Mapping the Sensory Landscape: Population-Scale Taste and Smell Data Across Asia-Pacific and Beyond
Background
Much of what we know about taste and smell comes from studies conducted in North America and Europe. The tools, frameworks, and assumptions that shape sensory science were largely developed within those cultural and dietary environments, and the instruments carry that origin with them. To function in Taiwan, the University of Pennsylvania Smell Identification Test (UPSIT) had to replace odorants including clove, cheddar cheese, gingerbread, and dill pickle with sandalwood, fish, rubber tire, and jasmine. In Australia, where the test was adapted by substituting response alternatives without changing a single odorant, healthy participants still scored below American norms, prompting researchers to propose a two-point correction factor. A test that requires a handicap to cross an ocean is measuring cultural familiarity alongside perception.
Across Southeast Asia, East Asia, and the broader Asia-Pacific region, food traditions have developed over thousands of years through distinct ingredients, fermentation practices, culinary philosophies, and sensory experiences. These regions represent some of the most differentiated flavor cultures in the world, and they remain thinly represented in the normative datasets behavioral sensory science runs on.
Four decades of cross-cultural work have narrowed this gap on the functional side, where culture-neutral instruments now exist, while leaving it open on the side that matters most for food: preference, choice, and the behavior built on top of them. A more complete view of the sensory landscape is a precondition for understanding how humans experience flavor, form preferences, and make food choices, and expanding research into these populations tests how much of what the field treats as perception is in fact culture.
-
What does population-scale behavioral taste data from underrepresented regions reveal about the distribution, clustering, and cultural patterning of human sensory preference — and what can that data contribute to emerging work in personalized nutrition, digital health, and sensory digital twin development?
-
Digitaste's Askyumi platform deploys behavioral sensory preference assessment in a consumer-accessible, multilingual format. Stimuli are presented as visual and verbal descriptors rather than delivered chemically, which situates the instrument alongside laboratory psychophysics rather than in place of it. The core instrument uses forced-choice paradigms with reaction-time measurement, complementing self-report with response-latency data. Whether latency carries incremental validity over rating scales in the food domain is itself an open question, and one a dataset at this scale is unusually well positioned to answer. Participants complete the assessment in approximately 20 minutes, producing individual profiles across two frameworks: the 16 Taste Types, indexing preference response at the sensory level, and Foodie Personality, mapping behavioral patterns in food choice. Both frameworks segment underlying continuous scores. Whether those segments correspond to genuine latent classes is treated here as an open empirical question rather than a property assumed by the instrument.
The critique this article opens with applies to any multilingual instrument, including this one. Cross-cultural comparisons are therefore gated on measurement invariance testing across language versions and item-level differential item functioning analysis. De-identified data, analyzed in aggregate, feeds into a shared research program with the World Taste & Smell Association across three domains: sensory cognition across demographic and cultural groups, sensory literacy and its relationship to dietary behavior, and the behavioral and quality-of-life dimensions of recovery in populations affected by taste and smell loss, measured alongside validated chemosensory testing rather than in place of it.
-
Hidden patterns in sensory preference. At sufficient scale, behavioral taste data can surface clustering that self-report is unlikely to detect: preference signatures cutting across demographic categories in unexpected ways, or sensory subpopulations within groups assumed to be homogeneous. Identifying those patterns in Asia-Pacific populations is a primary objective of this collaboration.
Cultural variance in flavor perception. The relationship between cultural food environment and individual sensory response is poorly characterized at the population level. Do people raised within high-umami culinary traditions show measurably different preference structures, meaning the relative weighting of taste, aroma and texture dimensions in their choices, than those raised in high-fat or high-acid traditions? Does early dietary exposure shift sensory response itself, or the behavioral pattern layered over it? Scale alone will not settle the second question, which is developmental and causal. The design that can approach it compares heritage, migrant and host populations within a single instrument, and that comparison requires exactly the cross-population coverage this collaboration is assembling.
Validation of digital twin frameworks. Sensory digital twins, computational models that simulate individual perceptual response for applications in personalized nutrition, food product development, and clinical dietary planning, require behavioral data to train against and independent criteria to validate against. The dataset being built through this collaboration serves the first function directly and the second through held-out partitions and external criterion measures, including laboratory sensory panels and observed food choice, so that validation does not reduce to testing a model against the data that produced it.
-
Data collection is ongoing across multiple markets under institutional ethics review, with participant consent covering secondary research use. Analysis of geographic and cultural patterning within the dataset is underway. Analysis plans are pre-registered, and findings will be reported through peer-reviewed publication rather than summarized here.
-
The scale achievable through consumer digital deployment differs categorically from what clinical or laboratory settings can produce. A dataset at this scale, generated through behavioral measurement, distributed across culturally distinct populations, and longitudinally extendable, is a resource sensory science has not previously had in this form or at this geographic scope. Scale is not representativeness. The sample is self-selected rather than probabilistic, and the analytical program relies on quota sampling, post-stratification weighting, and panel-based recontact to characterize and correct that skew.
The WTSA × Digitaste collaboration is designed to direct that resource toward questions with scientific, clinical, and commercial consequence: how taste and smell preference is distributed across populations that have been systematically underrepresented in sensory research, how cultural food environment shapes perceptual response at the individual level, and how that understanding can ground the next generation of personalized nutrition and digital health tools in evidence rather than assumption.
-
The World Taste & Smell Association contributes scientific oversight, clinical research infrastructure, and domain expertise in olfactory and gustatory science. Digitaste contributes platform methodology, behavioral data architecture, and the Askyumi assessment instrument. Joint outputs will include peer-reviewed research, public sensory literacy resources, and technical data partnerships supporting digital twin validation in food, health, and wellness applications.
All user data is aggregated and de-identified prior to research use. Individual Askyumi profiles are not shared. For research inquiries: Mindy@tasteandsmell.org.