Analytical Sensory Tests: Evaluating Food Quality with Precision

When you pick up a packet of biscuits from two different brands, both may list similar ingredients – but one tastes noticeably better than the other. That difference isn’t just about chemistry. It’s about sensory quality, and measuring it precisely is the job of analytical sensory tests. These are controlled, scientific methods that use trained human assessors to objectively evaluate specific attributes of food – flavour, texture, aroma, and appearance. Unlike consumer preference tests that ask “do you like it?”, analytical tests ask “how exactly does this product differ, and by how much?”

What are analytical sensory tests? 🔗

Analytical sensory tests are a category of sensory evaluation designed to produce objective, measurable data about the sensory properties of food products. They are carried out by trained or expert panelists – not regular consumers – who function much like calibrated instruments. The goal is not to capture personal likes or dislikes, but to identify, describe, and quantify sensory characteristics with precision and repeatability.

Broadly, analytical tests fall into two main groups: discriminative tests (which determine whether a sensory difference exists between samples) and descriptive tests (which identify and measure the nature and intensity of those differences). Within the discriminative category, there are further subdivisions – difference tests and sensitivity tests – each serving a distinct purpose.

Discriminative tests: detecting differences between samples 🔗

Discriminative tests are the most straightforward type of analytical test. Their primary question is simple: can the panelist tell two or more samples apart? These tests are widely used in quality control, ingredient substitution decisions, and process change evaluations.

Difference tests 🔗

Difference tests determine whether a perceptible sensory difference exists between two or more samples at a given level of statistical probability. Several well-established methods fall under this category:

Triangle test: Panelists receive three coded samples – two identical and one different – and must identify the odd sample. This is one of the most commonly used discrimination methods because of its strong statistical power. The probability of guessing correctly by chance is only one-third, which means fewer panelists are needed to reach reliable conclusions.

Duo-trio test: In this format, panelists receive a reference sample followed by two coded samples, one of which matches the reference. Their task is to identify which coded sample is identical to the reference. This method is particularly useful when the product is complex, as the reference anchor reduces panelist confusion and sensory fatigue.

Paired comparison test: Two samples are presented and the panelist is asked to identify which one has more of a specific attribute – for example, “which sample is sweeter?” This is a directional test, meaning it reveals not just whether samples differ but in which direction the difference lies.

Ranking test: Multiple samples are arranged in order of intensity for a particular attribute. This is useful when you need to compare several products or formulations quickly and understand their relative positioning.

Sensitivity tests 🔗

While difference tests focus on whether samples can be told apart, sensitivity tests focus on measuring how sensitive individual panelists are to specific sensory stimuli. These tests are essential for screening and training assessors and for understanding human perception thresholds.

Threshold testing is the core method within sensitivity tests. It determines the minimum concentration of a substance that a person can detect or recognise. There are several types of thresholds:

Absolute (detection) threshold: The lowest concentration at which a panelist can tell that something is present in the sample, even if they cannot identify what it is. For example, a panelist may sense that a water sample tastes “different” but cannot name the taste as sweet or salty.

Recognition threshold: The lowest concentration at which a panelist can correctly identify the nature of the stimulus. At this level, they can say “this is sweet” or “this is sour.”

Difference threshold: Also called the just-noticeable difference (JND), this is the smallest change in stimulus intensity that a panelist can reliably detect. It answers the question: how much must we change the salt content before someone notices?

Terminal threshold: The concentration above which no further increase in perceived intensity is noticed. Beyond this point, adding more of a substance has no additional sensory impact.

Threshold tests typically use a forced-choice ascending concentration series, where panelists evaluate a sequence of samples with increasing concentrations. Standards such as ASTM E679 provide standardised protocols for conducting these assessments.

Dilution tests are another form of sensitivity testing. Here, the product is progressively diluted until the panelist can no longer detect the target attribute. This method helps establish at what point a flavour, odour, or off-note becomes undetectable – valuable information for quality assurance and shelf-life studies.

Descriptive tests: profiling the sensory character of food 🔗

If discriminative tests tell us whether samples differ, descriptive tests tell us how and how much they differ. These are the most detailed and sophisticated of all analytical sensory methods. They require highly trained panelists and produce comprehensive sensory profiles that capture every nuance of a product.

Descriptive tests serve two functions simultaneously – they provide qualitative information (what attributes are present?) and quantitative information (how intense is each attribute?). The output is typically a detailed sensory fingerprint of the product.

Quantitative Descriptive Analysis (QDA) 🔗

Quantitative Descriptive Analysis is one of the most widely used descriptive methods in the food industry. Developed in 1974 by Herbert Stone and Joel Sidel at the Tragon Corporation, QDA was a breakthrough because it introduced individual data collection rather than group consensus.

In a QDA session, each panelist independently evaluates samples and rates the intensity of pre-defined attributes on an unstructured line scale – usually 15 cm long, anchored with terms like “none” at one end and “extreme” at the other. Panelists mark their perceived intensity by placing a mark on the line. This approach generates nuanced, continuous data that can be analysed statistically.

A typical QDA panel consists of 10-12 panelists who have been screened for sensory acuity and trained extensively – sometimes for 40 to 120 hours – before formal evaluations begin. During training, the panel collaboratively develops a standardised vocabulary (a sensory lexicon) for the product category and calibrates their intensity ratings using reference samples.

The results of QDA are commonly visualised using spider plots (also called radar charts), which display the intensity of each attribute as a point along a spoke of a wheel. These plots make it easy to compare the sensory profiles of multiple products at a glance.

Flavour Profile Analysis (FPA) 🔗

Flavour Profile Analysis was one of the earliest formal descriptive methods, introduced in 1949 by chemists at Arthur D. Little, Inc. Unlike QDA, FPA relies on a small panel of 4-6 experts who work collaboratively to build a consensus profile of the product.

FPA evaluates flavour across five dimensions: character attributes (specific flavour notes like fruity or smoky), intensity (strength of each note), order of appearance (sequence in which flavours emerge), aftertaste (lingering sensations after swallowing), and amplitude (overall impression of balance or fullness). This layered approach captures the temporal and compositional complexity of flavour in a way that single-point intensity ratings cannot.

Texture Profile Analysis (TPA) 🔗

Texture Profile Analysis focuses exclusively on the mechanical and structural properties of food. It evaluates attributes like hardness, cohesiveness, adhesiveness, viscosity, and elasticity, along with secondary parameters such as chewiness and gumminess. TPA is particularly important for products where texture drives consumer acceptance – baked goods, dairy products, confectionery, and meat analogues, for instance.

Spectrum Descriptive Analysis 🔗

The Spectrum method, developed by Gail Vance Civille, takes a more rigorous approach to standardisation than QDA. Panelists undergo a minimum of 100 hours of training per sensory modality and use a universal 15-point scale with fixed reference standards. A score of “10” for sweetness means the same thing whether it’s assigned to a soft drink, bread, or pizza sauce. This universality allows data comparison across products, categories, and time periods – something QDA’s relative scaling cannot easily achieve.

The role of trained panelists 🔗

The reliability of every analytical sensory test depends on the quality of the panelists. Unlike consumer panels that involve large groups of untrained individuals, analytical tests use small, carefully selected panels – typically 6 to 15 people – who undergo structured training programmes.

Training involves several stages. First, panelists are screened for basic sensory acuity – the ability to detect and differentiate fundamental tastes, aromas, and textures. Next, they learn the standardised terminology for the product category. Then, they practise using intensity scales with reference samples until their ratings are consistent and reproducible. Standards like ISO 8586 provide guidelines for panel recruitment, training, and monitoring.

Trained panelists essentially function as human analytical instruments. They are not asked whether they like a product. They are asked to measure its properties accurately and repeatedly – and their performance is validated statistically, much like the calibration of a laboratory instrument.

Controlled testing conditions 🔗

Analytical sensory tests are conducted under strictly controlled conditions to eliminate bias. Testing typically takes place in specially designed sensory laboratories equipped with individual booths to prevent communication between panelists. Lighting is controlled – often using red or green lights – to mask visual differences when only flavour or texture is being evaluated.

Sample preparation is standardised: all samples are served at the same temperature, in identical containers, and labelled with random three-digit codes. Presentation order is randomised or balanced to counteract position effects and adaptation. Panelists rinse their palates between samples – usually with water or unsalted crackers – to minimise carryover effects.

These conditions are not arbitrary. They are designed to ensure that the data reflects actual sensory differences in the products, not artefacts of how the test was conducted.

Applications across the food industry 🔗

Product development and reformulation 🔗

When a food company reformulates a product – say, reducing sugar or replacing an ingredient – analytical tests help determine whether the change is perceptible to trained assessors. A triangle test can quickly confirm whether the new formulation differs from the original. If it does, descriptive analysis can map exactly where the differences lie, enabling targeted adjustments.

Quality control and batch consistency 🔗

Manufacturers use analytical tests to monitor product consistency across batches, production lines, and factory locations. Descriptive profiling over time can detect gradual shifts in sensory quality – such as a slow increase in off-flavours – that might not trigger complaints immediately but could erode brand loyalty over time.

Shelf-life determination 🔗

Sensory evaluation is integral to shelf-life studies. Analytical tests track how sensory attributes change during storage. Trained panels can detect subtle deterioration – like early staleness in baked goods or the emergence of oxidation notes in oils – well before these changes would be obvious to consumers.

Ingredient and supplier evaluation 🔗

When sourcing raw materials from different suppliers, analytical tests can confirm whether the sensory impact of the new ingredient matches the existing one. This is critical for maintaining product identity while optimising cost or supply chain reliability.

Limitations to keep in mind 🔗

Analytical sensory tests are powerful, but they have constraints. Training a panel is time-intensive and expensive. Panelists can experience sensory fatigue during extended sessions, which reduces their accuracy. Results from trained panels do not directly predict consumer acceptance – a product may score well on a descriptive profile but still fail with consumers for reasons related to expectation, branding, or cultural preference.

This is why analytical tests are most effective when used alongside affective (consumer) tests and instrumental analysis. Together, these three pillars give a comprehensive view of food quality – what the product is, how it measures, and how it’s perceived.

What do you think? If you were developing a new food product, which analytical test would you start with – a quick discrimination test to check if your formulation change is detectable, or a full descriptive analysis to build a detailed sensory profile? And how would you decide when to rely on trained panels versus consumer feedback?

References

1. https://www.sciencedirect.com/topics/agricultural-and-biological-sciences/sensory-evaluation
2. https://pmc.ncbi.nlm.nih.gov/articles/PMC7922510/
3. https://www.foodresearchlab.com/blog/rte-rtc/sensory-evaluation-of-food/
4. https://store.astm.org/products-services/standards-and-publications/standards/sensory-evaluation-standards.html
5. https://pmc.ncbi.nlm.nih.gov/articles/PMC8834440/
6. https://foodsafety.institute/food-fundamentals-chemistry/descriptive-tests-sensory-evaluation-profiling/
7. https://pmc.ncbi.nlm.nih.gov/articles/PMC8001375/
8. https://www.eurofins.in/food-testing/services/sensory-evaluation/
9. https://www.tentamus.com/lab-analyses/sensory-testing/
10. https://pmc.ncbi.nlm.nih.gov/articles/PMC10527616/