What is sensory analysis
Sensory (organoleptic) analysis is the scientific discipline that studies how the properties of a product are perceived by the human senses — sight, smell, taste, touch and hearing — and turns sensations into quantitative data. The terminology of the discipline is defined by ISO 5492 (Russian adoption: GOST ISO 5492-2014).
The mission of sensory science is to translate subjective sensations ("tender", "aromatic", "crunchy") into objective metrics that a technologist, the quality department and marketing can understand. For example: "tender chocolate" is not a metaphor but a score of 6.5 on a meltability scale (0–10) plus a measurable time of complete melting in the mouth.
Product perception is the result of a triad: the product (chemical composition, texture, temperature), the person (perception thresholds, experience, genetics) and the context (time of day, environment, expectations). Sensory methodology exists precisely to control the last two members of the triad: people are calibrated (module 02), and the context is fixed by the protocol (module 03).
History: from psychophysics to digital sensory science
Gustav Fechner formalizes the link between the physical intensity of a stimulus and the subjective sensation — psychophysics is born. Ernst Weber describes difference thresholds (just noticeable difference, JND).
The US Army Quartermaster Laboratory (Natick): which rations will soldiers actually eat? Systematization of consumer testing, paired comparison, the triangle test.
The triangle test is developed at the Carlsberg laboratories (Bengtsson, Helm) — the first discrimination method with a rigorous probabilistic guessing model.
The 9-point hedonic scale by Peryam and colleagues — still the primary tool for measuring acceptability (see module 06).
The descriptive revolution: the Flavor Profile method (Arthur D. Little), then QDA (Quantitative Descriptive Analysis) from Tragon — Stone and Sidel: line scales, individual scores, statistics. In the USSR — the commodity-science school of sensory evaluation (G. L. Solntseva, VNIIMP, MGUPP) and GOST standards for organoleptic testing.
Integration with business: preference mapping, conjoint analysis. Sensory science becomes part of R&D at Nestlé, Unilever, Mars.
Temporal methods: TDS (Temporal Dominance of Sensations) — the dynamics of sensations over time. Rapid consumer methods: napping, CATA.
Digital transformation: the "electronic nose" and the "electronic tongue", neurophysiology (EEG, eye-tracking), machine learning for liking prediction, cloud data-collection platforms — including Sensostat.
The five perception channels
A product is perceived by all channels simultaneously, and disrupting one of them changes the overall impression. Tasting methodology is designed to control each channel separately.
Vision
Colour, gloss, shape, surface. The first impression forms in fractions of a second and sets expectations ("dark = bitter"). In taste tests, colour is masked (red light, opaque cups).
Smell
Orthonasal (through the nostrils — "odour") and retronasal (from the oral cavity — "aroma"). The retronasal route makes the main contribution to the product's "flavour".
Taste
Five basic tastes: sweet, sour, salty, bitter, umami. Plus chemesthesis: the burn of chilli pepper, the cooling of menthol, the astringency of tannins.
Touch and texture
Hardness, viscosity, meltability, firmness, granularity. The mechanics of chewing and tactile sensations in the mouth (mouthfeel) are a separate class of attributes (texture profile, ISO 11036).
Hearing
The "snap" of chocolate and biscuits is a marker of freshness and proper tempering; the rustle of packaging affects how fresh crisps are perceived.
Multisensory integration
The channels are not independent: a redder drink seems sweeter at the same sucrose level; boosting vanilla raises the perceived sweetness; the sound of crunch changes the freshness rating. That is why sensory tests control every channel, not just the one under study.
Psychophysics: how sensation relates to stimulus
Psychophysics is the mathematical foundation of sensory science. Three key laws:
- Weber's law: the just noticeable difference is proportional to the stimulus level: ΔI / I = k. For sweetness k ≈ 0.1–0.2: to notice a sweetness difference in a lemonade with 10% sugar, the concentration must change by 1–2 percentage points.
- Fechner's law: S = k · ln(I) — sensation grows as the logarithm of the stimulus.
- Stevens' power law: S = k · In — the exponent n depends on the modality: n < 1 (compression: odour, brightness), n ≈ 1 (sweetness, length), n > 1 (expansion: saltiness n ≈ 1.3, electrotactile stimuli n ≈ 3.5).
Practical takeaway: with a non-linear "stimulus → sensation" relationship, rating scales must be calibrated with references (module 05, Spectrum method), and recipe changes should be evaluated through the JND rather than by "noticeable / not noticeable to the eye".
Perception thresholds
- Detection threshold — the minimum concentration at which "something is there" is perceived, without identifying the quality.
- Recognition threshold — the concentration at which the quality is identified ("this is sour").
- Difference threshold (just noticeable difference, JND) — the minimum difference between two samples noticed in 50% of cases.
- Terminal threshold — the level above which increasing the concentration no longer intensifies the sensation.
Thresholds are measured with dilution series under forced choice (usually 3-AFC — "which of the three samples is different", ISO 13301). The psychometric function links concentration to the probability of a correct response; it is raised from the guessing level p₀, and the threshold is defined as the point halfway between chance and 100%.
Three classical psychophysical methods for measuring thresholds
All three methods grew out of Fechner's laboratory and still underpin assessor screening. Choosing a method is a trade-off between threshold accuracy and the subject's time.
Method of limits
Concentration is changed in steps — ascending and descending series; the threshold is the average transition point between "not perceived" and "perceived". The modern industrial version is ascending forced-choice series: at each step the assessor receives a 3-AFC triad, and the individual threshold (best estimate threshold, BET) is computed as the geometric mean of the last "missed" and the first "detected" concentration. This is exactly how taste (ISO 3972) and smell (ISO 5496) screenings are built.
Method of constant stimuli
The subject is presented with 5–9 fixed concentrations in random order, bracketing the expected threshold. From the proportion of correct responses at each step, a psychometric function (logistic or Weibull) is fitted, and the concentration corresponding to 50% chance-corrected detection probability is found. The most accurate method, but the longest — dozens of presentations per step are required.
Method of adjustment
The subject smoothly adjusts the stimulus intensity to "just noticeable". Fast and convenient for research tasks (e.g., matching reference standards of equal intensity), but the result depends heavily on the person's own criterion — a bold subject "tunes" lower, a cautious one higher. Hardly used in panel screening.
Typical evaluation errors — and how the protocol suppresses them
Halo effect
The overall impression "fills in" the scores of individual attributes: a liked product gets inflated scores on every scale. Countermeasures: separate questionnaires per attribute, panel training.
Contrast and convergence
After a very sweet sample, the next one seems bland (contrast); after a similar one — even more similar (convergence). Countermeasures: serving-order balancing, palate cleansers, breaks.
Position effects
The first sample is scored differently (primacy effect). Countermeasure: the Williams Latin square — every sample appears in every position equally often (module 03).
Expectation and suggestion effects
Knowing the brand or "what should come out" shifts scores. Countermeasures: blind presentation, three-digit codes, no discussions during the session.
Adaptation and fatigue
After 5–7 samples, receptor sensitivity drops. Countermeasures: a limit of 6–8 samples per session, breaks, spitting (expectoration) for strong-tasting products.
Central tendency error
Assessors avoid the ends of the scale. Countermeasures: reference anchors at the ends of the scale, training to use the full scale.
Further reading
- Lawless H. T., Heymann H. Sensory Evaluation of Food: Principles and Practices. 2nd ed. Springer, 2010 — the discipline's core textbook: chapter 2 covers the physiology of the senses, chapter 5 — threshold measurement. book
- Drake M. A. Sensory Analysis and Consumer Preference: Best Practices. Annual Review of Food Science and Technology, 2023 — a recent review of best practices from the editor of the Journal of Sensory Studies. review
- IFST Guidelines for Ethical and Professional Practices for the Sensory Analysis of Foods (2025 edition) — the ethical framework for tests involving people: consent, risks, confidentiality. guideline
Test yourself
Three questions on the fundamentals
Pass the quiz for the module to count towards your course progress.
1. Retronasal olfaction is the perception of aroma…
The orthonasal route is the "odour" from outside; the retronasal route is the "aroma" from the mouth, and it makes the main contribution to flavour.
2. Under Stevens' power law, an exponent n = 0.5 means that quadrupling the stimulus concentration makes the sensation grow…
S = k·In: 40.5 = 2. With n < 1 the sensation scale "compresses" high concentrations.
3. A sweet sample served after a very sweet one seems less sweet than it really is. This is…
Contrast is an order effect: the score depends on the previous sample. A balanced serving design and taste neutralization between samples work against it.
Next — people, not instruments
Module 02 looks at who may be admitted to tasting: screening to ISO 3972, training to ISO 8586:2023 and panel monitoring to ISO 11132.