Touching Emotions, Smelling Shapes: Exploring Tactile, Olfactory and Emotional Cross-sensory Correspondences in Preschool Aged Children

AI summary

Overview

• Research area: Human-Computer Interaction, specifically Child-Computer Interaction (CCI) and crossmodal/cross-sensory cognition, with relevance to multisensory technology design.
• Technical level: Intermediate. The design is a structured play-based empirical study analysed with Pearson's Chi-square tests and standardised residual analysis; no specialised machine learning or engineering background is required, but familiarity with HCI study conventions helps.
• Scope: A single study with 26 preschool children (ages 2 to 4.5) that tests whether young children form systematic mappings between tactile shapes, scents, and emotions, and examines the verbal strategies children use to explain those mappings.

What This Paper Is About

Most digital technologies engage vision, sound, and touch, while smell and taste remain rarely incorporated, partly because designers lack evidence about how chemical senses connect to other modalities. That evidence is especially thin for preschool-aged children (2-4 years), a period when sensory integration develops rapidly. The researchers ran a story-based, play-oriented study with 26 preschoolers to find out which tactile shapes children link to which scents and which emotions, and how children justify those links, in order to ground child-facing multisensory design in children's actual cross-sensory behaviour.

Key Contributions

1. Empirical evidence of smell–touch–emotion correspondences in preschool-aged children (2-4 years), a group the authors state has been largely unexamined for olfaction and emotion.
2. Methodological insights demonstrating the value of narrated, child-centred approaches (anthropomorphic characters, a "magic box" for unseen tactile exploration, a teacher present as co-participant, and a simplified binary-choice adaptation of the Self-Assessment Manikin) for eliciting reasoning from 2-4 year olds.
3. Design implications for cross-sensory technologies intended to support children's communication and emotional expression by grounding interactions in cross-sensory correspondences, including the recommendation that researchers adopt a more systematic approach to sensory stimuli.
4. A replicable method and materials for probing cross-sensory cognition in this age group, including 3D models of the tactile stimuli provided in the supplementary materials.

Main Findings

• Tactile–linguistic correspondence was statistically significant. The Chi-square test of independence indicated a significant association between label and tactile stimulus, χ²(4, N = 78) = 10.42, p = 0.034. Standardised residuals showed this was driven by strong associations between "Kiki" and the Spiky shape and between "Bouba" and the Round shape, plus a negative association (the text describing the negative association is cut off in the available content).

• "Kiki" was chosen most often for Spiky and "Bouba" most often for Round. In the reported counts, Kiki was given for Spiky 10 times and for Round 3 times; Bouba was given for Round 10 times and for Spiky 2 times. "Neither" was frequent across all shapes (14 for Spiky, 13 for Round, 16 for Cylinder).

• Tactile–olfactory correspondence was statistically significant. Spiky was most strongly linked with the Lemon scent (69.2 percent of responses), Round with Vanilla (65.4 percent), and Cylinder with Air (50 percent). The underlying counts were Lemon 18 for Spiky, 3 for Round, 6 for Cylinder; Vanilla 2 for Spiky, 17 for Round, 7 for Cylinder; Air 6 for Spiky, 6 for Round, 13 for Cylinder.

• Olfaction–emotion correspondence was statistically significant. Vanilla elicited the highest share of Calm responses (72 percent), followed by Air (52 percent) and Lemon (20 percent). Lemon provoked the strongest Excited response (48 percent), with Air at 16 percent and Vanilla at 12 percent. The underlying counts were Calm 5 for Lemon, 18 for Vanilla, 11 for Air; Excited 12 for Lemon, 3 for Vanilla, 5 for Air; Neither 8 for Lemon, 4 for Vanilla, 9 for Air.

• Tactile–emotion correspondence was not statistically significant. The paper states that Spiky shapes were associated with Calm and Neither at 44.4 percent each and Excited at 11.1 percent; Round shapes were predominantly associated with Excitement at 46.2 percent, with Calm and Neither both at 26.9 percent; Cylinder responses were more balanced, at 38.5 percent Calm, 30.8 percent Excited, and 30.8 percent Neither.

• A neutral stimulus anchored to "Air". The paper reports that the neutral Cylinder shape was most strongly associated with air.

• Children used a variety of explanatory strategies, but the specific results are not in the available content. The paper states it analysed cross-sensory association strategies using seven deductive coding categories carried over from prior work (Embodied Action, Grasping for Another Sense, Personal Connection, Sensory Features, Valence, Vocalisation, Familiar Experience), noting that strategies are not mutually exclusive and that children often layered them. The section reporting which strategies preschool children used most is truncated in the supplied text, so those specific frequencies are not reported here.

• Not every child explained a choice. Reasoning was not given by all children (n = 3 for tactile stimuli, n = 8 for olfactory stimuli).

• Prior benchmark context. The paper cites that 95 percent to 98 percent of the population associate a spiky shape with "Kiki" and a rounded shape with "Bouba", that preschoolers as young as 2.5 years show the Bouba–Kiki effect, and that prelexical infants do not.

Methodology in Plain English

The team recruited 26 children (16 females, 10 males) aged 2 to 4.5 years (M = 45.08 months, SD = 9.19 months) by asking four nurseries to circulate forms and information sheets to caregivers. All children were fluent in English. Sessions took place in a familiar nursery room with a member of staff always present.

Children were told a story in which three characters needed to be met, named, and helped. The characters were presented as invisible tactile shapes that children felt inside a closed wooden box, with a cloth-covered hole for the child's hand so the shape could not be seen. Three translucent 3D-printed stimuli were used: Round (54.7 mm × 54 mm × 55 mm), Spiky (56.7 mm × 57.2 mm × 66 mm), and Cylinder (43.7 mm × 43.7 mm × 43.7 mm). They were printed on a Form4 high-precision Stereolithography (SLA) printer with transparent resin at 0.025 mm resolution; sharp edges were sanded smooth; the shapes were designed to have approximately the same volume (65 mL) and the same weight (80 g) so children could hold them with a whole hand.

Scents (lemon, vanilla, and air) were delivered by a Hynt scent-dispensing machine controlled by a researcher from a laptop, with each scent presented for 4 seconds and breaks built in between deliveries because olfactory sensing declines with repeated stimuli. Emotion options were "excited", "calm", and "neither", framed as three villages rather than as pets, to reduce bias from children's prior experience. The story had four tasks: naming each tactile shape ("Bouba", "Kiki", "neither", or a self-created name); assigning an emotion to each tactile shape; assigning an emotion to each scent; and assigning a scent to each tactile shape. Associations did not need to be exclusive and children were never corrected. The naming section was always first and the conclusion always last, while the other three sections and the presentation order of tactile shapes were randomised per participant; the order of olfactory stimuli was kept consistent across participants because of limitations with the legibility of the system's interface, to support reliable transcription.

Each session was video and audio recorded with a GoPro positioned to hide the child's face while capturing interaction with the stimuli. The first author manually transcribed 2 hours and 29 minutes of recordings into a spreadsheet, capturing correspondences and children's reasons. Names other than "Kiki", "Bouba", or "Neither" were grouped as "Other". Analysis used descriptive statistics, Pearson's Chi-square tests, standardised residual analysis (residuals near zero indicate no strong preference, residuals above about 2 indicate a choice above chance, below about -2 below chance), and deductive coding of association strategies. The first author coded initially, the second author reviewed, and discrepancies were discussed among all authors until agreement.

Why This Matters

The paper argues that without a theoretical understanding of how olfaction integrates with other modalities, efforts to add smell to interactive systems risk remaining ad hoc. It positions cross-sensory correspondences as a principled foundation for integrating a wider range of senses into child-facing technology, and as a route to supporting children whose verbal language for emotion is still developing.

Real-world applications the work points toward:

• Emotion communication tools for young children: tangible or multisensory systems that help 2-4 year olds externalise and share feelings, similar in spirit to the cited ChaCha chatbot, MoodGems displays, and EmotionBlock tangible blocks, but grounded in smell and touch rather than words.
• Child-facing learning and play technology: scents and haptic objects paired according to measured correspondences (for example lemon with spiky shapes, vanilla with round shapes) rather than arbitrarily.
• Tangible user interfaces for preschoolers: the paper notes that tangible interfaces suit this age group and that physical interaction can support longer, more meaningful engagement than tablet-only interaction, making verified tactile–olfactory pairings directly usable in toy and exhibit design.
• Accessibility and inclusive design: the paper frames broader sensory engagement as a way to support accessibility, and tactile feedback as an inclusive channel highlighted in accessibility guidelines.

Industry relevance: the findings are directly usable by designers of toys, educational products, and interactive exhibits who want to add scent channels to products for young children, and by HCI researchers who need validated stimuli pairings and an age-appropriate elicitation method. The paper's headline recommendation, that researchers and designers adopt a standardised set of sensory stimuli, is aimed at making results comparable across studies.

Future Directions

• Determine which association strategies preschoolers actually favour. The paper's framing notes that children aged 8-11 relied most on Familiar Experience while young and older adults used a wider variety of strategies; how 2-4 year olds compare is the open question the truncated findings section addresses.
• Validate affect measurement instruments for preschoolers. The authors state there is relatively little evidence validating the Self-Assessment Manikin and its variants for pre-school-aged children, and that they adapted it into binary choices rather than validating it.
• Extend beyond three stimuli per modality and beyond lemon, vanilla, and air. The study deliberately limited each category to three stimuli to avoid overwhelming children; broader stimulus sets are needed to test the generality of the mappings.
• Address the methodological constraint on scent ordering. Because of system interface legibility issues, olfactory stimulus order was kept consistent across participants rather than randomised, which is a limitation for future replication.

Target Audience

Researchers and practitioners in Human-Computer Interaction and Child-Computer Interaction working on multisensory or olfactory interaction; designers of toys, educational technology, and interactive experiences for preschool children; and developmental or crossmodal cognition researchers interested in how touch, smell, and emotion become linked in early childhood. The paper is also useful for researchers who need a concrete, ethically informed protocol for running structured tasks with 2-4 year olds, including the use of storytelling, staff co-presence, and child assent.

Authors’ abstract

The use of a wide range of sensory modalities is increasingly central to technologies for learning, communication, and affective regulation. During the preschool years, sensory integration develops rapidly, shaping how children perceive and make sense of their environments. A key component of this process is cross-sensory correspondence: the systematic ways in which perceptions in different sensory modalities influence one another. Despite its relevance, little is known about cross-sensory correspondences in preschool-aged children (2-4 years). We present a study with 26 preschoolers examining smell-touch-emotion correspondences through playful tasks. We found significant correspondences both between sensory modalities and between sensory modalities and affective judgements. Further analysis revealed association strategies underpinning these mappings. We contribute empirical insights into cross-sensory correspondences in early childhood, design guidelines that align with how preschoolers relate sensory input, and a replicable method for probing cross-sensory cognition in this age group.