In this paper, we consider preference formation mainly focusing on emotional factors by analyzing a large dataset that includes information on experiences in life. Our approach is different from that of Tirunillai and Tellis  , who extracted the key latent dimensions of consumer satisfaction by using a unifying framework of unsupervised latent Dirichlet allocation. In their approach, rich data on product reviews across 15 firms in five markets over four years were used to reveal the evaluation of products, while our approach uses rich data on the accumulation of life experiences and consumption to examine preference formation. Our idea is that life experiences form an emotional framework, which produces certain kinds of emotional responses to stimuli such as advertisements or consumption experiences.
The article is organized as follows. Section 2 presents the literature review. Section 3 presents a model of preference formation focusing on unconscious processes in the brain. Section 4 gives an outline of the rich data on life experience and presents the results of the analysis of the relation between preferences and life experience. Section 5 discusses the implications of the analysis for the formation of preferences for consumption.
2. Literature Review
Consumer choice is crucially dependent on preference, which summarizes various kinds of information, including unconscious thought processes. In conventional studies of consumer behavior, preference is deemed to be exogenously given. Mainly, preferences have been estimated from realized consumption behavior. “Revealed preference theory” advocated by Samuelson  is the pioneering approach for examining preference states. However, revealed preference theory is not fully predictive: it only describes behavior under the current set of circumstances, such as income constraints. Thus it cannot be used to address, for example, the potential change in preferences for luxury goods when income changes. Although rich consumers are more likely to purchase luxury goods than poor consumers, this does not necessarily mean that a poor individual will increase their consumption of luxury goods if their income increases. The experience of having been poor may lead that individual to prefer simple goods through familiarity or the desire “to not waste money”, for example. Thus, in predicting consumer behavior, it is necessary to understand how preferences are formed. Preference formation depends partly on attributes such as age, sex, income, and education. In this paper, we consider how life experience influences preference formation.
In empirical studies, the Linear Expenditure System (Pollak and Wales  ) and the Almost Ideal Demand System (Deaton and Muellbauer  ) have been developed to analyze the demand system by using expenditure data. Extensions of these methods have been elaborated to capture the aggregate demand system in the market. These approaches are not aimed at analyzing preference formation. The work of Carpenter and Nakamoto  is more in line with our approach. They examined the reasons for the advantages of pioneering products in market competition not only during the initial stage but also during the maturing stage by focusing on the preference formation process.
In recent developments of marketing research, emotion is an explicit subject of focus. Govers and Schoormans  analyzed the mechanism for generating diversity in a demand system and showed that the diversity of images of products is generated by combining preference and personality. Xie et al.  examined the difference in the formation mechanism for preference between a local brand and a global brand and showed that preference for the global brand is based on the presence of some special attribute, while preference for the local brand is based on reliability and affection.
Jiménez and Voss  examined the concept validity of emotion and appeal (EA). By employing this new measure, they showed that EA and self-concept maintenance are related. Furthermore, they showed that willingness to pay is predicted by EA. Deshwal  focused on the role of emotions when a consumer purchases a product, and argued that emotional marketing can be used as a tool for increasing customer base. Majeed and Usman  conceptualized how women respond to emotional advertisements, and developed an ACE model, which is composed of appeal drivers (A), celebrity endorsements (C), and emotions (E). A partial least squares structural equation modeling analysis found that showbiz celebrities expressing the emotion of happiness make the most effective ACE mix for influencing the consumption behavior of women.
3. Model of Preference Formation
3.1. Importance of the Unconscious
As Winkielman and Berridge  argued, emotion is genuinely unconscious. They showed in laboratory experiments that this implicit emotion causes emotional reaction unconsciously. Preferences are, therefore, unconscious in the sense that implicit emotions are generated as the result of experiences throughout life. Winkielman et al.  suggested that preferences are jointly determined by basic affective processes and basic motivational processes, and a triggering affective stimulus can be unconscious. This statement is important in the sense that basic affect interacts with motivation.
We can interpret the above statement in a marketing context. Figure 1 illustrates the structure of a consumer’s decision making when he/she receives certain stimuli. The affective process and the motivational process refer to unconscious thought processes generated as the result of experiences in his/her life. The motivational process consists of the motivations for pursuing reward and for avoiding punishment. Dijksterhuis and Nordgren  stressed the importance of unconscious thought in decision making and argued that the precision of decision making improves by relying on the unconscious unless any applicable rules exist. The reasoning behind this statement stems from “the weighting theory”, which implies that some weightings are given unconsciously to stimuli experienced, and the affective and motivational processes interact with each other with these weightings (see Dijksterhuis and Nordgren  ).
To illustrate this structure, we consider the example of an advertisement for a
Figure 1. Structure of unconscious thought (color in print).
travel tour. Huge numbers of advertisements for travel tours can be seen in daily newspapers and magazines. It is expected that only rarely will consumers evaluate the price and quality of a travel tour by checking every advertisement. Whether the consumers pay attention to the advertisement or not depends on some unconscious decision making. When a consumer browses advertisements, some advertisement, such as one with a color picture of a beautiful mountain, may stimulate unconscious affection, provoking the consumer to check the price of the tour. Whether the consumer paid attention to the advertisement depends on unconscious decision making, although the behavior of checking price may stems from conscious thought. Checking the price is followed by thinking based on some calculation rule, and an evaluation of the tour could be attained consciously. However, it should be noted that one never evaluates a tour consciously without raising some kind of affection, such as excitement, when receiving stimuli from an advertisement.
In this paper, we hypothesize that unconsciousness is formed as the result of experiences in life. For example, a consumer who once had an experience of positive affection from beautiful scenery unconsciously seeks to repeat that experience, and unconsciously responds positively to similar stimuli. This positive affection is a kind of engine moving the consumer to check the advertisement with the picture of beautiful scenery. In the next subsection, we formalize the process of preference formation. Then, we examine the relation between experience and preference formation empirically.
3.2. Formal Model of Preference Formation
In this subsection, we formalize the process of preference formation based on the hypothesis that preference is formed through the accumulation of experiences in a life. Denote an i-type experience by , and the utility of some item generated by experience at time s by . Thus, the utility is given by
The overall utility formed through the accumulation of experiences up to time t is denoted by , and it is given by
where is the unconscious weight given to the i-type experience at time s, and I is the number of types of experience.
In the empirical implementation, we introduce a linear regression model in which preference is the explained variable and types of experience are the explanatory variables:
where is k-type preference of individual j, a and are parameters to be estimated, the error term, and n is the number of individual.
4. Empirical Analysis
For this analysis, I used a database compiled by the NTTDATA Institute of Management Consulting, Inc., from the results of a survey of human behavior (value judgments, experience in life, financial situation of the household, heath situation, personal information, etc.). The survey was conducted from February 24, 2017 to March 13, 2017 and the questionnaire was distributed to 20,955 monitors, 15,794 of which returned their answers (return rate, 75.4%).
The following is the basic information about the total sample. The male ratio to female ratio is 60.2%. The average age is 51.4 with a standard deviation of 12.8. The average personal annual income is 3.87 million Yen ($35,200) with a standard deviation of 3.73 million Yen ($33,900). The average household annual income is 6.09 million Yen ($55,363) with a standard deviation of 3.90 million Yen ($35,455). The proportion of university graduates is 50.7%.
The value of the average personal annual income of employee given by 2014 Statistical Survey of Actual Status for Salary in the Private Sector of Japan is 4.14 million Yen ($37,635). The value of the average household annual income given by 2016 Family Expenditure Survey of Japan is 5.54 million Yen ($50,353). The university entrance rate given by the 2015 Basic School Survey complied by Ministry of Education, Culture, Sports, Science and Technology of Japan is 51.5%. From these values, we can judge that the sample selected could represent the targeted population.
4.2. Steps of the Analysis
In the first step of the analysis, questions about experience are grouped into types of leisure and cultural experiences. In this step, factor analysis by the principal component method is repeatedly applied to these types of leisure and cultural experiences. In Appendix 1, one example of a factor analysis by the principal component method is shown.
In the second step, factor analysis by the principal factor method is applied repeatedly to a large number of questions about preferences. The questions are categorized into four groups: clothing, food, housing, and play. The results of the analysis are given in Appendices 2-5.
In the third step, we examined the relation between experiences and preferences. After examining the relation, we tried to increase its generality. For this purpose, we reduced the number of variables representing the types of experience by applying the factor analysis to the variables generated in the first step. The result is shown in Appendix 6.
4.3. Preference Formation and Experiences in Life
Using the variables on preferences generated by the factor analysis in Appendices 2-5, we examine the relation between preferences and life experiences. In the empirical implementation, we introduce a linear regression model based on the model given in Equation (3). The explained variable is the k-type preference of individual j such as preference for high quality and so on, those are denoted by . The explanatory variables are i-type life experiences of individual j, those are denoted by such as friendship experience and so on.
We conducted multivariate regression analysis and the standardized regression coefficients derived from the analysis are listed in Tables 1-4 and Figures 2-5. The liner regression procedure of SPSS version 24 is used for the estimation. By comparing the values of the standardized regression coefficients, we can determine the relative strengths of the influence of experiences on preference. The results of multivariate regression analysis could be judged to be quite robust, because there is no sign of multi-correlation. In addition, the stability of parameters is checked by changing the size of sample used in the estimation.
1) Wasteful habit
Musical experiences, such as “music and art experience” and “Soul and Reggae”, have a positive influence on wasteful habit. On the other hand, “friendship experience” and “light religious” experience have a negative influence on wasteful habit. These results suggest that “living for the moment” is promoted by music enjoyment and the “long-term-view way of living” is influenced by friendship and light religious experiences.
2) Preference for high quality
It is interesting that traveling for leisure has the strongest positive influence. Sensitivity to high quality is improved by having intense experiences while visiting museums or luxury hotels. Music and art experiences, photography experience, and social network service (SNS) experience also have the positive influence on the preference for high quality. This is interpreted as meaning that exposure to beauty or high-quality works forms the preference for high quality,
Table 1. Standardized coefficients for statistically significant experiences (5% level): clothing.
and SNSs stimulate conspicuous consumption. On the other hand, leisure experience (including play within the house), light religious experience, and friendship experience have a negative influence on the formation of preference for high quality. These experiences also have a negative influence on conspicuous consumption.
3) Ecology preference (i.e., preference for environmental friendliness)
Travel for leisure has the strongest positive influence on the ecology preference. The second strongest positive influence is deep religious experience. Travel is a good opportunity to experience the beauty and wonder of nature. Deep religious experience also strengthens the reverence for nature. It is interesting that only traditional Japanese music experience has a statistically significant effect on this preference. Traditional Japanese music places high regard on nature,
Table 2. Standardized coefficients for statistically significant experiences (5% level): food.
and the essence of the music stems from harmony with nature. It is interpreted that this stance for nature increases the preference for ecological options.
4) Preference for frugality
Leisure experience (including play within the house) has the strongest influence and exchange experience (such as meeting for promoting social human relation) has the second strongest influence on the formation of a preference for frugality. The third strongest factor is light religious experience and the fourth is club and friendship experience. The common feature of these influences would be the experience of joy without spending much money. The fifth strongest factor is family hardship experience, which implies that spending a lot of money is not allowed due to the family situation.
Table 3. Standardized coefficients for statistically significant experiences (5% level): housing.
5) Casual preference (i.e., preference for casual style)
Light religious experience has the strongest influence on the preference for casual style. Exchange experience has the second strongest influence. Exchange experience includes leisure and travel with friends and social exchange with a neutral relationship. This could be interpreted as meaning that neutral relationships make people casual. It is noteworthy that travel for leisure does not have a statistically significant effect on the preference for casual style. This means that the preference for casual styles formed through neutral rather than intense experiences.
1) Preference for gourmet food
Figure 2. Standardized coefficients for statistically significant experiences (5% level): clothing (color in print).
Figure 3. Standardized coefficients for statistically significant experiences (5% level): food (color in print).
Figure 4. Standardized coefficients for statistically significant experiences (5% level): housing (color in print).
Figure 5. Standardized coefficients for statistically significant experiences (5% level): play (color in print).
Table 4. Standardized coefficients for statistically significant experiences (5% level): play.
The strongest positive influence is travel for leisure experience, and the second is SNS experience. This implies that people form gourmet preferences as the result of excellent food experiences during traveling. SNSs stimulate conspicuous consumption by presenting photos of delicious food to the public.
2) Natural preference (i.e., preference for natural ingredients)
The most interesting point here is the lack of a specific experience influencing the formation of a preference for naturalness. Almost all the experiences except for photography experience and SNS experience have a positive influence on the formation of a preference for naturalness, and there is no negative experience. Among the experiences, light religious experience has the strongest influence. This can be interpreted as meaning that religion teaches us the importance of nature, which is deemed as a gift from the gods.
3) Preference for frugality
Leisure experience (including play within the house) has the strongest influence on the preference for frugality in this context. Those who enjoy staying home have fewer opportunities to eat excellent food and have less desire to enjoy good meals. On the other hand, those who seek to enjoy good meals through travel do not have a preference for frugality. Deep religious experience, music experience, and traditional Japanese music experience have a negative influence on the preference for frugality. These results imply that a preference for frugality is formed when one has fewer opportunities to eat in socially formal situations.
1) Preference for firmness and stability
Concerning preference for firmness and stability, we try to interpret the results using a different approach. In this category, we examine the experiences that have the strongest influence on preference formation. Leisure experience (including play within the house), light religious experience, exchange experience, friendship experience, Ballard music experience, and supplementary school or club activity experience have the strongest influences on preference for stability. One interpretation for this result is that the preference for stability is formed through weak social interaction, and a moderate stance toward community members.
2) Preference for frugality
Bullying and poverty experience have the strongest influences on the formation of a preference for frugality in this context. One interpretation for this result is that the preference for frugality is not formed as the result of former experience, but arises from the economic situation.
3) Preference for high quality
Travel for leisure has the strongest influence on the formation of a preference for high quality housing. Travel provides a good opportunity for viewing excellent houses around the world. The good impression produced by well-designed, high-quality houses remains for a long time and forms the desire to build such a house. The high value of the standardized coefficient for photography experience can be interpreted in a similar way.
4) Preference for natural materials
Friendship experience, travel for leisure experience, club and friendship experience, and photography experience all have a relatively strong influence on the formation of a preference for natural surroundings. These experiences promote sensitivity to nature.
5) Preference for used housing
This preference is similar to the preference for frugality. Bullying and poverty experience have the strongest influence. One interpretation for this result is that this preference is also formed from the economic situation rather than from former experience.
1) Preference for outdoor activities
Exchange experience, club and friendship experience, and friendship experience have a relatively strong influence on the formation of a preference for natural activities. One interpretation for this result is that enjoying nature through social activities such as hiking and camping promotes sensitivity to nature.
2) Preference for frugality
Leisure experience (including play within the house) has the strongest influence on the formation of a preference for frugality in leisure activities. Those who know how to enjoy themselves within house have less incentive to spend a lot of money on leisure activities.
3) Preference for high quality
Travel for leisure has the strongest influence on the formation of a preference for high quality leisure activities. Through travel, one gets to know various types of leisure activity. By experiencing large variety of activities, it becomes possible to select the high quality ones. Exchange experience can be interpreted in a similar way. Sophisticated leisure experiences, including music and photo experiences, promote the formation of a preference for high-quality leisure activities.
4) Preference for indoor activities
Leisure experience (including play within the house) has the strongest influence on the formation of a preference for indoor activities. Those who know how to enjoy themselves within the house have less incentive to go outside.
5) Preference for attending events
Exchange experience has the strongest influence, and club and friendship experience has the second strongest influence on the formation of a preference for attending events. Travel for leisure also has a relatively strong influence. These results are interpreted as meaning that good and enjoyable experiences at various types of events are experienced through social activity or travel.
6) Preference for second hand play tools or media
Leisure experience (including play within the house) has the strongest influence on the formation of this preference. Those who know how to enjoy themselves within the house have less incentive to spend a lot of money on leisure activities. For them second hand play is good enough.
7) Preference for firmness and stability
Friendship experience has a relatively strong influence on the preference for stability. Compared to other preferences, the influence of experience is weaker. This means the preference for firmness and stability is not formed through experience, but through social consciousness.
4.4. Correlation of Preferences among Items
In Tables 5-7, the coefficients for the correlations of various preferences are listed. The strength of correlation differs among items, but the pattern is similar for all preferences. For example, “clothing and food” and “housing and play” have strong correlations, whereas “clothing and play” and “clothing and housing” have relatively small correlations.
5. Discussion and Conclusion
In this paper, we focused on unconscious thought processes that are formed through experiences in life. Intuitively, we can expect that emotionally strong
Table 5. Correlation coefficients for a preference for a high grade experience among items.
Table 6. Correlation coefficients for a preference for frugality among items.
Table 7. Correlation coefficients for an ecological preference among items.
experiences have a strong influence on the formation of unconscious patterns of thought, and preferences are formed through the accumulation of experiences. The results of the empirical analysis are consistent with this intuition and give us various insights into the mechanism of preference formation. Firstly, it is important to note that around 60% to 80% of experiences are statistically significant in explaining the formation of preference, and the number of significant variables increases when the preference is influenced by intense experiences, as is the case for the preference high quality or ecological options. This supports the intuitive idea that emotionally strong experiences have a strong influence on the formation of unconscious biases.
Secondly, the correlation coefficients of preference among items suggest that preference depends on the items, but the pattern of difference is similar among preferences. This suggests that the possible differences in the strength of the relation between experience and unconscious thought are different among items. Alternatively, it might be possible that there are differences in the strength of the relation between unconscious thoughts and preferences among items.
This study gives us various kinds of insights into the marketing strategies. Especially, the analysis on the formation of preference by experiences in life helps to provide advertisement information with strong appeal to each consumer. For example, it might be possible for an apparel company to design an effective promotion film or picture for consumers with specific preference such as ecology by imaging beautiful scenery with ballad music. This example is a part of strategy for changing consumers’ behavior. The main force for changing consumers’ behavior would be emotional impact. This study could be developed further for designing this emotional impact.
Funding: This work was supported by a grant of Strategic Research Foundation Grand-aided Project for Private Universities from Ministry of Education, Culture, Sport, Science and Technology, Japan (MEXT), 2013-2018 (S1391010).
Appendix 1. Principal component analysis of experiences during elementary school days.
Note: Rotation method is the Equamax Method.
Appendix 2. Factor analysis of clothing preferences.
Note: Rotation method is the Equamax Method.
Appendix 3. Factor analysis of food preferences.
Note: Rotation method is the Quartimax Method.
Appendix 4. Factor analysis of housing preferences.
Note: Rotation method is the Equamax Method
Appendix 5. Factor analysis of play preferences.
Note: Rotation method is the Quartimax Method.
Appendix 6. (a)-(c) Principal component analysis of experiences.
Note: Rotation method is the Equamax Method.
 Tirunillai, S. and Tellis, G.J. (2014) Mining Marketing Meaning from Online Chatter: Strategic Brand Analysis of Big Data using Latent Dirichlet Allocation. Journal of Marketing Research, 51, 463-479.
 Xie, Y., Rajeev, B. and Peng, S. (2015) An Extended Model of Preference Formation between Global and Local Brands: The Roles of Identity Expressiveness, Trust, and Affect. Journal of International Marketing, 23, 50-71.
 Majeed, S., Lu, C. and Usman, M. (2017) Want to Make Me Emotional? The Influence of Emotional Advertisements on Women’s Consumption Behavior. Frontiers of Business Research in China, 11, 16.
 Winkielman, P., Berridge, K.C. and Wilbarger, J.L. (2005) Unconscious Affective Reactions to Masked Happy versus Angry Faces Influence Consumption Behavior and Judgments of Value. Personality and Social Psychology Bulletin, 31, 121-135.