Article Summary: Margulieux (2019) The Relationships between Spatial Skill and STEM Achievement


To propose a theory of the cognitive mechanisms responsible for the relationship between spatial skill and STEM achievement.

Spatial Skill and STEM Achievement

Decades of work show that high achievers in STEM, e.g., chemistry, physics, geology, computer science, have high spatial skill, e.g.,

  • spatial visualization, like mentally rotating or folding an object
  • spatial relations, like using a map to plan a route
  • spatial orientation, like following a route.

Moreover, improving spatial skill through training broadly improves STEM performance. This type of broad transfer from one type of cognitive training to many types of problem-solving tasks across multiple domains is exceedingly rare. I can’t think of a factual analogy for it, so I’ll give you a fake one. It’s like if memorizing the digits of pi helped you solve any problem that included a number, whether it was in solving equations in math, finding proportions in art, using measurements in science, or counting beats in music. We still don’t understand the cognitive mechanisms responsible for this relationship, though, so I pulled together literature from psychology, discipline-based education research, learning sciences, and neuroscience to propose a theory. I’ll summarize the main points from each area before presenting the theory.

Memory Systems (Psychology)

There are different memory systems for information that you are processing (working memory) and information that you are storing (long-term memory). Working memory is very limited, which is why it’s better to write down a group’s coffee order than try to remember it. To process more information, you can develop strategies for rapidly encoding information in long-term memory, like waiters who can remember long orders. People can successfully train these strategies to, say, memorize over 100 numbers at once, making them seem to have an expanded working memory capacity. However, that training doesn’t transfer to other, even similar, tasks. People who have trained to memorize numbers perform no better than others when memorizing letters.

Correlations with STEM (Discipline-Based Education Research)

As far as I know, any STEM field that has examined the relationship between spatial skill and achievement has found a positive correlation. People who perform better in STEM classes, earn a degree in STEM, and have a career in STEM are likely to have high spatial skill. This relationship makes intuitive sense for STEM fields that are based on physical phenomena with a spatial element (e.g., mentally orienting molecules in chemistry), but it also holds true in STEM fields that are more abstract like computer science. In addition, studies have suggested that the gender gap in STEM achievement is partially due to a gender gap in spatial skill. Female learners with low initial spatial skill, or anyone with low initial spatial skill, perform as well as those with high initial spatial skill after spatial training.

Spatial Training (Learning Sciences)

Training learners on spatial skill improves their performance in STEM on procedural tasks, like solving problems. In a meta-analysis of 217 studies, spatial training improved performance regardless of the training task (i.e., games, courses, or practice), performance task, time between training and testing, and learners’ gender or age (Uttal et al., 2013). This is the kind of robust effect that commercially- and research-produced brain training games have failed to achieve for over a decade. It suggests that spatial training improves a fundamental cognitive mechanism that is applicable to many types of tasks. 

Cognitive Mechanisms (Neuroscience)

The brain structures that process spatial information are very central in our brain. They’re just outside of the part of the brain that deals with subconscious reflexes, which makes sense evolutionarily because early humans and most animals need spatial information to find necessities–food, water, home. Recent neuroscience research found a particular pattern of neural firing associated with spatial navigation based on a 2D grid. More importantly, it detected these same 2D grid patterns when processing other types of non-verbal information that are not inherently spatial.

This finding suggests that we can use the same elementary cognitive mechanism that processes spatial information for processing non-spatial information by translating it to a 2D grid. We do this all the time when we create graphs with x- and y-axes. We take two dimensions that are not inherently spatial, like time spent studying and exam grade, and relate them in a 2D plane. Along with the 2D grid for navigating that is made of neurons called grid cells, we also have neurons that keep track of where we are in 2D space, called place cells. They are why you don’t bump into things in the dark at your house. It makes sense that, like grid cells, we could also use these neurons non-spatially to orient ideas in a 2D grid that is not based on spatial dimensions, but I don’t know of research that supports this hypothesis.

Spatial Encoding Strategy (SpES) Theory

SpES theory is based on

  • the transferability of spatial skill and training detected in discipline-based education research and learning sciences
  • the benefits of developing encoding strategies to minimize the limitations of working memory
  • the recent discovery of grid and place cells in neuroscience and their applicability to non-spatial information.

The theory is

Developing spatial skills helps people to develop generalizable strategies for 1) encoding mental representations of non-verbal information, including 2) identifying useful landmarks to orient the representation.

Having general strategies that can be applied to STEM problem-solving before learners have internalized domain-specific problem-solving strategies would help learners perform better early in their STEM classes. Better performance early on could lead to better performance later on, decreased likelihood of dropping out, and increased likelihood of achieving a degree or career in STEM.

Why this is important

Understanding the mechanisms responsible for any effect in education is important because it helps us to design interventions that are likely to improve learning. If we don’t know why something works, we can’t consistently make it work. To my knowledge, this is the first mechanism-based theory for the relationship between spatial skill and STEM achievement. It requires a lot of validation before we can justify basing design decisions on it. But if it were valid, it would mean that we could design better spatial training games for young kids to improve their STEM learning later in life. We could design better instruction in STEM fields to help learners identify dimensions for spatial encoding. And we could design better tools to support students with low spatial skill. For example, in programming, we could design better software visualizations or interactive development environments (IDEs). The connection between spatial skill and STEM achievement is so powerful that understanding how to make interventions more effective could have a significant impact on STEM education.

Margulieux, L. E. (2019). Spatial Encoding Strategy theory: The relationship between spatial skill and STEM achievement. In Proceedings of the Fifteenth Annual Conference on International Computing Education Research (pp. 81-90). New York, NY: ACM. doi: 10.1145/3291279.3339414

Uttal, D. H., Meadow, N. G., Tipton, E., Hand, L. L., Alden, A. R., Warren, C., & Newcombe, N. S. (2013). The malleability of spatial skills: A meta-analysis of training studies. Psychological Bulletin, 139(2), 352.

For more information about the article summary series or more article summary posts, visit the article summary series introduction.

2 thoughts on “Article Summary: Margulieux (2019) The Relationships between Spatial Skill and STEM Achievement

  1. Pingback: Social studies teachers programming, when high schools choose to teach CS, and new models of cognition and intelligence in programming: An ICER 2019 Preview | Computing Education Research Blog

  2. Pingback: Article Summaries: Series Introduction | Lauren Margulieux

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s