When doing sensory tests, the results largely depend on the study design. Important issues to take into account are blinding and coding, randomization, order effects, and frame of reference. As you have learned in the chapter multimodal interaction, people form expectations of the taste based on other cues than taste, for example, the color or description of the food. To get rid of this effect, samples are often standardized and blinded before they are judged. That is, all cues that can affect their perception and are not being investigated are kept constant for all samples, e.g. the temperature, color, and amount of sample (standardization). Also, no information is given about the differences between the food samples, and samples are often labeled with non-informative numbers and/or letters. The labeling will be done using coding that is not retraceable to the manipulation or manipulation order. So for example a 3-digit number in a random order is used, which only the researchers can understand.

Order effect

As you can imagine many factors affect the response of a panelist. For example, the order in which they receive the sample may affect the response, this is the so-called order effect. The order effect can be influenced by multiple factors:

• Adaptation. Sensory receptors may become less sensitive after continuous exposure to a stimulus, affecting the perception of subsequent samples;
• Contrast Effects. The perception of a sample can be influenced by the sample tested before it;
• Expectation. Panelists might develop expectations based on the order of presentation, which can bias their responses;
• Fatigue. Sensory fatigue can occur if panelists are exposed to too many samples in a short period, leading to less accurate evaluations;
• Memory. Panelists might remember previous samples and compare them, rather than evaluating each sample independently.

So, for example, the first sample can set a reference for the samples that are tasted later. Let’s imagine you have 4 different samples that are increasing in sweetness concentration. You can imagine that depending on the order in which you will receive these samples, you will use the scale that is presented differently. If you get the least sweet sample first, you may score this in the middle of the scale, as you do not know that the next samples will be sweeter. But let’s pretend that you will get the sweetest sample first. You may also put this in the middle of the scale (as you may not know that the others are all least sweet. So the first sample sets the expectations for further samples. To exclude order effects, you always need to randomize the samples that a panelist receives so that this leads to random error and will not systematically affect your results.

Frame of reference

An additional, strategy to deal with the effect of the first sample is by introducing a so-called warm-up sample. This is a sample can help set a standardized frame of reference for all panelists, and is given to each panelist before the study starts so that they all have the same starting point. When it is very important for your research another strategy of getting rid of this effect is by training the panelists on the use of the scale, but this is a lot of extra work if you want to do this in a good way, see also the Chapter about Panel training. Next to a frame of reference that is formed by the testing design, the panelists also have a frame of reference that is caused by their eating habits. People who eat more sweet foods will have another way of rating compared to people who eat less sweet diets. More about the taste of diets will be explained in topic 3 of this course.

Sample volume

Another issue that may occur during tasting sessions is sensory fatigue. That is, you are not able to sense small differences anymore after tasting a lot of samples, so make sure that the samples and your questionnaire do not require large volumes of consumption. Around 10-20 mL of a solution may already be enough to judge the taste intensity. Typically, sensory panelists can taste up to 40 samples, but this also depends on the number and complexity of the questions. To have valid data, try to find the balance in your design between the number of panelists, the number of questions, and the number of samples.

testing context

As mentioned before the context in which samples are tested affects rating. Testing in a laboratory setting will standardize the context. See below some pictures of sensory booths as we have them in the Helix building in Wageningen.