Numerical cognition
Do we learn that odd numbers 'go with' a colour without noticing?
Open access · cc by · source: Europe PMC
When even and odd (or small and large) digits usually appeared in particular colours, people doing a simple colour task learned these category-colour links, mostly without realising it.
How this study connects
Role on claims
Each row is a claim on a concept or method page where this paper supports, challenges, or qualifies the statement. Roles are hand-checked — not a model guess.
Not yet placed on a claim. This paper has study layers, but no concept page yet cites it as support, challenge, or qualifier.
Study at a glance
- Design
- Human experiment — Colour-naming task with high-probability (about 10:1) number-colour pairings organised by parity or magnitude (category-level group) or arbitrarily (item-level group); congruent vs rare incongruent trials compared, plus an explicit association report with novel two-digit numbers.
- N
- N=54 · 54 adults analysed (56 tested, 2 excluded for low accuracy), split evenly between category-level and item-level versions; each did both parity and magnitude sections. Only the first 36 answered the awareness questionnaire.
- Population
- Young adults (18-30) from the University of Luxembourg community, without synaesthesia or dyscalculia
- Outcome
- Accuracy and response time on congruent vs incongruent colour trials; accuracy of explicitly choosing the associated colour for novel two-digit numbers
Structured fields used in claim comparison tables when every cited study has a complete layer.
Inspect source passages
Do we learn that odd numbers 'go with' a colour without noticing?
These passages occur verbatim in the stored paper text. They support the points listed below; they do not verify every statement in the explanation.
54 participants analysed after excluding 2 of 56
A total of 56 participants were tested, from which two were excluded for low accuracy, below 80%, on control (congruent) trials, leading to a total sample of 54, pseudo-randomly, evenly assigned to the experimental (concept-level) and control (item-level) conditions (age range: 18–30 years; mean age = 22.4 years; 11 male, 1 non-binary; 6 left-handed).
3.06% accuracy cost for incongruent parity trials at the category level
In the parity experiment, there was lower performance for incongruent (M = 89.3%; SE = 1.69%) than congruent (M = 92.4%; SE = 1.17%) trials in terms of accuracy, i.e., a significant difference of 3.06%, t 26 = 3.67, p < .001, d = 0.71, as predicted (, upper panel).
Key findings
At the category level, accuracy dropped by about 3% on unusual-colour trials for parity and by about 1.3% (with about 9 ms slower responses) for magnitude, whereas the item-level version showed no significant effects. For parity, the congruency effect was significantly larger in the category-level than the item-level group, but for magnitude this interaction was not significant. In the explicit report with new two-digit numbers, category-level participants chose the matching colour above chance (about 63% for parity, 55% for magnitude) while item-level participants were at chance, and very few reported noticing the parity or magnitude rule.
Methodology
Participants pressed a key for the ink colour of digits 2-9 and were not told that each digit usually appeared in one colour. In the category-level version, colour tracked a number concept (even digits mostly blue and odd mostly yellow; small digits mostly green and large mostly red); in the item-level version, the same digits had equally consistent colours but with no conceptual pattern. Learning was measured as worse performance on the rare trials where a digit appeared in its unusual colour, followed by a task asking which colour went with new two-digit numbers, answered with the other hand and different keys.
Limitations
The effects are small (a few percent in accuracy, a few milliseconds in speed) and the authors say they need replication. Support for automatic magnitude processing is limited because the category-vs-item interaction was not significant for magnitude, and the item-level version was harder to learn (eight separate pairings vs two category rules), so it is not a perfectly matched control. Awareness was only probed in the first 36 participants with open questions, which may miss partial awareness, and the colours were vivid, prototypical ones that may be especially easy to associate.
Related papers in this topic
Same topic cluster — not a recommendation engine.