Lauren Henry of Sensory Dimensions discusses the merits of psychological testing to explore the emotional impact of food brands and products.
Traditional market research methods often rely on explicit feedback from respondents. These questionnaires ask direct questions and tap into System 2 thinking[1], which is relatively slow, logical and deliberate. Product developers have found, however, that despite products performing well in explicit market research, they often do not succeed in the market and this brings into question the validity of explicit methods in predicting consumer behaviour.
In the last few years there has been an interest in using well-established psychological tests within a market research context. These implicit tests are designed to tap into faster, instinctive, subconscious processing, known as System 1 thinking[1] and do this without overtly asking the question being investigated. This implicit response may be more representative of real-life decision-making processes at point of purchase because many of our decision-making processes occur unconsciously[2]. Unlike explicit tests, implicit responses remain free of bias and are neither influenced by social norms nor our desire to say what we think people want to hear[3]. For these reasons, implicit may be a better predictor of real-world behaviours and so of the performance of a product in the market.
Implicit methods are powerful for researching emotion because emotions are difficult to verbalise and simply being asked about our feelings changes the way we feel. Being able to identify the emotions evoked by a product is valuable as research has shown that generating emotion increases consumer engagement and encourages repeat purchase[4]. Emotions are instant and impact behaviour in an impulsive and unconscious way and so implicit methods offer a way to explore the emotional impact of brands and products.
Implicit testing methods
A wide variety of implicit methods are being explored in food and beverage research, including fMRI (functional magnetic resonance imaging), EEG (electroencephalogram), heart rate monitoring and eye tracking. This article focuses on reaction time-based tests. These tests are usually performed by a respondent with a keyboard.
The Implicit Association Test (IAT)[5] assesses associations between two targets (e.g. two brands) and a bipolar attribute (e.g. poor quality and good quality) using categorisation tasks requiring quick responses. The fundamental premise of the test is that faster responses are expected when strongly associated concepts are paired. One of the main barriers to using IAT for food and drink assessment is that it cannot be used for more than two stimuli or two attributes. This, and the multiple exposures to stimuli required, means it becomes laborious and repetitive leading to sensory fatigue and lack of attention by the respondent. Additionally, the IAT varies the association of the targets with the left and right key presses. Critics claim this creates confusion and noise in the data from excessive incorrect key presses.
Like the IAT, semantic priming relies on response speed to highlight congruent associations. It was originally used within market research to measure the strength of associations between psychological attributes and a range of media brands[6]. It revealed large statistical differences between the brands, which had not been demonstrated by explicit methodologies. The method uses the idea that priming (exposure to a stimulus) influences the reaction to a following stimulus at an automatic level: if the following stimulus is congruent with the prime, reaction to that stimulus is expected to be faster. The theory behind this is that concepts are stored in the brain across clusters of neurons. Closely related concepts will be stored across overlapping clusters of neurons, meaning that some neurons triggered by the second stimulus will have already been triggered by the prime and will therefore respond more quickly to the second stimulus. Semantic priming methods measure the strength of association between the prime and the following stimulus. A benefit of semantic priming over the IAT is that it allows the use of more than two stimuli. In addition, the quantitative output can be used to compare all stimuli in the test and allow them to be ranked or grouped to give a more multidimensional output than the IAT. Neither the IAT nor semantic priming are designed to measure attitudes at the level of an individual but rather averaged across the consumer group.
Sensory Dimensions has worked alongside Truthsayers on multiple implicit projects using semantic priming. This kind of test can be carried out on a computer or on a mobile device or tablet, and this combined with the fact that this kind of implicit test can incorporate multiple stimuli and many attributes means it is very versatile. A range of different kinds of attributes can be used as part of this test, for example, emotional, sensory, functional, descriptive, usage occasions, or a mixture of all of these. Stimuli being tested could be a brand logo, interacting with packaging, smelling a fragrance or tasting a food or drink product. You can compare how your product or brand is perceived compared to that of a competitor, find out the emotions your product generates to ensure that your marketing effectively communicates the product experience, or make sure your product, packaging and brand are all in line with one another.
Format for semantic priming tests
Implicit tests can be more fun for respondents than explicit tests because they are more game-like, requiring quick responses and giving feedback when incorrect responses are given. Respondents are instructed to hit a certain key on the keyboard when ‘Yes’ appears on the screen and a different key on the keyboard when ‘No’ appears. They are given an opportunity to complete this task as a warm-up before being given a stimulus to ensure they are familiar with the task. The pairing of ‘Yes’ or ‘No’ with the keys is randomised across the consumer group as a whole but kept consistent within an assessor to prevent confusion-based errors. Reminders of which keys to press are always on screen during the task.
After the warm-up, respondents are instructed to interact with the stimulus. What form this interaction takes will depend on what is being tested; it may be looking at a brand logo or tasting a drink for example. Then, for each stimulus the attributes being tested are flashed up on the screen (one at a time in a randomised order) followed by the word ‘Yes’ or the word ‘No’ and respondents should hit the corresponding ‘Yes’ or ‘No’ key on the keyboard (Figure 1). Their response is dependent on whether they saw the word ‘Yes’ or the word ‘No’, not the attribute that preceded it.
The attributes are repeated in blocks and between each block respondents are re-exposed to the stimulus to ensure it stays fresh in their minds. Each attribute is paired with ‘Yes’ and with ‘No’ in a randomised order. The reaction time between the appearance of ‘Yes’ or ‘No’ and the key press is recorded. Respondents are required to make a quick response within a time limit of less than a second in order to avoid allowing time for rational thought. The idea behind the test is that a congruent pairing of an attribute and ‘Yes’ or ‘No’ will elicit a faster response than an incongruent pairing[7].
![]() |
| Figure 1 A typical nsemantic priming test |
It is important to recognise that although there are a lot of reaction time-based tests around, many are not implicit. In many cases they are just speeded up versions of explicit tests. Fast explicit tests ask respondents to make conscious evaluations very quickly e.g. would you buy this? The answer here depends on the respondent – they have a choice. The fact that they have a choice means that they could lie, give a biased response, or choose any answer at random without thinking about their opinion. The fact they have a choice means their response relies on System 2 processing: it is an explicit response. In contrast, in implicit tests the respondent does not have a choice. If the screen displays ‘No’ they have to hit the key that corresponds with ‘No’.
Once all of the data have been collected, the differences in reaction time when each attribute is paired with ‘Yes’ and ‘No’ and the variance of the data can be analysed to give an idea of how strongly each attribute is associated with each stimulus and this information can be used to answer a variety of questions and objectives.
Case Studies
Emotional profiling of sweet snacks
Sensory Dimensions and Truthsayers used semantic priming to investigate the emotional profiles of different snacks (cinema popcorn, toffee popcorn, premium popcorn with chocolate drizzle, Maltesers and chocolate buttons).
Couples, who usually watched television together, were invited to take part in a study in which they ate the snacks while watching an episode of their selected TV sitcom in a typical living room (Figure 2). At each visit they ate a different snack and watched a different episode of their chosen sitcom.
![]() |
Figure 2 Implicit tests can be carried out in
a specialised sensory laboratory or in a more realistic context, such as a living room |
The semantic priming implicit task was used to investigate how strongly 28 emotional attributes were associated with each snack. The attributes were selected after a review of other literature involving emotional profiling of food products, including the EsSense profile[8] and best-worst scaling[9]. All attributes selected were positive emotions, which were not synonyms of one another. They were not specific to particular snacks but could be associated with consumption of any indulgent product.
The study found that each of the snacks had a distinct emotional profile. Maltesers made people feel sociable and happy; toffee popcorn made people feel free spirited, satisfied and comforted; cinema popcorn made people feel adventurous and proud; chocolate buttons made people feel happy and the premium chocolate-drizzled popcorn made people feel elated, joyful, sociable, proud and excited. The strength of association is described by a number between 0 and 100 (Figure 3). A score close to 100 indicates a strong positive association between the emotion and the snack, a score close to 0 indicates a strong negative association between the emotion and the snack, while 50 is the neutral point.
![]() |
Figure 3 Table with results of snack study – Strength of association of each emotional attribute with each snack |
Implicit methods offer an exciting and versatile tool that is relatively new to the sensory and consumer research field.
Emotional profiling of chocolates and their packaging
We have also used this method to explore the congruence of the descriptive, emotional and usage profile attributes associated with chocolates, their wrappers and their box. We found that the box was extremely well received and associated strongly with attributes, such as modern, desirable and generous. However, the box attributes were not supported by those of the wrappers and whilst the product itself was very highly liked, it too was not aligned with the premium expectations generated by the box and the brand. Overall the product was highly liked with a high propensity to purchase. It was considered fun and great for sharing, features which positioned it as a treat and everyday gift for younger people but made it less suitable for more sophisticated gifting occasions.
![]()
Emotional profiling of laundry liquids
Another study aimed to define the drivers of choice for two brands of laundry liquid so that these functional and emotional benefits could be incorporated into targeted on-pack messaging and brand communication that would better retain loyal customers and attract new ones. Liking for the two fragrances measured explicitly was very similar. However, implicit testing showed that the brands generated very different emotional reactions amongst their user groups and that the fragrances cued different functional benefits. One brand was associated with emotions, such as caring and friendly, and a long lasting, all day fresh, great fragrance. The other brand was more associated with functional attributes, such as strong on stains and dependable.
Conclusions![]()
Implicit methods offer an exciting and versatile tool that is relatively new to the sensory and consumer research field. Their application is the subject of several research programmes but it is clear that these techniques have great potential to complement our traditional explicit research methodologies and further our understanding of consumer choice and purchase behaviour.
Lauren Henry, Project Manager, Sensory Dimensions
5, Cutbush Industrial Park, Danehill, Lower Earley, Earley, Reading RG6 4UT
Email Lauren@ sensorydimensions.com
Web sensorydimensions.com
References
1. Kahneman. D. (2011) Thinking Fast and Slow, Farrar, Straus and Giroux
2. Dijksterhuis, A., Smith, P. K., van Baaren, R. B., & Wigboldus, D. H. (2005). The unconscious consumer: Effects of environment on consumer behavior. Journal of Consumer Psychology, 15, 193–202.
3. Vianello, M., Robusto, E. & Anselmi, P. (2010) Implicit conscientiousness predicts academic performance. Personality and Individual Differences, 48, pp. 452–457.
4. Dobele, Angela & Lindgreen, Adam & Beverland, Michael & Vanhamme, Joëlle & van Wijk, Robert. (2007). Why pass on viral messages? Because they connect emotionally. Business Horizons. 50. 291-304.
5. Greenwald AG, McGhee DE, Schwartz JL. (1998), Measuring individual differences in implicit cognition: the implicit association test. J Pers Soc Psychol. 1998 Jun;74(6):1464-80.
6. Calvert, G., Fulcher, E., Fulcher, G., Foster, P. and Rose, H. 2014. “Using Implicit Methods to Develop an Objective Measure of Media Brand Engagement.” International Journal of Market Research56 (1): 15–32.
7. Schiller, B. et al. (2016), Clocking the social mind by identifying mental processes in the IAT with electrical neuroimaging. Proc. Natl. Acad. Sci. 113, 2786–2791.
8. King, S.C., Meiselman, H.L. and Carr, B.T. (2010), Measuring emotions associated with foods in consumer testing, Food Quality and Preference, Vol. 21 No. 8, pp. 1114-1116.
9. Thomson, D. M., Crocker, C., & Marketo, C. G. (2010). Linking sensory characteristics to emotions: An example using dark chocolate. Food quality and preference, 21(8), 1117-1125. doi: 10.1016/j.foodqual.2010.04.011.
