An Exploration of Spatial Visualization Skills: Investigating Students’ Use of 3D Models in Science Problems during Think-Aloud Interviews

Abstract

Effective spatial visualization and reasoning skills are often credited for students’ success in science and engineering courses. However, students enrolled in these science courses are not always exposed to or trained properly on the best ways to utilize models to aid in their learning. Improving spatial visualization techniques with 3D models, such as molecular and DNA modeling kits, is often suggested to facilitate students’ ability to conceptualize compounds in two and three dimensions. Here, we investigate what techniques students use to conceptualize 2D representations of various biomolecules with the use of 3D models by interviewing undergraduate students from various natural science and engineering disciplines in task-based, think-aloud sessions. After scoring and analyzing the participant data we explored some of the techniques used among successful scoring participants, including the use of informal models to transition between 2D and 3D. Additional techniques used by students who were able to successfully conceptualize 3D images included starting with smaller, granular details to inductively make conclusions when thinking between two and three dimensions. We find that (1) students who anchor their thinking in 3D models show a deeper level of understanding in initially solving science problems successfully, and (2) proper 3D model use and spatial visualization techniques may improve students’ abilities to accurately visualize 2D and 3D representations of molecules in science courses. Our results demonstrate that implementing spatial visualization training to teach students how to effectively use 3D models may improve students’ problem-solving techniques in science curricula.

Introduction

Spatial visualization is the ability to mentally construct three-dimensional (3D) images from their corresponding two-dimensional (2D) representations.¹ It encompasses various types of skills used by students including mental rotations and reflections, which are applicable to chemistry (Table 1).² Mental rotations and reflections share the foundation of spatial visualization in that they describe the ability of students to mentally construct a 3D object from its related 2D image. However, mental rotations and reflections hold an added factor, which involves the ability to imagine the 3D object being rotated across different axes or the ability to construct a mirror image.³

Table 1. Definitions and Examples of Two Types of Spatial Visualization Skills: Mental Rotations and Reflections.

The Challenge of Spatial Visualization in Chemistry

Chemistry is a course that serves as the building block to various fields including medicine, pharmacology, materials science, and energy research, yet the nature of the content and exams pose an arduous challenge for many college students. Poor performance on chemistry exams usually stems from one of its most critical, foundational, and abstract concepts: molecules are 3D, take up space, and can rotate.⁴ Students are often presented with different 2D representations of molecules, such as Newman and Fischer projections, to encourage them to visualize the bonds between each atom in a molecule,⁵ as shown in Figure 1. From these images, students are then expected to rotate and reflect molecules, often without explicit instruction on how.

Figure 1.

Fischer projection (left) and Newman projection (right) of organic chemistry molecule example (created with ChemDraw).

Occasionally, students are also shown perspective drawings that depict bonds projecting behind (dashed bonds) and in front (wedged bonds) of the plane, as illustrated in Figure 2, to aid in mentally constructing the molecule. However, learners often encounter obstacles in translating and relating the 3D representations to the 2D information they are provided, then rotating or reflecting them in their minds.^4,6,7

Figure 2.

Example of perspective drawing of chemical molecule CCl₂F₂. The wedged C–F bond represents a bond that projects out of the plane, toward the viewer. The dashed C–Cl bond represents a bond that is pointing behind the plane, away from the viewer (created with ChemDraw).

With these various molecular representations, a strong foundation of spatial visualization skills is required by students on coursework and exams.⁸ Part of the ability to truly understand organic chemistry, for example, involves connecting an outcome of a reaction with a chemical explanation (i.e., how chemical structures and processes provide evidence for the products that were formed). It is critical that organic chemistry students understand how molecules can be rotated or reflected.^9,10 For instance, students first learn about reaction mechanisms, such as eliminations and nucleophilic substitutions, in organic chemistry.¹¹ Many students, however, find it challenging to visualize the mechanisms and products of nucleophilic substitution or elimination reactions.¹² A recent study found that using 3D visualization can improve students’ ability to think about how structural isomers or stereoisomers impact the mechanism and results of a reaction.¹³ Additionally, students with more developed spatial visualization skills can better explain how molecules that are rotated interact differently in reactions and how reflected molecules can be used to explain different chemical properties.¹⁴ Without these essential spatial visualization abilities, students may have trouble conceptualizing foundational organic chemistry concepts.

Underemphasis of the development of spatial visualization skills remains one of the pressing challenges that students face in science, technology, engineering, and mathematics (STEM) courses.¹⁵⁻¹⁷ Consequently, too many students solve problems by focusing on rote memorization techniques, which further leads to poor performance and retention in courses that could prepare them for various undergraduate fields and the workforce.^5,15,17 However, encouraging spatial visualization skills may promote students’ success in STEM classes, likely improving retention of students in STEM.^18,19

The Importance of Improving Spatial Visualization in Chemistry

Students often acquire spatial visualization skills through their practice of mental rotations and reflections²⁰ and by using 3D models to practice.^15,16,21,22 The use of informal (e.g., hands, arms, and toy building blocks) or formal (e.g., molecular modeling kits or molecule building software programs) 3D models in introductory courses benefit students by allowing them to manipulate objects and develop a strong understanding of how 2D images can be translated into 3D objects.²³

To better understand the importance of improving mental rotation and reflection skills within chemistry, cross-disciplinary studies provide crucial insight. Significant efforts in improving students’ spatial visualization skills have been made in engineering and related disciplines - with potential application of these techniques in the natural and physical sciences. These skills allow students to solve problems more creatively and efficiently as they enter the upper-division courses in their disciplines and the workforce.²⁰ Engineering students often hone their spatial visualization skills in prerequisite courses by learning how to solve puzzles with conceptual scenarios involving isometric and orthographic projections.²⁴ When practicing these projections, engineering students may be provided basic step-by-step approaches to help them learn how to approach problems and allow them to further strengthen their spatial visualization abilities,²⁴ which can positively impact their success in engineering. Work from Duffy et al. revealed that strong spatial visualization abilities were a significant predictor of engineering students’ success in multiple mathematical word problems.²⁵ These results further illustrate how these spatial visualization practices may be valuable in supporting and predicting students’ success on classroom assessments that incorporate spatial visualization problems.²⁵

Within chemistry, Hegarty and Waller acknowledged that spatial ability consists of several different abilities, which rely on numerous factors, including the ability to maintain spatial information and speed of processing.²⁶ Furthermore, their work reviewed past studies investigating how differences in individuals’ spatial abilities may be a result of “short-term” and “working” memory.²⁶ This provides valuable support in the importance improving students’ spatial skills to promote the use of their “working” memory, which may be improved by the use of 3D models.²³ Extending upon Hegarty’s and Waller’s findings, Newcombe and Shipley developed a framework, which encompasses four categories of spatial skills: intrinsic-static, intrinsic-dynamic, extrinsic-static, and extrinsic-dynamic.²⁷ Intrinsic characteristics are those that are inherent or within an object and extrinsic characteristics are spatial relations between objects.²⁷ Static and dynamic characteristics refer to spatial relations that are fixed or involve movement, respectively.²⁷ For example, a static task would involve the mental rotation of a still image, whereas a dynamic task would involve tracking moving objects. The chemistry education literature has provided strong evidence of using intrinsic and extrinsic spatial skills to support students’ conceptual understanding of chemical structures.²⁸ Additionally, Bodner and McMillen demonstrated the importance of intrinsic-static spatial skills to successfully translate between different diagrams of chemical structures.²⁹ Bodner and McMillen’s study also illustrated the importance of spatial skills in predicting students’ success in chemistry.²⁹

The evidence provided by these past studies further supports the necessity for exposing chemistry students to various spatial visualization practices to facilitate student success in courses that require strong spatial visualization abilities.

Theoretical Framework

We framed our work through the lens of representational competence to better understand STEM students’ spatial visualization abilities regarding biomolecules. Representational competence is defined as a set of skills exhibited by chemistry learners that supports them in understanding, applying, communicating, and transferring between different chemical representations.³⁰ Kozma and Russell proposed two levels of representational competence, which consist of lower-level and higher-level skills. Lower-level representational competence skills are surface-level and refer to the skills that chemists exhibit to interpret, generate, or translate between different representations.³⁰ For example, students demonstrate lower-level skills when translating between the Fischer and Newman projections (Figure 1) and recognizing that the representations are identical molecules. On the other hand, higher-level skills describe the deeper set of skills that students exhibit.³⁰ For example, higher-level representational competence skills may include the ability to discuss and describe the advantages and disadvantages of different representations (e.g., Fischer Projections, Lewis-Dot Structures, Newman Projections, etc.). Though both sets of skills are important to holistically describe students’ spatial visualization abilities, it is crucial to understand students’ strength in lower-level abilities before understanding their abilities to exhibit higher-level competence skills.³¹ This framework has been used by others in chemistry education research. For example, Gurung et al. investigated how organic chemistry textbook problems centered on lower-level and higher-level representational competence skills.³² Others have investigated how instructors teach molecular representations and how specific instructional strategies may help students improve their representational competence skills.^33,34 To date, however, there is limited work surrounding the use of this framework to investigate how students exemplify different spatial skills when solving spatial problems and visualizing molecules.

Thus, we set out to explore the various lower-level representational competence skills that students exhibit while solving unfamiliar science problems that require mental rotations, reflections, and translating images from 2D to 3D (or 3D to 2D). Our study explores the following research questions:

• 1.What techniques do students use to visualize molecules when translating between 2D and 3D chemical and molecular structures?
• 2.What spatial visualization strategies distinguish students who successfully translate between 2D and 3D chemical and molecular structures from those who do not?

Methods

Instrument Development

We designed a think-aloud task-based interview protocol de novo in order to investigate how students solve mental rotations, reflections, and translations of 2D science problems using informal and formal 3D models. The questions (see Supporting Information for activities provided during each module) were collaboratively developed by two members of the research team (S.A. and C.K.). To ensure validity of the questions, three members of the research team (D.N., K.W., and J.B.) who were not involved in writing the questions reviewed the instrument and provided feedback. The revisions refined the wording format of questions and improved the chemistry, DNA, and protein assessments based on the teaching experiences of three coauthors (D.N., K.W., and J.B.). Finally, updated versions were iteratively refined through gathering feedback from members of a biology education research lab group consisting of undergraduates in biology education research. Before beginning interviews, we further established the validity of our assessment questions, by conducting practice interviews with two lab members who were not involved in writing the assessments.

We specifically designed four sections of tasks (labeled Modules I, II, III, and IV) to assess foundational topics of reflecting familiar objects, reflecting chemical compounds, conceptualizing DNA structures in 2D and 3D, and rotating proteins, which are topics generally studied by introductory biology, chemistry, and biochemistry students.

Recruitment

We recruited 14 undergraduate students who were conducting summer research or taking summer classes at a large, private, comprehensive university in the Northeast. Students from both engineering and natural science related disciplines (Table 2) were recruited to obtain students with a broad array of spatial visualization skills that may be used by students across disciplines to solve the problems in each module. Participants were recruited via fliers posted around the campus. All participants had taken at least one year (2 semesters or 3 quarters) of introductory biology and one year (2 semesters or 3 quarters) of introductory chemistry. Prior to recruiting participants for the study, the university’s Institutional Review Board (IRB) approved the project (IRB #03051922). After completing the interview, participants were compensated with a $20 Amazon Gift Card.

Table 2. Demographic Information of Interview Participants.

Discipline	Number of Participants
Engineering: Biomedical	5
Engineering: Chemical	2
Engineering: Mechanical	1
Natural Science: Biochemistry	3
Natural Science: Biotechnology	1
Natural Science: Ecology	1
Natural Science: Medical Illustration	1

Interview Protocol

All of the think-aloud interviews were video and audio recorded with consent from all participants. Participants took between 35 and 60 min to complete the interview. Each interview incorporated the four modules briefly described below. Interviewees were all asked to work through the same tasks, in the same order, and were prompted to verbalize their thought processes as they were solving each problem.

Module I: Toy Building Blocks Reflection Warm-Up

Module I was designed to test students’ ability to differentiate between reflection and rotation. To do so, we provided participants with a simple, prebuilt toy building block model. Participants were given a pile of additional building blocks and asked to build a reflection of the original model and verbalize their thought processes (Figure S1). We encouraged participants to pick up the prebuilt model and rotate it to aid them in building the reflection. After building their models, participants were asked to verbally validate their answers.

Module II: Chemical Reflections

To test participants’ ability to apply reflections in chemistry, we asked participants to draw reflections of two molecules and verify their drawings with a 3D model. In Module II, interview participants were shown the bond-line representation of a chemical molecule (Figure S2). Participants were asked to draw a reflection of the original Molecule A they were given, analogously to the task given in Module I. Participants were then shown a second chemical molecule, Molecule B. Participants were challenged to draw as many distinct reflections of Molecules A and B as they could. We included two different molecules to test if students could apply concepts of reflections to more complex molecules.

After completing the reflections of Molecules A and B, participants were handed the Flexible Molecular Model Kit^© designed by Molecular Visions. Using the molecular modeling kit, participants were asked to build a 3D model of Molecule A and a 3D model of one of the reflections they drew for Molecule A. Participants were asked whether the 3D models they built of the original Molecule A and their reflection of Molecule A matched the 2D representations of the molecules they drew. If they agreed, participants moved on to Module III. If participants disagreed, they were encouraged to redraw their original reflections or rebuild the 3D models.

Module III: Identifying Correct DNA Representations

Participants were first provided a partially deconstructed 3D model of the DNA Discovery Kit^© (3D Molecular Designs) and were given a 2D representation of a DNA ladder. Participants were asked to build the 3D double-stranded DNA molecule and share their thoughts aloud in connecting the pieces to assess how participants identified details and related the 2D representations to their corresponding 3D models. While building the 3D model, they were provided a 2D representation of a DNA ladder to use as additional guidance.

After building the model, participants were given a series of six 2D representations of DNA (Figure S3). They were asked to identify whether the representation was correct or incorrect and encouraged to refer to the 3D DNA model they built. As participants identified the correctness of each of the six representations, they were asked to share their thought processes to justify their choices in order to assess whether participants can conceptualize parts of a 2D representation with its respective 3D model.

Module IV: Protein Rotation

Interview participants were shown two JSmol (https://jmol.sourceforge.net/) 2D images of the same protein, at two different viewpoints (Figure S4) to investigate how participants understood the relationship between the two models. Participants were asked to explain how to obtain the second view of the protein from the first view of the same protein and encouraged to use any of the available 3D models of their choice to aid in their response. Then, they were also asked to verbalize their thought processes. Participants were asked to complete this same task using a second protein model.

Rubric Development

The scoring rubric was developed based on the expected objective actions of each interview participant (see Supporting Information for detailed scoring rubric). The observations were coded in a binary manner for all tasks, apart from the tasks in the chemistry module. For example, if interview participants completed the action described by the rubric, the action was scored as ‘1.’ If interview participants failed to correctly complete the action described by the scoring rubric, the action was scored as ‘0.’ In the chemistry module, participants could score up to 2 points for each molecule, as there were two distinct reflections. To authenticate the scoring rubric, two authors (S.A. and C.K.) first scored the modules separately, then discussed scores together to ensure there was agreement on scoring. After scoring each module, the scores were totaled for a maximum of 13 points possible. Table 3 describes the scoring system and rubric used to assess the answers provided by the participants during the task-based modules.

Table 3. Scoring Rubric.

Module	Objective Actions	Points
Toy Building Blocks Reflection	Built one correct reflection of toy building block model provided	1
Chemical Reflection	Drew Two Distinct Reflections of Molecule A	2
Chemical Reflection	Drew Two Distinct Reflections of Molecule B	2
Identifying Correct DNA Representations	Identified sugar–sugar bond in incorrect representation of DNA	1
Identifying Correct DNA Representations	Identified phosphate-phosphate bond in incorrect representation of DNA	1
Identifying Correct DNA Representations	Identified correct representation of DNA	1
Identifying Correct DNA Representations	Identified correct representation of DNA	1
Identifying Correct DNA Representations	Identified one missing Carbon atom in incorrect representation of DNA	1
Identifying Correct DNA Representations	Identify correct representation of DNA	1
Protein Rotation	Identified 180° rotation of Protein 1 across z-axis	1
Protein Rotation	Identified 180° rotation of Protein 2 across x-axis	1

After gathering participant scores for each module, the weighted percentages were obtained by dividing the points each participant scored from the total number of points possible for each module.

Analysis

Upon completion of all interviews, the recordings were transcribed using Otter.IO. To ensure consistency in the scoring, three coders (S.A., C.K., and J.B.) watched the recorded interviews for Modules I, II, III, and IV independently. The participants were scored independently using the rubric with any discrepancies resolved through discussion. Finally, the three scorers noted and shared unique observations with each other about interview participants’ techniques for Modules II, III, and IV. Techniques for Module I were not discussed among the three scorers as the module was designed as a warm-up to assess whether interview participants could differentiate between rotations and reflections.

Results and Discussion

Overall, we found wide variation in performance across participants and modules. For example, although some participants achieved high scores across all the modules, other participants may have achieved a high score in one module but low scores in the others. Because of these findings, no correlations or associations with scores were observed between the modules. Instead, we classified each participant as earning a “high-score” or “low-score” based on their actions (Figure 3). In the chemical reflections module, participants who drew at least three of the four reflections total resulted in a high score. In the DNA representations module, participants who correctly identified at least four of the DNA representations earned a high score. Finally, correctly rotating at least one protein in the protein rotation module earned a high score.

Figure 3.

Percentages of participants who received a high-score or low-score based on their performance in Modules II (chemical reflection), III (identifying correct DNA representations), and IV (protein rotation).

Using the representational competence framework to guide our findings, participant strategies were explored as lower-level representational competence skills.³¹ Though all participants demonstrated lower-level skills, high-scoring participants had stronger abilities in those skills while participants who had low scores in modules demonstrated weaker abilities.

We identified a range of techniques used by participants to apply 3D models to 2D representations presented in the chemistry, DNA, and protein modules (Table 4). Similarly, we identified how students who did not successfully solve reflections and rotations or productively apply 3D models to their corresponding 2D images in the three modules (Table 4). As seen in the chemical reflections module, participants who successfully drew distinct reflections of the original molecule typically anchored their thought processes in a three-dimensional space by using informal models. In the DNA representations and protein rotation modules, successful participants followed systematic approaches when projecting details from the 2D representation to the 3D model. Participants who exhibited low scores in these modules ineffectively utilized the 3D model by projecting the 3D model onto the 2D representation or overlooking details that otherwise would facilitate them in connecting the 2D representation and the 3D model.

Table 4. Actions Depicted by High-Scoring and Low-Scoring Participants for Modules II (Chemical Reflection), III (Identifying Correct DNA Representations), and IV (Protein Rotation).

Module	High-Score Action	Low-Score Action
Chemical reflections	Establishes thought process in understanding three-dimensional space before the introduction of a formal 3D model	Unable to visualize 2D representation as a spaceoccupying object before formal 3D model is introduced
Identifying correct DNA representations	Projects the familiar 2D representation onto the unfamiliar 3D object in a step-by-step manner	Projects the unfamiliar 3D object onto the familiar 2D representation
Protein rotation	Leverages minor details to inductively understand the movement of a larger object	Focuses on the shape and structure of the larger object while neglecting more granular details

In our analysis below, we have highlighted four participants who most strongly represented the common techniques presented in Table 4, either leading these participants to successfully or unsuccessfully solve the questions introduced in Modules II, III, and IV. All participants were given pseudonyms.

Chemical Reflections Module: 3D Models in Drawing Reflections

For Molecule A in the chemical reflections module, all participants benefited from using the 3D model. Participants who exhibited high-score actions benefited in one of two ways: 1) if they initially drew the two correct reflections of Molecule A, they used the 3D model to confirm that their reflections were correct, or 2) if they initially only drew one correct reflection, the 3D model exposed them to a second reflection of Molecule A. Similarly, participants who exhibited low-score actions and only drew one correct reflection benefited from the 3D model in confirming that they had indeed drawn a reflection. Only one participant was not able to successfully draw at least one reflection for Molecule A. However, after using the 3D model in Molecule A, the participant then successfully drew one correct reflection for Molecule B. The results from the participant actions presented in the chemical reflections module reveal a key conclusion: 3D models can be sufficient tools in assisting students’ ability to reflect molecules that occupy space. In Figure 4, Taylor demonstrated two distinct reflections of the original molecule, as indicated by the blue box.

Figure 4.

Participant’s (Taylor) interview artifact from chemical reflections module. The two drawings boxed in blue are two distinct reflections of the original molecule (bolded) that Taylor drew.

Video and interview transcripts highlighted a critical action that Taylor used in order to successfully reflect the original molecule without the aid of a 3D molecular modeling kit. Taylor used their hands as an informal 3D model, explaining:

“If this were the ground [pointing to a table] and this were the mirror right here [using right hand as a mirror], the image of what you’d see down here would look··· dashed like that.”

This action illustrates that Taylor was not required to hold any prior knowledge of chemistry, with the exception of having a conceptual understanding that molecules are 3D and occupy space. Additionally, their actions emphasize that it is not crucial to memorize the appearance of molecular reflections. Rather, using hands as an informal guide to facilitate them in reflecting a molecule and ground their thought processes in a 3D space introduces a simple yet efficient way for Taylor to grasp how mirrors function and produce reflections. Taylor’s use of an informal 3D model leads us to conclude the importance of using 3D models to help students grasp how molecules occupy space. When exposed to the 3D modeling kit, Taylor verified the molecules they drew as correct and distinct reflections of the original molecule.

Alex’s interview artifacts further provide evidence of the importance of 3D models in guiding students to anchor their thought processes in three-dimensional spaces. In Figure 5, Alex initially illustrated a rotation (rather than reflection) of the original molecule, as highlighted by the image boxed in orange. This may imply that Alex exhibited a poor conceptual understanding of rotations and reflections. Despite their ability to build a reflection with the toy building blocks in Module I, Alex may have relied on memorizing how reflections look, initially resulting in the unsuccessful drawing. Additionally, when Alex was further prompted to draw more reflections of the original molecule, they stated that they “could not think of any other way to draw a reflection.” The exchange between the follow-up prompt and Alex’s response suggests that Alex may have felt restricted to the concept of the 2D image of the molecule and was unable to visualize it as a 3D object that occupies space. This may have initially hindered Alex’s ability to illustrate reflections of the molecule.

Figure 5.

Participant’s (Alex) interview artifact from chemical reflections module. The illustration boxed in orange was the first drawing that Alex produced. The illustration boxed in blue was the second drawing that Alex produced.

However, after being introduced to the 3D modeling kit, Alex drew a correct reflection of the original molecule, as highlighted by the blue box in Figure 5. Alex’s ability to draw a reflection of the original molecule illustrates the necessity of anchoring thought processes in a 3D space by using 3D models (formal or informal) in students’ comprehension of 2D objects. Further, Alex’s interview artifacts and transcripts add to critical points demonstrated in the literature that highlight the necessity of accompanying learners with 3D models in classrooms. Our results showing the benefits of 3D models for students in comprehending unfamiliar representations align with past studies that have also found that 3D models can help students successfully solve spatial problems by potentially lowering their cognitive load.³⁴ For example, Newman et al., suggest that learners may hold onto a mental image of molecular structures occupying space.³⁵ Holding onto mental images may decrease the amount of working space students have, further making it difficult for students to visualize how molecular structures may be rotated or reflected to obtain a different view in their minds. However, introducing 3D models that can be manipulated and observed can reduce the amount of information placed on learners when having to memorize a mental image of the object occupying space.³⁵ Furthermore, Katsioloudis et al. suggest that manipulating 3D models enables students to develop a robust comprehension in translating 2D images into 3D objects.²³ Thus, 3D models may further a learner’s ability to develop a strong foundation of understanding how molecules occupy space without the burden of holding onto a mental 3D image of the molecule.

Identifying Correct DNA Representations and Protein Rotation: 3D Models as Effective tools

The results from Module III depict two specific examples of high-scoring and low-scoring participants’ distinct utilization of the model resulting in their score differences. Video and interview transcripts of Charlie, a participant who exhibited a high-score in this module, underscore a common thought process observed among high-scoring participants. When asked to verify whether the 2D image provided was a correct representation of DNA, Charlie explained:

“I’m going to say that the phosphate group [points to the 2D structure] is represented by the red groups here [points to the red groups on the 3D model]... This is an incorrect DNA representation.”

Charlie exemplified a productive use of the 3D model by detailing each element on the 2D structure that corresponded to the 3D DNA model. Charlie established a method in pointing out each corresponding group gradually on the 2D representation which was then linked to the 3D model and concluded that the 2D representation was incorrect. When questioned about why the structure was incorrect, Charlie noticed the inconsistency between the incorrect sugar–sugar bond on the 2D image and the correct sugar–phosphate bond on the unfamiliar, space-filling 3D model.

Contrarily, Sam’s interview video and transcripts highlight a recurring thought process across low-scoring participants in the DNA module. As Sam examined the 2D DNA representation, they explained:

“The three bonds there are a good sign [points to the 3D model]. These have to be the phosphorus here [point to the the 3D model]... The phosphates are there, the triple and double bonds are there [points to the 2D structure]...”

The distinction between these two participants rested on how they initially guided their thought processes: high-scoring participants used 2D images of DNA, while low-scoring participants employed the 3D DNA model. In this example, Sam used the 2D DNA representation to verify the 3D model. Starting with the 3D model and verifying the model with the 2D image caused Sam to overlook the detail in the incorrect sugar–sugar bond portrayed in the 2D DNA representation, which led to an unproductive use of the 3D DNA model. In following this technique, Sam appeared to exhibit a less pragmatic and more haphazard method in connecting the 2D and 3D DNA visuals. Sam took a less systematic approach than Charlie by highlighting general groups on the 3D model first then pinpointing several corresponding groups on the 2D representation. In following this approach, Sam made a general assumption about the 2D representation being a correct form of DNA and overlooked the granular, incorrect bond between the two sugar groups.

Participants who looked at 3D models first may have been unsuccessful in identifying correct or incorrect DNA representations because of their attention to detail as seen in Charlie’s and Sam’s interview transcripts. The 3D models were more complex representations than the 2D images. While the 2D images of DNA structures were generally portrayed as flat objects on paper, the space-filling, colorful design of the molecules making up the DNA structure of the 3D model may have exhibited more complexity in visualizing each model. Thus, because the 3D model in this case was more complex due its space-filling molecules, participants may have been prone to overlooking specific details on the 3D model more often, causing them to make incorrect connections between the 3D DNA model and the 2D DNA representation. On the other hand, by making observations with the 2D representation first, participants may have been more familiar with the details of the DNA structure presented in a 2D format. This may have enabled participants who initially pointed to details on the 2D representation to be more closely attentive to identifying small details in two dimensions that were more commonly overlooked when using the 3D model first. Thus, it is critical to teach students a more practical approach when using 3D models to effectively relate to 2D representations.

Module IV highlights similar observances with drastic differences in the participants’ scores as a result of participants’ initial step in relating the 2D and 3D visuals. The high-score and low-score actions involved the use of a relatable 3D object, such as the toy building blocks, but the way in which participants used the 3D objects created a distinction in how they scored in this section. The spatial skills used by participants who were both successful and unsuccessful in the protein rotations module predominantly represent the intrinsic-dynamic category proposed by Newcombe and Shipley.²⁷ The distinction between successful and unsuccessful participants in this module, however, relied heavily on their competence of and recognition of “useful” portions of the representation.

To exemplify a high-score action of productively relating a familiar 3D object, Figure 6 illustrates the interview artifact from Taylor in the protein rotation module. As seen in Figure 6, Taylor circled three minor sections of the protein in both views. To help them understand the movement of the protein, Taylor constructed the toy building blocks to form the two views of the proteins. Their techniques involved matching the parts of the toy building blocks to the beige space-filling “horns” that were circled in blue as seen in Figure 6. Using the detailed pieces of the 2D protein image on the 3D model, Taylor inductively concluded that the second view of the protein was due to a 180° clockwise rotation, which resulted in this participant receiving a high score in Module IV.

Figure 6.

Participant (Taylor) circles and beige clusters (in blue) on Protein 1 in both views.

Another participant, Sam, who was identified as a low-scoring participant in this module reached an incorrect conclusion about the two views of the protein. As seen in Figure 7, Sam utilized the point of view of the purple DNA structure to demonstrate the movement of the protein.

Figure 7.

Participant (Sam) illustrates the point of view of the large purple DNA structure.

Despite using the toy building blocks as a 3D model similarly to Taylor, Sam misinterpreted the rotation of Protein 1. Sam instead used a “big picture” detail, which was the DNA structure in Protein 1, to rotate the protein. Sam explained that the view was slightly tilted to the right and therefore, did not construct a correct conclusion about the rotation of the protein from the first view to the second view. Consequently, Sam connected the model to the two views of the protein because the detail they used on the 2D image with the 3D model did not offer insight to the correct solution. Without identifying the informative landmarks of the 2D representation, Sam was unable to project the details of the 2D representation onto the toy building blocks model, thereby rendering the model ineffective for them. This further highlights the importance of knowing how to effectively utilize 3D models to more accurately relate them to their images portrayed in 2D.

Our work demonstrates the importance of using 3D models and, furthermore, the importance of preparing students to correctly correspond models to their 2D representations. Past studies have addressed the importance and benefits of 3D models in student learning and the importance of students’ use of representational competence skills,^14,36 which aligned well with the findings from this task-based research. Additionally, the findings from this study further demonstrate and imply the importance of teaching students how to correctly correspond 3D models with 2D representations. While models may assist learners by reducing the amount of information placed on the learner,³⁵ 3D models may also generate greater confusion for learners if they are used in an unmethodical or unreflective approach.³⁷ Even initially allowing students to use familiar 3D models (e.g., arms, hands, balloons, or toy building blocks, as presented in our research) may teach learners how to relate spatially occupying objects to complex molecular representations that are usually depicted as 2D images is a crucial first step.²³ Learners might better maintain a solid understanding of the overall molecular structure without the burden of maintaining a mental image of the molecule while they may still be learning the details of the structure.³⁸ Despite the inherent challenges associated with “real-life” 3D models, an initial exposure to familiar and simpler 3D models may assist learners in preparing to comprehend how 3D models correspond to their relative 2D representations. This proficiency may improve from their capacity to overcome the challenges associated with understanding spatial visualizations.

Limitations

The overarching theme that distinguished participants who performed well in the modules was not simply attributed to participants using an informal or formal 3D model but also required the participant to understand how to use the model. We recognize some critical limitations in the conclusions of this study. At the time of interviews, we did not collect demographic data regarding participants’ gender, race or ethnicity, and age. Therefore, we do not have information about how different backgrounds may have factored into participants’ experience with spatial visualization skills. Additionally, the sample size in the study was small and thus may not be representative of all the spatial visualization techniques that may be observed in a larger sample size. However, this study provides an important first step in guiding future research to explore common themes more deeply among larger sample sizes and further probe students’ spatial visualization techniques with the guidance of 3D models.

Implications for Instructors

Spatial visualization skills are crucial in chemistry to promote student success.²³⁻²⁵ Our findings underscore the necessity of instructing students in proficient interpretation of molecular representations within STEM courses, particularly by employing 3D models. Tibbell and Rundgren find that instructors’ abilities and their abilities to help students build their “perceptualizing” skills. Furthermore, our conclusions support this suggestion in that the mere use of a 3D model is insufficient for learners to comprehend its various representations.

To further enhance students’ abilities to conceptualize 3D structures, we propose the utilization of familiar items as a foundational approach. For instance, balloons can effectively illustrate why phosphorus pentafluoride adopts a trigonal bipyramidal shape before exposing students to the intricate details of the molecular composition. As students develop a spatial visual familiarity with these foundational models, instructors can progressively introduce more formal representations through molecular modeling kits. These modeling kits can then enable students to manipulate models, offering a more precise depiction of the molecular composition and bond angles due to the molecule’s shape.

While requiring modeling kits on chemistry exams or in laboratories may help students better connect to the material,³¹ the cost of physical modeling kits is often a burden to many students – especially when students are not exposed to the invaluable learning skills honed and developed with modeling kits. To combat this issue, introducing students to virtual modeling tools, such as PyMol, MolView, or PhET VSEPR simulations, can facilitate students in developing a more complete mental image of molecules and continue to construct their knowledge of molecular structures.³⁴ Stull and co-workers observed no differences in performance between students who use physical models and those who use virtual models.³⁹ Although this is not yet a well-supported finding, virtual models may be a suitable introduction to improving spatial visualization skills for students who do not have the resources to readily access modeling kits.³⁹

Implications for Research

As highlighted in the limitations, future work should consider how students from different demographic groups use molecular modeling and 3D models. For example, past empirical studies have investigated the role of gender on success in assessments that require spatial visualization techniques.⁴⁰ To date, however, there is little work examining differences in the skills used across gender. Thus, future directions should examine differences and similarities in the spatial visualization strategies used by participants of different gender identities, ages, or experiences with spatial visualization software (e.g., computer-aided design). Understanding the skills that are used across different demographics and implementing interventions to expose learners to those skills may lower or eliminate disparities in spatial abilities. Additionally, this study provides new insight for our understanding of the various skills students use successfully or unsuccessfully when given spatial visualization problems. Future work can expand upon our results by examining whether these spatial visualization strategies are consistent in broader populations of students at different institutions. Furthermore, understanding the “high-score” and “low-score” actions may provide valuable context in future research when investigating how science and engineering classrooms that use 3D modeling create an opportunity for students to address spatial visualization problems in a different manner.

The higher-level representational skills as described by Kozma and Russell³⁰ were not investigated in this study but can provide a more holistic view of the skills that students use when rotating or reflecting unfamiliar molecules. Additionally, studies on brief interventions that expose students to data-driven classifications, such as Hegarty and Waller’s 2 × 2 categorization of spatial skills,²³ may further provide students with foundations of how to approach various problems and organize their spatial skills and thoughts based on the problems they are given.

Conclusions

Overall, we conclude that 3D models serve as an integral component in comprehending and successfully translating 2D representations of various molecules into space. However, 3D models on their own are not sufficient tools to ensure understanding. For these tools to be useful, it is necessary to familiarize students with using 3D models methodically, especially as the models become more complex. Equipping students with adequate spatial visualization skills can have a tremendous impact on student learning and understanding of difficult course content if they are provided with the necessary techniques to facilitate them in the utilization of 3D models. Additionally, the proper use of 3D models provides greater accessibility for students to engage in creatively solving problems and diverge from merely memorizing 2D-3D translations in the form of rotations and reflections.

Supporting Information Available

The Supporting Information is available at https://pubs.acs.org/doi/10.1021/acs.jchemed.3c01355.

• Module tasks for toy building blocks reflection, chemical reflections, DNA representations, and protein rotations (PDF)

• Detailed scoring rubric for module tasks (XLSX)

The authors declare no competing financial interest.