2.1 Spatial Thinking

Key takeaways

• Spatial thinking is a learnable skill, not an innate talent.
• It combines visualisation, abstraction, reasoning about location, and understanding of scale.
• You can practise it without software: drawing, describing, and comparing spatial situations.

Introduction

The National Research Council's 2006 report Learning to Think Spatially made a strong claim: spatial thinking is a "missing link" in science education. Every GIS professional and spatial data scientist exercises it constantly, yet it's rarely taught directly. This lesson demystifies spatial thinking, breaks it into sub-skills, and gives you practice patterns you can repeat throughout the course.

What is spatial thinking?

Spatial thinking is the deliberate use of space — its properties, structure, and our models of it — to ask and answer questions. It has three ingredients:

1. Concepts of space — location, distance, direction, proximity, containment, scale, shape.
2. Tools of representation — maps, diagrams, tables, formulas, and data models that externalise our thinking.
3. Processes of reasoning — visualising, comparing, generalising, inferring.

When you look at a choropleth map of unemployment and instinctively wonder "is this pattern really about county size?", you're spatial-thinking. When you realise that a heatmap legend without a base layer is meaningless, that's also spatial thinking.

Four sub-skills to practise

1. Visualisation

The ability to form, manipulate, and compare mental images of spatial situations.

Exercise: Close your eyes and picture your daily commute. Which roads form the T-junctions? Which way does the sun rise relative to your route? How far is the nearest coffee shop from the half-way point? If you hesitate, you're exercising an underdeveloped skill — which is normal and trainable.

2. Abstraction

Distilling a messy real-world situation into a simpler spatial model.

Exercise: Describe a busy intersection in terms of only lines and nodes (a graph). Which details matter for a traffic-flow question? Which matter for a pedestrian-safety question? Same physical location, different models.

3. Reasoning about location

Drawing inferences from the where-ness of things.

Exercise: Given a map showing hospitals (points), roads (lines), and neighbourhoods (polygons), what does the clustering of hospitals tell you? Does it correlate with road density, population, historical land use?

4. Scale reasoning

Understanding that patterns at one scale may not exist at another.

Exercise: Cancer rates measured per census tract look different from cancer rates per county, even with the same underlying data. This is the Modifiable Areal Unit Problem (MAUP) — aggregation boundaries change statistical results. Module 2.2 covers scale in depth.

Tobler's First Law of Geography

"Everything is related to everything else, but near things are more related than distant things." — Waldo Tobler, 1970.

This single sentence underlies most of GIS. It's why:

• Interpolation works — unknown points are more like their nearby known points than distant ones.
• Spatial autocorrelation is everywhere — neighbouring areas share economic, environmental, and social characteristics.
• Local weights dominate global averages — the three nearest hospitals matter more for emergency response than the three farthest.

When you plan analyses, ask yourself: which nearby observations should influence this calculation?

The five conceptual questions

A spatial thinker habitually asks five questions about any dataset or problem:

1. Where? — what are the geographic units and their extent?
2. Why there? — what processes produced this spatial pattern?
3. How related? — do the features exhibit clustering, dispersion, or randomness?
4. What scale? — at which resolution does the pattern appear, and does it hold at coarser or finer scales?
5. How certain? — what's the uncertainty in location, attribute, and classification?

The discipline is asking these questions before writing a single line of code.

Common cognitive traps

Spatial thinking is sometimes tripped up by predictable traps:

• The ecological fallacy — assuming individual behaviour from aggregate patterns ("people in neighbourhood X voted Y" does not imply every person in X voted Y).
• The modifiable areal unit problem (MAUP) — results that depend on how you draw the zones.
• Cartographic lying with colour — choice of colour ramp and classification hides or reveals patterns.
• Map projection distortion — a world map in Mercator makes Greenland look the size of Africa.
• Scale ignorance — drawing inferences at a scale where your data cannot support them.

Awareness is the first defence.

A worked example

Suppose you're asked: "Does proximity to a highway correlate with childhood asthma rates?"

Applying the five questions:

• Where? Which city, which time frame, which age group, which highway network dataset?
• Why there? Plausible mechanism: vehicle exhaust, PM2.5, noise stress.
• How related? Before running a regression, plot a buffer of 500 m around highways and overlay asthma incidence to eyeball it.
• What scale? Individual residence addresses? Census blocks? Counties? Results differ.
• How certain? Asthma diagnoses underreport; addresses may be geocoded imprecisely; highway locations may be slightly outdated.

A regression p-value without this context is almost meaningless. A well-thought-out analysis accompanies statistics with maps, caveats, and alternative framings.

Drills you can do this week

• Mental mapping. Draw, from memory, a map of your neighbourhood. Compare to the real thing.
• Caption rewriting. Take three published maps and rewrite each caption to state explicitly: the unit of analysis, the time range, the classification method, the CRS, and one source of uncertainty.
• Operation hunting. Read any news article involving location ("crime increased in the southeast district"); identify which GIS operations would be needed to reproduce the claim.
• Tobler check. Every time you see a map of something, ask: does near-it look more like it than far-from-it? If yes, Tobler is at work.

Self-check exercises

Summary

• Spatial thinking combines visualisation, abstraction, reasoning about location, and scale sensitivity.
• Tobler's First Law — near things are more related than distant things — underpins most spatial methods.
• Five habit questions (where, why there, how related, what scale, how certain) structure any analysis.
• Watch for the ecological fallacy, MAUP, projection distortion, and cartographic misuse of colour.

Further reading

• National Research Council — Learning to Think Spatially.
• Tobler, W. (1970) — A Computer Movie Simulating Urban Growth in the Detroit Region.
• Monmonier, Mark — How to Lie with Maps.
• Downs & Stea — Image and Environment: Cognitive Mapping and Spatial Behavior.