The experimental assessment of sensory dominance in a product development context

J. Design Research, Vol. 8, No. 2, 2010

119

The experimental assessment of sensory dominance in a product development context Hendrik N.J. Schifferstein*, Jacco J. Otten, Fien Thoolen and Paul Hekkert Department of Industrial Design, Delft University of Technology, Landbergstraat 15, 2628 CE Delft, The Netherlands E-mail: [email protected] E-mail: [email protected] E-mail: [email protected] E-mail: [email protected] *Corresponding author Abstract: Although all senses are open to obtain information during userproduct interactions, some sensory information can have a larger effect on the product experience than others. We investigated an experimental approach to quantify the relative importance of the sensory modalities in user-product interactions. For each modality, two stimuli were selected with comparable differences in pleasantness scores. All possible stimulus combinations were evaluated for two products: a portable air purifier and a table lamp. Both studies found a significant effect on pleasantness only for different product colours, suggesting that vision was the dominant modality for both products. Interestingly, next to vision, smell affected responses on the activity dimension of product experience, whereas touch and sound affected the potency dimension. Because we found hardly any relationship between the pleasantness of a complex product and the pleasantness of its component stimuli, we discuss a number of factors that may have interfered. Keywords: multisensory; dominance; sensory integration; importance; product design; product experience. Reference to this paper should be made as follows: Schifferstein, H.N.J., Otten, J.J., Thoolen, F. and Hekkert, P. (2010) ‘The experimental assessment of sensory dominance in a product development context’, J. Design Research, Vol. 8, No. 2, pp.119–144. Biographical notes: Hendrik N.J. Schifferstein is an Associate Professor at the Department of Industrial Design of Delft University of Technology. After having worked in the food realm for several years, he now studies the multisensory experiences evoked by consumer durables. Among others, he published in Perception and Psychophysics, Acta Psychologica, Marketing Letters, The Design Journal, and Journal of Experimental Psychology: Human Perception and Performance. He is co-editor of the books Food, People and Society (2001; Springer Verlag) and Product Experience (2008; Elsevier). Jacco J. Otten has an MSc in Industrial Design Engineering from Delft University of Technology. He has worked both as a Designer and Researcher on various user interaction topics. While user interaction is often mainly visual, he had the opportunity to perform multi-sensory design and research. Currently, Copyright © 2010 Inderscience Enterprises Ltd.

120

H.N.J. Schifferstein et al. he works as Lead User Interaction Designer at Mavim, an innovative Dutch software development company. Fien Thoolen has an MSc in Design for Interaction from Delft University of Technology, where she focused on sensory research and design. At the moment, she works at the Center for People and Buildings. Here, she investigates suitability of new office environments, user satisfaction in offices and she helps companies to prepare staff for changes in their office environment. Paul Hekkert is a Professor of Form Theory at the Faculty of Industrial Design Engineering, Delft University of Technology and Head of the Section Design Aesthetics. He has published on product experience and aesthetics in major international journals and is co-editor of Design and emotion: the experience of everyday things (2004) and Product Experience (2008). Together with a colleague/designer, he has developed an interaction-centered design approach called Vision in Product Design (ViP). In addition, he is the Founder and Chairman of the Design and Emotion society.

1

Introduction

From an ecological point of view, the task of the perceptual systems is to obtain useful information about the environment. When people interact with products, all their senses are used to obtain information (Gibson, 1966). When a person uses a durable consumer product, sensory feedback is used directly to operate the product (e.g. Akamatsu et al., 1995). For instance, a woman who uses a vacuum cleaner can judge on the basis of what she sees (the carpet becomes clean) and hears (tickling of pieces of dirt in the nozzle) whether she has to clean the current section again or she can continue with another section. Furthermore, sensory inputs affect the way in which people experience and evaluate products. For instance, the noisy sound of the motor suggests that the vacuum cleaner is powerful and its metal finishing may suggest that it is a robust apparatus. These experiential aspects play a role, for example, in the Kansei engineering approach to design (e.g. Nagamachi, 1995; Schütte et al., 2008). Because all sensory inputs can affect the way in users perceive, make sense of, experience, and evaluate a product, several authors have proposed a multisensory approach to product design, in which all perceptual impressions elicited when a potential user interacts with the product are explicitly created and evaluated (e.g. Adank and Warell, 2008; MacDonald, 2002; Schifferstein and Desmet, 2008). During the interactions with a product, users need to integrate all different types of sensory information in order to perceive the product as a unitary whole. Nonetheless, some sensory information can have a larger effect on how a product is experienced than other. If we can determine how important people find a specific type of sensory input, this may help product developers and designers to set priorities in allocating time and effort during product optimisation. We define ‘sensory importance’ as the relative contribution of each sensory modality to the product experience. The dominant sensory modality is the modality that has the largest effect on the specific product experience. In the present paper, we try to quantify the relative importance of the sensory modalities in user-product interaction in a naturalistic context. To test realistic products,

The experimental assessment of sensory dominance

121

we develop concepts for new or modified products and we build product prototypes. During the design process, we evaluate materials and design options, some of which will be incorporated into the prototypes. This allows us to evaluate the rules through which the evaluations of single materials are related to evaluations of complex products. Because product development teams often have to make decisions on parts of the design during the development process, the outcomes of our study provide insight in the extent to which consumer evaluations of product parts predict their evaluation of the product as a whole.

1.1 Sensory dominance: beliefs and perception A general and popular belief seems to be that vision is the dominant sensory modality. When people are asked which sensory modality they would miss most if they lost it, the majority is likely to indicate vision (Fiore and Kimle, 1997; Schifferstein, 2006). In addition, when people are asked to describe objects, they primarily use adjectives that refer to the visual (60%) or tactual (32%) modalities (Stadtlander and Murdoch, 2000). Nonetheless, several anthropological studies have suggested that vision is not necessarily the dominant modality in all societies. In several non-Western societies taste, olfaction, or audition play dominant roles in important rituals (Howes, 1991). For instance, for the Songhay of Niger words are not only knowledge or information; they are not just representations of something. Instead, the Songhay believe that sound itself can be the carrier of powerful forces (Stoller, 1984). Among the Weyéwa in Eastern Indonesia taste plays an important role in a discrete, marked event called a social visit. The quality, size, freshness, quantity and variety of the ingredients presented to guests during the traditional betel and areca nut chew reflect on the wealth and prestige of the giver, and also perhaps on the disposition of the host towards the guest (Kuipers, 1991). There are indications that smell, taste and touch used to be more important in Western societies several hundreds of years ago (Classen, 1993; Classen et al., 1994) and that these societies have become more vision-oriented over the course of time. For instance, in early monastery gardens flowers were grown together with garlic, onions, and other herbs and vegetables, because flowers were grown mainly for their medicinal and culinary purposes. The walls surrounding the medieval garden intensified the scents by confining them to an enclosed space. From the 16th century, visual design became increasingly important in gardening. Topiary, the trimming of trees and shrubs into fanciful shapes, became a popular garden practice, as well as the planting of flowers in geometrical designs. This increased emphasis on visual beauty was originally accompanied by an increased appreciation of the aesthetic value of scents: flower gardens were now placed in front of the house and kitchen gardens to the side. In the 18th century, however, the primary role of the garden became to divert the mind by diverting the eye, for example, through perspectives and scenes, light and shade, line and form. The increasing popularity of the philosophy of the Enlightenment, which endorsed sight as the most important sensory modality for science and technology, increased the importance of sight also in other spheres of life and led to a decrease in the appreciation of smell. For instance, in the 19th century rose breeding mainly focused on size and colour, and flowers tended to be planted in bright masses of colour for maximum visual impact (see Classen, 1993). Even in the world of today, vision is not the dominant modality in the interaction with all products. For about half of the products investigated by Schifferstein (2006)

122

H.N.J. Schifferstein et al.

participants reported that they found one of the non-visual sensory modalities most important during usage. For example, for a computer mouse the tactual characteristics were most important, for a vacuum cleaner the sound it made, for a cleaning product its smell, and for a soft drink its taste. The relative importance of a modality may depend on various aspects, such as the availability of sources of sensory stimulation, the degree of variation in sensory stimulation over various products, the usefulness of the sensory information during functional use, the proportion of time a modality is used actively, and the role of the stimulation in enjoying the product. Hence, the roles of the modalities depend not only on whether a certain type of sensory information is present, but also on whether the information is perceived, how it is processed, and how people react to it emotionally. Schifferstein (2006) suggested that the visual system may be regarded dominant, because it plays a significant role in many daily activities. The importance of vision may have increased over the years due to the products that were created, such as books, television and computers that require major input from the visual modality. Alternatively, one could hypothesize that people report vision to be dominant, because people tend to be more aware of what they see than of what they hear, smell, feel and taste (Köster, 2003).

1.2 Sensory dominance: experimental studies Controlled experimental studies on sensory dominance have typically investigated the integration of conflicting information provided by two sensory modalities. In the case of two modalities, the degree to which one modality biases the other can be estimated on the basis of empirical data (see Welch and Warren, 1980). For instance, the degree to which touch is biased by vision in a localisation task can be calculated by:

Bias

ª rT – rVT º « » u100% ¬ rT – rV ¼

(1)

where rT is the location indicated when presented with the tactual stimulus alone, rV is the location indicated with the visual image alone and rVT refers to the position seen when visual and tactual information are both presented simultaneously. The importance of a modality has been found to depend on both stimulus properties and task instructions. For example, people use their modalities differently when they try to identify a product, judge a product’s roughness or evaluate its aesthetic appeal. Each type of task requires another type of information, and some modalities may be more suitable to provide that information than others (Lederman et al., 1996). Analogously, some experience dimensions may be affected more by information from one modality than the other. Below, we present an overview of the hypotheses that have been proposed to account for inter-modality differences in importance. Here, we focus mainly on the original ideas, without reviewing the empirical evidence that supports the various hypotheses. The most commonly offered explanations for intersensory interactions are based on three different mechanisms (Welch and Warren, 1980). First of all, the ‘modality precision’ hypothesis suggests that when two sensory modalities provide discrepant information, the resolution of the discrepancy will favour the more precise of the two in registering that event. This hypothesis has been investigated, for example, by relating the degree of bias to the standard deviations of unimodal control measurements (e.g. Welch et al., 1979).

The experimental assessment of sensory dominance

123

Typically, in performance tasks humans tend to integrate sensory information from two modalities in a statistically optimal fashion: when the variance of one information source increases, the impact of that source decreases (Ernst and Banks, 2002; van Beers et al., 1999). On the other hand, according to the ‘directed attention’ hypothesis (Canon, 1970; Kelso et al., 1975; Posner et al., 1976) the modality that receives the observers’ directed attention is the dominant modality. This hypothesis suggests that visual capture results from the fact that visual stimulation is typically more closely attended to than is, for example, proprioceptive or kinaesthetic stimulation. Studies in which the direction of attention was manipulated by changes in instructions alone have generally been unsuccessful in demonstrating the effects of attentional shifts on sensory dominance. Hence, an adequate test of this hypothesis seems to call for tasks that implicitly require attention to a specific sensory source for their completion (e.g. Warren and Schmitt, 1978). Third, the ‘modality appropriateness’ hypothesis (Freides, 1974; Welch and Warren, 1980) suggests that people weigh modality inputs according to their relative unimodal performance capabilities for a specific task. Each sensory modality is capable of performing a variety of functions, but it has one or more functions that it performs very well and even better than the other modalities. Hence, a modality will dominate the perception of a particular product aspect if it is the most suitable to perceive this aspect. For instance, if a task requires a spatial judgement, the visual modality is likely to have a biasing effect over the other modalities, because vision is more appropriately designed for spatial judgements. The ‘modality appropriateness’ hypothesis is somewhat broader than the ‘modality precision’ hypothesis described above: the fact that vision is more precise in performing localisation tasks can also be regarded as a manifestation of the superiority of vision as a localising modality, hence supporting the ‘modality appropriateness’ hypothesis.

1.3 The present study One of the simplest ways to investigate the importance of various modalities in product experiences is through asking persons directly to rate their importance (Schifferstein, 2006). However, these introspective self-reports can only reflect what people are aware of; they do not necessarily reflect the impact that the different senses have on the overall product experience (e.g. Heller et al., 1999; Myers and Alpert, 1977). Especially for the ‘lower’ senses olfaction and gustation, which are almost never in the centre of attention, their importance may be underestimated (Köster, 2003). Appropriately designed experimental studies allow the researcher to calculate the modality’s degree of bias, without asking participants to estimate modality importance (e.g. Welch and Warren, 1980). However, studies performed in the past have typically singled out only two sensory modalities under highly artificial laboratory conditions. Therefore, it remains uncertain how the results of these experimental studies translate to naturalistic user-product interaction situations in which multiple modalities are involved. Therefore, in the present study, we propose and evaluate a new stepwise approach that combines the advantages of the experimental approach with the use of realistic product prototypes. By using the experimental approach, we will not need to ask participants to estimate the importance they attach to various sensory aspects. Instead, we calculate their relative importances from the impact of changes in the corresponding

124

H.N.J. Schifferstein et al.

sensory product characteristics on the overall product evaluations. To create a realistic setting in which the participants are convinced that all stimuli come from the same source (Warren et al., 1981), we create 16 different product prototypes in which stimuli impinging on four different sensory modalities are integrated. By using product prototypes in the main experiments, we ensure that the results are applicable to real-life, multisensory product evaluation processes and are relevant for product design. Hence, we increase the ecological validity compared to previous experimental studies of sensory dominance. We use the term ‘product’ to refer to a source of multimodal stimulation (e.g. a vacuum cleaner), whereas we use ‘stimulus’ for a product aspect that stimulates a single modality (e.g. the sound of a vacuum cleaner). Our research approach is based on an additive model of sensory integration. According to the additive model, the participant’s evaluation of a combination of stimuli (i.e. a product) equals the sum of the (weighted) subjective values of the given stimuli: rVTAO

wV sV  wT sT  wA sA  wO sO

(2)

where rVTAO stands for the response to the multisensory product, and s stands for the subjective value of the visual (V), tactual (T), auditory (A) and olfactory (O) stimuli, respectively. We investigate whether this approach can be used to assess the relative importance weights (w) of the sensory modalities for product experiences. The additive model assumes that the subjective value of each stimulus is independent of the other stimuli it is combined with. The additive model has proven to be successful in many areas of psychology, including sensory perception and attitude formation (see Anderson, 1981; Fishbein and Ajzen, 1975) and the prediction of consumer preferences (e.g. Green and Srinivasan, 1978; Klitzner, 1975; Moskowitz and Krieger, 1993). We chose to start out from the additive model, because it is a very simple model. If the additive model proves to be invalid, the present study may indicate whether a different type of algebraic model may be more suitable. In the present study, participants evaluated 16 product prototypes on a number of rating scales. The contribution of each sensory modality to the final product design was assessed through analysis of variance (ANOVA). If the effects of the different types of sensory stimulation are independent and combine additively, the stimulus manipulations will lead to a main effect for each modality, and the size of this effect (either calculated as the shift in mean rating or as effect size Ș2) can be used as an estimate of the relative importance of that modality.

2

Study 1A

Study 1A tries to assess the impact of the various modalities on the selected experience dimension (pleasantness) of a portable air purifier. In a pre-study, participants rated the pleasantness of unisensory visual, tactual, auditory or olfactory stimuli. Subsequently, we selected four stimulus pairs (one for each sensory modality) that differed approximately to the same extent in pleasantness when evaluated in isolation. In the main study, these stimuli were combined to yield 16 different products, which were evaluated by naïve consumers using a between-subjects design.

The experimental assessment of sensory dominance

125

2.1 Participants The main study was completed by 161 men and 161 women, who were recruited in a museum (77.3%), in a library (15.8%) or at an open air market (6.8%). The men ranged in age from 14 to 70 years (mean 41.0, SD 12.0) and the women from 15 to 69 years (mean 40.7, SD 12.9). Each product was evaluated by 19–22 consumers.

2.2 Experience dimension The experience dimension and the measurement scale that are used to evaluate the importance of the modalities should not be biased towards any of the modalities. Concrete, descriptive attributes cannot be used for the present study, because they are likely to refer to a single sensory modality. Also, attributes that are typically used in a literal sense for one modality and in a metaphorical sense for other modalities (e.g. roughness and heaviness) are unsuitable, because their meaning is likely to be dominated by the modality for which the descriptor has a literal meaning (i.e. touch). Therefore, items that can be used to assess the degree of sensory dominance are likely to refer to abstract concepts. In the present study, we will assess all products on the evaluation dimension pleasantness represented by the items agreeable–disagreeable, pleasant– unpleasant and good–bad. These three items have been shown to be related to all four sensory modalities (vision, audition, touch and olfaction) to the same degree (Suzuki et al., 2006; Fenko et al., 2010).

2.3 Product The product chosen should also not favour one modality over the other ones. Therefore, we decided to choose a test product with which respondents were unfamiliar. Presenting a familiar product could instantaneously activate participants’ expectations of user context, which is likely to affect the evaluation of the importance of the senses in the user-product interaction. With an unfamiliar product, the participant would be more likely to approach the product open-minded, without preconceived ideas. To give the experiment a realistic character, the object should be credible as an addition to the person’s product environment. The product has to stimulate four sensory modalities: vision, audition, touch and smell and it should be possible to manipulate the stimuli for one modality independently of those for the other modalities. During the interaction, the user should use the four senses, without being explicitly instructed to do so. On the basis of these criteria, we decided to develop a portable air purifier, which is a small object that you can take with you as an accessory. It is said to provide fresh air by filtering the polluted air that surrounds you. The final design for the portable air purifier consists of a handheld device measuring about 8 × 4.5 × 2 cm (Figure 1). Its shape fits comfortably into a person’s hand. By pressing a large button on the curved side of the object, a small fan inside the purifier creates an airflow that disperses the fragrance. When the button is released, the fan stops. The object consists of four main parts: the base plate on which all elements are fitted, the two covers and the button (Figure 2).

126

H.N.J. Schifferstein et al.

Figure 1

The air purifier used in Study 1 (see online version for colours)

Figure 2

Schematic representation of the air purifier (see online version for colours)

The experimental assessment of sensory dominance

127

2.4 Stimulus selection Stimuli for Study 1A were selected in a separate pre-study, by having 30–31 undergraduate students evaluate how the unisensory stimuli looked, sounded, felt or smelled. The study was performed in a meeting room at the university, and was introduced as part of the development process for a new air purifier called the ‘Fresh 2 U’. The purpose of this device, the usage context, and the way it was operated were explained. The study consisted of four parts, one for each sensory modality. For each modality 7–11 stimulus variants were designed to vary in perceived pleasantness. We tried to present all stimuli in a way suggesting they were part of the final end product. The visual stimuli were right-half shapes of the final product attached to a white piece of foam board. They differed in colour and in the pattern of air holes. Different sounds were obtained by manipulating the voltage driving a small fan. Changes in voltage led to changes in pitch (change of rpm), as well as changes in volume. All auditory stimuli were presented in a prototype that was held at shoulder height. The tactual stimuli consisted of complete covers of the final product that were filled with parts of lead to obtain stimuli with different weights and weight distributions. To evaluate the tactual stimuli, participants inserted their hands in a ‘feel’ box that was completely covered with soundabsorbing material and that blocked visual and olfactory perception. Fragrances were created by perfumers from International Flavors and Fragrances (IFF) and were absorbed into polyethylene pellets (7 mm diameter and 6 mm height). For the olfactory stimuli, one fragrance pellet was placed in a 100 ml transparent plastic cup. Participants smelled the fragrance through three holes in the cap that could be closed by a sticker. They were instructed to allow for at least 20 sec rest between consecutive sniffs and 60 sec rest before sniffing a different sample. In addition, they were instructed to clean their noses by sniffing their wrist or the inner part of their elbow and to drink a sip of water between samples.1 Each stimulus was rated on six semantic contrast pairs that included two of the three pleasantness items (agreeable–disagreeable and good–bad) on a 7-point scale. We used only two pleasantness items in the pre-study in order to conceal the focus of the pre-study in the small-size questionnaire. Responses on the two pleasantness items (Pearson r = 0.78) were averaged. Subsequently, for each modality, we selected two stimuli that differed significantly and to approximately the same degree on the pleasantness scale (Table 1). The selected visual stimuli differed in colour only: ice blue (Munsell 5B 7/2) represented the pleasant (P) colour and grey (Munsell N 5.75) the unpleasant (U) colour. For the P sound stimulus, the voltage gently rose from 2 to 5 V over 2 sec when the apparatus’s button was pushed, and it decreased to 2 V likewise when the button was released. For the U stimulus, the voltage described a saw tooth pattern with a cycle of 2 sec, resulting in a continuously varying fan rpm. This fluctuating pattern created the impression that the product had trouble functioning properly. The voltage varied between 3 and 5 V. The tactual stimuli differed in weight: 105 g (P) or 50 g (U) placed in the centre of the object. The U fragrance is called fresh dewy fields 1; it contains cedar and musk base notes, watery, floral and herbal middle notes, and green, mint, ozoney, citrus and spicy top notes. The P fragrance is called water 1; it is composed of a musk base note, muguet, ozone, and apple middle notes, and mandarin, juniper, marjoram and melon top notes.

128

H.N.J. Schifferstein et al.

Table 1

Differences in pleasantness scores for the stimuli used in Study 1

Modality

Unpleasant

Pleasant

Pre-study

Study 1A

Study 1B

Stimulus

Mean Stimulus (SE)

Mean (SE)

Grey

4.21 (0.24)

Ice blue

5.26 (0.16)

1.05

0.000

í0.18

0.177

0.39

0.002

Audition Cycle 3–5V

3.31 (0.21)

Gradual 5V

4.23 (0.20)

0.92

0.002

0.03

0.825

0.10

0.420

Touch

50 g

4.48 (0.25)

105 g

5.23 (0.17)

0.75

0.001

í0.20

0.140

0.19

0.116

Smell

Fresh dewy 3.84 fields 1 (0.25)

Water 1

4.79 (0.21)

0.95

0.002

0.16

0.264

í0.13

0.298

Vision

Difference p-value Difference p-value Difference p-value

The four selected stimulus pairs (one for each sensory modality) were combined using a full factorial 24 design to yield 16 different products, varying from a product composed of all pleasant stimuli to a product composed of all unpleasant stimuli.

2.5 Procedure Participants for main Study 1A were recruited in a museum, a library, or at an open air market. They were taken into a separate room or to a quiet part of a hall, where they received one variant of the air purifier. The purpose of this device, the usage context, and the way the ‘Fresh 2 U’ was operated was explained. They were asked to rate the product on 20 bipolar items on 7-point scales presented in random order. Most of these items were related to the dimensions of affective meaning that were identified by Osgood and colleagues (Osgood et al., 1957): five items were related to each of the dimensions evaluation, activity and potency. Five additional items were included to reflect several other aspects of product experience (Table 2). The evaluation items included the three pleasantness items. These three target items were submerged in the larger set to increase the credibility of the test as a consumer test and to conceal the focus of the study, in order to avoid socially desirable responses. Although the ratings on the 17 non-target items do not contribute to the measurement scale with which the degree of sensory dominance can be assessed, they can be instrumental in determining whether the sensory manipulations lead to perceivable differences between the product variants. Furthermore, they provide additional information on the nature and size of these differences. Therefore, responses on the nontarget items were also analysed and reported in the result section. To evaluate whether the relative importances that were determined experimentally concurred with participants’ self-reports, participants were also asked to rank order eight product aspects with respect to their importance for using the Fresh 2 U. The most important aspect was given rank order 1, the next important rank order 2 and so on. Four aspects reflected the importance of the four sensory modalities (visual appearance, sound, feel and smell of the product), whereas the other four reflected general product evaluation dimensions (easy to operate, economical with energy, safe and does not break down easily). Participants completed the entire test in approximately 5 min.

The experimental assessment of sensory dominance Table 2

129

List of items used in the main studies

English translation

Original Dutch item

Evaluation Agreeable–disagreeable

Aangenaam–onaangenaam

Pleasant–unpleasant

Prettig–onprettig

Good–bad

Goed–slecht

Seductive–repulsive

Verleidelijk–afstotelijk

Sympathetic–unsympathetic

Sympathiek–onsympathiek

Activity Stimulating–relaxing

Stimulerend–ontspannend

Excited–calm

Opgewonden–kalm

Flashy–discrete

Flitsend–discreet

Conspicuous–inconspicuous

Opzichtig–onopvallend

Lively–quiet

Levendig–rustig

Potency Gentle–tough

Schattig–stoer

Masculine–feminine

Mannelijk–vrouwelijk

Strong–weak

Sterk–zwak

Weighty–humble

Gewichtig–nederig

Persistent–accommodating

Vasthoudend–meegaand

Additional Mature–youthful

Volwassen–jeugdig

Interesting–boring

Interessant–saai

Funny–serious

Grappig–serieus

Strict–mild

Streng–mild

Expensive–cheap

Duur–goedkoop

2.6 Results 2.6.1 Pleasantness scale 4

ANOVA of the full 2 factorial design with the mean of the three target pleasantness

attributes as the dependent variable (Cronbach’s Į = 0.85) and the four modalities as explanatory variables (at levels pleasant and unpleasant) yielded no significant main effects or interactions (all F(1, 303) ” 2.92, all p • 0.088, Ș2 ” 0.01) (see Table 1). The mean pleasantness scores for the 16 products differed only slightly, from 3.19 to 4.07. The difference in pleasantness scores between the air purifier with all four manipulated sensory stimuli at the pleasant level (4.00) and the one with all four at the unpleasant level (3.47) was not statistically significant (two-tailed t-test, t = 1.3, p > 0.20).

130

H.N.J. Schifferstein et al.

2.6.2 Complete item set A multivariate ANOVA of the full factorial design with all 20 items in the questionnaire as dependent variables and the four modalities as independent variables also yielded no significant effects (all F(20, 258) ” 1.61, all p > 0.050, Ș2 ” 0.11). This indicates that the manipulated variables had no systematic effects on the other dependent measures either.

2.6.3 Self-reported importances The self-reported rank orders for the importance of the various product aspects were submitted to repeated measures ANOVA with modality (four levels) as within-subjects factor and gender (two levels) as between-subjects factor (Labovitz, 1970). In this analysis, only the importance rankings for the four sensory aspects were used; results for the four non-sensory aspects were excluded. This yielded a significant modality main effect (F(3, 309) = 151.8, p < 0.001, Ș2 = 0.60). The gender effects were not significant (p > 0.20). The participants reported that they found smell (2.9) the most important aspect for this product, followed by touch (4.8), vision (5.4) and sound (6.4). All paired comparisons were significant in post hoc tests with Bonferroni adjustment (p < 0.001).

2.7 Discussion Although the self-reports indicate that consumers seem to have a clear idea of the relative importance of the different modalities for a portable air purifier, no significant effects were found in the evaluations of the different product prototypes. Possibly, our experimental design may be responsible for not finding any effects. When people are asked to evaluate an entirely new product, they lack a specific reference frame to base their judgements on. As a consequence, they may focus mainly on whether they like the product concept as a whole by comparing it to other types of products, without paying attention to the product details. Using a within-subjects design in which each participant evaluates multiple products might contrast and, thereby, amplify the differences between the eight product variants (e.g. Parducci, 1974). Therefore, we expect to find larger response differences between air purifier variants in an experiment using a withinsubjects design. Furthermore, it can be argued that some of the outcomes of the pre-study may not be completely valid for the main study, because the test locations differed considerably. The consumer test of Study 1A was performed in museums and libraries in order to recruit a large number of respondents for a small test. However, the pre-study was performed under laboratory conditions in a controlled setting. Therefore, the participants of the main study may have been in a different mood, and the environments may have been somewhat noisier or may have had a different background smell and lighting. Therefore, keeping the test environments more comparable could improve the validity of the test procedure. It can also be argued that the participants’ ages in the main study may have affected the results adversely. The air purifier is a gadget-like product typically designed for young adults of 20–30 years old. Because many participants in the main study were considerably older than this target group, perhaps the product did not appeal to them. However, when we repeated the ANOVA including only participants who were 40 years

The experimental assessment of sensory dominance

131

or younger, the outcomes were similar. Therefore, we think that the age of the participants was unlikely to have affected the outcomes of the study.

3

Study 1B

The aim of Study 1B was identical to that of Study 1A. However, we improved the design of the study to compensate for some of the shortcomings in Study 1A: participants were presented with multiple variants of the air purifiers, and they performed the test in a laboratory environment similar to the one used in the pre-study. Only specific differences with Study 1A are mentioned below.

3.1 Participants Forty-five naive undergraduate students from the Faculty of Industrial Design Engineering (21 male and 24 female) participated. All participants were paid for their participation. Ages varied between 18 and 28 years (mean 22.2).

3.2 Experimental design The 16 products were identical to those described in Study 1A. To limit the number of products in a single session, we presented each participant with a set consisting of eight products. In each product set, each of the eight stimuli (2 visual, 2 tactual, 2 auditory and 2 olfactory) occurred four times. Six different product sets were created and an approximately equal number of participants evaluated each set. Each product was evaluated by 22 or 23 participants.

3.3 Procedure Participants were presented with eight product variants, one at a time, in a laboratory environment. For each variant, they completed the questionnaire used in Study 1A. At the end of the study, they rank ordered eight product aspects with respect to their importance for product usage (see Study 1A). Participants completed the test in approximately 30 min.

3.4 Data analysis Because each participant evaluated only a subset of the products and these subsets differed between participants, we could not use within-subjects ANOVAs to analyse the complete dataset. Therefore, in the data analysis we disregarded the fact that each participant had evaluated eight products and we analysed the data as if each person had evaluated one product only. Hence, responses were subjected to between-subjects ANOVA with the four modalities (each two levels) as explanatory variables. In withinsubjects ANOVAs differences between conditions are not obscured by differences between participants, because participants serve as their own control (Winer, 1971, p.517). Therefore, within-subjects ANOVAs are usually more sensitive in detecting differences between conditions than between-subjects ANOVAs. Nevertheless, this change in data analysis does not completely eliminate the advantage of presenting

132

H.N.J. Schifferstein et al.

multiple products to the same individual, because we still benefit from the psychological effect that participants will try to differentiate their responses to the eight product variants (Parducci, 1974).

3.5 Results 3.5.1 Pleasantness scale Between-subjects ANOVA of the full 24 factorial design with the mean of the three target pleasantness attributes as the dependent variable (Cronbach’s Į = 0.88) and the four modalities as explanatory variables (at levels pleasant and unpleasant) yielded only a significant main effect of colour (F(1, 344) = 9.9, p < 0.01, Ș2 = 0.03; all other effects p > 0.10). On average, the ice blue air purifiers were evaluated as more pleasant (4.6) than the grey air purifiers (4.2) (Table 1). Similar to the previous study, the mean pleasantness scores for the 16 products differed only slightly, from 3.97 to 4.94. The difference in pleasantness scores between the air purifier with all four manipulated sensory stimuli at the pleasant level (4.77) and the one with all four at the unpleasant level (4.28) was not statistically significant (two-tailed t-test, t = 1.5, p > 0.10).

3.5.2 Complete item set A multivariate ANOVA of the full factorial design with all 20 items in the questionnaire as dependent variables and the four modalities as independent variables yielded significant effects for colour, weight and weight × sound (all F(20, 325) • 1.7, all p < 0.05). This indicated that the manipulations affected responses on the other items of the questionnaire as well. In fact, it seems that even more significant effects were found for the 17 non-target items than for the three pleasure attributes. This is an important difference with Study 1A, where no effects were found for the other items either. This suggests that presenting participants with multiple product variants indeed increased the perceived differences between the 16 products. To reduce the number of dependent variables in this analysis, we first performed a principal components analysis (PCA) with varimax rotation on the entire dataset. This yielded three factors with Eigenvalues above 1. We will only mention attributes here with factor loadings above 0.60 on the main factor and no cross-loadings above 0.30 on other factors. The first factor (22% of variance explained) was an evaluation factor, which contained the three items of the pleasantness scale together with sympathetic and seductive. The second factor (21%) was mainly an activity factor, which was best represented by the items lively, excited, flashy and conspicuous. The third factor (17%) was a potency factor, and was best represented by tough, masculine and strong. We calculated means over the items specified above to obtain scores on the evaluation, activity and potency scales. Cronbach’s Į values of the three scales were 0.89, 0.87 and 0.68, respectively, which appears acceptable for the early stages of basic research (Nunnally, 1967). The scores on these three scales were submitted to MANOVA. This yielded significant effects for colour, weight, colour × smell and colour × weight × smell (all F(3, 342) • 3.7, all p < 0.05). The colour difference was significant on all three scales (all F(1, 344) • 5.4, all p < 0.05): the ice blue air purifiers were evaluated more positively, and they were judged more active and less potent than

The experimental assessment of sensory dominance

133

the grey air purifiers. The weight effect was found only for potency (F(1, 344) = 33.0, p < 0.001): the heavier air purifier was judged to be more potent. The colour × smell (F(1, 344) = 11.2, p < 0.001) and the colour × weight × smell (F(1, 344) = 16.5, p < 0.001) interactions were only found for activity: when the 50 g air purifier was grey, participants found that the fresh dewy fields smell made it more active (3.8 vs. 2.9), whereas when it was ice blue the water smell made it more active (4.1 vs. 3.2). For the 105 g air purifier, the smell differences were negligible, with means ranging from 3.2 to 3.6.

3.5.3 Self-reported importances Repeated measures ANOVA of the self-reported importance orders for the various product aspects yielded a significant modality main effect (F(3, 129) = 6.9, p < 0.001, Ș2 = 0.14). The gender effects were not significant (p > 0.20). The participants reported that they found smell (2.6) the most important aspect for this product, followed by vision (3.2), touch (3.9) and sound (4.4). Almost all paired comparisons were significant in post hoc tests with Bonferroni adjustment (p < 0.05), except for the difference between smell and vision (p > 0.10), and the difference between touch and sound (p > 0.20).

3.6 Discussion According to our approach outlined in Section 1, we need to compare the sizes of the effects on the pleasantness dimension to determine a hierarchy of importances for the various sensory modalities. Our finding that only a single significant main effect of colour was found on the pleasantness dimension (column 11 in Table 1) suggests that vision is the most important sensory modality for this product. However, no hierarchy of effects could be derived, because the main effects for the other modalities were not significant. The scores on the evaluation, activity and potency scales do not allow us to determine an importance hierarchy of the sensory modalities, because the items of which these scales are composed are possibly biased towards one or more modalities. However, shifts in scores on these scales do indicate to what extent the underlying experiential dimensions are susceptible to manipulation of sensory information in the different modalities. These analyses show that, next to vision, touch is an important determining modality for potency, whereas smell may be an important determining modality for activity. In the current design, the product prototypes varied in the types of smells they emitted. We included smells, because air fresheners typically carry a smell and their olfactory characteristics are an important determinant of their appeal. In addition, the release of a smell provides feedback to the user, signalling that the product is actually functioning. On the other hand, people may not expect to perceive any smell at all from an apparatus, that is supposed to purify the air. In that case, both smells may have decreased the product’s appeal, hence decreasing any effect of the olfactory manipulation. This may have contributed to the discrepancy between the high importance people attributed to smell in their self-reports, and the absence of a smell manipulation effect in the pleasantness scores for the prototypes. One of the reasons why we found almost no differences between products on the pleasantness scale may be that the differences between the individual stimuli are quite

134

H.N.J. Schifferstein et al.

small (column 6 in Table 1). These differences may be rather small, because the freedom of designing stimuli was limited. On the one hand, the experimental requirement to use only stimuli that could be manipulated independently of one another made it impossible to use several stimulus dimensions such as size, shape and surface texture, which can be perceived by both touch and vision. Furthermore, loud sounds could not be used, because their vibrations can be perceived tactually. On the other hand, choosing a portable air purifier as the test product limited variation in product size and weight. Moreover, because we wanted to develop a plausible and realistic product concept that could be introduced on the market, we were limited in the sounds the product was likely to make and the types of smells it would disperse. Another aspect that deserves attention is the involvement of product designers in our research approach. A professional designer developed the stimuli for the pre-studies that were expected to differ in pleasantness, and of which two per modality were incorporated in the final product. We used designers’ input, because we wanted to derive knowledge, that is not only theoretically interesting, but also relevant for design practice. However, we noticed in Study 1 that the differences between the stimuli proposed by the designer were quite subtle and possibly not radical enough to affect general consumer evaluations. Possibly, a designer who has the design of the final end product in mind is likely to come up with subtle differences between variants, because larger differences in stimulus properties may be regarded as obvious, extreme, over-the-top manipulations. Nevertheless, the latter differences are more likely to result in statistically significant effects and, therefore, tend to be used in experimental research. Hence, we explicitly instructed the designer in Study 2 to create larger differences between the stimuli.

4

Study 2

One of the reasons why we found no significant effect for audition, touch and smell on the pleasantness scale in Studies 1A and 1B may be that the manipulations were not strong enough. The stimuli that were selected in the pre-tests differed by 0.75–1.05 units on the pleasantness scale (column 6 in Table 1). This difference between the unisensory stimuli may have been too small to evoke a pleasantness difference in the complex, multisensory product. Therefore, we decided to replicate the study for a new product (a table lamp), in which the pleasantness differences between the unisensory stimuli would be larger. The approach was similar to that used in Study 1B; only specific differences will be mentioned.

4.1 Participants Forty-seven undergraduate students of Industrial Design Engineering (27 male, 20 female) with a mean age of 22.0 (range 18–29 years) took part in the main study.

4.2 Product Because we wanted to develop a product for which pleasure is an important aspect, we decided to develop a new product for a hotel room setting: a table lamp. When people stay in a hotel, they often have some free time available to have fun and to explore the objects in the room. In addition, we expect that many people have already stayed in a

The experimental assessment of sensory dominance

135

hotel room or could easily imagine staying in a hotel. The lamp we designed had to be turned upside down to switch it on. This movement ensured that the tactual modality would be stimulated and it created a possibility to generate sounds. While the lamp was burning, a scent would be dispersed in the air by the heat generated.

4.3 Stimulus selection Stimuli were selected in a separate pre-study by having 30 students evaluate how the unisensory stimuli looked, sounded, felt or smelled. The study was introduced as part of the development process for a new lamp. Participants received a description and pictures of the lamp and the hotel context (Figure 3). Participants were asked to imagine that they were in the hotel. As visual stimuli, we used ten upper halves of the lamp that were painted in different colours. Participants were instructed to imagine that the whole lamp had the same colour. The auditory and tactual stimuli were similar in shape to the final lamp and were painted black. Participants were asked to take hold of the stimulus at the centre, and turn the lamp upside down. The tactual stimuli differed in weight, weight distribution and surface texture (smooth or rough). The sounds were produced by moving elements that were hidden in one half of the lamp. For the olfactory stimuli, three polyethylene fragrance pellets were placed in a white, plastic jar (height 33 mm, diameter 43 mm) and were covered with cotton wool. Participants were instructed to open the jar and keep it at a distance of about 20 cm from the nose. They could move the jar closer to the nose if they could not perceive the fragrance clearly enough. Between different fragrances participants waited for at least 50 sec. Figure 3

The table lamp presented in the hotel room context (see online version for colours)

136 Table 3

H.N.J. Schifferstein et al. Differences in pleasantness scores for the stimuli used in Study 2

Unpleasant Pleasant Pre-study Study 2 Modality Stimulus Mean (SE) Stimulus Mean (SE) Difference p-value Difference p-value Vision

Fluorescent 2.65 (0.26) Dark green blue

4.55 (0.24)

1.90

0.10): the blue lamps were

138

H.N.J. Schifferstein et al.

judged to be less active (3.3) and more potent (4.5) than the fluorescent green ones (4.7 and 4.0, respectively). The tactile effect was found only for potency (F(1, 360) = 50.0, p < 0.001): the heavier lamp was judged to be more potent (4.6 vs. 4.0). The sound effect was found for potency (F(1, 360) = 75.0, p < 0.001) and evaluation (F(1, 360) = 6.5, p < 0.05), with the nut sound being evaluated more negatively (4.2) and more potent (4.7) than the bell sound (4.5 and 3.9, respectively). The smell effect was only present for the activity dimension (F(1, 360) = 6.0, p < 0.05) with the patchouli lamp (4.1) being perceived as more active than musk (3.9).

4.5.3 Self-reported importance and order Repeated measures ANOVA of the self-reported importance rank orders of the various product aspects yielded a significant modality main effect (F(3, 135) = 28.7, p < 0.001, Ș2 = 0.39), but no gender effects (p > 0.10). The participants reported that they found vision (1.9) the most important aspect for this product, followed by sound (4.2), touch (4.9) and smell (5.3). All paired comparisons with vision were significant in post hoc tests with Bonferroni adjustment (p < 0.001), but not between the three remaining modalities. Participants also wrote down which differences they had perceived between the lamps. For each person, we determined the temporal sequence in which product differences were attributed to the various sensory modalities. Subsequently, we performed a Ȥ2-test for each modality to determine whether the distribution of rank orders deviated from a uniform distribution. These tests indicated that product colour was mentioned significantly earlier than the other aspects (Ȥ2(3) = 41.7, p < 0.05); no other significant differences were found. After completing the study, we asked a subsample of the participants which stimuli they preferred for the different lamps. In contrast to our expectations, between 36% and 47% of the participants actually preferred the unpleasant stimuli for vision, touch and sound. For smell, no reversed preferences were reported, but several participants indicated that they did not prefer either of the two stimuli.

5

General discussion

5.1 Evaluation of the proposed approach to assess sensory dominance The first objective of the current studies was to determine the relative importances of multiple sensory modalities in product experiences, based on an additive model for the integration of sensory information. The approach consisted of selecting a dimension that was associated with all the tested sensory modalities to the same extent (pleasantness), selecting stimuli that varied to the same degree on this dimension, and assessing people’s responses to products composed of all multisensory stimulus combinations. Unfortunately, in all studies the variation in scores on the pleasantness scale between the product variants was small, indicating that the effects of the manipulations were small and, in most cases, not statistically significant. Therefore, an importance hierarchy could not be established. In comparing the air purifier study with the lamp study, it is interesting to note that the experimental outcomes of the two studies were quite similar, although the two

The experimental assessment of sensory dominance

139

products were very different. In contrast, participants’ self-reports showed very different importance rank orders between studies. Perhaps this suggests that the outcomes of the experimental studies were not so much determined by the products and their usage context, but by the way the experiment was designed? Possibly, an experimental design focusing on differences in pleasantness for individual stimuli is unlikely to yield substantial pleasantness differences at the level of the complete product. Note that both studies only found a convincing main effect of colour on the pleasantness scale. In addition, in Study 2 participants reported colour differences between products earlier than auditory, tactual or olfactory differences. The latter result suggests that the visual information was more readily available and hence more accessible during the experiment than the other types of sensory information. Possibly, this higher accessibility is related to the relatively large effects of the colour manipulation on the pleasantness judgements. A possibly important aspect, that is ignored by the proposed approach is that interactions may occur between stimulus attributes, whereas an additive model assumes no interactions. Especially in judgements of pleasantness for complex stimuli, the extent to which individual stimuli fit together or are congruent is important for the overall pleasantness judgement (e.g. Schifferstein and Verlegh, 1996; Veryzer and Hutchinson, 1998). This could decrease the likelihood of finding modality main effects in ANOVA. However, note that no interaction effects were found on the pleasantness dimension in our studies. Therefore, a simple additive model with interaction effects is unlikely to account for the current data. Instead, future models might need to incorporate additional subjective evaluations of the congruency of stimulus combinations to improve their explanatory power (e.g. Schifferstein and Verlegh, 1996). Despite the current findings, the additive model may work very well in predicting sensory integration for other dimensions of product experience than pleasantness (see Fenko et al., 2009). However, because other experience dimensions tend to be biased towards one or more sensory modalities (Fenko et al., 2010), they cannot be used to derive a hierarchy of sensory importances.

5.2 Single stimuli vs. complex products The second objective of the current study was to investigate the relationships between evaluations of single materials and evaluations of complex products. However, the situational context in which the data were obtained sometimes differed between pre-study and main study. As concerns Study 1A, the pre-study was conducted in a laboratory, while the main study was conducted in museums and libraries. In Study 1B and Study 2, the effect of context was reduced by performing both parts of the study under laboratory conditions. Nevertheless, for both studies it can still be argued that some of the outcomes of the pre-study may not provide a completely valid basis for comparison with the main study. This is illustrated by the post-experimental responses of participants in Study 2, of which a large proportion appeared to prefer the unpleasant stimuli. How can these outcomes be accounted for? First of all, in the pre-studies all sensory stimuli were singled out from the other modalities and were sometimes presented in a different format than in the main study. For example, in the pre-studies the odourants were presented in plastic cups, which had to be opened to smell the sample, whereas in the main studies smelling the odourants was an

140

H.N.J. Schifferstein et al.

integral part of the user-product interaction and did not require a separate action. These different ways of presenting the stimuli may have affected the evaluation of the stimuli. Furthermore, even if all stimuli had been presented as part of a finalised end product, the task instruction to focus on one particular aspect may have activated a different frame of reference for judging stimuli than when end products are evaluated as wholes. When people evaluate complex products, they may employ different strategies that differ in the extent to which they pay attention to and actively process the various product aspects (e.g. Carbon and Leder, 2005; Sujan, 1985). Different information processing strategies usually result in different product evaluations. In the current pre-studies approximately 10 different stimuli were presented for each modality, whereas in the main studies only two stimuli per modality were used. This may have made the target modality much more salient in the pre-studies than in the main studies. If this is the case, it will be extremely hard to obtain comparable data for stimulus and product judgements. As a consequence, our results suggest that the validity of laboratory tests that focus on a single product aspect (e.g. optimise the sound) in an attempt to improve the experience of the entire product may be rather low (for another example see Ludden and Schifferstein, 2007). This has important implications also for product development processes, where separate components often need to be evaluated and decided upon when the total end product is not yet finished. This outcome may also have consequences for methods used in other areas of psychology that investigate how information from various sources is integrated into an overall evaluation, such as in attitude formation (e.g. Fishbein and Ajzen, 1975). Another aspect of interest here is the test location. The disadvantage of using laboratory conditions is that they deviate from the final product usage contexts. For instance, stimulus preferences revealed in the pre-study of Study 2 in some instances deviated from reports from prospective users (hotel guests). This suggests that participants in the laboratory were unable to take the usage context fully into account while evaluating the stimuli and the products. Hence, although we informed participants of the context of the study, they may have disregarded it during the tests. Especially for olfaction, contextual factors may be of utmost importance for pleasantness evaluation (Herz and von Clef, 2001; Stevenson, 2001). Therefore, it may be necessary to conduct both pre-study and main study in the final usage context even if this causes practical problems, such as a large degree of distraction, high cost, and difficulties in recruiting participants. Alternatively, the context might be made more dominantly present in the laboratory setting by using larger photographs of the context, by projecting pictures of the context on the wall, by displaying objects that belong to the context, or by rebuilding the entire context in the laboratory. It is an empirical question whether these measures will be sufficient to recreate the feeling of being in the target context. For now, we propose that the main implication of the present outcomes for product development research is that home use tests seem to be essential in assessing the market potential of newly developed or improved consumer products. To strengthen the relationship with the overall evaluation, individual target aspects should preferably be evaluated as part of the entire product in its natural usage context. In addition, the test should avoid focusing the participants’ attention on a single target aspect, either by the products involved in the test (e.g. presenting multiple variants that differ only in sound characteristics) or by the questions asked (e.g. ‘what do you think of the sound?’).

The experimental assessment of sensory dominance

141

5.3 Other experimental limitations Some of the effects we found may be partly due to the specific stimuli we selected. Because the tactual manipulations varied primarily in weight and the sound manipulation in Study 2 varied in loudness, it may not be surprising that both these modalities affected responses on the potency dimension. However, the effects of the various smells on the activity dimension were less straightforward: in the air purifier experiment, the combinations of stimuli (colour–weight–smell) seemed critical for the occurrence of an effect, whereas in the lamp study there was a small general difference between patchouli and musk. It remains to be established whether all these effects are reliable and, if they are, whether they are modality-specific or stimulus-specific. Furthermore, it should be noted that the participants in most of our experiments were students from an Industrial Design program. Possibly, these students are more consciously aware of the stimuli they perceive than average consumers, because they are trained to select materials for design assignments. Therefore, we expect that the effects found might be even smaller in the general population. In fact, the only study in which we used participants that were more representative of the general population was Study 1A, and here no effects of our manipulations on overall product pleasantness were found. However, because Study 1A suffered from several other methodological problems, it would still be interesting to replicate Study 2 with a sample representative of the general population.

6

Conclusion

We proposed an experimental approach to determine the relative importance of the sensory modalities, based on an additive integration model of the product evaluation process. Unfortunately, our approach did not allow us to derive an importance hierarchy for the different sensory modalities from the current experiments. This may be due to the limited extent to which the pleasantness of a unisensory stimulus influences the pleasantness of a complex multisensory product, to effects of stimulus (in)congruity among the various stimuli, or to differences in situational context and in directed attention between pre-studies and main studies. Future research will need to show whether these outcomes are specific for the current products and for the selected experience dimension (pleasantness) or whether they are a general finding.

Acknowledgements The Netherlands Organisation for Scientific Research (NWO) is gratefully acknowledged for funding this research (MAGW VIDI Grant 452-02-028 and MAGW OC Grant 400-03-131). We thank Samantha Hosea for her assistance in organising and conducting Study 1A, Rob Luxen and Marijke Melles for technical assistance during the development of stimuli and product prototypes, and IFF (Hilversum) and Senta Multisensory Concepting (Amersfoort) for providing the odor stimuli. We thank three museums (Museon, Legermuseum and Kunsthal) and the public libraries of Delft and Rotterdam for providing facilities to perform Study 1A, and we are indebted to Hotel de Plataan in Delft for their cooperation in Study 2.

142

H.N.J. Schifferstein et al.

References Adank, R. and Warell, A. (2008) ‘‘Five senses testing’ – assessing and predicting sensory experience of product design’, In P.M.A. Desmet, J. van Erp and M. Karlsson (Eds.), Design and Emotion Moves. Newcastle upon Tyne, UK: Cambridge Scholars Publishing, pp.35–58. Akamatsu, M., Mackenzie, I.S. and Hasbroucq, T. (1995) ‘A comparison of tactile, auditory, and visual feedback in a pointing task using a mouse-type device’, Ergonomics, Vol. 38, pp.816–827. Anderson, N.H. (1981) Foundations of Information Integration Theory. New York: Academic Press. Canon, L.K. (1970) ‘Intermodality inconsistency of input and directed attention as determinants of the nature of adaptation’, Journal of Experimental Psychology, Vol. 84, pp.141–147. Carbon, C.C. and Leder, H. (2005) ‘The repeated evaluation technique (RET). A Method to capture dynamic effects of innovativeness and attractiveness’, Applied Cognitive Psychology, Vol. 19, pp.587–601. Classen, C. (1993). Worlds of Sense: Exploring the Senses in History and Across Cultures. London: Routledge. Classen, C., Howes, D. and Synnott, A. (1994) Aroma: the Cultural History of Smell. London: Routledge. Ernst, M.O. and Banks, M.S. (2002) ‘Humans integrate visual and haptic information in a statistically optimal fashion’, Nature, Vol. 415, pp.429–433. Fenko, A., Otten, J.J. and Schifferstein, H.N.J. (2010) ‘Describing product experience in different languages: the role of sensory modalities’, submitted. Fenko, A., Schifferstein, H.N.J., Huang, T.C. and Hekkert, P. (2009) ‘What makes products fresh: the smell or the colour?’ Food Quality and Preference, Vol. 20, pp.372–379. Fiore, A.M. and Kimle, P.A. (1997) Understanding Aesthetics for the Merchandising and Design Professional. New York: Fairchild. Fishbein, M. and Ajzen, I. (1975) Belief, Attitude, Intention, and Behavior: an Introduction to Theory and Research. Reading, MA: Addison-Wesley. Freides, D. (1974) ‘Human information processing and sensory modality: cross-modal functions, information complexity, memory, and deficit’, Psychological Bulletin, Vol. 81, pp.284–310. Gibson, J.J. (1966) The Senses Considered as Perceptual Systems. London: George Allen & Unwin. Green, P.E. and Srinivasan, V. (1978) ‘Conjoint analysis in consumer research: issues and outlook’, Journal of Consumer Research, Vol. 5, pp.103–123. Heller, M.A., Calcaterra, J.A., Green, S.L. and Brown, L. (1999) ‘Intersensory conflict between vision and touch: the response modality dominates when precise, attention-riveting judgments are required’, Perception and Psychophysics, Vol. 61, pp.1384–1398. Herz, R.S. and von Clef, J. (2001) ‘The influence of verbal labeling on the perception of odors: evidence for olfactory illusions?’ Perception, Vol. 30, pp.381–391. Howes, D. (1991). The Varieties of Sensory Experience: A Sourcebook in the Anthropology of the Senses. Toronto: University of Toronto Press. Kelso, J.A.S., Cook, E., Olson, M.E. and Epstein, W. (1975) ‘Allocation of attention and the locus of adaptation to displaced vision’, Journal of Experimental Psychology: Human Perception and Performance, Vol. 1, pp.237–245. Klitzner, M.D. (1975) ‘Hedonic integration: test of a linear model’ Perception and Psychophysics, Vol. 18, pp.49–54. Köster, E.P. (2003) ‘The psychology of food choice: some often encountered fallacies’, Food Quality and Preference, Vol. 14, pp.359–373.

The experimental assessment of sensory dominance

143

Kuipers, J.C. (1991) ‘Matters of taste in Weyewa’, In D. Howes (Ed.),.The Varieties of Sensory Experience: a Sourcebook in the Anthropology of the Senses, Toronto: University of Toronto Press, pp.111–127. Labovitz, S. (1970) ‘The assignment of numbers to rank order categories’ American Sociological Review, Vol. 35, pp.5151–524. Lederman, S.J., Summers, C. and Klatzky, R.L. (1996) ‘Cognitive salience of haptic object properties: role of modality-encoding bias’, Perception, Vol. 25, pp.983–998. Ludden, G.D.S. and Schifferstein, H.N.J. (2007) ‘Effects of visual-auditory incongruity on product expression and surprise’, Int. J. Design, Vol. 1, No. 3, pp.29–39. MacDonald, A. (2002) ‘The scenario of sensory encounter: cultural factors in sensory-aesthetic experience’, In W.S. Green and P.W. Jordan (Eds.), Pleasure with Products: Beyond Usability, London: Taylor & Francis, pp.113–123. Moskowitz, H. and Krieger, B. (1993) ‘What sensory characteristics drive product quality? An assessment of individual differences’, Journal of Sensory Studies, Vol. 8, pp.271–282. Myers, J.H. and Alpert, M.I. (1977) ‘Semantic confusion in attitude research: salience vs. importance vs. determinance’, Advances in Consumer Research, Vol. 4, pp.106–109. Nagamachi, M. (1995) ‘Kansei engineering: a new ergonomic consumer-oriented technology for product development’, Int. J. Industrial Ergonomics, Vol. 15, pp.3–11. Nunnally, J.C. (1967) Psychometric Theory. New York: McGraw-Hill. Osgood, C.E., Suci, G.J. and Tannenbaum, P.H. (1957) The Measurement of Meaning. Urbana, IL: University of Illinois Press. Parducci, A. (1974) ‘Contextual effects: a range-frequency analysis’, In E.C. Carterette and M.P. Friedman (Eds.), Handbook of Perception Vol. II. Psychophysical Judgment and Measurement. New York: Academic Press, pp.127–141. Posner, M.I., Nissen, M.J. and Klein, R.M. (1976) ‘Visual dominance: an information-processing account of its origins and significance’, Psychological Review, Vol. 83, pp.157–171. Schifferstein, H.N.J. (2006) ‘The relative importance of sensory modalities in product usage: a study of self-reports’, Acta Psychologica, Vol. 121, pp.41–64. Schifferstein, H.N.J. and Desmet, P.M.A. (2008) ‘Tools facilitating multisensory product design’, The Design Journal, Vol. 11, No. 2, pp.137–158. Schifferstein, H.N.J. and Verlegh, P.W.J. (1996) ‘The role of congruency and pleasantness in odorinduced taste enhancement’, Acta Psychologica, Vol. 94, pp.87–105. Schütte, S., Eklund, J., Ishihara, S. and Nagamachi, M. (2008) ‘Affective meaning: the Kansei engineering approach’, In H.N.J. Schifferstein and P. Hekkert (Eds.), Product Experience. Amsterdam: Elsevier, pp.477–496. Stadtlander, L.M. and Murdoch, L.D. (2000) ‘Frequency of occurrence and rankings for touchrelated adjectives’, Behavior Research Methods, Instruments, and Computers, Vol. 32, pp.579–587. Stevenson, R.J. (2001) ‘Perceptual learning with odors: implications for psychological accounts of odor quality perception’, Psychonomic Bulletin and Review, Vol. 8, pp.708–712. Stoller, P. (1984) ‘Sound in Songhay cultural experience’, American Ethnologist, Vol. 11, pp.559–570. Sujan, M. (1985) ‘Consumer knowledge: effects on evaluation strategies mediating consumer judgments’, Journal of Consumer Research, Vol. 12, pp.31–46. Suzuki, M., Gyoba, J., Kawabata, H., Yamaguchi, H. and Komatsu, H. (2006) ‘Analyses of the sensory-relevance of adjective pairs by the modality differential method’, Shinrigaku Kenkyu, Vol. 77, pp.464–470. van Beers, R.J., Sittig, A.C. and Denier van der Gon, J.J. (1999) ‘Integration of proprioceptive and visual position-information: an experimentally supported model’, Journal of Neurophysiology, Vol. 81, pp.1355–1364.

144

H.N.J. Schifferstein et al.

Veryzer, R.W. and Hutchinson, J.W. (1998) ‘The influence of unity and prototypicality on aesthetic responses to new product designs’, Journal of Consumer Research, Vol. 24, pp.374–394. Warren, D.H. and Schmitt, T.L. (1978) ‘On the plasticity of visual-proprioceptive bias effects’, Journal of Experimental Psychology: Human Perception and Performance, Vol. 4, pp.302–310. Warren, D.H., Welch, R.B. and McCarthy, T.J. (1981) ‘The role of visual-auditory ‘compellingness’ in the ventriloquism effect: implications for transitivity among the spatial senses’, Perception and Psychophysics, Vol. 30, pp.557–564. Welch, R.B. and Warren, D.H. (1980) ‘Immediate perceptual response to intersensory discrepancy’ Psychological Bulletin, Vol. 88, pp.638–667. Welch, R.B., Widawski, M.H., Harrington, J. and Warren, D.H. (1979) ‘An examination of the relationship between visual capture and prism adaptation’, Perception and Psychophysics, Vol. 25, pp.126–132. Winer, B.J. (1971) Statistical Principles in Experimental Design. New York: McGraw-Hill.

Note 1

More information on the stimuli evaluated in the pre-studies is available from the first author upon request. 2 Analyses performed on the scales based on the complete item sets for Study 2 yielded comparable results.