Rest breaks aid directed attention and learning

ABSTRACT

Objective

When students learn or solve problems, attentional resources are depleted; rest breaks may restore cognitive functioning in support of learning. Research framed by attention restoration theory holds that exposure to natural environments may be another means to restore attentional resources. The study investigated the effects of alternative rest break formats on learning a challenging mental mathematics strategy.

Method

Students first completed a series of timed arithmetic tests expected to deplete attentional resources. Students in the control condition proceeded directly onto a mental mathematics lesson, while students in the unstructured rest and nature-based rest conditions took a 5-min break before the lesson. All students then completed a self-reported questionnaire on directed attention levels during the lesson, then completed a problem-solving post-test.

Results

The unstructured rest condition reported higher levels of directed attention during the lesson than the control condition; no other comparisons were statistically significant. The unstructured rest condition solved more post-test problems than the control condition, and the nature-based rest condition also solved more problems than the control condition. The post-test score difference between the two rest conditions was not statistically significant.

Conclusions

The study provided clearer evidence for the general benefits of rest than for the additional benefits of nature-based rest.

KEY POINTS

What is already known about this topic:

• (1) Attentional resources depleted by challenging tasks can be restored by unstructured rest breaks.

• (2) Rests based on exposure to nature may also restore attentional resources.

• (3) Both actual and video-based nature exposures have restorative effects.

What this topic adds:

• (1) This study compares the effects of unstructured and video-based nature rest on learning a complex cognitive skill.

• (2) Both unstructured and nature-based rest breaks enhanced learning.

• (3) Instructional designers should plan for rest breaks in lessons on complex topics.

KEYWORDS:

• Cognitive resource theory
• attention restoration theory
• nature exposure
• worked examples
• directed attention
• problem-solving

Introduction

One of the ongoing challenges for students of all ages is sustaining attention to the complex task of learning (Shell et al., 2010). To support this goal, simple “brain breaks” or “focused attention practices” are making their way into classrooms, as they are argued to restore attentional resources and assist in emotional regulation (Desautels, 2019). While there is a sound empirical basis for the effects of short breaks on laboratory-based measures of attention, there is less experimental evidence for the effect of short breaks on learning itself. The present study draws on cognitive resource theory (Flanagan & Nathan-Roberts, 2019) and attention restoration theory (R. Kaplan & Kaplan, 1989) to test the effects of alternative brief rest break formats on learning a complex cognitive skill.

Rest breaks in applied settings

According to cognitive resource theory (Flanagan & Nathan-Roberts, 2019), “humans have a limited but renewable source of cognitive resources that is slowly depleted as we perform cognitive tasks” (p.1639); as cognitive resources are depleted, performance on tasks requiring those resource declines. Research in applied settings has often focused on vigilance task performance, with a decline in vigilance resulting in mind wandering, fatigue, and ultimately cognitive resource depletion (Warm et al., 2008). Rest breaks have been found to reduce the vigilance decrement (Helton & Russell, 2015; Ross et al., 2014) as well as reducing subjective fatigue and increasing vigour for both laboratory tasks (Finkbeiner et al., 2016; Gilsoul et al., 2021) and in work settings (Hunter & Wu, 2016; Zacher et al., 2014). Recent neuroscientific findings also suggest that breaks of short duration are effective in consolidating declarative and procedural knowledge (Brokaw et al., 2016; Bönstrup et al., 2019; Wamsley, 2019).

Educational and work settings share a concern with sustained on-task attention as a foundation for learning (Cowan, 1995; Shell et al., 2010). Indeed, long-standing criticisms of lesson formats, such as lectures, have noted the difficulty students experience in concentrating for more than 10–30 min (Frederick, 1986; MacManaway, 1970; Stuart & Rutherford, 1978). Such criticisms have led to calls from influential higher education practice guides such as Bligh (2002) and McKeachie and Svinicki (2013) to incorporate student-based activities (e.g., buzz groups and think-pair-share) and rest breaks into lectures. Blasche et al. (2018) tested effects of alternative forms of rest breaks during a university lecture, finding clear effects on self-reported vigour and fatigue compared to a no-break condition; however, the study did not investigate the effects on what students learned from the lecture. In a recent large multi-country study, Mok et al. (2020) investigated the impact of 3–5-min physical activity breaks during school lessons. Across a 4-month intervention, elementary school students who engaged in such breaks twice a day, each school day, reported a range of improved attitudes to physical activity, but the impacts of student learning were not investigated.

In contrast to active breaks during lessons, short mindfulness practices, including breath awareness and body scans, are increasingly used in schools as part of broader interventions based on findings from positive psychology (Shankland & Rosset, 2017). Mindfulness protocols typically work by encouraging students to direct their attention to the present moment. As part of this process, focused attention is often taught by encouraging students to pay attention to the breath coming in through the nostrils, being passed through the body, and then being released. As an example of such a school-based intervention, the Mindfulness Education programme includes mindful attention training exercises that are practiced in classrooms for at least 3 min three times a day (Schonert-Reichl & Lawlor, 2010).

In the field of musical skill learning, Cash and colleagues (Cash, 2009; Duke et al., 2009; Simmons et al., 2019) have investigated the degree to which musical skill performance is advantaged by short break intervals interposed at different points in practice sessions; for example, Cash (2009) found nonmusicians who rested for 5 min in either the early or late stages of practice showed large gains in performance, with participants in the early rest condition continuing to improve across training. Investigating skill development in video game play, Johanson et al. (2019) compared a no-rest condition against five break lengths (no break, 2 min, 5 min, 10 min and 1 day) interposed through 5-min play sessions, finding that all rest conditions improved more during game play than the continuous play condition. Taken together, these results generally support the use of rest breaks in educational settings, but also point to a lack of research on the effects of such breaks on cognitive skill acquisition. A qualitatively different form of rest break to both of the above forms of rest, which may also have the potential to assist learning, is based on exposure to nature, which we now discuss.

Attention restoration theory and the restorative effects of nature

Attention restoration theory (ART; R. Kaplan & Kaplan, 1989) argues that exposure to natural environments may reduce attentional fatigue. Different attentional processes are proposed in ART to identify the specific effect exposure to nature has on restoring attention. S. Kaplan and Berman (2010) differentiate between two types of attention based upon the effort required for their use. The first type, involuntary attention, occurs when attention is directed towards stimuli that have a directly exciting or engaging quality and thus requires no effort. In contrast, directed attention is used when stimuli that are not particularly interesting require extensive effort to maintain attention. Directed attention fatigue occurs when a person must focus on a particular stimulus and allocate increasing amounts of cognitive effort to suppress other potentially distracting stimuli. According to ART, exposure to and immersion in natural environments may act to restore the capacity for directed attention as a result of psychological distance from everyday concerns (sense of “being away”), along with unforced, interest-motivated attention (“fascination”), and reinforced by an environment of wide scope (“extent”) (Anderson et al., 2017).

Attention restoration theory’s notion of directed attention should be of particular interest to educators, as it is held to support executive functioning and self-regulation processes in cognition (Pearson & Craig, 2014). In a review of 14 studies involving students across a range of educational levels, Mason et al. (2021) found that in 12 of the 14 studies students benefited from nature exposure (ranging from 10 to 90 min) through restoration of directed attention. However, Mason et al. noted a need for future research to investigate the impacts of nature exposure on educationally relevant variates such as text comprehension or calculation alongside the attention and memory tasks typically used in basic research in this field. In addition, the effects of breaks shorter than 10 min have not yet been investigated.

As people’s daily lives are increasingly constructed around urbanised living and, in some cases, geographic location restricts access to natural environments, “virtual” nature approaches might provide an effective and accessible alternative to physical nature exposure. The mass migration of online learning in countries around the world during the COVID − 19 pandemic, including lockdowns further restricting access to nature for many, has sharpened the need to identify viable alternatives for nature exposure. Participating in simulated nature breaks in the form of video can provide an immersive and accessible rest experience, with distinct psychological benefits compared to less immersive methods, e.g., still images of natural settings. Movement and audio embedded into videos, presented on a computer screen or in virtual reality, should allow for a superior sensation of immersion and escape compared to still-photos to promote attention restoration (Anderson et al., 2017; Kjellgren & Buhrkall, 2010; Moreno et al., 2018; Snell et al., 2019).

Taken together, the theory and research reviewed above suggest that when directed attention is depleted, attention restoration through a short break should improve subsequent learning. However, it is less clear from the literature whether a nature-based break should be more effective than an unstructured break with no particular design elements; simply taking a short break may provide sufficient opportunity for directed attention to recover after depletion. Finkbeiner et al. (2016) compared groups taking different forms of 45-s breaks (watching YouTube videos with video scenes involving dogs or robots playing in various environments, versus an on-screen countdown) to a no-break condition on performance on the vigilance decrement task. Groups taking breaks performed similarly on the vigilance task, and all were better than the no-break condition; however, participants exposed to dog videos reported more positive moods. Nonetheless, in the present study, we hypothesised that rest involving nature exposure would generate larger effects on directed attention and subsequent learning than an unstructured break, because the nature break video’s design was more consistent with the central tenets of ART than those used by Finkbeiner et al., as well as being longer in duration.

The present study

This study examines the effects of computer-based rest-break interventions in an educational context. We hypothesise that, following an intellectually demanding problem-solving task expected to deplete attentional resources:

Hypothesis 1:

Participants’ reports of directed attention will follow a positive gradient across conditions. Specifically, participants in the No Rest group will report lower directed attention than during the lesson than those in the Unstructured Rest group (H1a) and those in the Nature Video Rest group (H1b); in addition, participants in the Unstructured Rest group will report lower directed attention than the Nature Video Rest group (H1c).

Hypothesis 2:

Participants’ post-test problem-solving performance will follow a positive gradient across conditions. Specifically, participants in the No Rest group will solve fewer problems than those in the Unstructured Rest group (H2a) and those in the Nature Video Rest group (H2b); in addition, participants in the Unstructured Rest group will solve fewer problems than those in the Nature Video Rest group (H2c).

Method

Participants

Seventy-two university students (42 females, 30 males) aged between 18 and 25 years old (M = 21.56, SD = 1.51) voluntarily participated in this experimental study. Participants were recruited via Facebook posts to University of Sydney degree programme groups to which the second and third authors belonged, as well as email invitations to peers within the desired 18–28 age range. Each participant was randomly and evenly assigned to either the No Rest, Unstructured Rest, or Nature-Based Rest conditions. The initial planned sample was 30 participants per condition, which would provide sufficient power (80–95%) to detect a standardised mean difference between conditions of 0.65 and above based on a one-tailed test. This was considered reasonable given “laboratory” studies in education tend to generate larger effects than field studies (Cheung & Slavin, 2016). The data collection was completed with 24 participants per condition due to the deadlines of the present study as an Honours project. With this sample size, a post hoc power analysis indicated there was sufficient power – ranging from 80% to 95% – to detect true effects between 0.73 and 0.96. The study received ethics clearance by the University of Sydney Human Research Ethics Committee (HREC protocol 2019/022) under Australia’s National Statement on Ethical Conduct in Human Research and was pre-registered through aspredicted.org (#20467).

Materials and procedure

The study material consisted of a 28-page paper booklet, divided into multiple phases. Mathematics was used as the lesson medium. Participants completed the study individually in a quiet meeting room for approximately 50 min. For consistency, verbal instructions were read off a script. The study began with a short demographics questionnaire, followed by a pre-test on mental mathematics. As well as providing an indication of group equivalence prior to the intervention, the pre-test used in this phase was designed to deplete attentional resources, given that mathematical problem-solving is known for its high cognitive demands (Ashcraft & Kirk, 2001; Swanson & Beebe-Frankenberger, 2004). The use of the pre-test for this latter purpose is consistent with research on mental fatigue following prolonged exposure to a cognitively challenging task (e.g., Lorist et al., 2005). The participants then engaged in their randomly assigned rest condition. Following this, participants took part in a self-study lesson entailing a mental strategy to solve 2-digit multiplication. After the lesson, participants then completed a self-reported survey on concentration levels during the lesson, as a measure of directed attention. The study concluded with a post-test on the materials studied in the self-study lesson. Upon completion of the study, the participants were thanked for their time and debriefed.

Phase 1: pre-lesson

Participants were given 20 min to complete a series of four pre-tests (5 min per test) adapted from Ekstrom et al. (1976) number facility tasks. Each test assessed one or more arithmetic skills including addition, subtraction, division and multiplication. Each test began with a page of test instructions accompanied by four practice problems (1 min to complete), followed by a series of test questions that were presented in two parts, each marked correct or incorrect. Following Ekstrom et al.’s instructions for these tests, participants had 2 min to complete each part of the test. Participants were required to solve the problems mentally, i.e., without writing out steps, and write the final answer for each question only in the spaces provided. Each of the four tests had 120 items; thus, the total pre-test score was a mark out of 480.

Phase 2: rest interventions

The form of rest participants received constituted the independent variable of this study. Participants were randomly assigned to 1 of 3 conditions, as follows.

No rest

Once participants completed phase 1, participants in this condition immediately moved on to phase 3. Participants did not receive a rest break of any form. This condition therefore functioned as the control group for this study.

Unstructured rest

Upon completion of phase 1, the first experimental group remained quietly seated for 5 min. Based on the unstructured rest condition used by Helton and Russell (2015), every 19.7 s, participants were informed of the time remaining via a 23-inch desktop monitor positioned in front of the participant.

Nature-based rest

Experimental group 2 received a 5-min rest in the form of viewing a nature video. The video was presented to the participants via a 23-inch desktop monitor with audio. The video was taken from McAllister et al. (2017) study, which is available on YouTube: https://www.youtube.com/watch?v=qHZ3rV6TzMs. The video displays wild nature featuring rainforest scenes located in Mount Glorious, Brisbane, Australia. A variety of native tree and epiphyte species were presented over 4–5 horizontally panned scenes, with birdsong audio embedded. The only evidence of human activity was gravel walking tracks.

Phase 3: lesson phase

The lesson phase involved a short lesson on a mental mathematics strategy to solve 2-digit multiplication, and a concentration self-report.

Mental mathematics lesson

This phase involved a 10-min short self-study lesson on a mental strategy to solve 2-digit multiplication. The contents of the lesson included a page of instructions and four worked problems adapted from Ginns et al. (2019). After reading the instructions, participants were given 2 min to study the worked example on mental 2-digit multiplication. The study problems were provided in colour and presented in a “step by step” format. The difficulty of the worked examples increased over the progression of the lesson. Once the 2 min had elapsed, participants then attempted to solve a practice problem using the strategy they had just learnt. The practice problems reflected the difficulty of the worked problem previously studied. Participants were given 20 s to attempt the problem. If an incorrect solution was given, participants were encouraged to try again until a correct solution was given or 20 s had elapsed. A correct solution was awarded a score of 1; if a correct solution was not given, a score of 0 was awarded and the correct solution was demonstrated to the participant using the strategy.

Directed attention self-report

Immediately after the mathematics lesson, participants completed the “concentration” self-report scale taken from the Dundee State Stress Questionnaire (Matthews et al., 2002), as a measure of directed attention. Participants responded to each of seven items (“My attention was directed towards things other than the lesson”; “I found physical sensations such as muscular tension distracting”; “My performance was impaired by thoughts irrelevant to the lesson”; “I had too much to think about to be able to concentrate on the lesson”; “I found it hard to maintain my concentration for more than a short time”; “My mind wandered a great deal”; “My thoughts were confused and difficult to control”) using a Likert scale of 0 (not at all) to 4 (extremely), in relation to their experience during the mathematics lesson. Given the negative valence of items, participants’ responses were reverse-coded prior to statistical analysis so that a positive score reflected higher self-reported directed attention during the lesson. Construct validity of the scale was established by testing a congeneric (single factor) model of the data using confirmatory factor analysis through jamovi 9.5.17 (The jamovi project, 2019). When a good model fit is achieved for a congeneric model, this indicates that all indicators are suitable indicators of the hypothesised latent trait (Holmes-Smith & Rowe, 1994) – in this case, “Directed Attention”. The model had a good fit to the data, χ² = 18.50, df = 14, p = 0.141, CFI = 0.98, RMSEA = 0.067; reliability as estimated using McDonald’s (1999) omega was 0.87.

Phase 4: post-lesson test

Phase 4 immediately followed the self-report. It consisted of one page of instructions and a test on the mental 2-digit multiplication strategy learnt in phase 2. The post-test was presented in two sections. The first section was comprised of 10 problems followed by 10 slightly harder problems, totalling 20 assessable items. After reading the instructions, participants were allocated 20 s to mentally complete each question and write the final answer in the spaces provided. Parallel to phase 2, participants were able to make multiple attempts within 20 s. After every attempt, participants were informed whether their answer was correct and to move on to the next problem or to try the problem again until the 20 s elapsed. Again, if the participants correctly solved the problem within 20 s, they were awarded a score of 1. If the problem was not correctly solved within 20 s, a score of 0 was awarded.

Data analysis

Data screening consisted of calculation of standardised scores on each variate. Standardised scores greater than ± 3 standard deviations would be classified as potential outliers. Using this decision rule, one outlier (z = 3.22) was apparent on the pre-test, and one outlier (z = 4.23) was apparent on the concentration self-report. To test the assumption of normally distributed scores in each condition, the Anderson–Darling test of normality was used, with values over 0.20 indicating sample distributions were non-normally distributed (Keselman et al., 2013). Levene’s test was used to test the homogeneity of variance assumption across experimental conditions. While Levene’s test was not statistically significant for any of the variates under analysis, results of both outlier analysis and Anderson–Darling tests indicated substantial deviations from normality across variates (see Table 1). Subsequently, tests of hypotheses utilised non-parametric rather than parametric statistical analyses.

Table 1. Means (M), Standard Deviations (SD) and Anderson–Darling Test p-Value (A-D) for Total Mathematics Pre-Test Scores, Acquisition Phase Directed Attention Self-Report Ratings, and Total Test Questions Correct.

	No Rest Group			Unstructured Rest Group			Nature Video Rest Group
	M	SD	A-D	M	SD	A-D	M	SD	A-D
Pre-lesson
Mathematics Pre-Test questions correct (/480)	130.46	48.54	0.150	123.63	43.35	0.571	131.54	50.44	0.028
Acquisition Phase
Directed attention self-report (/4)	3.05	0.73	0.005	3.40	0.79	<0.01	3.40	0.56	<0.01
Post-lesson Test Phase
Total test questions correct (/20)	7.58	4.10	0.085	10.58	4.10	0.259	11.92	4.71	0.636

As noted in the study pre-registration, the present study included tests of hypotheses with sequence order (i.e., condition 1 > condition 2 > condition 3). Inferential statistics that incorporate information about the hypothetical rank order generally result in higher power compared to conventional analysis of variance (McKean et al., 2001). Based on the above hypotheses specifying score gradients across conditions for each of the variates under analyses, the first set of analyses present bootstrapped estimates of Spearman’s rank-order correlation coefficient (r) between the independent variable and median scores on dependent variables for each condition. Values of Spearman’s r are supplemented by the 95% confidence interval for r. Since hypotheses were directional, bootstrapped p values are one-sided.

The second set of analyses consisted of pairwise comparisons of results across the three conditions, to provide a more fine-grained analysis of between-group differences. As the pre-registration stipulated directional hypotheses, analyses used one-tailed statistical tests. The exception was for analyses of pre-test scores, where two-tailed tests were used since no specific directions in average scores were predicted. Given the extent of non-normality present in the data, rank-based multiple comparisons were tested using Noguchi et al.'s (2020) method. These analyses controlled the familywise error rate across the three comparisons at the stipulated rate (in this study, alpha = 0.05) while also generating an effect size similar to the standardised mean difference (Cohen’s d) and 95% confidence intervals. Because analyses are one-tailed, the upper limit of each confidence interval is given as infinity (∞; see Cumming, 2012, pp. 109–113).

Results

Results for the no rest (control), unstructured rest and nature-based rest conditions across the pre-lesson phase, acquisition phase and post-lesson test phase variates are given in Table 1.

Were the groups equivalent in mathematics pre-test prior knowledge?

For two out of three experimental conditions, the mathematics pre-test scores were found to be non-normally distributed. As a result, a Kruskal–Wallis test was used to test for equivalence across conditions. The differences in the mean ranks of the three experimental conditions were not statistically significant, χ² (2) = 0.18, p = .916, indicating random assignment to conditions created by groups that were equivalent in terms of arithmetical problem-solving ability. Follow-up pairwise comparisons found that the difference in pre-test scores between the unstructured rest and no rest conditions was not statistically significant, d = −0.08 (two-sided 95% CI − 0.56, 0.40), p = .916. The difference in pre-test scores between the Nature Video Rest and No Rest conditions was also not statistically significant, d = −0.02 (two-sided 95% CI − 0.51, 0.47), p = .994. Lastly, the difference in directed attention self-reports between the Nature Video Rest and Unstructured Rest conditions was also not statistically significant, d = 0.06 (two-sided 95% CI − 0.44, 0.55), p = .955.

Did rest affect self-reported directed attention during the acquisition phase?

Testing the hypothesised gradient across experimental conditions (H1), the gradient for self-reported directed attention was statistically reliable, r = 0.21 (95% CI 0.00, 0.42), p = .04. Pairwise comparisons between the conditions found only one statistically significant result: the unstructured rest condition reported higher levels of directed attention than the no rest condition, d = 0.48 (one-sided 95% CI 0.05, ∞), p = .030, supporting H1a. The difference in directed attention self-reports between the Nature Video Rest and No Rest conditions was not statistically significant, d = 0.36 (one-sided 95% CI − 0.05, ∞), p = .083; thus, H1b was not supported. The difference in directed attention self-reports between the Nature Video Rest and Unstructured Rest conditions was also not statistically significant, d = −0.12 (one-sided 95% CI − 0.54, ∞), p = .981; thus, H1c was not supported.

Did rest affect problem-solving performance during the Post-lesson test phase?

Testing the hypothesised positive gradient across experimental conditions (H2), the gradient for the post-lesson test performance was statistically reliable, r = 0.37 (95% CI 0.16, 0.57), p < .001. Pairwise comparisons between experimental conditions found two statistically significant results: the Unstructured Rest condition solved more problems than the No Rest condition, d = 0.40 (one-sided 95% CI 0.01, ∞), p = .043, supporting H2a, and the Nature Video Rest condition also solved more problems than the No Rest condition, d = 0.63 (one-sided 95% CI 0.24, ∞), p = .002, supporting H2b. However, the difference in number of problems solved between the Unstructured Rest and Nature Video Rest conditions was not statistically significant, d = 0.23 (one-sided 95% CI − 0.21, ∞), p = .301; thus, H2c was not supported.

Discussion

Learning complex concepts and skills generally requires sustained, directed attention. The tendency for these attentional resources to be depleted over time presents a challenge for educators and instructional designers. Educational researchers have advocated for the deliberate use of a range of rest break formats, such as mindfulness activities, physical activities, or exposure to nature, to refresh these resources and thus support ongoing learning. The present study drew on cognitive resource theory and attention restoration theory to generate predictions regarding the effects of alternative rest format periods on learning. Cognitive resource theory (Flanagan & Nathan-Roberts, 2019) predicts a rest break interposed between two cognitively challenging tasks – in the present study, arithmetic tasks performed under time pressure, followed by learning a challenging mental mathematics strategy – would restore attentional resources available for subsequent learning. Attention restoration theory (Kaplan & Kaplan, 1989) can be used to further predict rest based on nature exposure will be more effective than unstructured rest in restoring students’ directed attention. Specifically, we predicted the use of a first-person perspective video of a walk in a natural setting would promote a greater sense of psychological distance from the challenges of the pre-lesson phase (sense of “being away”), along with promoting effortless, interest-driven attention (“fascination”) in an environment of substantial scope (“extent”)” (cf. Anderson et al., 2017) than a simple “featureless” rest-break.

Hypothesis 1 was partially supported: a significant positive gradient was observed for self-reported directed attention, but mean scores indicate that the bulk of this effect was due to a difference between the no-rest and rest conditions, with the unstructured rest and nature-based rest conditions returning the same mean score. Follow-up multiple comparisons found only one statistically significant difference, favouring the unstructured rest group over the no-rest group. It is possible that ceiling effects limited the sensitivity of these analyses, with a large proportion (83.5%) of participants’ self-reports accumulating at the higher end (3–4) of the (reversed) 0–4 Likert scale. Hypothesis 2 was also partially supported: a significant positive gradient was also found for post-lesson test scores. Specifically, a significant difference was found between comparisons of the no-rest and unstructured rest conditions, and also between the no-rest and nature-based rest conditions, but not between the unstructured rest and nature-based rest conditions. These findings provide initial evidence for the benefits of alternate forms of computer-mediated rest in restoring directed attentional capacity depleted through pre-lesson activities.

The present study contributes to research on the educational benefits of rest breaks in several ways. On balance, the present study provides clearer evidence for a cognitive resource theory-based prediction of general benefits of rest than for the attention restoration theory-based prediction of substantial additional benefits of nature-based rest. These results support taking a brief break to support learning but do not clearly favour watching video-based natural scenes over unstructured rest. Nonetheless, results indicate that both forms of rest yield substantial impacts on learning. The study also demonstrates the utility of a self-report measure of directed attention developed by Matthews et al. (2002) for instructional design research on rest breaks. Much of the research literature on the effects of rest has used laboratory measures such as vigilance tasks (e.g., Finkbeiner et al., 2016; Helton & Russell, 2015), but in educational research, a student’s attention is typically on the contents of a lesson. Incorporating secondary tasks (such as a vigilance task) to measure attentional hypotheses thus brings substantial measurement challenges, including the potential for interference with the primary task. In comparison, retrospective self-reports of cognitive constructs such as mental effort and cognitive load have been found to yield straightforward and sensitive tests of hypotheses in instructional design research (Paas et al., 2003). The directed attention self-report used in the present study may thus have broader applications in instructional design research, given the increasing interest in cognitive load theory in working memory models based on attentional control (e.g., Sepp et al., 2019).

The present study’s results can inform instructional design and teaching practice in several ways. Rest breaks were tested prior to “self-study” of a lesson; thus, results are applicable to students engaged in extended, “homework”-type learning where attention is likely to be depleted. Students may benefit from explicit reminders from educators that such rest breaks are beneficial, including to use timers and calendar reminders as prompts for rest. From a practical standpoint, the present results would support students choosing either a “featureless” break using a timer or accessing video recordings of natural scenes (such as through YouTube), in the expectation that both would support subsequent learning. We speculate that many students would find video recordings of natural scenes preferable to an unstructured alternative. Extrapolating from self-study settings to classroom settings, educators might also experiment with either unstructured or nature-based video breaks projected on interactive whiteboards or lecture screens to refresh attention. However, it should be noted that many students may prefer to use such breaks for socialising, which, while providing a break from a lesson, is likely to engage qualitatively different cognitive processes than the non-social processes engaged by the rest formats investigated in this study.

This study’s limitations provide several directions for future research. Participants did not have to show any workout for answers or explain their answers to test questions verbally; as a result, it is possible that some participants used a solution strategy other than that demonstrated in the lesson phase. Future studies using this testbed might use a think-aloud protocol (Ericsson & Simon, 1993) to ensure the intended strategy is being used. The use of a self-reported measure of directed attention carries the risk that some students would report being on-task due to social desirability pressures. Studies would benefit from triangulating self-reports of directed attention with other objective measures, such as measures of attentional processes (for an ART-based review, see Ohly et al., 2016), or real-time physiological measures, such as EEG (Antonenko et al., 2010; Castro-Meneses et al., 2020) and eye-tracking (Miller, 2015; Zu et al., 2020). In addition to cognitive measures, ART researchers have often considered the effects of nature exposure on emotions (e.g., Finkbeiner et al., 2016; White et al., 2010). In the context of learning under cognitive load, Plass and Kalyuga (2019) have argued emotions may relate to cognitive load in several ways, including positive moods acting to increase available resources, enable creative thinking, and facilitate executive processes (p.349). Inclusion of such measures of mood or stress would support investigation of such potentials in the context of nature breaks.

As noted above, contrary to expectations, differences between the rest conditions on the variates were not statistically reliable. The present study contributes to educational research on nature exposure by investigating a briefer exposure than those identified in Mason et al. (2021) review, but also speaks to the importance of an active control group – in this case, a rest break not involving nature exposure. Future investigations of nature breaks to support learning might explore more immersive experiences, such as virtual reality (cf. Anderson et al., 2017) or a walk in a natural setting, to investigate if such experiences generate stronger effects; the impact of shorter versus longer breaks might also be investigated. Lastly, larger samples would also support investigations of process models including aptitude–treatment interactions; for example, in the context of the present study, the effects of nature-based breaks may be more pronounced for students with higher levels of mathematics anxiety.

Disclosure statement

No potential conflict of interest was reported by the author(s).