1
Introduction

Sarah E. Kemp1, Tracey Hollowood2, Joanne Hort3 and Lauren Rogers4

1 Consultant and formerly Head of Global Sensory and Consumer Guidance, Cadbury Schweppes, UK

2 Formerly Managing Director of Sensory Dimensions Ltd, UK

3 Food Experience and Sensory Testing Laboratory (Feast), Massey University, Palmerston North, New Zealand

4 Freelance Sensory Scientist, UK

1.1 Introduction

Sensory discrimination tests are methodologies used in fields such as foods and beverages, home and personal care, cosmetics, and consumer goods to assess whether two samples are perceptibly different. The two samples being tested generally come from two production batches. These different batches might use, for example, different ingredients, different manufacturing processes, or different packaging. Discrimination tests are also used to evaluate and quantify individuals' abilities to perceive and differentiate various sensory attributes or a holistic difference between products. Participants are typically presented with pairs of products and are tasked with identifying differences or similarities in sensory characteristics, such as appearance, aroma, taste, flavor or texture. These tests play a crucial role in quality control, product development, and market research, for example, providing valuable insights into consumer preferences and aiding in the improvement of product formulations based on perceived sensory attributes.

1.1.1 Discrimination Testing as a Tool

Imagine that you are the new product development manager for a major snack manufacturer and have been tasked with an innovative global wellness project to further reduce the levels of salt and fat in the company's flagship product. Several designs of experiment studies are currently being discussed by the team and the next step is to discuss the project timings for the assessment of the sensory aspects of the products arising from these experiments. You have a meeting with the sensory manager to discuss the plans and the sensory methods available to help meet the project's action standards. There are several classes of sensory methods available to you, including descriptive, temporal and consumer methods, however, you agree with the global sensory manager that discrimination tests1 are the best choice for this stage of the project as the differences between products should be slight: there is no point in conducting a discrimination test on products that are obviously different. The sensory manager has explained that if you need to know how products are different, you will need a different method (for example, quantitative descriptive analysis or free choice profiling). The sensory discrimination tests will allow you to determine if consumers will notice the difference between the chemically and physically different products from the experimental design studies and the current gold standard products.

To use a sensory discrimination test, three things are usually required: people to assess the samples, the choice of the test and some samples to assess. Obviously, a sensory scientist is also required to choose and administer the test, as well as to analyze and report the results. The person who takes part in the test is often someone who has been screened for their sensory acuity (ISO 2023) and trained in the test method, although they might also be naïve consumers. By training in the method first, the participant is focused on the items in the test and not the execution of the test itself. This person assesses the items presented to them using the training they received, answering the question posed by the test; for example, selecting the odd sample in a discrimination test known as a triangle test (Sinkinson 2017) or rating the difference between samples on a scale (Whelan 2017). The results from all participants in the test are then analyzed to make a final decision based on the objectives and action standard(s).

There are many different sensory discrimination tests available to the sensory scientist to help meet the types of objectives they might encounter. The choice of test will depend on several factors, with the primary considerations being the type of product being tested and the objectives of the test. There is a tendency for businesses to habitually use one discrimination test for all objectives, but not all objectives are equal: for an interesting discussion about the choice of sensory test related to the objective(s) of the research, see Lawless and Heymann (2010, pp. 8-10).

When the focus of the discrimination test is to answer a product-related question, the intended use of the results and the test's position within the project stages are important factors in the choice of test (Meilgaard et al. 2016).

1.2 Developments of Discrimination Testing

Sensory discrimination testing began early in the 20th century and was originally referred to as difference testing (Meiselman et al. 2022), however, the initial "discrimination tests" were conducted by Weber and Fechner in the early 19th century to examine the relationship between physical stimuli and sensory experience (David 1963). This field of study, now referred to as psychophysics, was characterized by the researchers' belief that sensations could not be measured directly. Consequently, they employed a variety of indirect methods and experiments to gauge perception. Weber's experiments in 1834 explored the "just noticeable difference" between different weights using blindfolded subjects (Holden et al. 2011). He found that the detectable difference was proportional to the initial weight. This method has since become known as "the method of constant stimuli" and bears resemblance to paired comparison methods such as the 2-alternative forced choice test commonly used in sensory science today.

Fechner expanded on Weber's findings, formulating Weber's law and introducing two additional methods: "the method of limits" and "the method of adjustment or average error." Although the latter method is challenging to implement in food research, as it requires subjects to adjust levels of ingredients such as salt, the method of limits - which involves changing the stimulus incrementally and asking subjects if they perceive any sensation - is akin to same-different tests and the ascending forced-choice method of limits employed to determine thresholds today. Thus, the technique of comparing pairs of stimuli began with Weber and was further refined by Fechner (Fechner 1860; Lawless and Heymann 2010).

Louis Thurstone (1927) also made extensive use of paired comparison tests in his psychophysical work to describe the discrimination process, which has been of great interest in the field of sensory discrimination testing. Researchers employ Thurstonian modelling and signal detection theory (SDT) to study differences in people's sensitivity to specific stimuli.

Perhaps the earliest example of a sensory discrimination test in the literature with a focus on food is Fisher's famous article on tea tasting, which would become the cornerstone reference for hypothesis testing. The test originated in the 1920s in Cambridge, when a group of friends, including Ronald Fisher, were debating the proper way to pour tea and whether the milk should be added to the cup before or after the tea. One member of the group claimed that she could discern whether the milk had been added first or last. To test this claim, Fisher designed an experiment to determine whether she could accurately identify the order in which the milk was added (Fisher 1935). The original documentation provides an interesting account of the experiment's design, statistical analysis, and results. Although the method employed in the tea testing was not given a name, Gridgeman (1959) referred to it as a "double-tetrad sorting design." Today, we would likely classify it as an octad or under the M + N tests (Lockhart 1951).

Of course, food discrimination testing has been in use for much longer than the past nine decades. Sensory tests, though informally, have been employed to assess food edibility and check drinking water quality for centuries. Additionally, nonfood testing has also been conducted throughout history, such as evaluating the suitability of housing, e.g. caves and weapons, e.g. assessing the sharpness of flint tools (Meilgaard et al. 2016).

Ranking is possibly the earliest sensory discrimination method, with references to its use for assessing preferences for eggs and sweetcorn in very early studies (as reported by Morse 1942). This type of grading, although not a formal sensory test, was one of the first methods used to assess the quality of a product. Experts would assess the quality based on a grading system, and these grading experts were possibly the first example of panel training and maintenance in the literature. However, there was no mention of screening for sensory acuity, and experts checked their assessments against those of their colleagues or standard samples. Some of these grading methods are still used for tea, wine, and coffee (Kilcast 2010).

The first named discrimination testing method was published in 1936 by Sylvia Cover, who wanted to determine if the cooking temperature of roasts made the meat more or less tender (Cover 1936). The new method was referred to as "the...