Colour psychology occupies an odd position. It is simultaneously one of the most confidently asserted bodies of knowledge in marketing and one of the shakiest in the laboratory. Infographics circulate claiming that blue signals trust, that orange drives impulse purchases, that a particular shade of red lifts conversion rates by a specified percentage. Meanwhile the peer-reviewed literature is a patchwork of promising early results, awkward replication attempts and effects that shrink the harder anyone looks.
There is real science here. It is just considerably more modest, and more interesting, than the version that turns up in brand guidelines.
Sorting the claims into piles
Broadly, assertions about colour and mind fall into four groups, and they are not equally supported. Physiological claims concern light acting on the body directly. Associative claims concern which feelings people link to which hues. Behavioural claims concern colour changing what people do. Diagnostic claims propose that colour preference reveals something stable about who you are.
The first pile is on firm ground but is not really psychology. Short-wavelength light suppresses melatonin and shifts circadian timing through specialised retinal cells discovered around the turn of the century, which is why screen manufacturers added warm evening modes. That is photobiology, and it concerns the wavelength of ambient light reaching your eye, not the colour a designer painted a button.
The part that holds up: colour and feeling
The associative pile has genuinely good data behind it. In 2020 Domicele Jonauskaite and colleagues published a study in Psychological Science covering 4,598 participants across 30 nations and 22 languages, each asked which emotions they associated with each of twelve colour terms. The pattern was strikingly consistent worldwide. Red drew both anger and love. Yellow and orange drew joy. Black drew sadness and fear. Overall similarity between countries was high, and it declined in an orderly way with linguistic and geographic distance, so that neighbours with related languages agreed more closely than distant strangers did.
This is a real and replicable finding, and it deserves to be stated carefully. It shows that colour terms carry shared emotional connotations across much of the world, probably arising from a mixture of common physical experience, such as blood, sunlight and darkness, and shared cultural transmission. It does not show that seeing a colour causes the associated emotion. Asking someone what red makes them think of and measuring what red does to them are different experiments, and the first is much easier to run.
The red problem
The behavioural pile is where things get uncomfortable, and red is where the trouble concentrates. In 2005 Russell Hill and Robert Barton reported in Nature that across four combat sports at the 2004 Olympics, competitors randomly assigned red kit won more often than those assigned blue. It was a memorable result with an evolutionary story attached, about red as a signal of dominance across various species.
Then in 2007 Andrew Elliot, Markus Maier and colleagues reported a different red effect: briefly seeing red before a test impaired performance, apparently because red is associated with danger and failure in achievement settings and triggers avoidance. A separate line of work reported that red increases men's ratings of women's attractiveness.
Each of these has run into difficulty. A 2024 meta-analysis pooling 6,589 contest outcomes from 1996 to 2020 found red winning 50.5 per cent of the time, a difference that did not reach statistical significance. Analyses of ultimate fighting and of judo have similarly failed to recover the original advantage. On the romance side, Gabrielle Lehmann and Robert Calin-Jageman ran two pre-registered replications with the original materials and considerably larger samples, and found effects hovering around zero, one slightly positive and one slightly negative. Statisticians examining the red-attractiveness literature have also noted that its published studies succeeded suspiciously often given how small they were, a signature of selective reporting.
Not everything has collapsed. A pre-registered replication published in Acta Psychologica in 2019 found that people marking work with a red pen flagged more errors and completed more word stems with failure-related words than those using blue, supporting a narrow version of the red-and-failure association. The pattern across the field is consistent with small, context-bound effects that early small studies inflated into general laws.
Context does most of the work
Elliot himself proposed the sensible resolution, in what he called colour-in-context theory: the meaning of a colour depends entirely on the situation in which it appears. Red on an exam paper means error. Red on a dress in a bar means something else. Red on a warning label means stop. Red at a Chinese wedding means good fortune. There is no context-free effect of red on human behaviour, because there is no context-free red.
Cross-cultural variation reinforces the point. White is the colour of weddings in much of Europe and of mourning in parts of East Asia. Purple carried imperial connotations in the Mediterranean world for the mundane reason that the dye was extracted from sea snails at ruinous expense. These meanings are learned, they differ, and they change. Any theory that treats a wavelength as having a fixed psychological payload is going to keep being embarrassed by the data.
Expectation is doing more than pigment
One of the more reliable colour effects has nothing to do with colour acting on the nervous system and everything to do with what people expect. A 1996 systematic review in the BMJ by Anton de Craen and colleagues examined the effect of the colour of medicines and found that people judged blue and green tablets more sedative and red, yellow and orange ones more stimulant. The authors also noted that hypnotics and sedatives on the Dutch market were disproportionately likely to be green, blue or purple, suggesting manufacturers had absorbed the same intuition.
This is a genuine finding with practical consequences for drug trials and adherence, but the mechanism is belief rather than biology. The tablet is not doing anything different. The person taking it is.
The same logic accounts for the most famous colour anecdote of all. In the late 1970s a shade of pink was reported to reduce aggression in prisoners, and holding cells were duly repainted around the world. Subsequent controlled studies largely failed to find the effect, and some found the reverse. The original observations came from a setting where everyone involved knew what was expected of the paint.
Your favourite colour reveals nothing about you
Which brings us to the diagnostic claims, and here the verdict is straightforward: there is no serious evidential basis for the idea that colour preference reveals personality. Nothing resembling a validated instrument exists. The proposed links are generated by free association, then confirmed by describing the interpretation to people in terms flattering and vague enough that almost anyone would nod.
There is also a plain statistical obstacle. Blue is the most popular colour by a wide margin in survey after survey, across countries and across sexes, typically preferred by something like a third to a half of respondents. A trait marker chosen by a third of the population cannot distinguish between people. Whatever colour preference tracks, and it does correlate weakly with things like age and with the emotional associations described above, it has nowhere near the discriminating power a personality measure requires.
What to do with all this
The defensible position is narrower than the marketing and more useful than blanket scepticism. Colours carry shared associations that are broadly consistent across cultures, and those associations are worth respecting when you design something: red for warnings, calm hues for a clinic waiting room, high contrast wherever legibility matters more than mood. Colour also affects attention and visibility in ways that are measurable and uncontroversial, which is probably the real explanation behind many a triumphant button-colour experiment.
What the evidence does not support is the idea that a hue reaches past your judgement and operates you directly, or that your preference among twelve swatches is a window onto your character. Colour is a language, learned in a place, read in a context. That is a more demanding thing to work with than a lookup table of effects, and considerably closer to the truth.
