Keywords

1 Introduction

Global market for mobile learning products and services will reach $37.60 billion in 20201 with revenues doubling in more than 66 countries (Lim and Churchill 2016).

Mobile learning, also referred to as m-learning (Biggs and Justice 2011) is defined as learning with the use of mobile or portable devices anywhere and at any time (Costabile et al. 2008; Deegan 2015; Sinclair 2011). By introducing new and unique ways of learning, m-learning is now replacing traditional learning facets and improves learning performance (Sinclair 2011; Wang and Shen 2012). It has reached a mature stage and became “mainstream” in countries such as the United States, Japan, South Korea, Singapore, and in Northern Europe (Sanakulov and Karjaluoto 2015). Mobile network connectivity provides fast access to current learning materials and enables quick communication and collaboration among the users (Biggs and Justice 2011; Sinclair 2011). However, poorly designed systems that do not incorporate learning environment present certain challenges. Often times which leads to decreased learning performance. These challenges can range from technical issues (slow performance, poor internet connection etc.), general usability issues (Deegan 2015), design of learning materials (Wang and Shen 2012), and also the learner’s environment (Deegan 2015). While usability and designing appropriate material has received considerable attention, much less focus has been paid to learning environment (Hoehle and Venkatesh 2015; Adipat et al. 2011).

Deegan emphasizes that the main source of possible distractions and interruptions for the learner exists in the environment (Deegan 2015). According to a study, interruptions impose a higher cognitive load on the learner and, thus, impair his or her learning performance (Costa and McCrae 1990). With increased cognitive load also comes less efficiency and accuracy while performing tasks (Wickens et al. 2000; Albers and Kim 2000). Furthermore, the ability to resume the main task post interruption is impaired by interruption characteristics. It was found that 41% of tasks remain un-resumed after interruptions (Mark et al. 2005). Some task takes up to 25 min to be resumed (Mark et al. 2005). Thus mobile applications that are designed for mobile learning have to address the issue of interruptions faced learning. Therefore, our main research question is:

RQ:

What are the design requirements of learning based mobile applications that support interruptions for higher learning success?

Design is one of the main components of learning environments. Gamification has been a major influence in the education context (Nah et al. 2014; Nah et al. 2013). Studies have looked at interface aesthetics (Bhandari and Chang 2014; Cyr et al. 2006; Sonderegger and Sauer 2009), color and menu icons (Sonderegger et al. 2012), and graphics (Wells et al. 2011), but enough focus has not been paid on design from an interruption perspective. One of the major design guideline for developers by Apple is to be prepared for interruptions (Apple 2012). Very few studies have synthesized the existing body of research on learning interruptions in order to derive theory-based IS design recommendations.

Our study looks at long term use of technology and since interruptions have known to effect task performance (Speier et al. 2003), decision making (Speier et al. 1999; Xia and Sudharshan 2002), usage (Rennecker and Godwin 2005) and thus is of extreme important to the area of information systems. By exploring design principles specifically for m-learning domain we are contributing to the theory of interruptions for learning success and practically guiding developers to design for interruptions.

The paper proceeds as follows: we discuss relevant prior literature in the next section. After that we present research model and hypothesis. Post that we discuss research methodology and finally present the potential contributions.

2 Theoretical Background

2.1 M-Learning

Learning has been defined in literature as “the acquisition and development of memories and behaviors, including skills, knowledge, understanding, values, and wisdom” (Deegan 2015). Psychology take a different perspective and defines it as commitment of changes to the long-term memory (Craik and Lockhart 1972; Eysenck and Keane 2000).

Even though research in the domain of mobile technologies an devices and it associated theories has been progressing, it is very disparate (Alrasheedi et al. 2015). Scope dependency is a major factor in defining learning (Alrasheedi et al. 2015; Trifonova and Ronchetti 2003). Since e-leering was the predominant domain when studying learning the definition is also heavily borrowed dorm this discipline. It can also be described as fusion of E-Learning and mobile technologies: “mobile devices are a natural extension of e-learning” (Kossen 2001).

Thus we adapt the definition of mobile learning by (Deegan 2015), who defines it as learning with the aid of mobile or portable devices anywhere and at any time. Mobile devices are often connected to a mobile network and, thus, enable an on-demand access to learning materials and collaboration with other learners (McQuiggan et al. 2015).

2.2 Interruption and Interruption Complexity

Interruption is defined as an “externally-generated, randomly occurring, discrete event that breaks continuity of cognitive focus on a primary task” (Botha et al. 2010; Coraggio 1990). In order to understand the design guidelines for learning based mobile apps it is important to understand what factors lead to decreased performance in learning and how to counteract them. Interruptions in e-learning environment and m-learning environments are quite unique. While in e-leering interruptions can be more task based, in m-learning lot of interruptions are environment based. For e.g. halting the learning process due to noisy background or having a phone in between and resuming subsequent task on the mobile device. M-leaning has its advantages in the form of freedom and portability in learning and choosing the environment, the inherent issues remain. This causes extra cognitive load on the learner as he needs to recuperate from these environmental interruptions (Botha et al. 2010; Deegan 2015; Costabile et al. 2008).

When considering in depth what kind of negative effect interruptions has on lining we can refer o study by Morgan et al. 2009 where they categorized the negative effects into following broad categories: forgetting to resume from an interruption, delays in resumption of the primary task, decreased efficiency, decreased accuracy, and stress elevation (Morgan et al. 2009). Even when user intends to resume the task after being interrupted t is in most cases unsuccessful (Einstein et al. 2003). A study showed that because of delay in retrieval of required information, the task resumption is enhanced (Hodgetts and Jones 2006). Resumption of a task can be measured by the time lag between interruption of primary task and resumption of primary task. This effects performance and accuracy (Eyrolle and Cellier 2000) (Flynn et al. 1999). Study also showed that the effect of interruption could be seen in elevated levels of mental stress (Zijlstra et al. 1999).

According to Botha, interruptions decrease the learner’s attention span. In this context, this paper defines interrupted learning as learning that occurs at irregular intervals, each of which receives only a short attention span by the learner (Botha et al. 2010).

2.3 Designing for Interruptions: Memory for Goals

We store information in two forms of memory: namely short term memory and long term memory. Information that we perceive with our perceptual senses is stored in short term memory. However information that receives our attention passes from short term to long term memory. Tasks that occur at regular interval can eventually be stored in long-term memory, while small interruptions like a phone call while reading, knock in the door while playing a game are short term interruptions (Paas et al. 2004; McQuiggan et al. 2015).

In terms of executing complex task requiring cognitive resources short term memory is not the adequate part of memory to be used, the long term working memory is more useful to retrieve and then process the information required for completing cognitive complex task. (Eysenck and Keane 2000; Shrager et al. 2008). Distractions have been found to interrupt the primary task and thus take up working memory resources (Sweller 1988).

Memory for Goals (MFG) is “a theory of cognitive control that explains goal memory in terms of general declarative memory constructs, such as activation and associative priming, rather than using a special goal memory or control structure, such as a goal stack” (Altmanna and Trafton 2002). Goals stored in the memory compete for higher control of cognitive resources. Decision of which goals gets activated depends on which goal holds the highest instantaneous activation value. Activation is a function of total number of times an item from memory has been retrieved in given time frame and the length if the above mentioned timeframe. Thus it is combining usage history and the current requirements together so that the cognitive system can deal with memory decay and keep the information needed alive.

Specifically for interruptions, the memory for goals theory suggests that there are two important ways of reducing memory decay with respect to goals: rehearsal and using environmental cues. Rehearsal is done in two ways, retrospection (e.g., “What was I doing till now?”) or in a prospection (e.g., “What was I about to do?”) (Altmanna and Trafton 2002). Both are important, however people prefer to use prospective introspection when needed (Trafton et al. 2003). Another factor that can help in activation of a goal in our model is to prime using external mnemonics. External cues when added, they further add to the activation to any goals with which they are associated (Dodhia and Dismukes 2009). This expansion of activation is added to the activation produced using heuristics. This is based on assumption that the goal will have decayed during the interval of the interruption. Studies have shown that system designers should keep “interaction chains” (the number of interface actions that lead to a goal or sub goal) quite short (Oulasvirta and Saariluoma 2006). The amount of time is not guided by theory, but 20 s currently used by designers.

3 Research Model and Hypothesis

3.1 External Mnemonics/Metacognitive Processes Support

Using prospective memory theory framework, it can be inferred that reminders causes a person to explicitly encode an intention to resume, which should facilitate performance (Trafton et al. 2003). Studies found that the use of a blue dot cue improves performance upon resumption of the task (Finstad et al. 2006). People who are constantly interrupted by to real-time interruptions and tasks that need prospective memory can great benefit from reminders. To set free working memory resources, the mobile application should provide an implementation of an external prospective memory (Einstein et al. 2003; Morgan et al. 2009) (Fig. 1) .

Fig. 1.
figure 1

Research framework

Full size image

Consistent with this predictions studies found that reminders facilitate resumptions. Resumption is defined as abandoning/attending the interruption task and returning to primary task (Dodhia and Dismukes 2009; Dismukes 2010). The external prospective memory is an information storage which the learner can use to outsource information – in contrast to reminder cues relatively high quantities – from his working memory for a later use (McDaniel et al. 2004). Following the above, we hypothesize:

  • H1 (a): For information browsing via mobile applications, design features supporting external mnemonic will lead to increase in learning success.

  • H1 (b): For information browsing via mobile applications, design features supporting external mnemonic will lead to increase in resumption success.

3.2 Contextual Cues/Interactive Immediacy

Memory for goals theory explains that when the learner’s environment alters or there is little information from environmental context, the goal will be more difficult to resume (Altmanna and Trafton 2002). However, highlighting the context via environmental cues can facilitate resumption. The theory provides a very specific process description: the environmental cue adds activation to the goal that was suspended by the interruption. For example, a user is switching between two tasks (e-mail and making a graph in Microsoft excel). When switching to the graphing task after an interruption, it would be helpful to present some environmental or contextual cues (Dodhia and Dismukes 2009). To aid in reorientation in such a scenario information like “tasks already completed”, “next task to do” is helpful.

This is in line with activation based memory for goals model which focuses in the resumption of an interrupted task. By activating the specific goal needed to ensure resumption of primary task this framework highlights the importance of contextual cues (Altmanna and Trafton 2002). For e.g. three colors coded arrows can show the user’s previous three steps: light red (top arrow) to bright red (bottom arrow). This type of color order has been shown to be a natural way of presenting ordered and quantitative data (much like task statistics) for task performance (Spence et al. 1999; Breslow et al. 2009). Interface facilitation in earlier empirical work has been strongly recommended to support resumption (Trafton et al. 2003), although the task was quite different from traditional office work. This is supported by the memory for spatial location paradigm where participants generally are able to return to the location of where they left for.

Following the above our hypothesis is proposed as follows:

  • H2 (a): For information browsing via mobile applications, design features supporting contextual cues will lead to increase in learning success.

  • H2 (b): For information browsing via mobile applications, design features supporting contextual cues will lead to increase in resumption success.

Memory of goals theory suggests that the length of interruption can determine how adversely it impacts the outcome. The more complex an interruption the more rehearsal or external cues it will be needed to slow down the decay. Also that includes higher external mnemonics to support resumptions. It was found that subtle external mnemonics like cursor pointing to last point of action was ineffective. This calls for combining design strategies that can combat the effect of complex interruptions (Trafton et al. 2003). Finally, as the complexity of interruption during browsing task increases (e.g., frequency of interruption, duration of interruption etc.), the need for information scents to help locate desirable information will be stronger (Coraggio 1990). As discussed earlier, we predict that interruption complexity positively moderates the individual effect of external mnemonic and contextual cues on users’ learning performance and perception. Therefore, we expect that as interruption complexity increases, the effect of adding integrated adaptation of external mnemonic and contextual cues to basic design will be stronger than that of adding external mnemonic and contextual cues separately. Our hypothesis is proposed as follows:

  • H3: The greater the interruption task complexity, the greater the positive effect of adding both external mnemonic and contextual cues, as compared to adding only one of them on

    1. (a)

      Learning success

    2. (b)

      Resumption success

4 Research Methodologies and Data Analysis

4.1 Design Principles for Interruptions in M-Learning

This is a research-in-progress paper that uses an experiment to examine influence of interface design guided by memory for goals theory and interruption theory on learning success. Majority of users use mobile application simply for content viewing. These applications are popularly known as “content aggregators”. They provide magazine style interface to view information in articles. Users also browse through multiple pages in mobile learning applications. These are more goal-oriented applications. The information contained can vary from text, images, hyperlinks etc. Supporting for interruptions is a challenging task for both these categories of learning applications. In this study we focus on the first type of applications simply because they are more affected by lack of support for interruptions, as intrinsic motivation to resume the task is low. Also they are more popular type of applications.

We first analyzed learning based applications in the appstore from category “Education”. (https://itunes.apple.com/us/genre/ios-education/id6017?mt=8). We excluded free applications from our set, as it is much harder to control for quality in this category. Meta level design requirements were extracted specific to interruptions in m-learning, which then resulted in design considerations corresponding to these meta-requirements. The aim is to ensure that our selection of interruption support principles as guided by memory for goals theory and prospective framework theory map to the real world scenario. The meta-level design requirements are mapped to the extant literature and broadly divided in to two categories (a) meta-cognitive process support and (b) interactive immediacy.

4.2 Experiments

We aim to test the hypothesis in laboratory experiments with a 2 × 2 × 2 factorial design. Interruption support design factors: meta-cognitive support (high/low) and interactive immediacy (high/low) is within subject factor. Interruption complexity (high/low) will be a between subjects factor. Participants will be asked to perform multiple tasks while manipulating levels of interruption design support. A different mobile application is used for each of the conditions. These applications will be experimentally designed guided by literature and current app development design guidelines. We intend to use repeated measures ANOVA to analyze the effect of interruption support design and interruption complexity. We will also check for Bonferroni multiple comparison because it is a robust multiple comparisons method suited for within-subjects design (Maxwell and Delaney 2000).

Dependent and Independent Variables.

For manipulating meta-cognitive support, we follow studies mentioned in Table 1. Information is broken down into easy reading chunks rather than long running paragraphs. No scrolling is involved in condition of high meta-cognitive support (all the information is contained in one page). Easy navigation capabilities are provided and are placed at the bottom rather than at the top.

Table 1. Design guidelines synthesized from existing literature and existing mobile applications applying them.
Full size table

For interactive immediacy, intention based reminder cues are presented as user finishes the interruption task. Lack of such reminders would be the low manipulation of immediacy feedback. Also external mnemonics like a colored arrow is provided referring to steps completed further steps to be completed and corresponding details.

In this study users are asked to find answers to simple fact based questions (Adipat et al. 2011). Accuracy is the objective measures which captures whether the number of correct responses. Search time is measured as time taken to finish the task. Time to resumption is the lag between suspension of primary task and resuming form interrupted task. After the user finished the task on his/her assigned application, they proceed to fill a survey-based questionnaire measuring perceived ease of resumption. Perceived ease of resumption is subjective measure capturing how easily they think they resumed the primary task.

5 Contributions

The study has contributions to both theory and practice. Theoretically we develop a unique set of design factors specific to learning on mobile platforms using memory for goals theory. Further interruption theory is used to understand complexity as a potential moderator for the relationship between interruption design principles and learning success. Practical contributions include guiding practitioners to understand how to tackle interruptions as a major source of hindrance in learning on mobile apps. Also design principles like meta-cognitive support features and immediacy are useful for developers in ensuring long term learning success for their mobile applications.