Colours and emotions.
Crossmodal: what the studies say
Crossmodal research studies how the senses correspond: which sound goes with which colour, which smell with which shape. Most people make these links in the same way. It is not synaesthesia: nobody sees a colour when smelling an odour, but when asked to choose one, people's answers converge. The module brings together what published studies measured: smells, tastes, emotions, music, and the conventions of film. In the last tab, “Your colours”, choose up to three colours, or pick them from an image: each one is read with the same data, side by side.
“A compatibility effect between attributes or dimensions of a stimulus […] in different sensory modalities.”
Charles Spence, Crossmodal Research Laboratory, University of Oxford · Crossmodal correspondences: A tutorial review, 2011
OKLCH: lightness rises from black to white, hue turns around the axis, saturation grows outwards.
What remains are the references: the grey axis for lightness, the band of hues at the equator.
What to remember
- Light goes with positive. It is the link that comes back most across the senses: fresh smells get light colours, joy goes with light and vivid colours, and light, saturated colours also feel more lively.
- Hue is shared far less. Coffee and lemon reach consensus, lavender and pine divide; white goes with sadness much more in China (31%) than across the 37 countries (13%).
- Some links hold from study to study, others are conventions. Yellow and green for sour come back in 6 studies out of 7; the colours of film are a craft convention, not a measurement.
- To set the mood of an image, work on lightness and saturation first, then choose the hue with your audience in mind.
Glossary
- Crossmodal correspondence
- A link most people make between two senses: a high pitch seems light, a lemon smell seems yellow. Charles Spence (2011) defines it as a compatibility effect between attributes of stimuli in different senses.
- Synaesthesia
- An automatic perception in another sense, specific to a minority of people: seeing a colour when hearing a sound. Crossmodal correspondences, on the other hand, are shared by most people and produce no perception: they appear when people are asked to choose.
- Valence
- How positive or negative an emotion is. In the study by Lahdelma & Eerola, chords are rated from 1 (negative) to 7 (positive).
- Arousal
- How activated an emotion is, from calm to lively. With valence, it is one of the two dimensions measured by Wilms & Oberfeld (2018).
- Isolated colour
- A swatch or a colour name, shown alone: this is what the studies in this tutorial measure. In an image, the subject, the contrast and the placement of the colour change how it reads.
- OKLab, OKLCH
- The colour space by Björn Ottosson (2020), designed so that equal differences look roughly equal. OKLCH is its polar form: lightness, chroma, hue. Every result in the tutorial is converted into it.
- Lightness
- From 0 (black) to 1 (white): the height in the bicone, along the grey axis.
- Chroma
- The intensity of the colour, zero on the grey axis. The highest displayable chroma depends on lightness and hue: a vivid yellow is light, a vivid blue is dark.
- Relative saturation
- Chroma as a share of the highest displayable for that lightness and hue. It sets the distance from the axis in the bicone: the surface is the maximum, hence a regular shape.
- Hue
- The angle around the axis: red, yellow, green, blue, purple. The coloured band at the equator is the reference.
- Bicone
- The shape of the space in the tutorial: two cones joined at the base. At the poles, black and white have no chroma; the most intense colours sit towards the middle.
- ISCC-NBS
- A system of 267 colour names (Kelly & Judd, 1955). “Your colours” takes the closest category, from the centroids computed by Paul Centore, then reduces it to one of the 12 names of Jonauskaitė's survey.
- Ward et al., 2022
- 68 participants choose a colour for each of 10 smells (Heliyon). The individual answers are public (Zenodo, CC BY 4.0): the Smells tab shows the median and all 68 answers.
- Spence & Levitan, 2021
- A review of 7 colour-taste studies published between 1983 and 2016, 3,468 participants from the United States, Germany, the United Kingdom, 17 countries and 4 countries (i-Perception). Yellow and green for sour come back in 6 studies out of 7.
- Wan et al., 2014
- 428 participants analysed in China, India, Malaysia and the United States sort 11 colours under 5 tastes (Frontiers in Psychology). Bitter-black and sweet-pink hold in all four countries; no colour goes with umami.
- Jonauskaitė, 2024
- 8,615 participants from 37 countries say, for 12 colour names, which emotions they link to them among 20 (Psichologija; OSF data, CC BY 4.0). The tutorial recomputes the shares, country by country.
- Geneva Emotion Wheel
- The list of 20 emotions used by Jonauskaitė's survey: joy, love, sadness, anger, fear, shame and the others, with the option of choosing none.
- Lahdelma & Eerola, 2016
- 410 participants rate the valence of single chords played on the piano (i-Perception). The original recordings (CC0) are those of “Your ear”.
- Temperley & Tan, 2013
- Melodies played in different modes, compared in pairs: which one sounds happier (Music Perception). Ionian (major) comes first, Phrygian last.
- Palmer et al., 2013
- The link between music and colour goes through emotion: happy music makes people choose lighter, more saturated colours (PNAS). The tutorial uses it to compute the position of modes and chords.
- Wilms & Oberfeld, 2018
- 62 participants rate colours varying in hue, saturation and brightness (Psychological Research): lighter and more saturated, they feel more positive and more lively.
- Bellantoni, 2005
- If It's Purple, Someone's Gonna Die sorts over 60 films under six colours, each with two faces. It is a craft reading, not a measurement: the Cinema tab shows it as flat swatches, at indicative hues.
- Median
- The middle value: as many answers above as below. For a smell, the median of lightness and of the two colour axes gives the main ball.
- Spread
- The mean distance of the answers to their median. Coffee reaches consensus (33 CIELAB units), pine divides (61): clustered, the balls show agreement; scattered, they do not.
- Share of participants
- The percentage of people who made a choice: 31% of Chinese participants link white to sadness, against 13% across the 37 countries.
- Colour name, typical shade
- Several studies ask for a colour name (“red”), not a shade. The tutorial places these names at a typical shade, the RGB values of Wan et al. (2014).
- Computed position
- No study has measured the colour of a mode or a chord. Their lightness and saturation are computed from their measured emotion; hue is not fixed, hence rings rather than balls.
- Craft convention
- A habit of the profession, passed on through films and books, not a study result: this is the status of the colours of film.
- “Your colours” thresholds
- Reading marks chosen for the tutorial, not those of the studies: a colour is called light from a lightness of 0.65 and dark below 0.45, vivid from a chroma of 0.12 and dull below 0.06. A smell is called nearby when at least 10% of Ward's answers fall within 0.08 in OKLab.
Ten questions on smells, tastes, emotions, music and what the studies really measure.
Sources: Ward et al. (2022), Heliyon 8:e09284, raw data Zenodo (CC BY 4.0) · Spence & Levitan (2021), i-Perception 12(3), review of 7 colour-taste studies · Wan et al. (2014), Frontiers in Psychology 5:1365 · Jonauskaitė (2024), Psichologija 70:8, data OSF (CC BY 4.0) · Palmer et al. (2013), PNAS 110:8836, for the computed position of modes and chords · Temperley & Tan (2013), Music Perception 30:237 · Lahdelma & Eerola (2016), i-Perception 7(3), data and recordings Harvard Dataverse (CC0) · Spence (2011), Attention, Perception, & Psychophysics 73:971 · Wilms & Oberfeld (2018), Psychological Research 82:896, lightness and saturation · colour names: ISCC-NBS system (centroids by P. Centore) · colour names (tastes, emotions) are placed at a typical shade: the RGB values of Wan et al. (2014) · cinema: P. Bellantoni, If It's Purple, Someone's Gonna Die (Focal Press, 2005), a craft convention, words from its chapter titles, indicative hues · OKLab / OKLCH: Björn Ottosson (2020).
