Students need more practice with spatial thinking in geoscience education: a systematic review of the literature

ABSTRACT

Myriad research in a variety of contexts shows spatial skills benefit students; however, they are not given enough attention in classroom instruction. In this review we systematically explore geoscience education literature focusing on spatial interventions to answer research questions on trends in spatial skills and other characteristics. We narrow our attention to studies published since numerous calls to action to teach more spatial skills in STEM education, resulting in 28 articles for review. To analyse and compare these studies, we organise the literature into a framework of geoscience-relevant spatial skills. We reviewed interventions and assessments to determine the aligning spatial typology skills. Themes of coursework, mapping, and modelling emerged; sub-themes include sketching, gestures, physical models, computer models, and curricular interventions. In the articles reviewed, just over half of the skills identified were intrinsic skills. Future geospatial research should explore how best to incorporate spatial skills into the classroom over long time periods and should focus on the process of spatial reasoning and the strategies students use when problem-solving about spatial phenomena, especially at the elementary and secondary school level. Educators can use the resources outlined in this review to engage in spatialising their curricula.

KEYWORDS:

• Geoscience
• geoscience education
• spatial learning
• geospatial education
• spatial typology

There are a plethora of challenges involved in assuring the students of today are prepared to tackle the problems of the natural world that lie ahead of us. Addressing these issues requires students to think about Earth processes in a way that incorporates the spatial nature of the problem and of the geoscience discipline (Herbert, 2006; Turcotte, 2006). However, educators and researchers may struggle to understand the existing spatial thinking literature without a consistent framework to draw useful conclusions for practices and future research directions for spatial curricular activities. This inconsistency highlights the need for a review of the literature in terms of a single framework.

There has been a great deal of research on spatial thinking skills, from which we can draw three conclusive statements to ground our understanding and approach to synthesising the literature we reviewed. First, spatial skills are strongly related to success in science, technology, engineering, and mathematics (STEM) fields (Buckley et al., 2018; Wai et al., 2009). Early in the 20th century, spatial skills were recognised as important in engineering and other trade fields (Smith, 1964). These skills have been shown to have a positive statistical relationship with success and retention in STEM fields (Shea et al., 2001; Super & Bachrach, 1957; Wai et al., 2009). For example, a longitudinal study called Project Talent that followed students from high school through their careers identified a clear connection between early spatial abilities and selection of and success in STEM fields (Super & Bachrach, 1957). These findings increased interest in the already developing field of research on the relationship between skills in STEM and spatial skills.

Spatial thinking is particularly important in geoscience given the spatial nature of the discipline (Dutrow, 2007; Kali & Orion, 1996; Kastens & Ishikawa, 2006; Liben & Titus, 2012; Manduca & Kastens, 2012; Manduca & Mogk, 2006). The value of spatial skills in geoscience has been recognised by the geoscience community as exemplified by the two books of special papers published by the Geological Society of America to address the spatial nature of the discipline and how to support these skills in the classroom (Kastens & Manduca, 2012; Manduca & Mogk, 2006) and recent recognition of spatial thinking as an important component for future geoscience education research (Ryker et al., 2018; St. John et al., 2020). Furthermore, Colaianne and Powell (2011) showed that students in geoscience perform higher on measures of spatial thinking compared to students in liberal arts classes and Hegarty et al. (2010) found that geoscientists report higher use of spatial skills than professionals in other STEM fields. Studies by Orion et al. (1997), Titus and Horsman (2009), and Ormand et al. (2014) all reported that students improved on measures of spatial abilities after regular instruction in a geoscience course, which further suggests that geoscience is a particularly spatial discipline.

Second, research has shown that there are individual differences in spatial thinking (Liben et al., 2011; Sanchez & Wiley, 2014). These differences often manifest as gender differences, but they are not biological. No genetic cause has ever been identified (Newcombe, 2010; Newcombe & Stieff, 2012), however there is ample evidence of gender differences in the literature (e.g., Baenninger & Newcombe, 1989; Buckley et al., 2018; Dyar, 2012; Linn & Petersen, 1985; Lowrie & Diezmann, 2011; McGee, 1979). Newcombe and Stieff (2012) comment on the presence of gender differences and provide research that suggests these differences do exist for two spatial thinking skills in particular – mental rotation and drawing a line in an orientation against a conflictingly tilted background – but that these skills are not fixed. Research has found that gender differences can be eliminated with practice (Dutrow & Ormand, 2019, September 22-25; Uttal et al., 2013). There is also evidence that gender has no influence on the magnitude of improvement (Bitting et al., 2018). Gold et al. (2018b) found that differences in students’ spatial ability may relate to activities they participated in as children, which further supports the notion that spatial skills are experientially built rather than genetic.

Third, spatial thinking skills can be improved with practice (Newcombe & Stieff, 2012; Uttal et al., 2013). Not only can practice with spatial skills improve spatial thinking ability but such practice may also be able to support a scientifically accurate understanding of geoscience concepts. For example, Black (2005) found poor spatial understanding may influence Earth science misconceptions and Vosniadou and Brewer (1992) found that students can experience conceptual change due to the development of correct mental models of Earth that require that students imagine the shape of Earth in their mind’s eye, which requires spatial thinking.

Despite the importance and wealth of research on spatial thinking, the U.S. National Research Council (National Research Council [NRC] & Geographical Sciences Committee, 2006) in summarising the previous literature decidedly stated that even though spatial skills are ‘fundamental to everyday life, the workplace, and science’ (p. 229), they are not being taught in the elementary and secondary classrooms. Interestingly, there is not enough geoscience education in elementary and secondary classrooms either (Orion, 2019). For example, Lewis and Baker (2010) commented on the lack of qualified geoscience teachers at the middle and high school level and sought ways to address this concern. In a compelling editorial, geoscience education researcher Karen McNeal (2010) laments:

One cannot help to make the connection between the lack of geoscience curriculum in K-12 schools, especially at the secondary level, and the low recruitment of students, especially of minorities, into the geoscience fields; short-sighted public policy and management decisions; limited teacher knowledge of the geosciences; and the overall limited public understanding of the spatial and temporal complexity of complex Earth science systems and processes. (p. 197)

These limitations can be helped by the use of reform-based strategies in undergraduate coursework, which models student-centred instruction for future teachers (Manduca et al., 2017). Unfortunately, at present, it seems geoscience education is still falling short in spatial instruction (Ormand et al., 2017) and value in education and society (Orion, 2019).

In line with K. S. McNeal’s (2010) commentary, Wilson (2014) projected a shortage of about 135,000 geoscientists by 2020, despite an increase in recent graduates by the end of the decade. This decline in geoscientists has led to some concern over whether there will be enough qualified geoscientists to fill the number of career opportunities that will be available (Perkins, 2011). It is also important to consider the growing need for solutions to the imminent existential threats posed by the global climate crisis and natural disasters, which require geoscientists’ perspectives. These two issues, namely instruction in spatial skills and preparation of geoscientists, may seem unrelated; however, they are similar in that neither of them is given enough attention in elementary and secondary classrooms and would likely equally benefit students’ learning since research suggests that geoscience is a particularly spatial discipline compared to other STEM fields.

Exploring the factors that attract minority students to geoscience, Levine et al. (2007) found that students pursuing geoscience tended to have more experiences with engaging curriculum and more experiences outside in nature, among other factors. Levine et al.’s (2007) research on students’ choice of geoscience majors seems to align with K. S. McNeal’s (2010) belief that a strong curriculum that engages all students is needed. Incorporating activities that require spatial thinking into the curricula may be one way to engage students in geoscience and support their pursuit of geoscience majors and careers.

There is research that can help improve the instruction of spatial skills in geoscience (and other STEM fields), yet these studies are not as useful to elementary and secondary educators because they do not necessarily draw either instructional lessons learned or pedagogically relevant conclusions, and therefore perpetuate the lack of spatial skills in elementary and secondary education. Materials for teachers exist, including resources such as Spatial Ability: A Handbook for Teachers (Clausen-May & Smith, 1998), Visualisation in Undergraduate Geology Courses (Reynolds et al., 2005), visualisations available through NOAA (Steffen et al., 2014), materials that have been produced by partnerships with the Spatial Intelligence and Learning Center (SILC),¹ and lesson plans in practitioner journals such as The Earth Scientist (e.g., Burck & Soeffing, 2019; J. D. Moore et al., 2015a, 2015b; Olds & Charlevoix, 2019). Websites such as the Science Education Resource Center (SERC),² GET Spatial,³ GeoMapApp⁴ (Goodwillie & Kluge, 2013), and USArray⁵ (Butler et al., 2011) among others are useful resources as well. Resources to teach more spatially are available but they are few. It can be challenging to find these resources without already knowing they exist. Furthermore, many activities that have been developed at the undergraduate (tertiary) level may be adaptable to K-12 education.

Newcombe (2010) addressed the lack of readily available resources and discussed the extent to which spatial skills truly influence skills in STEM fields, particularly for students in secondary education. She points out that the National Research Council & Geographical Sciences Committee’s (2006) report noted that research has yet to determine what kinds of experiences improve spatial skills, how to infuse spatial thinking into curricula, and how best to use GIS and similar technologies with young children (Newcombe, 2010). She also outlined three practices teachers can follow at all levels of education to improve their inclusion of spatial skills in the classroom. Summarised here, she explained that teachers need to (1) be aware of what spatial skills are and resources they can use to support their development, (2) avoid making students nervous about spatial practices, and (3) increase the amount of time students and children engage in modelling and age-appropriate spatial activities (Newcombe, 2010, p. 33). This is one example of many suggestions in the research to make learning more spatial and improve students’ spatial skills, highlighting the growing recognition of the importance of spatial skills – however, these suggestions do not seem to be actively incorporated into instruction.

Despite increased attention, spatial skills are still not easily defined. Newcombe and Shipley (2015) created a framework, further described below, that provides a holistic understanding and elaboration of a structure for understanding spatial skills, but no clear and concise definition. This lack of a definition for spatial thinking may cause confusion about how to include it in instruction. Uttal et al. (2013) note that neither spatial thinking nor spatial ability is easily characterised; both comprise a variety of specific skills, processes, and ways of thinking. This is evidenced by the myriad uses of terms across the literature – such as spatial thinking, spatial ability or abilities, spatial skills, spatial cognition, or spatial learning – and in many cases, these terms are used interchangeably. Table 1 shows three traditions and definitions of different research perspectives that have been taken to develop an understanding of spatial thinking skills. Newcome and Shipley draw on all of these research traditions to develop their framework.

Table 1. Spatial thinking research traditions with a description of their background and purpose and a sample definition from an exemplar article.

Discipline	Background and Purpose	Sample Definition
Cognitive Psychology	Most general; not tied to a particular discipline but applicable to many (physics, mathematics, geoscience, etc.) Some approaches focus on spatial thinking as factors of general intelligence (e.g., Buckley et al., 2018) to better understand its role in intelligence whereas other approaches focus on real world applications and practices such as use of maps (e.g., Jirout & Newcombe, 2014).	‘Spatial thinking involves thinking about the shapes and arrangements of objects in space and about spatial processes, such as the deformation of objects, and the movement of objects and other entities through space. It can also involve thinking with spatial representations of nonspatial entities, for example, when we use an organizational chart to think about the structure of a company or a graph to evaluate changes in cost of health care’ (Hegarty, 2010, p. 266).
Geoscience Education Research (GER)	Specifically tied to spatial skills commonly used in various geoscience practices or tasks, such as recognising, describing, and classifying the shape of an object; describing the position and orientation of objects; making and using maps; envisioning processes in three dimensions; and using spatial thinking strategies to think about non spatial phenomena (Kastens & Ishikawa, 2006, p. 53).	‘ … thinking about objects or processes or phenomena in space, the kind of thought processes that we broadly call spatial thinking’ (Kastens & Ishikawa, 2006, p. 53).
Cognitive Neuroscience	This definition assumes spatial thinking is composed of intrinsic and extrinsic components, and only defines these two components. This line of research connects intrinsic and extrinsic spatial thinking to physiology within the brain, therein providing biological evidence of the difference in these skills.	‘Objects in space have intrinsic and extrinsic properties. Intrinsic properties refer to the object itself, whereas extrinsic properties refer to the object in relation to an external referent’ (Chatterjee, 2008, p. 227).

In this paper, we adopt the definition from the NRC (2006):

Spatial thinking, one form of thinking, is a collection of cognitive skills. The skills consist of declarative and perceptual forms of knowledge and some cognitive operations that can be used to transform, combine, or otherwise operate on this knowledge. The key to spatial thinking is a constructive amalgam of three elements: concepts of space, tools of representation, and processes of reasoning. It is the concept of space that makes spatial thinking a distinctive form of thinking. (p. 12)

This approach acknowledges the composite nature of the construct that is recognised by many researchers (e.g., Buckley et al., 2018; Newcombe & Shipley, 2015; Uttal et al., 2013) while still attempting to create a clear definition. Although we aim to focus our understanding through the NRC’s definition, when discussing specific research, we use the original terms from the study under consideration.

The purpose of this paper is to systematically review research on spatial geoscience interventions in classrooms and psychology laboratories that would be useful to elementary, secondary, and undergraduate (tertiary) science educators in terms of a contemporary spatial framework. We look for trends, patterns, relationships, and an overall picture from the rapidly increasing knowledge base (Borrego et al., 2014, p. 46) to identify gaps and future directions for research on spatial geoscience interventions, as well as to draw pedagogically relevant conclusions for educators. We specifically review studies published since Wai et al. (2009), who concluded that ‘ … the kind of research that is needed now is in how to utilize spatial ability for student selection, instruction, and curriculum design and in how to refine educational interventions and procedures on the basis of individual differences in spatial ability’ (p. 817). These researchers’ conclusions suggest that spatial research ought to focus on how to implement these skills into elementary and secondary STEM classroom instruction to support all students’ spatial thinking skills. This type of research is still needed (Ryker et al., 2018; St. John et al., 2020). To act on these researchers’ suggestions for future research on the development of instructional and curricular materials in geoscience, it is important to systematically review the literature in order to understand the research that has been done. We have developed two questions to guide our literature review on spatial geoscience interventions:

Research question 1: What trends are observed in recent literature on spatial intervention studies in geoscience in terms of the purpose of the study, the geoscience practice identified, strategies used to practice spatial thinking, and setting and level of education of each study?

Research question 2: How and to what extent does the existing literature on spatial geoscience interventions align with the Newcombe and Shipley spatial typology framework, and what gaps remain?

Theoretical framework

The spatial thinking typology developed by Newcombe and Shipley (2015) provides an organisational structure to understand spatial skills research in geoscience education. This framework was developed through the study of geoscience practices and is the result of research at the intersection of geoscience education and cognitive psychology supported by SILC, which has championed collaborations among Earth scientists, teachers, educational researchers, and cognitive psychologists. Their work and that of others, such as researchers affiliated with SERC at Carleton College, have provided a great deal of insight on the connection between spatial thinking and geoscience.

In their typology, Newcombe and Shipley argue for a top down analysis of the nature of spatial thinking, rather than a bottom-up inductive approach; they begin by asking what spatial thinking really is (p. 2). Drawing from the fields of cognitive science, neuroscience, and linguistics through the work of Chatterjee (2008), they argue for a distinction between intrinsic and extrinsic aspects as well as static and dynamic aspects. Newcombe and Shipley (2015) define intrinsic-extrinsic static-dynamic skills as follows: (a) intrinsic-static spatial skills involve the ability to identify objects as members of categories; (b) intrinsic-dynamic spatial skills involve the ability to transform objects in some way and imagine the resulting configuration of the object under consideration; (c) extrinsic-static spatial skills involve the ability to recognise the spatial location of objects with respect to a frame of reference; and (d) extrinsic-dynamic spatial skills involve the ability to maintain a stable representation of the world during navigation and to enable perspective taking (p. 5).

From their research at the intersection of cognitive science and geoscience education research (GER), Newcombe and Shipley (2015) identify 11 spatial skills specifically used in geoscience and categorise them according to the above definitions. Table 2 outlines these geoscience-specific spatial skills paired with brief descriptions as they appear in the original text along with our abbreviated description, when applicable, for the sake of discussion.

Table 2. Geology-specific skills organised into more general spatial skills as outlined by Newcombe and Shipley (2015, p. 10).

Within – Object (Intrinsic) Spatial Relations		Abbreviated Terms for the Present Study	Example of geoscience practice that requires each spatial skill
Static	Disembedding: Isolating and attending to one aspect of a complex display or scene.		Identifying fossils in a matrix.
Static	Categorisation: Learning categories based on spatial relations.		Identifying the type of fossil.
Dynamic	Visualising 3D from 2D: Understanding 3D spatial relations presented in a 2D image or drawing.	Visualising	Visualising the 3-dimensional shape and structure of a fossil from a 2D image (or shallow impression).
Dynamic	Penetrative thinking: Visualising spatial relations inside an object.		Imagining the internal structure of the fossil or a cross section.
Dynamic	Mental transformations: Visualising how an object will change over time.		Imagining the change in the shape and structure of the fossil since it first deposited.
Dynamic	Sequential thinking: Visualising the product of a series of transformations.		Imagining the changes over time experienced by a fossil since it was deposited.
Between – Object (Extrinsic) Spatial Relations
Static	Locating self and other objects: Identifying the past or present position of objects in real space and on maps.	Locating objects	Identifying where you found a fossil.
Static	Alignment: Reasoning about spatial and temporal correspondence (two important cases are scaling and the use of space as a proxy for time).		Aligning two rock layers because they have the same fossil assemblage.
Dynamic	Perspective taking: Visualising the appearance of a scene from a different vantage point.		Imaging where the fossil is located from different locations.
Dynamic	Relations among objects, including self, in space: Visualising the spatial relations defined by multiple locations (e.g., distance between 2 points and angle formed by 3 points; important for making and using maps).	Relations among objects	Imagining where the fossil is in relation to a certain location or another fossil.
Dynamic	Updating movement through space: Visualising movement of an object relative to other objects (for self this would include route planning).	Updating movement	Imaging the path you take to find the fossil.

Note. Adapted from ‘Thinking About Spatial Thinking: New Typology, New Assessments’ by N. S. Newcombe and T. F. Shipley, 2015, in Studying Visual and Spatial Reasoning for Design Creativity, pp. 188–189. Copyright 2015 by Springer.

We selected this framework due to its relevance to our topic of study. This recent framework is created from the SILC collaborations and therefore aligns with our geoscience education focus. As noted by P. M. McNeal and Petcovic (2020), other frameworks exist, yet their structure and organisation were not as relevant to our focus on spatial skills in geoscience education. Kastens and Manduca (2012) developed an organising framework that identified skills in geoscience and aligned them with cognitive science research. However, this approach would not allow us to organise the existing literature into useful spatial categories for comparison purposes, and would instead result in geoscience skill categories, which through their framework, would then connect to spatial thinking skills. Two compelling frameworks for spatial skills, McGrew (2009) and Buckley et al. (2018), were developed with respect to general intelligence, which is not the focus of this review. A summative framework to guide future education research by St. John et al. (2020), which included spatial and temporal reasoning as one future research direction, supports this review but was not used as a guiding framework because it does not offer a way to categorise spatial skills research.

Methods

To begin our systematic review of the geospatial literature, we brainstormed ten journals that would likely be read by geoscience educators from a variety of levels of education and backgrounds. These journals cover fields of psychology and education research and journals that publish articles that combine elements of both fields. We hand searched (Borrego et al., 2014) the following journals, which we think are representative but not necessarily exhaustive: International Journal of Science Education, Journal of Geoscience Education, Journal of Educational Psychology, Science Education, Journal of Research in Science Teaching, Research in Science Education, The Earth Scientist, Geosphere, Cognition and Instruction, and Cognitive Science. We also used a ‘snowballing’ effect to look at additional articles, in that after identifying a potential article, we looked within its reference list for other relevant articles and used Google Scholar to help identify subsequent articles that reference the earlier work (Borrego et al., 2014).

Through a general search of the university library knowledge web, online databases were used to search the ten journals we identified. Keyword searches were conducted in each journal using the keywords ‘spatial skills,’ ‘geoscience,’ and ‘geospatial’ followed by ‘intervention.’ We limited our search to articles published from 2009 to 2020 with the intent of focusing on articles that could have addressed Wai et al.’s (2009) call. We also included only peer reviewed studies that investigated the efficacy of a spatial intervention, rather than those that simply presented an intervention for use.

Article titles were screened for topics related to spatial thinking in geoscience and related fields, which resulted in 84 studies. These studies were further reviewed for a description of an intervention in a psychology laboratory or geoscience classroom in their abstract. If the study explored a spatial intervention, we included it, even if there was no spatial measure. This resulted in 38 studies. Further consideration resulted in a narrower focus on only spatial interventions in geoscience topics and contexts, such as but not limited to structural geology, topographics maps, water and the water cycle, landform identification, atmospheres, hydrogeology, sedimentology, fossils and past environments, and human influence on the environment. Using this method, 28 articles met our criteria. These various boundaries (i.e., geoscience topics, peer-reviewed journals, etc.) provide a relatively large-grained view of the field compared to one looking at more specific skills or situations (e.g., a review specific to use of gestures in geoscience or of geoscience fieldwork.)

We included psychology studies because they provide evidence of the effectiveness of various spatial strategies in a controlled environment. It has also been argued that psychology students have a similar level of understanding of geoscience concepts and spatial thinking as novice geoscience students (Myer et al., 2018). Although not a perfect alignment – after all, students who are drawn to geosciences may in part enjoy it because of prior experience and success with spatial thinking skills – such studies can still inform development of future instructional materials.

In order to limit the scope of the review, we focused only on in-class interventions. Some past reviews have focused on fieldwork either in general (e.g., Munge et al., 2018, for outdoor fieldwork in multiple disciplines; Mogk & Goodwin, 2012, for geosciences-specific fieldwork) or with more specific foci (e.g., situated learning by Donaldson et al., 2020). We did not explicitly search for articles pertaining to spatial thinking with representations, diagrams, or analogical reasoning. These topics are strongly related to both spatial skills and elementary and secondary science education, however, a more in-depth consideration of these branches of research is beyond the scope of the present study. The resulting articles predominantly took place at the undergraduate level. We also found that there are many practitioner articles that outline spatial interventions for use in K-12 classrooms. Many of these articles focus on geoscience activities that are inherently spatial, yet do not explicitly discuss the spatial nature of the activity nor focus on systematically investigating any changes in spatial thinking the activity may offer. There are also many dissertations and theses that focus on improving spatial thinking in geoscience, however they are not as easily accessible to readers and therefore we focused on peer reviewed journal articles.

Our review also excluded early childhood education books and studies, which restricted our search of journals (e.g., Davis & Spatial Reasoning Study Group, 2015; Piaget, 2013; Verdine et al., 2017). Additionally, there have been a great deal of map-related psychology laboratory studies that were not reviewed here due to the limitations we set by our journal selections. While studies from a wider range of journals could inform the development of geoscience interventions, they were beyond the scope of our review. For example, Journal of Experimental Psychology, Cognitive Research: Principles and Implications, Visual Cognition, and Current Directions in Psychological Science, to name a few, all contain articles that may offer insight into the efficacy of different strategies and students’ spatial ability with various skills related to the typology, but they are less likely to be read by geoscience educators. It is in these journals that many researchers and educational partners of SILC, including Newcombe and Shipley who created the typology we use, have published (e.g., Atit et al., 2016; Blacker et al., 2017; Cohen & Hegarty, 2014; Weisberg & Newcombe, 2016, 2017, 2018). We also excluded research from fields that were similar and related to, but different from, geoscience, such as geography and soil science.

As studies were collected, we abstracted the information to facilitate comparison (Borrego et al., 2014). This process included information about the setting, number of participants, purpose, intervention, procedure, measures, intrinsic and extrinsic spatial skills that align with the typology, and results. For each task that students were required to complete in a given study, the first author identified the corresponding spatial skills in the typology and discussed any uncertainties or concerns with the second author. The spatial skills outlined in the typology are complex and typically more than one skill is used in completing a task. For example, mapping requires a suite of intrinsic and extrinsic spatial thinking skills. The intrinsic skills are visualising, disembedding, and categorisation; the extrinsic skills are locating objects, relations among objects, and updating movement. So, a study that sought to improve students’ mapping abilities might focus on any one or more of these skills depending on the intervention. Furthermore, most authors did not classify their work in terms of Newcombe and Shipley’s typology (many could not, given the studies’ respective publication dates).

During the process of aligning studies with typology skills it became apparent that some researchers focused on the influence of coursework and whole curricula, some generally focused on mapping practices, and others focused on modelling of geoscience related concepts. Map-related studies that were included involved various uses of maps, such as wayfinding, topographic map reading and use, and topographic and geologic map creation. These studies were limited to those in classrooms, rather than in situ work of map reading in the field. Studies that were categorised as modelling involved students creating or working with a model of geological phenomena, such as a geologic block diagram, sketching of geological phenomena, and frequently computer models of geologic phenomena. Coursework studies could include elements of mapping and modelling studies, but were specifically described as being conducted in a classroom. In all three cases (mapping, modelling, and coursework), there is significant variation of detail across the different studies (e.g., in terms of what types of maps, models, or activities used).

These emergent themes, which we call geoscience practices (e.g., modelling), provided an organisational structure for our review instead of the typology skills (e.g., mental transformations) because the typology skills varied greatly and often multiple were relevant to each study. We also identified five sub themes of strategies used by researchers to improve students’ spatial thinking skills, which were sketching, gestures, physical models, computer models, and curricular interventions.

While reading the results, we encourage readers to visit the Appendix (Table A1), which contains the citation of each GER and psychology laboratory study included in the review. It also contains a number of details for each study that would be too cumbersome to include within the main text of this piece, including the level of education addressed, the setting, the geoscience practice, the types of intrinsic and extrinsic geospatial skills as outlined by the typology, and a brief description of the activity, assessments, and conclusions. Although we were as consistent as possible in what information we provided in Table A1, we note that not all articles reviewed provided the same level of detail as others.

Results

Research question 1: trends in the literature

Our first research question asks, What trends are observed in recent literature on spatial intervention studies in geoscience in terms of the purpose of the study, the geoscience practice identified, strategies used to practice spatial thinking, and setting and level of education of each study? Table 3 shows the frequency and percent of each type of level of education, setting, and intrinsic and extrinsic skills for all the studies reviewed (these are listed per article in the Appendix, Table A1). First, we discuss the trends in these studies with respect to these factors, then we present the results for research question 2.

Table 3. Frequency and percent of each type of level of education, setting, geoscience practice, intrinsic, and extrinsic skills, and strategy used in the studies reviewed.

Category	Description	n	%
Level of Education	Elementary	1	3
	Secondary	6	21
	Undergraduate	22	76
Setting	Psychology Laboratory	5	18
	Classroom	23	82
Geoscience Practice	Coursework	4	14
	Mapping	9	32
	Modelling	15	54
Typology Skill	Intrinsic	92	55
	Extrinsic	74	45
Strategy	Sketching	7	18
	Gestures	3	8
	Physical Models	7	18
	Computer Models	18	48
	Curricular Interventions	3	8

Note: The study by Mead et al. (2019) included both high school and undergraduate students, therefore the total for level of education does not equal the total number of studies included in the review (n = 28). Additionally, multiple strategies applied to all studies, therefore the total intrinsic and extrinsic skills is greater than the number of studies included in the review.

Coursework

Coursework studies included a variety of different interventions (i.e., activity types) to meet numerous stated or unstated learning goals (see the Appendix, Table A1, for details of any given study). Overall, four studies (14%) focused on coursework related interventions, and three intentionally sought to improve spatial thinking. All the coursework studies took place in an undergraduate classroom setting, and Mead et al. (2019) also implemented their intervention in a high school classroom. Mead et al. (2019) used a virtual reality activity related to fossils and found they improved students’ content knowledge in both settings. They also found that high school students took slightly longer (about a half hour) to complete the activity compared to undergraduate students.

Three of these four studies focused on a curricular intervention that supported students’ spatial thinking throughout the duration of the course, while the fourth only focused on a single activity and only measured knowledge gains (Mead et al., 2019). Of the curricular interventions, two focused on one strategy. Gold et al. (2018a) used a computer program and Titus and Horsman (2009) used sketching to provide students with practice with many spatial skills in the context of geoscience coursework. Both studies observed improvement in students, but Gold et al. (2018a) used spatial assessments to measure spatial thinking, while Titus and Horsman (2009) used exam scores to measure improvement. The other curricular intervention, developed by Ormand et al. (2017), consisted of two dozen activities for upper level geoscience classes such as mineralogy, structural geology, and sedimentology and stratigraphy which they found led to significant gains in spatial thinking.

Mapping

Like coursework, the details of the mapping interventions vary across the different studies (see Appendix, Table A1). Mapping related intervention studies comprised 32% (n = 9) of the studies reviewed. All mapping interventions were implemented in one class or lab period (n = 7) or took place in a psychology laboratory (n = 2). The psychology laboratory studies each looked at the use of two strategies to explore students’ abilities to perform mapping-related tasks and offer insight on the value of these interventions in education from cognitive science. In seeking to understand the efficacy of gesturing and using physical models, Atit et al. (2016) provide compelling evidence for use of gestures to teach topographic map reading comprehension. To gauge students’ abilities related to sense of direction, Liben et al. (2011) had students practice tasks such as pointing in the location of a well-known building and pointing in the north direction as well as measuring and mapping strike and dip. They found students varied very widely in their ability and explain that these differences should be considered during instruction.

Most of the mapping studies used the strategy of computer programs (n = 6) to help students learn, and these all took place in a classroom setting. A variety of types of computer programs were used in these studies. Half of the mapping studies that used computer programs used Google Earth in their intervention. These studies sought to improve students’ learning of content knowledge (Benson, 2010; Bitting et al., 2018; Bodzin, 2011; Hedley et al., 2013; Robinson et al., 2017) or spatial thinking skills (Giorgis, 2015). In the study by Lazar et al. (2018), the researchers sought to build community and promote geoscience practices but did not explicitly seek to measure student improvement in any spatial or geoscience skills. The other mapping studies used combinations of tools such as geographic information system (GIS) and global positioning system (GPS; Hedley et al., 2013) or Google Earth and remote sensing (RS; Bodzin, 2011).

No mapping studies used a curricular intervention. The three studies that did not use a computer program used a combination of two other strategies. The study by Benson (2010) was the only classroom-based study that did not use a computer program. Instead, students used sketching (creating a map) and physical models (index cards around the building with information) to sketch a mine. Benson (2010) measured learning using the final map students turned in for the assignment and students’ self-reported feedback via survey. The other two studies, which are from psychology, had different goals and used different strategies, although both used spatial measures drawn from the cognitive science literature. Liben et al. (2011) was an exploratory study to understand students’ spatial thinking, whereas Atit et al. (2016) included a particular intervention after which they found improvement on spatial abilities.

All mapping studies that used a computer program reported that they observed improvement using their measures, however here again, the assessments used as well as what was measured vary greatly. Giorgis (2015) explored improvements in spatial skills and others looked at improvement in geoscience content knowledge (Bitting et al., 2018; Hedley et al., 2013; Robinson et al., 2017). Assessments used to measure student improvement also vary. While Giorgis (2015) used a measure from previous literature (Titus & Horsman, 2009) and Bitting et al. (2018) used the Geoscience Concept Inventory (GCI; Libarkin & Anderson, 2005), most of the studies used their own measures of success, which depended on their intervention and purpose. These measures took the form of final written responses (Bodzin, 2011), self-reported enjoyment and perceived knowledge gains (Lazar et al., 2018), or a concept knowledge assessment (Hedley et al., 2013; Robinson et al., 2017).

Modelling

As described above, the modelling studies could include a number of different approaches, including virtual models, physical models, gestures, and more. We do not attempt, here, to directly compare virtual and physical models, unless describing that aspect of a particular study. Additionally, our discussion of gestures is limited to those studies using gestures in association with modelling; this may exclude other embodied learning such as that addressed in fieldwork, for example.

Most of the studies reviewed were modelling related interventions (54%, n = 15). Two-thirds of these studies implemented the strategy of computer programs (n = 10) to help students learn spatial concepts or phenomena. All other studies used only one strategy. Three modelling studies took place in a psychology laboratory setting and 12 took place in a classroom setting. All the psychology lab studies (Atit et al., 2015; Gagnier et al., 2017; Sanchez & Wiley, 2014), and nine classroom studies (Bursztyn et al., 2017; De Paor et al., 2012; Giorgis et al., 2017; Jackson et al., 2019; Lally & Forbes, 2019; Reusser et al., 2012; Singha & Loheide, 2011; Woods et al., 2016), took place at the undergraduate level. Three studies involved secondary school students (Markauskaite et al., 2020; De Paor, 2012; Virk et al., 2014) and one study involved elementary school students (Forbes et al., 2015).

Computer programs took the form of modelling concepts such as topographic map reading (Giorgis et al., 2017; Jackson et al., 2019; Woods et al., 2016), the carbon (Markauskaite et al., 2020) and water (Lally & Forbes, 2019) cycles, and introductory topics (Bursztyn et al., 2017) and modelling phenomena such as volcanic processes (Sanchez & Wiley, 2014), plate tectonics (De Paor et al., 2012), and groundwater flow rates (Singha & Loheide, 2011). Measures varied widely, as did the studies’ purposes.

The only psychology study that used a computer model looked at the efficacy of various representations to support learning of volcanic processes (Sanchez & Wiley, 2014). Their qualitative content knowledge-related assessment showed that animations were more useful than static images. De Paor et al. (2012) focused on a model of the process of hot spots and sea floor spreading, used content-related assessments, and found that students experienced learning gains. Bursztyn et al. (2017) also used content knowledge assessments to measure gains in student performance but focused on an augmented reality (AR) phone application of a field trip to the Grand Canyon. They observed learning gains on their assessments, which were developed using materials from SERC resources, the GCI, and ConcepTests.

Of the modelling computer program studies reviewed, four also included a physical model. One of these studies focused on matching a mathematical representation from a software (computer) program to a sand tank (physical model) in a hydrology class (Singha & Loheide, 2011), and the other three implemented an AR sandbox (an interactive tool using projection and sand to model topographic features) into a lab session on topographic map reading (Giorgis et al., 2017; Jackson et al., 2019; Woods et al., 2016). In the study by Singha and Loheide (2011), students’ self-reported reactions and short answer responses suggest that students were better able to make useful predictions about the sand tank model.

In the AR sandbox model studies, both Giorgis et al. (2017) and Jackson et al. (2019) used spatial measures and found that students did not experience learning gains on the lab activities nor the spatial measures as a result of working with the AR sandbox; however, they showed high self-reports of engagement from students. They both suggest that more time be given to AR sandboxes in class since these studies only implemented the activity in one class session. The study by Woods et al. (2016) exploring the use of an AR sandbox used exit surveys to assess the success of the activity. Their results suggest the activity was helpful for understanding topographic maps, surficial features, and processes and they suggest that the AR sandbox could be valuable in introductory classes to demonstrate concepts.

Four modelling studies used sketching to support student learning. One sketching study took place in a psychology laboratory and the other three took place in a classroom setting. The psychology study by Gagnier et al. (2017) sought to determine if sketching could be used to improve 3D diagram understanding in science settings. They found that students who compared their answers to the correct answer improved their ability to complete the Geologic Block Cross-sectioning Test (GBCT; Ormand et al., 2013) compared to the group who did not sketch and the copying group. Their findings suggest that students should be provided with feedback about their models and encouraged to improve their work and that guided sketching can improve spatial thinking in STEM.

The sketching studies focused on a range of geoscience topics as well as various levels of education. Gagnier et al. (2017) studied undergraduate students’ ability to predict the blank side of a geologic block diagram and Reusser et al. (2012) focused on geomorphology students’ understanding of fluvial and hillslope processes. Assaraf and Orpaz (2010) focused on junior high school students’ perceptions of the polar regions, and Forbes et al. (2015) focused on third grade students’ models of the water cycle.

To assess students’ improvement, these studies used various measures. In the study by Gagnier et al. (2017) students completed a pre- and post-assessment that was composed of select GBCT items. Assaraf and Orpaz (2010) used a pre-post assessment that consisted of Likert scale items, as well as interviews with five students to qualitatively assess student improvement. They found that students significantly improved their understanding of Earth systems and human’s impact on the environment. Forbes et al. (2015) also used a pre-post assessment that consisted of student modelling tasks and student interviews to gauge improvement and found students did not substantially refine their mental models over the course of the unit. They explain that the students may not have had the ability to pair the representation of mechanisms of the water cycle with verbal and written forms. To support elementary students’ ability to accurately model the water cycle, learning environments should use more appropriate representations that include all of the cycle’s components.

One study, a psychology laboratory study by Atit et al. (2015), focused on gestures. The researchers explored how gestures aid penetrative thinking by undergraduate psychology students. Their pre-post assessment measured improvement in spatial thinking using GBCT items. The researchers found that students who were able to gesture performed better than students who did not gesture. They suggest that students should be encouraged to use their hands to reason about and communicate 3D spatial relations and suggested that instructors also use more gestures and be aware of their gestures during instruction. Future research on gestures for modelling should explore the value of gestures in the classroom.

Research question 2: alignment with the Newcombe and Shipley framework

Our second research question asks, How and to what extent does the existing literature on spatial geoscience interventions align with the Newcombe and Shipley (2015) spatial typology framework, and what gaps remain? Table 4 shows the frequency and percent of each typology skill. We first present trends in the typology skills observed overall, then we present trends in each geoscience practice as well as trends in strategies. We then identify gaps in the literature based on this framework. A detailed description of how the typology aligns with each study can be found in the Appendix (Table A1), as can a short description of each study that provides context.

Table 4. Frequency and percent of each category of studies that address each type of spatial skill outlined in Newcombe and Shipley’s (2015) typology.

Intrinsic/ Extrinsic – Static/ Dynamic	Typology Skills (n = total number of studies investigating the skill)	Coursework		Mapping		Modelling
		n	%	n	%	n	%
Intrinsic – Static	Disembedding (n = 22)	2	9	9	41	11	50
	Categorisation (n = 22)	1	5	9	41	12	54
Intrinsic – Dynamic	Visualising (n = 20)	2	10	7	35	11	55
	Penetrative Thinking (n = 9)	3	33	2	22	4	45
	Mental Transformation (n = 6)	3	50	2	33	1	17
	Sequential Thinking (n = 13)	1	8	2	15	10	77
Extrinsic – Static	Locating Objects (n = 22)	2	9	9	41	11	50
	Alignment (n = 0)	0	0	0	0	0	0
Extrinsic – Dynamic	Perspective Taking (n = 20)	3	15	7	35	10	50
	Relations Among Objects (n = 25)	2	8	9	36	14	56
	Updating Movement (n = 7)	0	0	3	43	4	57

Overview

The typology skill that applied most frequently was relations among objects (n = 25) and was found most frequently in modelling studies. An example of this skill in modelling would be when students consider a water molecule of groundwater in relation to a water molecule of atmospheric water in a water cycle model (Lally & Forbes, 2019) or components of a hot spot in a Google Earth model (De Paor et al., 2012). Three skills tied as the second highest frequency (n = 22) of application to these studies: disembedding, categorisation, and locating objects. These three skills are common and basic geoscience skills that are widely applicable as evinced by the high frequency of each of these skills in each geoscience practice identified. Two skills were each found in 20 studies: visualising and perspective taking. No studies reviewed address the spatial typology skill of alignment, which is the skill of reasoning about spatial and temporal correspondence (Newcombe & Shipley, 2015, p. 189).

Coursework

Since coursework studies made up the fewest number of studies, they also generally made up the smallest percentage of studies for each typology skill. These studies tended to focus on intrinsic-dynamic skills. Table 4 shows that these studies make up a higher percentage of all the studies of each type of intrinsic-dynamic skills compared to the intrinsic-static skills. These studies also tended to align with extrinsic-dynamic skills more often than extrinsic-static skills (Table 4).

In terms of specific skills, coursework studies align with mental transformation more often than the other geoscience practices and comprise 50% of all applications of mental transformation. These studies made up 33% of studies that aligned with penetrative thinking, and 10% of the studies that aligned with visualising. No coursework studies aligned with either alignment or updating movement.

Mapping

Mapping studies tended to make up a greater portion of the intrinsic-static skills and generally aligned with more extrinsic-dynamic skills compared to extrinsic-static. Mapping studies most frequently aligned with visualising compared to the other intrinsic-dynamic spatial skills. All mapping studies aligned with the extrinsic-static skill of locating objects and the extrinsic-dynamic skill of relations among objects.

These studies tended to focus on four skills, namely: Disembedding, categorisation, locating objects, and relations among objects. Even though mapping studies make up 41% of the disembedding and categorisation skills, all nine of these studies included practice with these two skills. All mapping studies also align with locating objects and relations among objects and make up a similar percentage of the studies for each of these typology categories. Certain discipline-specific activities were sometimes implemented in addition to mapping which provided practice with visualising, such as sketching various locations in relation to one another to create a map (e.g., Benson, 2010) or a cross section profile (e.g., Giorgis, 2015). These practices also provided practice with perspective taking. Mapping studies comprised seven of the studies that aligned with both visualising and perspective taking.

Modelling

Because most of the studies reviewed were modelling studies, modelling studies also made up the highest percentage of studies that aligned with many of the typology skills. These studies tended to make up a higher percentage of the intrinsic-dynamic and extrinsic-dynamic skills. These studies tended to comprise upwards of 50% of the studies that aligned with a given skill, with the notable exception of mental transformation, with which only one study aligned (17%).

The most frequently aligned skill to modelling studies was relations among objects, where all but one study aligned (n = 14, 56%). Notably, modelling studies had the highest percentage of the studies that align with sequential thinking (n = 10, 77%). Only four studies aligned with both penetrative thinking and updating movement. Because so few studies aligned with updating movement, the four modelling studies make up 57% of these studies while only comprising 45% of studies that align with penetrative thinking. This is expected given the dynamic nature of the content and the ability of strategies like computer models to display and allow for interaction with dynamic phenomena.

Strategies

It is useful to report the trends in typology skills that aligned with studies that implemented particular strategies regardless of geoscience discipline, as in some cases there was an overlap in skills and tasks students completed among the three geoscience practices. For example, one student may practice disembedding when using Google Earth to answer map related questions (e.g., Atit et al., 2016), and another student may use disembedding when using the same tool to observe a modelled phenomenon (e.g., De Paor et al., 2012). These trends of spatial skills for each strategy can provide insight into how to create curricular interventions and spatial activities for geoscience classrooms of all levels to support greater discipline-specific skill or concept knowledge development.

Sketching appears to be useful for the skill of penetrative thinking and perspective taking. Computer programs appear to frequently provide practice for all spatial skills. Penetrative thinking and mental transformation were practiced in three studies using computer programs and three studies using a curriculum intervention. One of the curricular interventions was a computer program and the other used sketching throughout a curriculum (Titus & Horsman, 2009). This further shows the ability of these strategies to improve these particular skills, however sketching can support the development of other skills as well (e.g., mental transformation and relations among objects, among others).

Gestures supported practice with visualising, locating objects, and relations among objects. Physical models, which can take the form of a sandbox, sand tank, play doh model, or map, for example, tended to align with map related spatial skills, that is, disembedding, categorisation, visualising, locating objects, and relations among objects. Physical models also provided support with perspective taking since three of these studies involved sandboxes in which students could use this skill.

Discussion

The first three sections below focus on aspects of research question 1. We discuss what geoscience practices have been studied in the literature on spatial thinking in geoscience, then we focus on the strategies used to improve students’ spatial thinking in the studies reviewed, and third we explore the inconsistency with measures of success for these studies. In the fourth section of the discussion, we explore research question 2 by looking at the typology skills that were applied to each study and from this we identify gaps. The last section discusses non-geoscience literature reviews that focus on spatial thinking in other STEM fields.

Geoscience practices

We felt it was useful to use geoscience practices as an organisation theme to map existing literature to support our investigation of trends in this body of knowledge (Borrego et al., 2014). These themes emerged from our review, and there is support in the spatial literature for the distinction between mapping and modelling studies. Map related tasks range from using maps (e.g., Bluestein & Acredolo, 1979; Diana & Webb, 1997; Gilhooly et al., 1988; Liben & Downs, 1993; P. J. Moore, 1993; Schofield & Kirby, 1994; Sholl & Egeth, 1982; Warren & Scott, 1993) to making cognitive maps (e.g., Baker et al., 2012; Blaut et al., 1970; Bryant, 1982; Cooperrider et al., 2017; Gagnon et al., 2014; Holden et al., 2016; Newcombe et al., 2015; Schultheis et al., 2014; Scott & Schwartz, 2007; Thorndyke & Hayes-Roth, 1982; Weisberg & Newcombe, 2016) to how to best help students understand maps (e.g., Clark et al., 2008; Ishikawa & Kastens, 2005; Petty & Rule, 2008; Rapp et al., 2007).

Studies pertaining to modelling range from mentally visualising objects (e.g., Al-Balushi & Coll, 2013; Balliet et al., 2015; Chan & Wong, 2019; Dolphin & Benoit, 2016; Halpern et al., 2015; Hinze et al., 2013; Imhof et al., 2012; Kali & Orion, 1996; Lord, 1985; Lowe, 1993; Mathewson, 1999; Oh et al., 2016) to creating cross sections (e.g., Alles & Riggs, 2011; Cohen & Hegarty, 2012; Eley, 1981; Kali & Orion, 1996; Kastens et al., 2009; Khooshabeh & Hegarty, 2010) to mental rotation (e.g., Atit et al., 2013; Stieff, 2007; Wiedenbauer & Jansen-Osmann, 2008). Modelling studies and mapping studies tended to focus on the same skills but in different contexts and with different goals.

Coursework studies were classified as such if they consisted of an intervention that lasted longer than one session, therefore this category consisted of both mapping and modelling activities to varying extents. These were curricular interventions and a unit long intervention. A number of studies we reviewed suggested that researchers implement interventions for longer periods of time; in light of this, it appears the field of geoscience education would benefit from more long-term interventions similar to what Titus and Horsman (2009) or Gold et al. (2018a) implemented. An example of a successful geospatial curriculum intervention outside of geoscience can be seen in a study by Bodzin et al. (2013) who developed an energy literacy curriculum where students use geospatial technology to explore socio-scientific problems. The success of these studies and the potential value of more time with spatial practice highlights the importance of the need for more spatial curricular materials for geoscience.

Strategies

Trends in methods implemented to practice spatial thinking in geoscience education suggest five methods are frequently used: computer programs, sketching, gesture, physical models, and curricular interventions. Computer programs in the form of geospatial technologies have been highlighted by the NRC (2006) as an excellent way to bring spatial thinking into the geoscience curriculum. There have been a number of studies that address the use and benefits of this method of instruction (e.g., Bezzi, 1991; Bice, 2006; Brindisi et al., 2006; Höffler & Leutner, 2007; Kastens et al., 2001; Lindgren & Schwartz, 2009; J. D. Moore et al., 2013, 2015a, 2015b; Piburn et al., 2005; Reynolds et al., 2006; Taber & Quadracci, 2006; Van der Meij & de Jong, 2006; Whitmeyer, 2012). We also found that computer programs are useful tools to improve students’ spatial thinking skills, which was not surprising since the NRC (2006) identified GIS, GPS, and other geospatial information technology (GIT) as effective tools for classroom instruction for many fields, but especially geoscience.

According to a review of 61 studies exploring the influence of computer simulations on learning science, Smetana and Bell (2012) found that the success of these interventions had more to do with how the computer simulations were used than their inherent designs. Geoscience has not explored how computer simulations are used, rather, they are implemented in ways that target students’ attention on certain aspects of a larger system or scene. For example, Virk et al. (2014) implemented a simulation of glacial melting and target student attention on sea level rise.

Numerous studies have also shown that sketching is a useful strategy to improve spatial thinking (Ainsworth et al., 2011; Johnson & Reynolds, 2005; Shepard & Metzler, 1971; Yin et al., 2010). Gestures have also shown promise (Herrera & Riggs, 2013; Liben et al., 2010; Matlen et al., 2012; Stieff et al., 2016a), as have physical models (e.g., Casey et al., 2008; Gogolin & Krüger, 2018; Kastens & Rivet, 2010; Sell et al., 2006; Stieff et al., 2016b). Using more than one of these strategies (i.e., multiple strategies in the same intervention) has also proved to be beneficial to student learning (e.g., Ainsworth, 2006; De Vries, 2006; Seufert, 2003), which we observed in two of the curricular interventions (Gold et al., 2018a; Ormand et al., 2017). In addition to these strategies to practice spatial skills, research on the assessment of undergraduate (Rivet & Kastens, 2012) and secondary students’ spatial thinking skills (Ben-Chaim et al., 1986; Ramful et al., 2017) all provide the tools for implementation in classroom settings.

The classroom studies in mapping that used computer programs suggest that these strategies can be used to improve students spatial thinking skills in the context of mapping (Giorgis, 2015) and to improve a variety of geoscience skills such as land use change (Bodzin, 2011) or atmospheric concept knowledge (Hedley et al., 2013), as well as identification of land surface features (Robinson et al., 2017). These gains were seen in a variety of levels of education, where most researchers focused on undergraduate students (e.g., Bitting et al., 2018; Giorgis, 2015; Robinson et al., 2017) while other researchers focused on middle and high school students (e.g., Bodzin, 2011; Hedley et al., 2013). Despite the NRC’s (2006) call to action, it seems fifteen years later, more research on the value of computer programs in improving spatial skills using measures of spatial skills is necessary at all levels of education.

We found that sketching tended to provide practice with visualising and penetrative thinking and were found to be effective (Benson, 2010; Reusser et al., 2012; Titus & Horsman, 2009). This was often the case when students had to sketch a cross section either as an activity in regular class instruction or as an intervention to improve their ability to perform this commonly practiced geoscience skill (e.g., Benson, 2010; Gagnier et al., 2017; Giorgis, 2015; Ormand et al., 2014, 2017; Reusser et al., 2012; Titus & Horsman, 2009; Woods et al., 2016). Despite the value in geoscience there was a gap in the literature on studies using gestures in the classroom. The studies that did focus on this skill highlight its value in understanding challenging spatial concepts (e.g., Atit et al., 2015, 2016).

Measures of success

Generally, the studies reviewed reported that they observed improvements in their measures with the notable exception of Giorgis et al. (2017) and Jackson et al. (2019) who indicate the lack of learning gains in their titles, as well as Forbes et al. (2015) who also found underwhelming results. Overall, studies used a wide variety of assessments to measure their students’ success. All studies measured content learning gains in some way, either through pre-post assessments or solely post-intervention measures.

The pre-post assessments covered content knowledge (Bitting et al., 2018; Bursztyn et al., 2017; De Paor et al., 2012; Lally & Forbes, 2019; Markauskaite et al., 2020; Mead et al., 2019; Robinson et al., 2017; Singha & Loheide, 2011; Titus & Horsman, 2009; Virk et al., 2014), spatial thinking skills (Giorgis, 2015; Gold et al., 2018a; Liben et al., 2011), and often both (Atit et al., 2015, 2016; Gagnier et al., 2017; Giorgis et al., 2017; Hedley et al., 2013; Ormand et al., 2017; Sanchez & Wiley, 2014). Assaraf and Orpaz (2010), Gold et al. (2018a), and Singha and Loheide (2011) included self-report survey components in their assessments. Titus and Horsman (2009) surveyed students six months later and found students who engaged in the activity showed better retention.

Studies that included solely post-assessments took the form of content knowledge measures (Benson, 2010; Bodzin, 2011; Reusser et al., 2012) or spatial skills measures (Jackson et al., 2019). Some of these post-activity assessments also consisted of students’ self-report interest (Jackson et al., 2019) and interest and learning gains (Lazar et al., 2018; Woods et al., 2016), as well as interviews (Forbes et al., 2015). Therefore it appears that nine studies actually showed their intervention improved students’ spatial thinking as a result of the intervention, which is about 32% of the studies reviewed (Atit et al., 2015, 2016; Gagnier et al., 2017; Giorgis, 2015; Gold et al., 2018a; Hedley et al., 2013; Liben et al., 2011; Ormand et al., 2017; Sanchez & Wiley, 2014).

This wide variety of assessment types and focuses hampers our comparison of these studies and identifies a gap in the literature in terms of reliable and valid assessments for measuring spatial thinning in geoscience. This finding agrees with other recent literature identifying future lines of geoscience education research (Ryker et al., 2018; St. John et al., 2020). There are some geoscience assessments (e.g., Crystal Slicing Test, Geologic Block Diagram Test, Topographic Map Assessment, and more) and some research into development of more assessments (e.g., Bursztyn et al., 2017; Myer et al., 2018; Robinson et al., 2017) but the full suite of spatial typology skills that align with these assessments is not known.

Interestingly, Ormand et al. (2014), which preceded and laid groundwork for Ormand et al. (2017), found no correlation between success in STEM and spatial skills, whereas Titus and Horsman (2009) did find a correlation between these variables. Both Ormand et al. (2014) and Titus and Horsman (2009) explore the influence of regular instruction on spatial skills. Ormand et al. (2014) used GPA and final grades to determine their correlations while Titus and Horsman (2009) used GPA alone. These studies not only used a different way to measure success in STEM, they also focused on how instruction influences different skills: Ormand et al. (2014) sought to measure the spatial skills mental rotation, penetrative thinking, and disembedding, while Titus and Horsman (2009) sought to measure spatial relations, spatial manipulation, and visual penetrative ability.

Both of these studies used a pre-post assessment design to measure improvement and draw from the same psychometric measures frequently used in cognitive science research on spatial thinking (e.g., Crawford & Burnham, 1946; Ekstrom & Harman, 1976; Guay, 1976; Myers, 1953). It is unclear whether they used the same items from these tests to build their respective measurement instruments. Ormand et al. (2014) explain the fact that they did not see a correlation suggests that it is possible for students to perform well enough on other aspects of the course and how there is likely a threshold level of spatial reasoning, the latter point having also been made in previous research (Titus & Horsman, 2009; Uttal & Cohen, 2012).

Spatial thinking typology

Inspired by Cole et al. (2018) who aligned astronomy education to the intrinsic/extrinsic static/dynamic framework, we apply the more specific, geoscience related skills discussed by Newcombe and Shipley (2015). It was difficult to apply the typology at this level because it required consideration of a wide variety of studies, an ontological consideration of the goals of the studies, and various criteria to determine whether and how to include a study. The typology turned out to be difficult to use as an organisational tool, however it did provide us with useful information to understand gaps in the geospatial literature. Notably, there were no intervention studies identified that addressed the skill of alignment. Furthermore, application of the typology presented a major challenge in that students may use different spatial strategies to solve a given problem. A major gap in the literature is that no studies interviewed students about their problem-solving strategy. For example, a student could solve a problem about rotating a cube by considering the cube in their head and rotating it, therein engaging in mental transformation, or they could consider the cube from an external frame of reference and turn the cube so as to use perspective taking. Instead, those studies that included interviews with students focused their questions on students explaining the evidence that supported their claims (Forbes et al., 2015; Lally & Forbes, 2019).

Non-geoscience reviews

We include here a brief discussion of reviews of spatial thinking in other disciplines to draw comparisons and glean relevant future directions for geoscience; a more detailed review of this literature is beyond the scope of this paper. We noted that our review was inspired by Cole et al.’s (2018) review of spatial thinking in astronomy. They recommended that researchers indicate what assessments they used for content knowledge spatial skills. We saw a similar pattern (i.e., that this specification is lacking) in the geoscience studies, and strongly recommend that more researchers be attentive to both this need for detail but also commonality in factors such as learning goals and assessments in order to facilitate more cross-study comparisons. Cole et al. also noted the challenge in defining spatial thinking, and suggested that researchers state what discipline-specific practices students engage in. Lastly, they highlighted the need for more research to see what types of spatial thinking is required for different astronomy related skills as well as the need for more assessments. A brief review of spatial thinking in biology by Milner-Bolotin and Nashon (2012) highlighted that biology students often lack the sufficient skills to understand 3D and 4D visualisations, which was also seen in a geoscience study by Liben et al. (2011). They also concluded that these skills can be improved with interventions, using tools such as GIS, and we agree.

In chemistry, Harle and Towns (2011) provided a useful detailed review of the history of spatial skills. They concluded that spatial ability literature can provide chemistry faculty with the tools they need to make their instruction more spatial. A review of spatial thinking in geography education by Wakabayashi and Ishikawa (2011) showed that empirical testing of the use of GIS is (or at least was, at that time) insufficient. Recent GER has included a number of studies that empirically test the use of GIS but such research now needs to move to more targeted use in order to support specific spatial skills in geoscience contexts. Wakabayashi and Ishikawa also found that the relationship between discipline specific skills and spatial skills needs to be explored and detailed more clearly. We feel we have made an effort towards this goal in our review of spatial geoscience education literature. Mohler (2008) reviewed spatial thinking in engineering. In his review, he outlined the various approaches to spatial thinking, which is useful for a more elaborate background than we provide here; however, he did not draw pedagogically relevant conclusions as we attempt to do below.

Conclusions

There is a wide variety of types of studies reviewed here in terms of design, audience, discipline, field, and settings; the research community is still working to understand spatial skills in geoscience and how best to teach and learn spatial thinking in the classroom. This variety of differences in aspects such as learning goals and measures makes this work difficult to compare. However, the present study brings these works together to some extent and introduces consistent terms by way of the typology for future research. Based on the literature reviewed here, we posit that the kind of research that is needed now is how to incorporate spatial thinking skills into existing curriculum design and support individual differences in instruction. This notion was raised by Wai et al. (2009), and twelve years later we believe that it is imperative that we begin the process of spatialising the school curriculum.

Trends in the categories of the spatial thinking typology suggest that research on spatial thinking in geoscience has neglected the skill of alignment and provides relatively little practice with the skills of mental transformation and updating movement. It appears that the skills that align with common practices in the research are disembedding, categorisation, visualising, locating objects, perspective taking and relations among objects.

The geoscience literature reviewed here supports the earlier conclusion that spatial activities can improve students’ spatial thinking (e.g., Gold et al., 2018a, b; Newcombe & Stieff, 2012; Uttal et al., 2013). As such, it is important to consider how we might incorporate spatial thinking practice into instruction in a meaningful way. We draw implications for future research and implications for instruction for each conclusion below.

• Incorporate both abstract and discipline-specific practice. Instructors should incorporate ways for students to engage as much practice as possible with spatial activities, whether they are directly related to course content or abstract spatial puzzles (e.g., Newcombe & Stieff, 2012). There is a great deal of evidence in this review and in previous work that students would benefit from practice with spatial skills in ways that allow them to apply these skills to other STEM fields and life skills, in addition to spatial problem solving in the context of a specific discipline task (Buckley et al., 2018; Newcombe & Stieff, 2012; Wai et al., 2009). Researchers should focus on developing ways to incorporate field-based spatial skills activities since we found a dearth of research on spatial skills in field-based contexts (but see also Mogk & Goodwin, 2012). These basic skills ought to be developed more decontextualised before applying contexts that use these skills (Uttal et al., 2013, p. 369).

• Incorporate long term implementation. Researchers should develop new interventions being aware that multiple studies suggested that interventions be implemented for longer than one class or laboratory period (e.g., Giorgis et al., 2017; Jackson et al., 2019). Ideally interventions and spatial activities would span a curriculum rather than one topic and/or one class or lab session. Ormand et al. (2017) suggest interventions be implemented over a degree program to support students’ spatial thinking. Instructors should be aware of the finding that short interventions with new and engaging spatial materials may cause students to be distracted by the novelty so long-term implementation could avoid any problems novelty may cause (Jackson et al., 2019).

• Explore students’ problem-solving strategies. Instructors should ask students how they solved the problem, particularly about the mental processes they engaged in to come to their answer. Instructors could require an explanation of problem solving or conduct interviews. Researchers can use student interviews, such as those conducted by Czajka and McConnell (2018), to explore various ways of solving a problem. This line of research may also further inform possible sources of misconceptions since many errors students make have perceptual origins (Myer et al., 2018). Knowledge of common spatial or perceptual misconceptions students hold could inform the development of instructional materials.

• Assess spatial thinking and content knowledge. Researchers should be aware that in many cases spatial measures were not used (e.g., Lally & Forbes, 2019; Lazar et al., 2018; Robinson et al., 2017; Singha & Loheide, 2011; Woods et al., 2016) and content knowledge assessments were not always clear (e.g., Lally & Forbes, 2019; Mead et al., 2019; Robinson et al., 2017). To further geoscience education and research on this field with respect to spatial thinking, a suite of geoscience assessments ought to be developed, not dissimilar to, though greater in volume than, Ormand et al. (2017). Gagnier et al. (2016) provide a review of a number of tests and tools in geosciences that can be considered. Instructors should consider using assessments developed by researchers more frequently to better understand how their students perform on spatial thinking tasks.

• Better align learning goals, instruction, spatial skill development, and assessment. One limitation of the literature reviewed here is that few articles explicitly included the learning goals for the intervention. The learning goals should drive the subsequent decisions such as instruction, spatial skill development, and assessment. Although we previously addressed these opportunities separately, instructors should also consider how geoscience instruction – particularly that which contributes to students’ spatial thinking skills – is aligned with disciplinary goals and practices and its assessment, especially over the long term (Ormand et al., 2017). Research into such well-aligned instruction, such as through design-based approaches (Brown, 1992; Sandoval & Bell, 2004), can help to maximise its effectiveness.

We hope that this review provides a valuable background on the current spatial geoscience education literature from which future research can grow guided by grand challenges outlined by Ryker et al. (2018). Our themes of geoscience practices identified echo the first grand challenge of Ryker et al. (2018), which asks what skills are essential to different geoscience fields. These themes are practices that are widely used in many geoscience fields. In our discussion of the measures of success in the studies reviewed, we found measures varied widely. This also echoes the need for assessments that accurately measure spatial skills in geosciences, which is the second grand challenge Ryker et al. (2018) outlined. The strategies we identified may serve as a useful way to foster spatial and temporal reasoning skills in geoscience, which responds to Ryker et al.’s (2018) third grand challenge.

This review is limited in our search criteria, where we only included peer reviewed journal articles that described the efficacy of a spatial intervention in geoscience education. We did not include research on pattern identification, scale, time, diagram reasoning, or analogical reasoning, among other topics important within geosciences. For more detailed reviews of some of these topics, see the following and references therein: temporal and spatial scale (Cheek et al., 2017); diagram reasoning (Gagnier et al., 2017); and analogical reasoning (Jee et al., 2010). Likewise, reviews of field-related studies within and beyond geoscience may be of interest, though they do not necessarily focus on spatial interventions within fieldwork (Mogk & Goodwin, 2012; Munge et al., 2018; Riggs, 2005; Whitmeyer et al., 2009).

Informed by implications and working together, educators and researchers can develop curricular materials that build students’ spatial thinking to improve students’ ability to model and transfer spatial skills to other activities that they will encounter throughout STEM instruction. More spatial activities such as those studied in research presented here hold promise for increasing effective instruction of geoscience and spatial skills in elementary and secondary classrooms, therein curtailing the projected shortage of geoscientists and more STEM majors overall.

Disclosure statement

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

Additional information

Notes on contributors

Jessica A. McLaughlin

Jessica A. McLaughlin has a Master’s degree in Educational Psychology. She has worked on understanding students’ perceptions of Earth’s subsurface and developing instructional materials that contributed to understanding how students better understand topographic maps.

Janelle M. Bailey

Janelle M. Bailey’s research interests include astronomy education as well as both in service and preservice science teacher education and spans a range of K-16. Most recently, she is co-principal investigator of a US$2.3M grant from the National Science Foundation on helping high school students better evaluate the connections between evidence and explanations in Earth science. Bailey is a past president of the American Association of Physics Teachers a member of the editorial boards for Astronomy Education Journal (current) and three other journals (former), and a guest editor for a focused collection on astronomy education research for Physical Review Physics Education Research published in 2018. She teaches classes in secondary science education and has taught elementary science methods as well as introductory astronomy in the past.

Notes

1. https://www.silc.northwestern.edu/

2. https://serc.carleton.edu/index.html

3. https://serc.carleton.edu/getspatial/index.html

4. http://www.geomapapp.org/

5. http://www.usarray.org/

Appendix

Table A1. Geoscience studies reviewed, with applied Newcombe and Shipley (2015) typology.

Article	Level of Education/Research Setting	Geoscience Practice/Strategy	Within – Object (Intrinsic)	Between – Object (Extrinsic)	Description/Geoscience Field
Assaraf and Orpaz (2010)	Secondary/Classroom	Modelling/Sketching	By thinking about Earth systems and their interactions, students practiced sequential thinking.	By locating specific locations and objects, students practiced locating objects. When considering aspects of different environments in various ways, students practiced perspective taking. By taking a global view of events and identifying the influences on specific regions in relation to one another, students practiced thinking about relations among objects.	This study focused on 21 junior high school students’ perceptions of the environments in Earth’s polar regions. Students took a pre-assessment to measure their knowledge, then completed the ‘Life at the Poles’ unit intervention spanning 30 hours of teaching over three months. Students sketched concept maps and other components in the unit to promote systems thinking. Students then took a post-assessment. The pre- and post-assessment consisted of Likert scales and short response items that were developed based on in-depth interviews with five students. The researchers found that students significantly improved their Earth systems understanding and awareness of human influence on the global environment.
Atit et al. (2015)	Undergraduate/Psychology Laboratory	Modelling/Gestures	Students practiced visualising when imagining the 3D nature of the block diagram from a 2D image to answer the questions. Imagining the inside of the diagram provided practice with penetrative thinking. To actually construct the diagram with Play-Doh and consider what a face would look like after a certain amount of the Play-Doh structure is removed, students practiced sequential thinking.	Locating specific features of the diagrams in the activity and pre-post-assessment provided practice with locating objects. Considering layers of Play-Doh in relation to one another provided practice with relations among objects.	Researchers explored how gestures aid penetrative thinking of 92 undergraduate psychology students. Students were assigned to one of three groups: a gesture group, a gesture prohibited group, and a test-retest group. The gesture group worked through the intervention diagrams using their hands to build the structure from flat layers of Play-Doh and the gesture prohibited group sat on their hands and provided verbal instructions to build the structure. The test-retest group played the card game SET for the duration of the experiment. Students’ abilities were measured with a pre- and post-assessment by taking the Geologic-Block Cross-Sectioning Test (GBCT; Ormand et al., 2013) which consisted of seven problems. The GBCT and experimental activity only required students to practice intrinsic skills. The researchers found that the gesture group improved from pre- to post-assessment and the test-retest group did not. The gesture prohibited group also did not show significant improvement. The researchers concluded that students should be encouraged to use their hands to reason about and communicate 3D spatial relations and suggested that instructors also use more gestures and be aware of their gestures during instruction.
Atit et al. (2016)	Undergraduate/Psychology Laboratory	Mapping/Gestures; physical model	Completing the Topographic Map Assessment (TMA; Newcombe et al., 2015) provides practice with disembedding when recognising the corresponding topographic pattern that represents the feature and categorisation by classifying features by their shape. When considering the 3D shape that the topographic map lines represent, students likely practiced visualising.	Students needed to locate features on the map, and therefore practiced locating objects. To identify the shortest path, students practiced relations among objects. The TMA questions provided practice with perspective taking and updating movement.	In a two-part experiment, the researchers explored the efficacy of gestures and language to communicate about topographic maps. Due to differences in performance between males and females in previous research, these researchers focused only on female undergraduate students. In Experiment 1, 272 psychology students completed a suite of assessments including the Map Experience Survey (MES; Weisberg et al., 2013), the Water Level Test (WLT; Piaget & Inhelder, 1956), and the Spatial Orientation Test (SOT; Hegarty & Waller, 2004). Students were then assigned to one of four groups: pointing and tracing, 3D gestures and models, test-based, and no instruction. After participating in the intervention, students completed the TMA. The researchers found that using gestures of pointing and tracing individual contour lines facilitates topographic map reading. In Experiment 2, 58 students took the same assessments as in Experiment 1 and were assigned to one of three groups: Elevation language, Shape language, and no instruction. The materials from Experiment 1 were modified to match gestures and specific language to prompt students to think about topographic contour lines in different ways. The researchers found that the information conveyed in the language that accompanied the gestures influenced the information students focused when learning topographic maps.
Benson (2010)	Undergraduate/Classroom	Mapping/Sketching; physical model	Recognising features in the index card sketches provided practice with disembedding, and classifying those features provided practice with categorisation. The activity provided practice with penetrative thinking when imagining the ways the rocks between the flashcards connect and visualising when imagining the mapped location as a whole, independent of the building it is projected onto.	Students practiced locating objects by identifying the card locations and geologic formations on their map, perspective taking by considering the mine from different flash card locations, and relations among objects when considering how new information fits into their existing map and how flash card locations and geologic formations relate to each other.	In a hands-on mapping activity for field methods students, Benson (2010) used project-based instruction where flash cards were scattered around the geology building to superimpose layers of a mine on/in the campus building. By sketching and connecting the various locations, the 10 students enrolled in the course mapped a mine system which they used to understand mine geology and practiced determining the location of resources. There was no additional assessment beyond the quality of responses to the prompts and completed map. Benson focused on students’ impressions of learning outcomes. Student feedback suggested this activity helped them with visualising geologic projections and visualising and mapping 3D information.
Bitting et al. (2018)	Undergraduate/Classroom	Mapping/Computer Program	Students were required to record their observations of regions on Google Earth in response to specific prompts, which provided practice with disembedding, categorisation, and visualising.	Due to the Google Earth format and ability to rotate the map and manipulate the perspective, we determined that students practiced locating objects, perspective taking, and relations among objects.	Bitting et al. (2018) implemented Google Earth activities in undergraduate introductory geology classrooms to improve students’ conceptual understanding of the coursework. In this study, 204 students responded to all parts of the assignment. The activities were assigned to be completed outside of class time. Students were assessed using the Geoscience Concept Inventory (GCI; Libarkin & Anderson, 2005), which is a multiple-choice assessment of various geoscience concepts. Pre- and post-test results show that Google Earth activities improved students’ performance on the GCI with a moderate effect size of 0.64.
Bodzin (2011)	Secondary/Classroom	Mapping/Computer Program	To identify important information in the map students used disembedding and classifying that information requires categorisation. Considering land use change over time, students practiced sequential thinking.	Finding geologic features for the activity requires locating objects and being able to move the map around to see different perspectives provides students practice with perspective taking. Through activities such as considering fault lines in relation to the location of a volcano, the skill of relations among objects would be practiced.	Using Google Earth and remote sensing (RS), Bodzin (2011) studied 110 eighth grade students’ learning of land use change. Students analysed photographs of land use and development to understand concepts of heat Islands, land use change over time, urban sprawl, and other human activities’ effect on natural environments. Students’ potential learning gains were measured through an open-ended written response assessment based on common misconceptions of interpretations of the same aerial photographs used in the activity. Bodzin found that the activity improved students’ understanding of land use change issues and appeared to enhance students’ spatial thinking skills through their use of the photographs to interpret and identify objects. Bodzin notes that these findings suggest the activity is effective for teaching middle school students spatial thinking.
Bursztyn et al. (2017)	Undergraduate/Classroom	Modelling/Computer Program	Identifying certain features provided practice with disembedding and classifying them provided practice with categorisation. Considering the processes that occurred to create the various features provided practice with sequential thinking.	Students practiced locating objects by observing their location on a map. Given the moveable view of the augmented reality (AR) program, students also practiced perspective taking at each location in the Grand Canyon they visited using AR. When considering their location related to others, students practiced relations among objects.	Bursztyn et al. (2017) studied the use of an AR phone application field trip for undergraduate students in introductory Earth science classes at different institutions with a varied population of STEM and non-STEM majors. Class enrolments varied from 20 to 300 students depending on the institution. The smaller class sizes participated in the AR field trip and the larger class sizes participated in the traditional curriculum. Typical introductory geology topics were covered, such as stratigraphic principles, geologic time, surface water processes, and other concepts within those topics. The computer program focused on understanding geological features of the Grand Canyon via a virtual reality phone application. Students were assessed at the beginning of the class and two weeks after each intervention (i.e., AR field trip versus traditional curriculum). The assessments were created using SERC resources, the GCI, and ConcepTest questions and tested for validity and reliability. At the start of the class students also completed a demographic and interest survey. Learning gains were observed in both the AR and the regular instruction section, though for the geologic time and hydrologic processes components, the AR program appeared to show larger gains than regular instruction. The researchers also found that the AR activities increased the students’ interest.
De Paor (2012)	Undergraduate/Classroom	Modelling/Computer Program	By recognising certain features as important students practiced disembedding. Classifying the identified features provided practice with categorisation. By considering the 3D nature of the object they are viewing and imaging a hot spot below a surface location, the students also practiced penetrative thinking. Considering the whole process of formation with the help of Google Earth images, students practiced sequential thinking.	Identifying the area of interest provided practice locating objects. Viewing the region from different perspectives using Google Earth provided practice with perspective taking. Recognising where certain features were in relation to others, provides practice with relations among objects.	De Paor (2012) studied the influence of a computer program on 127 undergraduate non-science majors’ learning of plate tectonics. They focused on mid-ocean ridge spreading and hot spots, using Google Earth to show the 3D structure and temporal evolution. They conducted their study in a single laboratory class and included activities addressing the Iceland spreading ridge and hotspot. Pre- to post-assessment scores suggested students experienced overall learning gains that were not experienced differently by students with different levels of experience, nor between males and females. It is unclear if the pre- and post-assessments addressed any spatial skills directly.
Forbes et al. (2015)	Elementary/Classroom	Modelling/Sketching	When considering how to organise the information, students practiced categorisation. Through modelling these processes, students practiced visualising when they considered diagrams and how they occur in the 3D space we live in, and sequential thinking when considering the journey of a water droplet.	Students practiced perspective taking when viewing the water cycle from different focal points and contexts, and relations among objects when considering different elements of the model related to each other.	In a study by Forbes et al. (2015), 30 students in a third-grade class engaged in a curriculum intervention based on Construct-Centered Design and Construct Modelling to learn about the water cycle. Through the intervention, students were required to meet the same standards as the original unit. The researchers argue this format afforded opportunities to generate and use models of the water cycle to formulate evidence-based explanations for the target concepts. Students worked with scaffolded activities and received feedback to revise their models. Their success was measured based on interviews with students to empirically define upper and lower levels of understanding of multiple construct maps related to model-based reasoning about the water cycle. The researchers found underwhelming learning gains and modelling ability, though they noted this is likely due to the little experience students had with this type of learning.
K. M. Gagnier et al. (2017)	Undergraduate/Psychology Laboratory	Modelling/Sketching	Filling in blank sides of a GBCT involves imagining the inside of the block, which provided students with practice visualising and penetrative thinking.	Students did not necessarily practice extrinsic spatial skills.	K. M. Gagnier et al. (2017) conducted a psychology laboratory study to determine if sketching can be used to improve 3D diagram understanding in science settings. Participants were 105 undergraduate students completing an introductory psychology course requirement. To measure success, students completed a Geologic Block Cross-sectioning Test (GBCT) as a pre- and post-assessment, where they had to project and sketch the layers that exist on the blank sides of the block using the filled-in sides of the block as a guide. For the activity, students were placed into three conditions: one group completed activities like the GBCT tasks, a second group predicted the sides but did not sketch, and a third group copied the answers from a complete one. The copying group was meant to simulate classroom practice of students taking notes. The researchers found that students who accurately sketched diagrams and compared their answers to the correct answer improved their ability to complete the GBCT compared to the group who did not sketch and the copying group.
Giorgis (2015)	Undergraduate/Classroom	Mapping/Computer Program	Identifying the different geologic units provided practice with disembedding and classifying those rock types in various ways provided practice with categorisation. Students practiced visualising when imagining the map they created for the activity in their head and penetrative thinking when considering what the subsurface will look like. The assessment provided practice with mental transformation.	Students practiced locating objects when locating aspects of the map, perspective taking when moving around using Google Earth, and relations among objects when considering two aspects in relation to each other (i.e., the end points of the cross-sectional area).	Giorgis (2015) studied 75 undergraduate geoscience students in structural geology classes at a liberal arts college over five years to test the hypothesis that implementing Google Earth exercises improves the 3D visualisation skills of students. The activity required students to create a map and cross section profile through an area of folded rock. Giorgis used an assessment created by Titus and Horsman (2009) that measured students’ ability to understand spatial relations, spatial manipulations, and Visual Penetrability Ability (VPA) as a pre-post assessment. Although Giorgis found considerable variation in students’ abilities, students improved on all three measures from pre- to post-assessment. He also found that female students improved more than male students, but the difference in improvement between the genders was not statistically significant. The improvements observed suggest Google Earth activities are a useful tool to improve spatial thinking in the classroom.
Giorgis et al. (2017)	Undergraduate/Classroom	Modelling/Computer Program; physical model	Students practiced disembedding when recognising important information on the map such as contour lines on the map, and categorisation when recognising features as hills and valleys. Students practiced visualising when using the interactive map to understand how the 2D topographic map looks when changes are made to the 3D sand.	Locating hills and other features provided practice with locating objects. By considering multiple profiles from a perspective until one aligned with what one would see, and through the assessment questions related to topographic profiles and line of sight judgements, students practiced perspective taking. Students practiced relations among objects when considering relations among features in the sandbox; in the assessment, questions related to line of sight judgements. When considering how surface water flows and pools and how the landscape changes the corresponding topographic map students practiced updating movement.	To explore the efficacy of a sandbox in an undergraduate class with a total of 176 students, these researchers implemented a sand tank component into their regular pencil and paper lab. All students completed the pencil and paper lab, but students in the experimental group worked with a sandbox for 20 minutes as they answered some of the lab questions. One week after completing the laboratory, the researchers assessed students via Google Forms. The questions included 10 Purdue Visualisation of Rotations Test (PVRT; Guay, 1976; Yoon, 2011) questions like Titus and Horsman (2009) and some questions about topographic maps covering landform identification, topographic profiles, and line of sight judgements for a total of 15 questions. The researchers determined that sandboxes are effective at generating interest and enthusiasm towards understanding and using topographic maps, but the sandbox did not influence students’ performance on the lab activity. They suggest that future researchers implement the sandbox over a longer period of time.
Gold et al. (2018a)	Undergraduate/Classroom	Coursework/Computer Program; physical model; Curriculum Intervention	The modules consisted of various activities of isolated spatial skill practice that focused on disembedding, penetrative thinking, and mental transformation.	The modules provided practice with perspective taking.	In an experimental study by Gold et al. (2018a), 592 undergraduate students in introductory and advanced geology classes were placed into one of two conditions: an experimental group that completed 11 weeks of short weekly spatial training computer modules in addition to regular coursework or a control group that received instruction as usual and did not complete training modules. Introductory geology students in the experimental groups completed the modules as homework and upper level geology students completed the modules during class time. The modules consisted of isolated spatial skill practice that focused on mental transformation, penetrative thinking, and disembedding. These skills were measured by a suite of pre- and post-assessments drawn from the literature. Mental transformation skills were assessed using the PVRT, disembedding was assessed using the Hidden Figures Test (HFT; Ekstrom & Harman, 1976), and penetrative thinking was assessed using items from the Planes of Reference Test (PRT; Ormand et al., 2014). The results showed that these short regular interventions throughout the semester improved students’ spatial thinking skills significantly with a moderate to large effect size. Hands on activities that were incorporated later did not have a significantly different effect on student learning compared to online training modules, though they note there was some evidence that hands on training may benefit skills beyond the effect of the online training modules. Self-report data from students showed that 40% perceived the training as beneficial to their coursework.
Hedley et al. (2013)	Secondary/Classroom	Mapping/Computer Program	Students practiced disembedding when looking for meaningful patterns in the data and when identifying relevant aspects of the map. By classifying aspects of the map, students practiced categorisation.	Freely using these tools to investigate their own interests and goals provided students with opportunities to practice locating objects and practice with the relations among objects, which may have been useful for drawing comparison in locations between or among different objects of interest.	Studying atmospheric concept knowledge, Hedley et al. (2013) used the geospatial technology of RS, GIS, and GPS in a variety of classrooms with a total of 303 middle and high school students. The instructor provided an introduction and demonstration of the tools and instructed the students to create projects using these tools to explore atmospheric concepts. Students collected data for their projects and the data was shared with the university researchers. The students presented to these researchers for the culminating event. Throughout their projects, the students were supported by a partnership between their classroom and university researchers. The structured and guided practice with these tools helped students build their skills with these tools. The researchers state they observed significant gains from pre- to post-assessment, which focused on atmospheric concept knowledge. Spatial skills were not measured in this study, but the researchers suggest this open-ended activity is useful for middle and high school students learning to use GIT as well as atmospheric concept knowledge.
Jackson et al. (2019)	Undergraduate/Classroom	Modelling/Computer Program; physical model	Students practiced disembedding when recognising important information on the map such as contour lines on the map, and categorisation when recognising features as hills and valleys. Students practiced visualising when using the interactive map to understand how the 2D topographic map looks when changes are made to the 3D sand.	Locating hills and other features provided practice with locating objects. By considering multiple profiles from a perspective until one aligned with what one would see, and through any assessment questions related to topographic profiles, students practiced perspective taking. Students practiced relations among objects when considering relations among features in the sandbox. When seeing the corresponding changes in topography with changes in sandbox shapes and when answering some TMA questions students may have practiced updating movement.	Jackson et al. (2019) implemented a sandbox to explore questions related to engagement and knowledge gains into 40 sections of an introductory geology laboratory. The experimental group worked with the sandbox to recreate a simple topographic map in 3D in groups of three to four students within a six-minute time frame. Students completed the TMA as a post-assessment and there was no pre-assessment. Students reported high self-engagement, but both groups scored the same on the TMA. The comparison group had a slightly higher mean than the experimental group. The researchers suggest this may be because the novelty of the sandbox was a distraction. They suggest more experiences with the sandbox so that it is not novel may lead to learning gains.
Lally and Forbes (2019)	Undergraduate/Classroom	Modelling/Computer Program	Students practiced disembedding when looking for meaningful patterns in the data and when identifying relevant aspects of the model. By classifying aspects of the model, students practiced categorisation. Through modelling these processes, students practiced visualising when they considered diagrams and how they occur in the 3D space we live in, and sequential thinking when considering the journey of a water droplet.	Students practiced perspective taking when viewing the water cycle from different focal points and contexts, and relations among objects when considering different elements of the model related to each other.	To improve undergraduate introductory water course students’ model-based understanding of the water cycle, the researchers implement a flipped classroom and active learning opportunities related to water symptoms simulation modelling. They sought to compare student performance from year 1 (n = 38) to year 2 (n = 53). Students engaged in a three-part activity involving making farming decisions based on evidence from data collected from various parts of the farm and general area, including graphs such as contour maps of rainfall among others. They were able to make changes or explain changes they would make based on the evidence. Students were assessed using a 41 item pre-post-course concept inventory. These questions were water concept-related and consisted of multiple choice and short answer question formats. Results show students experienced content knowledge gains. No spatial thinking was measured, but they noted that evaluation of the model, rather than use of the model, may better support student learning.
Lazar et al. (2018)	Undergraduate/Classroom	Mapping/Computer Program	Students practiced disembedding when recognising important information on the map such as the representations on a map, and categorisation when recognising features such as the geocache. Students practiced visualising when imagining the location of the geocache and/or any points along the way.	Locating the geocache and other features provided practice with locating objects. Students practiced relations among objects when considering relations among features on the map. When considering possible paths to the geocache, students practiced updating movement.	In an effort to build community and promote geoscience practices among 202 students, these researchers implemented geocaching extra credit opportunities in one large and one small physical geology course and one water sustainability course. Students participated in twelve geoscience geocache activities or similar created as extra credit to increase engagement in the introductory classes. Students were asked to rate Likert type statements about enjoyment and perceived knowledge gains, and probed further interest. No content knowledge nor spatial measures were implemented.
Liben et al. (2011)	Undergraduate/Psychology Laboratory	Mapping/Sketching; Gestures	When taking strike and dip, students practice disembedding when they determine what is important to record and how. Determining what the information means and identifying information on the map as rock formations or other features provided practice with categorisation. Students practiced visualising when they considered what the representations on the map should look like. Students practiced mental transformation when completing the shoreline task, since they imagine the same area after some change. Students practiced mental transformation and sequential thinking when completing the WLT since they would imagine what the glass looks like when tilted at a certain angle.	To complete the task where students point to both the north direction and a nearby well-known building on campus, students practiced locating objects. By considering different locations around campus students practiced perspective taking. Considering themselves in reference to different locations and considering the outcrop in relation to the wooden rod in the map provided practice with relations among objects. In the drop task, students practiced the skill of updating movement as they considered the path of the water drop.	The WLT was administered to 125 undergraduate psychology students. These students participated in activities where they had to measure and map the strike and dip of an artificial outcrop on campus, they had to represent a wooden rod sticking straight out of the ground on the map, and they had to point to the North direction and to a local campus building, which was meant to gauge place awareness. After these activities students were asked about their confidence in their responses on a 7-point scale and to self-report their sense of direction. Students then participated in laboratory activities where they completed a shoreline and drop task. The shoreline task asked students to draw the line of the water if water came up to about halfway. The drop task asks students to draw the path a drop of water would take if it fell on a tilted piece of paper. In the lab, students also took the Santa Barbara Sense of Direction Test (SBSOD; Hegarty et al., 2002). They found that students who were more aware of the broader framework, or had more place awareness, also had an advantage on at least some of the tasks here. Men outperformed women on the strike and rod task and had smaller errors in the pointing tasks. They conclude that instructors cannot assume that all students have the same amount of knowledge about spatial concepts.
Markauskaite et al. (2020)	Secondary/Classroom	Modelling/Computer Program	Students practiced disembedding when recognising important information on the map such as the representations on a map, and categorisation when recognising features such as the carbon atoms. Students practiced visualising when imagining the location of the carbon atom at some point in the cycle the computer program represented.	Students practiced locating objects when they located the atoms and different environments in the program. Students practiced relations among objects when considering carbon and other elements in different spheres of the system, and by comparing different spheres of the system in the model. When considering the path of a carbon atom or similar moving through the system, students practiced updating movement.	Working with 60 high school students, Markauskaite et al. (2020) explored improving students’ climate systems understanding and how students ground their reasoning when working with computer models. Students completed a pre- and post-assessment that centred around the question of system equilibrium. The researchers found significant increases between pre- and post-assessment scores and described students’ styles of grounding their reasoning in three general categories: direct observation, straight abstraction, and generalisation, which made up a continuum of abstraction that they argue originated in students’ perception of what occurred in the models.
Mead et al. (2019)	Secondary and Undergraduate/Classroom	Coursework/Computer Program	Students practiced disembedding when attending to features of the complex display that may be fossils, and categorisation when classifying the fossils.	To find locations and features to attend to as directed by the activity, students practice locating objects. Looking around the spherical images, students were able to practice perspective taking. Considering the placement of fossils in relation to one another within, between or among beds students practiced relations among objects.	The researchers sought to explore the efficacy of virtual reality field trips with 153 high school AP Biology students and 437 students in two undergraduate courses: biology and astrobiology. The activities focused on life and environments during the Ediacaran Period (about 550 million years ago). Students explored a series of fossil beds and worked with 360 degree spherical images for 1.5 to 2 hours. After a short text introduction, students completed the work independently. A six-item quiz was built into the virtual field trip to serve as a pre-post-assessment. The questions covered content knowledge through multiple choice and multiple selection. The high school students completed a delayed follow up quiz, which showed retention, which the authors state is notable because of the short period of time their intervention was implemented. Both high school and undergraduate students showed large statistically significant improvements from pre- to post-assessment after a t-test analysis. Results were observed to be equally true for all students.
Ormand et al. (2017)	Undergraduate/Classroom	Coursework/Curriculum Intervention; Various Strategies	Students practiced visualising when using 2D representations of phenomena to visualise the 3D object or area, penetrative thinking when engaging in exercises where they imagined the shape and internal structure of an object, mental transformation when considering geologic processes over time based on a 2D diagram/image, and sequential thinking when considering the process that had occurred that led to the structure they currently see. The measures focused on mental transformation by the mental rotation test and penetrative thinking by the two discipline specific and one more general measures.	Students practiced perspective taking when taking the WLT to measure their understanding of planes of reference.	A team of geoscientists and cognitive scientists collaborated to create two dozen activities to support students’ spatial thinking. The materials covered topics in mineralogy, structural geology, and sedimentology and stratigraphy courses at the undergraduate level. Students enrolled in these courses in three different institutions over three years participated in the intervention. Students completed a wide variety of measures of spatial thinking, including a measure of mental rotation (or mental transformation), a few measures of penetrative thinking (where two were discipline specific and one was general), and a planes of reference test. Results from 181 students were reported and showed that their curricular intervention led to significant gains in spatial thinking. They note that the small improvements seen in one semester suggest larger improvements are possible with increased implementation, such as over a degree program.
Reusser et al. (2012)	Undergraduate/Classroom	Modelling/Sketching	Students practiced disembedding when identifying relevant features to sketch, categorisation when students named features in their explanations, visualising when the feature to be sketched was presented as an image, penetrative thinking when drawing fluvial system cross-sections, and sequential thinking when explaining the processes occurring in these sketches.	Locating spatial field data provided practice with locating objects. Considering streams from various angles for their sketches provided practice with perspective taking, and when including multiple locations and features in the same sketch students practiced with relations among objects.	In this experiment researchers sought to improve 19 geomorphology students’ understanding of concepts related to fluvial and hillslope processes using concept sketching. Students were required to write an explanation that included four levels of thinking that generally follow Bloom’s Taxonomy: identification, process, linkages or interactions, and predictions for the future. The sketches served different purposes during the class. In lecture, students sketched to engage in lecture and share ideas; in lab students sketched to explain material and spatially locate field data. Sketching allowed students to revise old work when considering new material as well. The researchers did not use a pre-post assessment for concept knowledge and nor any spatial assessment. Concept sketching was used as a final assessment. The researchers’ observations and students’ feedback suggest that concept sketching is a pedagogically useful tool, particularly for summative assessments. The researchers posit that the sketches clearly showed the depth of student knowledge and rapidly allowed them to identify what part of the process of understanding students had achieved according to Bloom’s taxonomy, explaining that over the course of the semester students’ sketches improved greatly.
Robinson et al. (2017)	Undergraduate/Classroom	Mapping/Computer Program	By recognising certain features as important students practiced disembedding. Classifying the identified features provided practice with categorisation. When imagining the landforms in their mind’s eye, students practiced visualising.	Students located landforms in the various images, which provided practice with locating objects. Viewing the region from different perspectives using Google Earth provided practice with perspective taking. Considering features in relation to one another provided practice with relations among objects.	In a two-part study, Robinson et al. (2017) explored the efficacy of using Google Earth images compared to lidar-derived High-Resolution Topography (HRT) images to identify topographic features on a landscape. Part 1 involved two classes of undergraduate students (n = 46). One class was given lidar images and a second class was given Google Earth images, and within each class students received one of two locations. Students answered questions about their observations and were prompted to use the image to support their answers. Results showed that students were able to recognise familiar landscape features using both methods, however looking at Google Earth images more students focused on distractions such as land use, vegetation, and infrastructure, while the students who used HRT images (which only showed topography of the land surface) were better able to recognise important topographic features. In Part 2, the researchers used this data to develop and assess instructional materials in an introductory geoscience course. The intervention consisted of a video about lidar and an activity related to earthquakes where students measure the offset of the San Andreas Fault and engage in subsequent calculations. The assessments were related to content knowledge and were developed by the researchers. No spatial measures were used. The results show an increase in scores from pre- to post-assessment for the experimental group and no statistically significant difference for the control group.
Sanchez and Wiley (2014)	Undergraduate/Psychology Laboratory	Modelling/Computer Program	By recognising certain features as important students practiced disembedding. Classifying the identified features provided practice with categorisation. When imaging the process that led to Mt. St. Helens erupting based on the model they had seen, students practiced visualising. Considering the processes related to plate tectonics, students also practiced sequential thinking.	When considering the different features of a volcano and the locations of objects during the process of eruption, students practiced relations among objects.	This psychology laboratory study used a computer program to study individual differences in spatial thinking, specifically visuospatial skills. Drawing from a pool of psychology undergraduates, 162 students who had taken a geoscience course participated in the study. For each experimental condition, participants viewed a webpage that showed various presentations of information. Students were put into one of three conditions: text with no illustrations, text with illustrations, and text with animations. Students completed two spatial measures: One was the paper folding task which they completed before the activity and the second was the intercept task which they completed afterwards. Students were also assessed on their comprehension by the quality of their response to ‘What caused Mt. St. Helens to erupt?’ In a separate session, students also took an assessment to measure working memory. Results show that students were better able to comprehend the text describing physical phenomena that change over time when it was accompanied by an animation.
Singha and Loheide (2011)	Undergraduate/Classroom	Modelling/Computer Program; physical model	Identifying what is important in the sand tank amid a messy background provided practice with disembedding. Classifying those features provided practice with categorisation. Thinking about the movement of water and resulting processes provided practice with sequential thinking.	Locating specific features of the sand tank provided practice with locating objects. Understanding objects in relation to one another provided practice with relations among objects.	Singha and Loheide (2011) explored the value of using a physical model and computer model together to help eight undergraduate students better understand hydrogeology concepts. The topography and rates of flow in the 2D sand tank (‘ant farm’) were shown on the computer model, which allowed students to practice numerical modelling. In a one-hour lecture, students engaged in a 2D sand tank experiment online and were provided with some relevant equations. Then in a three-hour lab, students were introduced to the COMSOL software that allowed them to engage in numerical modelling. This was followed by a model that prompted them to compare the realistic model and the numerical model. Students completed a pre- and post-assessment that focused on their reactions to the class activity and non-graded short answer prompts. There was no spatial measure. The researchers found that comparing these models allowed students to better predict the behaviour, test for outcomes, and explore how best to produce those results numerically. The activity provided a forum to confront and break down misconceptions about subsurface processes and improve students’ ability to estimate rates and controls of groundwater flow and transport.
Titus and Horsman (2009)	Undergraduate/Classroom	Coursework/ Sketching	The skill puzzles required students to imagine the inside of the block diagram and how layers align on various sides which provided practice with penetrative thinking. To check their cross section with the diagram, students engaged in visualising. The stereonet projection activities also required visualising. Rotating the geologic block diagram provides practice with mental transformation.	The stereonet projection activities provided practice with locating objects as students identified important features of the diagram. When considering these points in relation to each other, students practiced relations among objects.	Seeking to explore and characterise and improve spatial visualisation skills in geoscience students, Titus and Horsman (2009) conducted a two-part study. The first part was exploratory. The researchers sought to measure the spatial visualisation skills of students in various geoscience classes. In the second part, they implement instructional materials they developed to increase students’ practice with spatial visualisation in a structural geology course. There were two conditions based on the lab section, with seven students in each lab. The experimental lab section completed exercises related to stereographic projections, while the control group did not. In lecture, all students completed skill puzzles. The skill puzzles provided practice with spatial visualisation. An example skill puzzle would be drawing a cross section of a geologic block diagram. Exam scores were used to determine the effectiveness of the experiment. Results show that students in the lab section that engaged in stereonet related activities performed better on the exams. Improvements are seen over the entire course. The researchers suggest this general trend of improvement may be due to the skill puzzle activities in lecture. A follow up survey with visualisation and stereonet questions conducted six months later suggested that students who engaged in the activities showed better retention and performance.
Virk et al. (2014)	Secondary/ Classroom	Modelling/Computer Program	Students practiced disembedding when recognising elevation from contour lines on a map, and categorisation when recognising features such as the floating ice tongue of the glacier. Students practiced visualising when imagining the glacier from the map and how the removal will influence the simulation. Imagining this change also provided practice with mental transformation. Considering the process of losing parts of the glacier and resulting rising sea level over time, students practiced sequential thinking.	Locating the glacier and Manhattan in the map and simulation provided practice with locating objects. Considering the glacier from the simulation view and the map view provided practice with perspective taking. Considering the change in Manhattan caused by the change in the glacier supported practice with relations among objects.	The researchers sought to determine the effectiveness of an interactive simulation and use the data to improve its design. Participants viewed a short video introducing them to the topic of glacial melting and providing background information about the specific glacier in the activity. The video also provided a demonstration of the simulation. Participants could freely click the contour map on the glacier in the simulation to see the resulting sea level rise in Manhattan. Next, participants interacted with a specific glacier to see how removing chunks of the glacier causes changes in the structure. The simulation was connected to a map that showed the reduction in the glacier’s size with removal of chunks of the glacier. The assessment was a paper and pencil questionnaire with five short answer comprehension questions that were completed without access to the simulation. The small sample size resulted in high engagement and learning gains by participants. They suggest that students be able to access the simulation to help answer questions and provide some suggestions for implementation.
Woods et al. (2016)	Undergraduate/Classroom	Modelling/Computer Program; physical model	Students practiced disembedding when recognising important information on the map such as contour lines on the map, and categorisation when recognising features as hills and valleys. Students practiced visualising when using the interactive map to understand how the 2D topographic map looks when changes are made to the 3D sand.	Locating hills and other features provided practice with locating objects. By considering multiple profiles from a perspective until one aligned with what one would see, and through any assessment questions related to topographic profiles, students practiced perspective taking. Students practiced relations among objects when considering relations among features in the sandbox. When seeing the corresponding changes in topography with changes in sandbox shapes students practiced updating movement.	Woods et al. (2016) conducted a pilot study using an AR sandbox focusing on topographic map reading with nine undergraduate non-geoscience major students in an introductory geology class. The sandbox was connected to a projector that showed the sandbox topography in a 2D topographic map view on a wall of the laboratory classroom. Students used the sandbox to complete scaffolded activities to practice creating models over a six-week summer geology lab course. Student learning outcomes were measured by completing exit surveys. No spatial measures were implemented. Student responses were mostly positive. The researchers report that 97% of the students indicated that the activity was helpful for understanding topographic maps and surficial features and processes. Woods et al. (2016) concluded that the activity appears to be an extremely valuable addition to introductory geology labs particularly for clarifying concepts of contour lines, terrain steepness, rule of Vs, and topographic profiles.