The Alternative Uses Test: Measuring Divergent Thinking
The Alternative Uses Test (AUT) is the most widely used measure of divergent thinking in creativity research. Developed by J.P. Guilford in the 1950s as part of his Structure of Intellect model, it has survived as a primary research instrument for over 70 years — an unusual lifespan in cognitive science.
The test is simple: given an ordinary object, generate as many different uses as you can within a time limit, typically two minutes. The objects are deliberately mundane. A brick. A newspaper. A paperclip. A shoe. The mundaneness is the point — it eliminates domain expertise and forces purely associative thinking.
What the Four Scoring Dimensions Actually Measure
The power of the alternative uses test isn't in the task itself. It's in the scoring framework Guilford developed to extract structure from open-ended responses.
Fluency counts the total number of valid responses generated. More responses indicate broader search through the associative network. Fluency is the easiest dimension to improve with practice, and the first to respond to training.
Originality measures the statistical rarity of each response relative to population norms. If the prompt is a brick and you say "doorstop," that response scores near zero — almost everyone says doorstop. If you say "a heat sink for small electronics," that scores much higher, because very few people access that category. Originality scoring requires comparison against normative databases, which is why it's harder to self-score informally.
Flexibility measures how many different semantic categories your responses span. Ten responses all related to construction score low for flexibility. Responses spanning construction, art, cooking, and measurement score high. Flexibility is what distinguishes a genuinely broad search from a fluent search within a single domain.
Elaboration measures the level of detail added to each response. "Use it as a grinder" scores lower than "Use it as a surface to powder spices — hold the brick steady and scrape dry ingredients across the rough face to break them down." The elaboration dimension captures depth of processing, not just breadth.
These four dimensions don't move together. High fluency doesn't guarantee high originality. A responder who generates 40 uses for a newspaper, all of which are paper-related, scores high on fluency and low on flexibility and originality. Understanding which dimension is weak is more useful than knowing the composite score.
What the AUT Measures (and What It Doesn't)
The AUT captures one specific component of creative cognition: the ability to generate multiple responses to an open-ended prompt while moving past conventional associations. It's a measure of divergent thinking — deliberately isolated from other cognitive capacities.
This is distinct from convergent thinking, which identifies the single correct answer to a constrained problem. It's also distinct from creative synthesis, which combines disparate ideas into a coherent structure. The AUT isolates raw idea generation.
Research by Evangelia Chrysikou at Drexel has clarified some of the neural mechanisms. Studies using transcranial direct current stimulation show that AUT performance improves when activity in the left prefrontal cortex is temporarily suppressed. That region handles semantic control — the active filtering of word meanings. Reducing that filtering appears to widen the associative network that the brain searches during the task. In other words, the AUT is partly measuring the ability to relax semantic constraints.
A separate line of research connects high AUT scores to default mode network (DMN) connectivity. The DMN is active during internally directed thought — what researchers used to call mind-wandering. People with high divergent thinking scores tend to show stronger functional connectivity between DMN nodes involved in memory retrieval and semantic processing. The broad search the AUT requires appears to recruit the same network the brain uses for self-directed imagination.
Why It Has Lasted 70 Years
Many cognitive assessments from the mid-20th century didn't survive methodological scrutiny. The AUT has. A 2019 meta-analysis by Sowden and colleagues confirmed that AUT scores correlate significantly with self-reported creative achievement across domains including science, visual arts, writing, and everyday problem solving. The correlation with real-world creative output — not just other test performance — is what gives the measure validity.
It's also fast, cheap, and standardizable, which matters for research. The Torrance Tests of Creative Thinking (TTCT) extend the AUT framework into figural domains — completing incomplete drawings, finding uses for shapes — and take considerably longer to administer. For studies where time is limited, the AUT provides a valid divergent thinking estimate in under 10 minutes.
The measure has well-documented limitations. Originality scoring depends on population norms, which are culture-specific and potentially outdated. Elaboration scoring is partially subjective. The 2-minute constraint creates ceiling effects for very fluent responders. More recent work addresses some of this: semantic distance measures that use natural language processing tools can calculate originality without normative databases, offering more objective scoring that generalizes across populations.
How Training Changes AUT Performance
The AUT isn't just a measurement tool. Regular practice with alternate uses tasks produces measurable improvements in the dimensions it measures.
A study by Scott, Leritz, and Mumford found that deliberate divergent thinking practice produced effect sizes of 0.8 to 1.1 on fluency and originality measures — large effects by cognitive training standards. The training effect is largest for originality and flexibility, which is notable because those are the dimensions most relevant to real creative output.
The mechanism appears to be extending the retrieval search — training the brain to continue looking past its first cluster of associations. Most people's first five responses to any AUT item come from a small, similar cluster. The original and flexible responses come from pushing into more distant semantic territory, which requires inhibiting the first-cluster responses and forcing continuation.
This is why timed conditions matter. Time pressure prevents extended evaluation of each response, which forces faster cycling through the associative network. Generating more responses faster is not about quality in the moment — it's about forcing the brain past the familiar territory where quality eventually lives.
The AUT vs. Similar Measures
The AUT sits in a family of creativity measures that each capture something different.
The Remote Associates Test (RAT) is essentially the inverse of the AUT. Where the AUT rewards breadth and volume, the RAT rewards precision: finding the single word that connects three apparently unrelated words. AUT and RAT performance aren't perfectly correlated, reflecting genuinely different cognitive operations. High AUT fluency doesn't guarantee high RAT performance.
The Torrance Tests of Creative Thinking extend the AUT framework but add figural components and longer administration time. Research consistently finds that verbal and figural creativity aren't perfectly correlated — administering both provides a more complete picture than either alone.
The word association game as typically used isn't a creativity measure, but the cognitive capacity it trains — fast retrieval across a broad associative network — overlaps significantly with what the AUT measures. The difference is that the AUT explicitly tracks originality; free word association doesn't.
Training the Capacity Directly
If you want to improve the cognitive capacities the AUT measures, practicing the task itself is the most direct route. The divergent thinking exercise uses the same structure: a specific object or prompt, a time limit, and a goal of generating varied responses across as many semantic categories as possible.
For improving originality specifically: after generating your initial list, categorize every response by semantic domain. Identify the two categories that contain the most responses. Then generate five more responses that don't fit any of the existing categories. This forced category exclusion directly trains flexibility and pushes originality upward — you can't reach new categories without leaving the familiar ones behind.
For improving elaboration: take three of your most original responses and expand each to two or three sentences. Describe exactly how the use would work in practice. This builds the depth-of-processing habit that elaboration scoring rewards.
The divergent thinking examples post has additional worked examples showing what high-quality AUT responses look like across different objects — useful for calibrating what "original" and "flexible" actually mean in practice.
Ready to train your creativity? Try science-backed exercises that measure and improve your creative thinking. Start a Free Exercise