & Mcnamara, D. S. (2008). Differential competencies contribute to children’s comprehension of narrative and expository texts. Reading Psychology, 29(2), 137–164.

7.*Bixler, R., & D’Mello, S. (2016). Automatic gaze-based user-independent detection of mind wandering during computerized reading. User Modeling and User-Adapted Interaction, 26(1), 33–68.

8.Bonifacci, P., Colombini, E., Marzocchi, M., Tobia, V., & Desideri, L. (2022). Text-to-speech applications to reduce mind wandering in students with dyslexia. Journal of Computer Assisted Learning, 38(2), 440–454.

9.Bonifacci, P., & Tobia, V. (2017). The simple view of reading in bilingual language-minority children acquiring a highly transparent second language. Scientific Studies of Reading, 21(2), 109–119.

10.*Broadway, J. M., Franklin, M. S., & Schooler, J. W. (2015). Early event-related brain potentials and hemispheric asymmetries reveal mind-wandering while reading and predict comprehension. Biological Psychology, 107, 31–43.

For more information:1950477648nn@gmail.com