Future Scenarios for Mobile Science Learning

This paper adopts scenario planning as a methodological approach and tool to help science educators reconceptualise their use of mobile technologies across various different futures. These ‘futures’ are set out neither as predictions nor prognoses but rather as stimuli to encourage greater discussion and reflection around the use of mobile technologies in science education. Informed by the literature and our empirical data, we consider four alternative futures for science education in a mobile world, with a particular focus on networked collaboration and student agency. We conclude that ‘seamless learning’, whereby students are empowered to use their mobile technologies to negotiate across physical and virtual boundaries (e.g. between school and out-of-school activities), may be the most significant factor in encouraging educators to rethink their existing pedagogical patterns, thereby realizing some of the promises of contextualised participatory science learning.


Introduction
Mobile learning (m-learning) considers the process of learning mediated by handheld devices such as smartphones, tablet computers and game consoles (Schuler et al. 2012).The ubiquity, flexibility and increasingly diverse capabilities of these technologies have created considerable interest amongst science educators (Aubusson et al. 2012;Cheng and Tsai 2013;Foley and Reveles 2014;Johnson et al. 2013;Marty et al. 2013;Song 2014) who have begun to

Background Theoretical Framework
Research studies have examined m-learning through various theoretical perspectives and frameworks such as activity-based approaches, authentic learning, action learning, and experiential learning (Sharples et al. 2007).More recently, Kearney et al. (2012) developed a pedagogical framework of mobile learning, which draws on socio-cultural understandings.This framework privileges three distinctive features of m-learning: personalisation, collaboration and authenticity (see Fig. 1).The rationale behind these scales is provided through the use of subsidiary themes under each of the central features, which pinpoints the critical features of m-learning from a pedagogical perspective.How learners ultimately experience these pedagogical characteristics is influenced by the 'time-space' configuration of the learning context (Ling and Donner 2009): the organisation of the temporal (scheduled/flexible, synchronous/asynchronous) and spatial (formal/informal, physical/virtual) aspects of the m-learning environment (Traxler 2009) as depicted in Fig. 1.This configuration is often described in the literature through words such as 'anywhere, anytime', 'on the move' and 'multiple contexts' (Mifsud 2014).
Firstly, the personalisation feature through its sub-themes has strong implications for ownership, agency and autonomous learning.It consists of the sub-themes of agency and customisation.High levels of personalisation would mean the learner is able to enjoy a high degree of agency in appropriately designed m-learning experiences (Pachler et al. 2009) together with the ability to customise and tailor both tools and activities, leading to a strong sense of ownership.Secondly, the collaboration feature captures the oft-reported conversational, connected aspects of mobile learning.It consists of conversation and data sharing subthemes, as learners engage in negotiating meaning, forging networked connections and interactions with other people and the environment and sharing information and resources across time and space through rich collaborative tasks (Wang and Shen 2012).Finally, the authenticity feature highlights opportunities for contextualised, participatory, situated learning.Radinsky et al. (2001) espoused two models of authentic learning environments: a simulation model and participation model.Tasks that fit a simulation model of authenticity use the learning space (e.g.classroom) as a 'practice field' (separate from the 'real community') but still providing contexts where learners can practise the kind of activities they might encounter outside of formal learning settings.Alternatively, under a participation model of authenticity, students participate in the actual work of a professional community, engaging directly in the target community itself.Hence, the sub-themes of contextualisation and situatedness bring to bear the significance of learners' involvement in rich, contextualised tasks (e.g.realistic setting and use of tools), involving participation in real-life, in situ practices.
This framework has recently been used to inform research on m-learning in school education (Burden et al. 2012;Kearney et al. 2015), teacher education (Kearney and Maher 2013) and other areas of higher education (Kinash et al. 2012).For example, Green et al. (2014) used the framework to inform the development of their own instrument-the 'Mobile App Selection for Science' (MASS) rubric-to aid teachers' rigorous selection and evaluation of K-12 science applications (or 'apps').In this paper, we will rationalise a focus on the two constructs of personalisation and collaboration to explore how handheld technologies might influence future science learning.
Fig. 1 Framework comprising three distinctive characteristics of mobile learning experiences, with sub-scales.From Kearney et al. (2012, p. 8) Learning Science 'Seamlessly' Across Contexts Studies of mobile learning in science education have typically focused on informal learning contexts (Aubusson et al. 2012), promoting science 'on the move'.The portable, flexible nature of mobile devices is well suited to these contexts and can facilitate location-based (or place-based) learning (Jones et al. 2013).However, given the ongoing physical realities of formal schooling and higher education, recent studies have focussed on the notion of using handheld devices to provide 'seamless learning' tasks (Rushby 2012;Toh et al. 2013), supporting a continuity of learning across contexts and devices and transitions between episodes of formal and informal learning, for example, connecting learning in/out-of-class, in/out-of-school, between curricular/co-curricular, social/personal or academic/recreational boundaries between physical/virtual contexts and across times and locations (Wong and Looi 2011).In science education, 'seamless' learning might connect learning in classrooms and science museums, provide a bridge between lab-based inquiry to be continued in a more realistic setting or connect an 'in situ' learning episode (possibly personal and informal) to be used as a resource for formal learning at school.Mobile devices might mediate this 'flow of learning' between formal and informal contexts, for example, using microblogging, social networking platforms, specific science tools, simulations or games (Lai et al. 2013).

Promoting Inquiry Across Authentic Contexts
Digital technologies have typically been promoted in science for many purposes, from tools for instructional delivery to student research, communication and presentations.Recent studies have focused on digital learning environments that Bemulate the activities of practising scientists^ (DeGennaro 2012(DeGennaro , p. 1319)), where learners' use of technology becomes an integral part of their task.For example, visualisations, animations, participatory simulations and multiuser virtual environments have been used to actively immerse students in realistic scientist roles.In response, m-learning studies in science education have advocated a more participatory authenticity (Radinsky et al. 2001), whereby tasks are embedded in real-life, connected, community-based science projects (e.g.Jones et al. 2013;Scanlon et al. 2014).In the same way as real scientists are 'connected to a broad community of other scientists who share information and co-construct knowledge and ideas' (DeGennaro 2012(DeGennaro , p. 1321)), such mlearning tasks allow students to participate in authentic ways in real-life, project-based pursuits.
The importance of student inquiry and student-driven questions has long been advocated in science education (Krajcik et al. 2000).Consequently, there has been a burgeoning interest in exploiting mobile devices to mediate inquiry-based learning, mirroring the types of investigative processes carried out by real scientists.These include support of question generation, planning and implementing investigations, data collection, observation, analysing and interpreting data, constructing evidence-based explanations and arguments (Herodotou et al. 2014;Wilson et al. 2013).Mobile devices are ideal tools for supporting the inquiry process, with their ability to support multimedia access and collection, communication, representation, information sharing, knowledge construction, connectivity, reference and analysis (Song 2014).However, they are not yet used to their full potential in science education for inquiry, particularly in support of measuring and investigating real-world phenomena Also, many science students currently carry out inquiry tasks in relative isolation (individual or pairs, small groups) and in a minimal number of locations such as classrooms and excursions (Herodotou et al. 2014).Lui et al. (2014) argue a need for expanding these typical inquiry experiences, with less abstract, contrived forms of interactions, for example through digitally augmented physical spaces (mixed-reality environments).
For example, Herodotou et al. (2014) presented a toolkit (the Sense-it app) to support measuring and investigating real-world phenomena.It combines and customises data from a full range of sensors into new or existing citizen science projects.Nonprofessional members of the public can use these toolkits to collaborate with scientists, contributing to observation and measurement data in science projects such as species identification and air/water pollution monitoring.The app allows users to create their own personally relevant science investigations and offers instant feedback on how their own sensor recordings relate to other users' data.Jones et al. (2013) compared two case studies to explore the different ways mobile devices can support inquiry learning in semiformal and formal settings.One study explored the science learning by 14-15-year-old students using web-based software in a semiformal context.The other study looked at informal adult learners using their own devices to learn about landscape.Looking at these studies together allowed the researchers to focus on both the use of mobile devices in situ and how the devices supported choice and learner control.In the first case of semiformal learning, Jones et al. (2013) found that mobile devices with dedicated software supported the science students to choose and take personal responsibility for their inquiries without adult help.These inquiries were engaging and personally relevant.They also discussed their nQuire software tool and how it was used to support the inquiry process seamlessly across different contexts (an afterschool club and home).They found the tool used location-based awareness facilities to support the inquiry process, including information sharing and collaborative activities, communication between learners, other observers and experts.They illustrate ways of supporting personal inquiry learning with m-devices (locationbased inquiries), accessing resources and information in situ.As nQuire is an open software resource, it is also developing a strong community of users.Scanlon et al. (2014) presented a similar tool, the iSpot application, allowing users to participate in location-based activities akin to real scientific pursuits, in informal science settings.This UK initiative also uses an inquiry learning approach and aims to create and inspire a new generation of nature lovers to explore, enjoy and protect their local environment.Members of the public can use this tool to work in combination with science researchers.For example, their (location-based) observations of animals and plants became 'shared, social objects amongst associated groups, networks and collectives' (p.60).Indeed, selected observations are used in biodiversity monitoring and research, essentially enabling learners to actively contribute to knowledge building as a community activity.
Song (2014) made a 1-year case study in a primary school science inquiry context using bring-your-own devices (BYOD).Students developed a positive attitude to science inquiry and demonstrated improved understandings of the topic (the anatomy of a fish).Song (2014) emphasised 'affordance networks' (p.60) as a key aspect to making optimal use of m-devices for knowledge construction across constantly changing contexts such as digital and physical environments at home, school and other spaces.Finally, another example of seamless learning in primary school science contexts was reported by Marty et al. (2013).Their project aimed to develop inquiry skills and digital literacies using an app called the 'Habitat Tracker'.These m-learning experiences provided a link between formal and informal contexts, including the classroom and excursions to science museums and wildlife centres.

Use of Augmented Reality and Immersive Simulations
Augmented reality (AR) is an emerging technology that 'utilizes mobile, context-aware devices (e.g.smartphones, tablets), which enable participants to interact with digital information embedded within the physical environment' (Dunleavy and Dede 2014, p. 735).Cheng and Tsai (2013) distinguish two types of AR: image-based and location-based.Through a scan of existing studies, they found image-based AR was beneficial to students' spatial abilities, practical skills and conceptual understanding, while location-based AR was beneficial to scientific inquiry learning.Location-based AR is usually underpinned by a situated learning perspective, emphasising authentic contexts, inquiry with real-time data and other virtual information in a real context.Students may also communicate with avatars and peers to collaboratively hypothesise, reason and solve problems.
AR-based tasks typically take the form of participative simulations, using fictional scenarios added to a local setting, allowing learners to connect science ideas to community-based experiences.For example, Wong and Looi (2011) report on games played in a physical environment but augmented by virtual artefacts (what they called 'mixed reality learning').Mobile devices with location-based sensors allowed users in the study to interact with explorations, experiments and challenges for inquiry and games-based learning.Another example is Kamarainen et al. (2013) pilot study for the EcoMobile (Ecosystems Mobile Outdoor Blended Immersive Learning Environment) project (http://ecomobile.gse.harvard.edu), exploring children's use of a smartphone AR application (FreshAiR) for blended learning across virtual and natural (pond) ecosystems.Combining this application with environmental probeware allowed students to take samples of pond water, gain increased understanding of the ecosystem and interact with each other in student-centred ways that resembled scientific practice.
Immersive and participative simulations have been used as platforms to engage learners in inquiry-based approaches.Lui et al. (2014) described an immersive, cave-like rainforest simulation (called EvoRoom) and a mobile inquiry platform (called Zyeco) that enabled users to collect and share data.Students are co-located in an immersive and physical digital space, collecting observational data from both the classroom itself (Evoroom) and out-of-class settings (such as parks or museums) and exploring peers' data using large visualisations displayed at the front of the room.This arrangement allows students to pose questions, collect observation data, review and share data and use it to form evidence-based arguments.Foley and Reveles (2014) presented a 'connected classroom' that used online resources to engage students in inquiry, creating authentic science learning experiences.They emphasised the connection between students' handhelds and the Internet to 'share information instantly and enable computer supported collaborative learning' (Foley and Reveles 2014, p. 4).Students' data from experiments and simulations was pooled across classes or schools, allowing them to compare and analyse across larger data sets and collaboratively identify trends as a community of science learners.Collaborative tools such as Google Education suite then allowed for further discussion and feedback on ideas and consensus building.
Location awareness is an aspect of AR that Zimmerman and Land (2014) use to explore the principles of place-based education (PBE) for teaching science in an era of mobile devices.For a decade, PBE has provided a way of engaging out-of-school students with the issues, artefacts, cultural practices and natural histories of their local communities.To accommodate the location-awareness features of mobile devices in PBE, Zimmerman and Land developed empirically derived guidelines for research and design for outdoor informal mobile computing (p.82), emphasising participation in disciplinary conversations and practices within personally relevant places; amplification of observations, in liaison with experts, to understand the disciplinary-relevant aspects of a place.Students gain value from experts who can illustrate aspects of a place and capturing, sharing and reflecting on knowledge artefacts found in local settings to explore new perspectives.
In summary, the contemporary m-learning literature in science education mainly comprises case studies of innovative mobile applications exploiting authentic, connected, participative inquiry-based approaches.Research has explored the possibilities for science learning across formal and informal contexts, making seamless links between virtual and physical environments, and particularly using participatory simulations and augmented reality technologies.Mindful of these current research directions and informed by an established framework of mobile learning, this paper explores possible future scenarios for science education.

Designing the Scenarios
In this section, we describe the scenario planning procedures underpinned by future scenario thinking (Snoek 2013).Drawing on our empirical data, we provide a rationale for the nomination of the two dimensions (or 'drivers') for our scenario planning, student agency and collaborative networking, before illustrating their utility for developing the four future scenarios.

Scenario Planning
Scenarios have been described as 'presentations of multiple possible futures' (Snoek 2013, p. 311), which are widely used in businesses (e.g.Shell 2003) and the military (Cann 2010) but until recently, less common in education.This may be changing with some high profile scenario planning exercises commissioned by organisations like the OECD (2001) and teacher futures special editions in international education journals (e.g.Aubusson and Schuck 2013).This recent surge of interest amongst educators is not surprising given the complexity and unpredictability of the environments within which they operate, since scenario planning is seen as a more suitable alternative to traditional prediction methods, which depend on greater levels of stability and more predictable contexts (Snoek 2013).Indeed, Snoek identifies this as one of two major problems associated with traditional approaches to planning and predicting the future, pointing out how this has a tendency to produce a single future prediction when in fact there are likely to be many.To compound this tendency, policy-makers and governments are also guilty of believing they can realise a single prediction of the future by mandating change ignoring the 'fundamental unpredictability of the future and the possibility of different futures that need to be taken into account' (Snoek 2013, p. 308).
Scenario planning is positioned as a viable alternative to the traditional 'rational-centralrule' approach (Van Gunsteren 1976) since it accepts the inherent unpredictability and complexity of modern society and seeks to identify multiple possible futures enabling greater scope for discussion and alternative perspectives.Put another way, traditional approaches are akin to 'forecasting', which leads to future predictions, compared to 'foresighting', which leads to alternative scenarios for the future (Codd et al. 2002).
It is important to note however that the primary purpose of scenario planning is not to second guess or predict an absolutely accurate future.Indeed, fidelity is not the primary concern of scenario thinking.Rather the purpose is to ferment discussion and reflection, encouraging perspectives and positions that might otherwise be overlooked or under represented.In this sense, they have an important role to play both at policy and practitioner level since they engage more creative lateral thinking processes that can generate new ways of seeing that might not otherwise be imagined (Snoek 2005).

Data Collection and Analysis
Data were collected in 2013 in an international survey on m-learning aiming to identify how educators use the distinctive mobile pedagogical features.The 30-item survey instrument was informed by our theoretical framework of mobile learning (Kearney et al. 2012) focusing on the three themes of personalisation, collaboration and authenticity.Participants were asked to identify and analyse a specific learning task in which they had recently used mobile technologies.Each item usually contained three response options that corresponded to 'low' or 'none', (depending on the context) of the item, 'medium' and 'high' ratings for a particular construct.Findings from school teacher participants in the survey, including a reliability analysis of the survey instrument, are reported elsewhere (Kearney et al. 2015).
Although the total number of survey participants consisted of 195 educators, this paper draws upon a purposive, original data set of 54 participants who identified themselves as teaching in science education (31), engineering (17) or health sciences education (6), since these areas were considered to be most relevant for our future scenario development.The 54 teacher participants were mainly from Australia (43 %) and Europe (25 %), where the researchers' institutions were located.Thirteen percent of these participants taught in primary/elementary school contexts, 33 % taught in secondary school contexts and 49 % in tertiary education.Participation in the survey was voluntary, and there was a diverse range of experience levels identified in the participants' background data.Seventy-four percent of the survey participants had been teaching for more than 10 years, while 9 % had been teaching for less than 2 years.Also, 52 % of participants perceived themselves as experienced users of mobile devices in their teaching-defined as more than 2 years of experience-while 19 % said this was their first attempt at implementing a mobile learning task.Participants chose a range of task contexts.Eighty-five percent of the teacher participants described a formal task that was classroom/campus-based.Only 7 % of teachers reported on a task that was situated in an 'extra-mural' context (school playground, excursion site, museum, home) and only one task was set in a totally informal location such as a cafe or public transport (three tasks involved a combination of locations).Most tasks involved use of an iPad (30 %), laptop (26 %) or mobile phone (15 %), with 30 % of tasks integrating a mixture of devices.Forty-one percent of tasks involved use of institution-owned devices (22 % restricted to on-campus use only), while only 30 % of tasks involved student-owned, BYOD.
Data from the personalisation and collaboration constructs were compelling, and these two areas proved to be problematic for the science educators.Responses to items relating to personalisation revealed that teacher participants did not generally design learning episodes which granted their science students high or even moderate levels of decision-making with regard to the context of their learning (e.g.where or when it occurs).For example, just over one quarter of these teachers perceived their science task as lending no student control over aspects such as the learning context-where and when the activity occurs (32 % of teachers), task pacing (26 % of teachers), task content and learning goals (28 % of teachers).Conversely, only 19 % of tasks gave full control to students of the task content and learning goals and only implemented.This lack of opportunities for students to enjoy autonomous learning tasks is particularly surprising given the general commentary around enhanced agency in m-learning environments (see for example, Burden et al. 2012).
Data in the collaboration construct showed that the majority of m-learning tasks described by the 54 teacher participants were highly social and collaborative in nature, involving a high level of face-to-face conversation at the device, usually in the classroom.Most teachers prioritised students working in small groups around their device, with 70 % ranking their task as 'medium' or 'high' for face-to-face collaboration.Whole-class discussions were frequently mentioned; however, levels of online conversation through the device (Crooks 1999) were generally ranked low (63 %).In tasks that included online discussion, communications were mainly between class peers (44 %) or between students and their teachers (20 %).Only 4 % of tasks involved 'extra-mural' communications with participants outside their immediate peer/ teacher class network.Although student generation of digital content was a feature of many tasks, there were lower rates of sharing through networked interactions.Only 44 % of tasks involved a networked exchange of digital data and information or networked interactions (e.g. via blogosphere, Twitter, multi-player games etc.), while any online interactions were almost exclusively on an asynchronous basis with limited opportunities to exploit any of the real time benefits afforded by mobile technologies for communication and networking.
In summary, the overall picture from these 54 participants was one in which science students were granted relatively limited autonomy over when they used mobile technologies and were restricted to largely face-to-face interactions within their own classroom or with their teachers.Given the high level of formal, classroom-based tasks in this data set (85 %), these results support the contention that many of the characteristics of m-learning are foreign to traditional classroom-based learning (Mifsud 2014;Traxler 2009).

Identifying the 'Drivers' for the Scenarios
A critical stage in the production of scenarios involves the identification of key trends or 'drivers', which shape the development of society such as environmental change, social inequality, demographic shifts and technology itself, the last of which is a broad focus of this paper.Although identified trends are recognised as important drivers of change, it is only those defined as 'unpredictable' which are selected since these serve as vectors inviting debate, discussion, difference and ultimately polarities.Technology meets this criterion well as it generates considerable debate and difference at both the micro and macro level.
However, this study is not primarily driven by an exploration of technology per se but rather by a socio-cultural investigation of the signature pedagogical affordances associated with the use of mobile technologies and their particular relevance for science educators in the future.Therefore, our first task was to examine our model to identify potential 'drivers' within the field of mobile learning.Table 1 identifies these possible drivers, all of them being capable of generating dichotomous positions, as illustrated below.
We therefore followed the recommendations in the literature on scenario planning to scrutinise these positions to identify those drivers considered to be amongst the most impactful and unpredictable (Van Der Heijden 2005).All of them are capable of generating dichotomous positions, as illustrated above, but in light of our previously discussed survey data, we deemed two themes of collaborative networking and student agency as more unpredictable in the sense that the science educational community is divided or unclear about how these constructs/ drivers might be applied in practice (Aubusson et al. 2012;Foley and Reveles 2014).
The sub-themes of conversation and data sharing were originally part of a broader category termed 'collaboration' which described those pedagogical affordances which enable individuals to engage in greater levels of networked sharing, exchange and collaborative discussion mediated through the mobile technologies.This notion of networking is similar to what Park (2011) refers to as the 'social nature of learning', which measures the degrees to which learning is an entirely independent or entirely social enterprise.Although the 54 survey participants described tasks or activities that were ranked relatively high for face-to-face conversations and discussion, they ranked online networking and data sharing as relatively low.We therefore identified collaborative networking (or 'connectedness') as one of the potential drivers to adopt in this exercise, since it offered considerable scope for alternative practices and thinking in science education around virtual and multiple conversations and collaborative data collection and exchange.
We selected student autonomy/agency as the second potential driver or variable, since this had also emerged as a problematic area in the survey.The 54 survey participants reported surprisingly low levels of student autonomy and choice (goals, content, etc.) given the dominant discourse in the literature which portray digital technologies as vehicles for greater learner agency (Burden et al. 2012;Pachler et al. 2009).
Since the purpose of the scenario planning methodology is to stimulate discussion and thinking about possible futures in science education, we identified these two potential drivers, student agency and collaborative networking, as ideal candidates for generating our scenarios.

Confirming the Drivers
Qualitative survey data from the 54 teachers was analysed according to their match with the polarities of the two nominated dimensions of student agency and collaborative networking.The survey did not mandate participants to provide an actual example of their m-learning task but 27 of the 54 participants did so for the following open-ended survey questions: & What was the topic of your learning task/activity?& What were the objectives of the topic associated with the task you have described?& What did the students do during the task using mobile technologies?& What was your role as the teacher during the task?
All 27 examples were analysed by the two researchers using the two dimensions of agency and collaborative networking to rate these critical features of the m-learning activities.Differences between researcher ratings were resolved through group consensus and consultation with other expert colleagues.Table 2 illustrates a selection of these tasks, providing a snapshot of the qualitative survey data.
Following the recommendations of others who have adopted this approach (Schuck and Aubusson 2010), we endeavoured to generate a two-dimensional model of the scenarios with four separate quadrants.Examples were then plotted according to their rankings along each axis of these two dimensions, with lower ratings at the bottom of the vertical axis or left-hand side of horizontal axis and higher ratings towards the top or right-hand side of each axis.From this analysis, each of the seven examples in Table 2 above is shown in one of the four quadrants of Fig. 2: The examples in this exercise are contained and can be described by a relatively small 'footprint' in the lower two quadrants of the diagram.Despite two outliers (examples 3 and 5), the other examples (1, 2, 4, 6 and 7) were plotted within a consistent pattern, showing limited interactions beyond the physical boundaries of the classroom and relatively solitary in nature, with little opportunity for students to share data or engage in conversations.Ratings for all 27 examples revealed similar patterns, with 10 classified in quadrant A, 12 in quadrant B, 5 in quadrant C and 0 in quadrant D.
In this way, we confirmed our two nominated drivers as entirely suitable variables to generate a two-dimensional futures model that would promote discussion and yield four sufficiently different and provocative scenarios.

The Scenarios
On the basis of the research literature, our own theoretical framework and empirical data presented in previous sections, we present four futures scenarios in the form of persuasive narratives or stories.The scenarios are deliberately written in a compelling and persuasive fashion, and all four are written with a positive perspective since scenario planning is designed to encourage consideration of alternatives that might not otherwise appeal.

The Four Futures Scenarios
In this section, we describe the four narrative scenarios, as depicted in the quadrants of Fig. 3, which present alternative possible futures for mobile technology-enhanced science learning.Each quadrant is labelled according to the agency and collaborative networking (or 'connectedness') characteristics: quadrant A: guided and scaffolded; quadrant B: simulatory and autonomous; quadrant C: connective and directed; and quadrant D: participative.Each scenario is rooted and grounded in the data we have described previously, but these are not intended to be merely descriptions of the data.Rather we have used the data as starting points to extrapolate the possible futures.Each scenario has been developed in a way that is consistent and recognisable with the data set to ensure it is plausible yet sufficiently challenging to encourage new patterns of thinking.Following the methodology recommended by Snoek (2005), each scenario is described in an extreme manner in order to differentiate them.

Scenario A: Guided and Scaffolded Science Learning
In this scenario, mobile technologies are used by science educators to underpin and reinforce traditional practices of science education (i.e. the status quo) where science is taught as a  In this way, students can return to their personal store of notes for revision purposes after the lesson is complete.Mobile technologies are seen as a highly effective and efficient way to better prepare students for high-stake testing.2)

Scenario B: Simulatory and Autonomous Science Learning
In this scenario, mobile technologies are appropriated by science educators to mediate autonomous but largely isolated learning by students whereby the device acts as an 'intellectual partner' and cognitive tool for the students (Jonassen et al. 1998).Students typically use mobile technologies to mediate relevant science processes and tasks, depicting a simulation model of authenticity (Radinsky et al. 2001), making use of class-based investigations and fieldwork as a 'practice field', albeit separate from a real science community.Use of the mobile device gives students the ability to control tools to collect data, manipulate a range of scientific variables and make predictions, thus encouraging them to think and behave like real scientists.
Students are given varying degrees of freedom and choice to explore a scientific problem or issue, and the teacher adopts the role of facilitator or guide.Rather than tightly scaffolding the learning of science to the entire class, the teacher allows students to use their mobile device to explore simulations and other resources (depending on the problem), such as animal dissection apps and 3D views of the periodic table.In this way, students work more at their own pace on a challenging, self-selected problem or issue.They use a wide range of personally chosen apps and tools to observe phenomena and collect and analyse data in and outside of the classroom, for example, to measure sunlight, gauge sound levels and observe the night sky using location-based AR apps such as Skyview (http://tinyurl.com/lonln3j).Many experiments that cannot be undertaken physically are simulated using mobile apps such as Wind Tunnel Pro (http://tinyurl.com/p3ohqmh), to gain a more accurate understanding of how scientists think and behave.In some cases, such as the use of augmented reality apps, observations provide the Fig. 3 The four scenarios intrinsic feedback on students' earlier predictions.Students typically work in small groups to tackle a science-based problem and are encouraged to use a range of generic media capture and editing tools such as the camera, the audio recorder and the video editing and animation apps to produce high quality representations of their current understandings.Assessment is based on these authentic representations rather than simple tests.

Scenario C: Connective and Directed Science Learning
In this scenario, teachers use mobile technologies to liberate students from the physical boundaries of the formal classroom, enabling them to work and interact with peers and experts beyond the classroom to articulate their ideas and negotiate shared meanings.Teachercontrolled sites, such as class blogs and wikis, microblogging services such as Todaysmeet and discussion forums in learning management systems, are used to ask questions, receive responses and exchange ideas.The teacher uses the technology as a starter to carefully scaffold and monitor realistic explorations, often based beyond the classroom.Students use their devices to collect data and to analyse it, often in teacher-selected contexts and under the careful guidance of the teacher or an external expert.Students behave more like scientists to the extent that they are working collaboratively, undertaking problem-solving activities and real-time data collection and exercises, such as the use of Bluetooth-enabled data collection tools to undertake a beach survey.Data and findings are shared with peers and teachers in externally controlled cloud-based documents.However, projects are carefully selected and externally designed to ensure students cover curriculum content.Although collected data may be shared beyond the class, it does not contribute to any wider science community projects.Most of the activities undertaken are likely to be highly scaffolded inquiry projects, or tightly controlled multi-player games or simulations, making greater use of the networking features of mobiles and the ability to tap into real-time data.Assessment is more collaborative in nature with more emphasis on peer assessment.Teachers also use data analytics to monitor students' activities in these mostly institutional spaces.

Scenario D: Participatory Science Learning
In this scenario, mobile technologies are a dynamic and reciprocal conduit to live time data, expertise and a community of real scientists that enable students to think and behave as part of the real scientific community (e.g. as citizen scientists).This is not simulated and the students are seen to have equal status and act as co-constructors of knowledge with their teachers and members of the scientific community, akin to the notion of participative authenticity espoused by Radinsky et al. (2001).Science tasks are likely to be multidisciplinary in nature, mirroring the complex, interdisciplinary processes of real-life science.Indeed existing formal school curricula may not be recognisable in this scenario.Students are immersed in real scientific communities and areas of interest (e.g. a nature reserve) where they undertake an extended in situ work experience using the technology to share, analyse and interpret their own and others' data and maintain contacts with their peers and with experts in the real world, who validate and credential the learning.Students are asked to think and behave as scientists, and their findings are used and valued by the scientific community (e.g. in collecting real time data as citizen scientists).Students in this scenario use networking tools and social media apps like Facebook, Instagram and Twitter to pose questions and share their predictions and interpretations with peers (in and beyond their own cohort) and with other scientific experts.Connective, augmented reality apps, multi-player games and immersive learning tools enable students to understand complex ideas and concepts at their own pace and in many cases these act to mediate student's learning, independently of the teacher.Examples include use of the previously discussed Sense-it app (Herodotou et al. 2014), nQuire app (Jones et al. 2013) and iSpot app (Scanlon et al. 2014).Teachers may use data analytics to monitor students' activities in these spaces and assess their development in real-time.Self and peer assessment become critical, as learners engage in more complex environments and receive increasing opportunities for formative feedback.Learners self-monitor real-time data to self-assess their progress, for example, data from multi-player games or data from activity tracking apps monitoring their movement or performance.

Discussion
The previous section presented four different alternatives for the use of mobile technologies in science education and therefore addressed the main research question of this paper: What possible futures might present themselves to science educators interested in harnessing the potential of mobile technologies?In this section, we discuss the implications of these four scenarios and consider how they might be used to encourage science educators to exploit more fully the affordances of mobile technologies beyond current patterns of usage.The research literature in mobile learning makes a persuasive case for using mobile technologies to support collaboration and networking between learners and to overcome many of the problems and restrictions associated with traditional practices of education that are bounded and parochial in character (Norris and Soloway 2011;Parry 2011;Peluso 2012).It also highlights how mobile technologies offer learners opportunities for much greater autonomy and independence in relation to their own learning, reducing the need for the teacher to control and direct students (e.g.Beauchamp et al. 2015;Burden et al. 2012).Despite these twin opportunities for greater collaboration and autonomy mediated by mobile technologies, and in contrast to the innovative science education case studies cited earlier, the data presented in this study indicates that students are learning science in relative isolation (individual or small groups) and in a minimal number of locations (e.g.classrooms, excursions), typically operating in quadrants A or B from our four scenarios.This is in agreement with other researchers who highlight that teachers use a relatively limited range of mobile learning pedagogies, largely within traditional face to face learning contexts (Herodotou et al. 2014), and are designing tasks that use mobile technologies to 'fit' into traditional notions of formal, scheduled, institution-based learning (Cochrane and Antonczak 2014;Rushby 2012)essentially pointing to a future of science learning in quadrants A and B. These trends raise a number of questions that we condense into one main focus for the remainder of this discussion: What are the inhibitors and enablers for science teachers to use a wider range of connected m-learning practices akin to quadrants C and D?
What Limits Science Educators from Using a Wider and More Connected Range of m-Learning Practices?
The practice of inquiry-based learning is well established amongst science educators (Krajcik et al. 2000).This suggests a willingness to grant students some degree of autonomy and independence in their learning, as evidenced in the high frequency of science tasks from the data that were categorised in quadrant B. Therefore, we may assume that science educators are not averse to using m-learning to support student agency but rather to extending its use beyond the traditional face to face contexts and settings they are currently familiar with.This may be linked to the fact that science educators, like educators from other disciplines reported elsewhere (e.g.Kearney et al. 2015), are limited by their personal inexperience and skills deficit in the use of networking tools such as social media and, more crucially, application of these tools to current pedagogical practices such as inquiry-based learning.One of the few examples of connected pedagogies cited in this study featured the dynamic use of the social media platform Twitter which science students used to share the findings from their laboratory work with the wider world (see Table 2, example 3).In this example, the teacher was himself an active user of Twitter, supporting the contention that educators are more likely to understand and exploit the opportunities of new technologies from prior experience and personal use of the technology (Schuck and Kearney 2008), in this case for connected science learning.
This reluctance to embrace the connected and virtual affordances of m-learning may also be associated with the dominant transmissionist paradigm of the institutional virtual learning environment (VLE) or learning management system (LMS), where it might be considered the default position for non face-to-face teaching (Repetto 2013).Although they may be virtual in the sense of being online spaces, VLEs are essentially bounded, closed and contrived environments, which bear little resemblance to the open, flexible and transitory virtual spaces that characterise mobile learning.The mindset which underpins the use of the institution VLE is at variance from that which supports the more anarchic, user-oriented mobile world with which students are more familiar and more likely to use if they are given a choice (Erstad and Sefton-Green 2012;Ito et al. 2010).If science educators conceptualise collaboration through the lens of the VLE, it is hardly surprising that they avoid using m-learning to enhance connectedness, since community learning (Wenger 1998) is informed by an entirely contrary set of learner-centric values and pedagogical philosophies which are averse to the bounded, teacher-dominated pedagogies of institutional online learning environments.
Following on from this premise, it may also be the case that science educators may be constrained in their wider use of m-learning by epistemological concerns around what constitutes 'valid' scientific knowledge.For example, despite the proven value of using mobile technologies to support crowdsourcing of data and citizen science projects (e.g.Newman et al. 2012;Pecl et al. 2015), concerns have been raised about the reliability and trustworthiness of such projects (e.g.Alabri and Hunter 2010).This includes scepticism about data reliability from professional scientists themselves, though interestingly these concerns diminish when they are actively involved themselves in these types of citizen science projects (Pecl et al. 2015).
These constraints may reduce the willingness of science educators to develop more extended and connected pedagogies that extend beyond their own visible orbit, thus depriving students of opportunities to undertake educational activities in networked online spaces, despite evidence which demonstrates these same students are already active in such contexts (Erstad and Sefton-Green 2012;Ito et al. 2010).Indeed, these networked practices are becoming increasingly authentic in science education as they mirror the connected behaviours of real scientists who invariably work in teams across the globe in private, semi-private and public networks, using a wide variety of tools to both collect and disseminate their findings.
What Is Needed to Encourage Science Teachers to Extend Their Students' Connectedness Behaviour in Ways that Would Include Practices in Quadrants C and D?
Whilst the data from this study and other literature (e.g.Herodotou et al. 2014) indicate that current use of m-learning by science educators is rooted primarily in face-to-face contexts, it also demonstrates how with relatively simple adjustments to the task designs, teachers could broker more opportunities for students to cross the boundary between their digital worlds and the physical realities of formal education (Royle and Hadfield 2012), effectively moving practices into quadrants C and D. For example, by empowering the students to use an interactive app or social media tool such as Twitter, in example 5, students would enter the more collaborative and participative quadrant D by posting real questions and problems for real scientists to respond to, rather than simply consuming their expertise in a passive manner as they appeared to be doing.Indeed, many of the examples from the data set had a similar potentiality to be shifted from the lower two quadrants to the upper two quadrants (i.e. the boundaries are permeable), usually by considering opportunities to collaborate and network and by thinking about learning tasks as multi-staged events to be completed in more than one place or time (for example, 'seamlessly' linking an in-situ field investigation with networked sharing of data and follow-up learning conversations).This picks up on the 'seamless' learning theme covered in the literature which indicates how mobile technologies have the potential to assist teachers and students in crossing boundaries between various settings and contexts to extend and continue their science learning beyond the formal, physical classroom (Toh et al. 2013) and beyond the familiar online institutionalised VLE or LMS.
Indeed, we propose that flexible time/space configurations could be applied to any of the four scenarios, particularly multi-staged tasks across a blend of contexts.For example, teachers following a flipped learning pedagogy (Herreid and Schiller 2013) might encourage students to view their instructional podcast (quadrant A features) using a negotiated time/space configuration, 'at their own time, pace and place' before class.The rationale for this type of pre-class task is to reduce the need for instruction in subsequent classes, allowing for precious, formalised classroom-based time to be used for more active, autonomous, inquiry-based work (e.g.quadrants B or D).
In order for science educators to recognise these opportunities, however, they need to be cognisant of the various pedagogical affordances that characterise m-learning and this is more likely to transpire if teachers are themselves connected and active in these online contexts.This has major implications for the preparation and in-service training of educators to use mobile technologies more seamlessly, in line with the practices and habits that characterise students' personal use of their own devices.A recent study which investigated models of professional learning associated with the use of iPads by primary school teachers in the UK concluded that the intuitive nature of the device precluded the necessity for formal training (e.g.courses) and revealed how educators are more likely to learn through more informal approaches such as discussion with their peers, playfulness (e.g. at home) and often by learning alongside their own students.This would appear to have relevance for educators across all sectors and suggests that science educators could use mobile technologies in this fashion.
Perhaps most challenging for all educators, not just science teachers, is to consider what elements of quadrant D might be achievable and desirable, given that students are demonstrably undertaking informal learning in this quadrant already (e.g.participation in citizen science projects), and the fact that this quadrant is predicated on participation in communities of practice that are not rooted, or even grounded, in traditional school structures or practices.For science educators to move into this quadrant, they may need to think outside the existing curricular practices and debates that characterise institutional learning and this is both risky and confronting (Repetto 2013).They need to reconsider concepts of space and time which are the defining boundaries of institutional learning (Traxler 2009) and consider the ubiquitous learning that is already apparent in game-based learning (e.g.Minecraft), where learners are accustomed to high degrees of independence, autonomy and agency but are equally aware of the importance of collaboration, teamwork and interdependence.In these fluid, non-hierarchical spaces, learning cannot be controlled or regulated in the same way classrooms are, and therefore teachers have to be willing to cede some authority and control to learners to take more responsibility for their own learning, including control of the context and resources used.There needs to be acknowledgement that science learning can be self-initiated and possibly unplanned and may occur without a teacher in the presence of peers or science experts in unpredictable local and global communities.There is also a need to understand how m-learning can provide instantaneous, 'just in time' feedback (e.g.personal health data provided through wearable activity-tracking devices), allowing learners to select and apply information to self-assess their own progress.

Conclusion
We acknowledge that other emerging technologies may well have a profound influence on science learning in the future, for example, learning analytics, 3D printing, games-based learning and wearable technologies (Johnson et al. 2013).However, given the current interest and investment in mobile technologies, it is timely to explore the future of science learning in light of the distinctive features of mobile-intensive pedagogies.In some ways, the tendency of using mobile technologies to duplicate traditional pedagogical strategies has been influenced by the underpinning design of educational apps that typically informed by an information transmission model of learning, or behaviourist, drill and practice approaches (Murray and Olcese 2011).Indeed, Mifsud (2014) and Traxler (2009) argue that many of the features of m-learning are in conflict with traditional classroom-based learning, making the effective use of m-learning a challenge for educators.In this study, we address this problem by rationalising and developing four future scenarios that help science educators forecast how they might choose to exploit two distinctive pedagogical aspects of m-learning: student agency and collaborative networking.Unlike some macro-level driving forces that cannot be easily influenced by teachers (e.g.national policy or global trends), each of these two micro-level variables falls within the locus of control of individual teachers.The scenarios reveal a range of pedagogical opportunities for science education, highlighting connection between peers and the science community, participative authenticity and student autonomy.There is a need for science educators to understand the nature of learning in mobile contexts, especially the use of 'mobile technologies to provide a participatory structure and architecture to support communities of learners' (Lai et al. 2013, p. 421).This paper advocates further studies into how informal science learning can complement formal science learning, the changing nature of teacher and student roles in these blended environments, and use of emerging mobile technologies to engender agency and networking of science learners.
-based subject and technology is employed to make teaching and learning more effective and efficient.The main emphasis lies with the transmission of accepted scientific principles and knowledge, and this is undertaken most effectively through teacher-directed access to information sources such as YouTube video demonstrations, podcasts, e-textbooks and the use of 'skill and drill' apps such as science quiz apps.Mobile devices are used extensively in the classroom and laboratory to free students from traditional note-taking and drawing exercises, and these are replaced by camera capture and digital annotation tools, such as stand-alone mind-maps, electronic worksheets and e-books.Teachers control the content, objective and pace of lessons, including tightly scaffolded, recipe-style science investigations.They administer live polls to students to test their immediate understanding of a concept (e.g. through an app like Socrative) and to gain feedback about what students know or need to know better.Teachers present and explain scientific ideas and principles using whole-class presentation apps such as ShowMe, Explain Everything, Nearpod and Zeetings which enables them to scaffold the content delivery, ensuring all of the class are working at the same pace, schedule and place.Students work mainly with the teacher and their classroom peers, only using the Internet to access recommended information or to email the teacher their work.In class, students are encouraged to use their mobile device to capture and annotate notes made by the teacher on the interactive whiteboard or examples of experiments or demonstrations which cannot be undertaken by the students for reasons of efficiencies of time or health and safety.

Fig. 2
Fig. 2 Qualitative data plotted against twin variables (numbers refer to the examples in Table2)

Table 1
Potential 'drivers' identified for this study

Table 2
Sample m-learning tasks from study derived from qualitative survey data