Neuromarketing: Using Fmri to Study the Brains of Consumers

NEUROMARKETING: USING FMRI TO STUDY THE BRAINS OF CONSUMERS

BY KIM DO, YINUO HAN, CASSIDY HARDIN, YANA PETRI, AISWARYA SANKAR

Ming Hsu is an Associate Professor at the izona, but at the same time I was work- Haas School of Business and Helen Wills ing in a cognitive neuroscience lab. Tat : What kind of informa- Neuroscience Institute at the University of was when folks in the social sciences frst BSJtion can neuromarketing ex- California, Berkeley. He received his Ph.D. started working with functional magnetic tract from data that could be of inter- in Economics from the California Institute resonance imaging (fMRI). It was around est to people in the feld of marketing? of Technology and now heads a neuroeco- that time that I saw a fascinating presenta- nomics lab at Berkeley that studies consumer tion from a group of economists who used : Te f eld is moving incredibly choice and social behavior. In this interview, fMRI to study how people made econom- MHfast, so it’s hard to say something we discuss how fMRI can be used to study ic decisions. I thought that this was one of that isn’t at risk of becoming obsolete in the consumer brain and brand personali- the coolest things that I had ever seen and the near future. But in general, I would say ty without resorting to biased self-reports. soon started working with them. I had no that we can use neuroscience to address idea at the time that these were economists some of the long-standing skepticism that : Given your background in Politi- of some renown, especially Vernon Smith, people have toward self-report measures, BSJcal Science and Economics, how did the director of the lab who won the 2002 like focus groups and surveys. We can use you get into the feld of neuromarketing? Economics Nobel Prize. Te second major neuroscience to validate whether what con- infuence was in my PhD studies work- sumers are telling us refects what is going :I think the common thread is the ing with my thesis advisor Colin Camer- on in their brains. In the same way that MHneed for all of these felds to un- er. He has this sixth sense for knowing people don’t necessarily trust interviews, derstand human decision-making. What what is going to be the next generation of there is a lot of skepticism surrounding distinguishes my work and that of those cutting-edge research questions. For ex- focus groups, so not everyone will report like me is that we also look into the bio- ample, before Caltech even had an fMRI their true thoughts. We can use neuromar- logical mechanisms in addition to behav- scanner, he would tell me to design exper- keting to avoid bias in consumers’ answers ior and societies. How I got into the feld iments applying brain imaging to econom- and confrm that the information we derive is a bit more circuitous. As you mentioned, ic questions, because he thought this was from consumers’ brains is consistent with I studied Economics and Political Science going to be a hugely important set of ques- their self-reports. We can also use neuro- as an undergrad at the University of Ar- tions. Turns out he was completely right. scientifc tools to increase the precision of

32 Berkeley Scientifc Journal | FALL 2017 our measures; for example, consumer engagement. Finally, we can use neural measures to forecast how consumers will react to a branding or an adver- tising campaign. Can we scan a brain and fgure out signatures that are going to allow us to project what’s going to happen afer the product is launched?

: In your research you talk about BSJseveral methods that you use for brain response imaging. Can you go Figure 1: Difference between (a) the range of consumer choice in the laboratory over briefy what these methods are and and (b) the range of consumer choice in the real world.1 why you chose to use fMRI specifcally?

: Tere are three primary im- regions are contributing to your signal. So controversial because, beyond self-reports, aging methods used in human MH if you care about where things are happen- there are few ways to actually validate what neuroscience: fMRI, electroencepha- ing in the brain, fMRI is the better tool. consumers think about a brand. For exam- lography (EEG), and positron emission ple, a skeptic can argue that people don’t tomography (PET). PET is largely re- : In one study,3 you used machine spontaneously think of these attributes served for medical imaging because it learning and fMRI to study brand without being explicitly prompted by the involves injection of radioactive tracers. BSJ personality—human-like characteristics questionnaire. What’s missing is the pos- So for neuromarketing we are typically that consumers associate with brands, like sibility for objective independent verifca- talking about fMRI and EEG only. In excitement, competence, sincerity, etc. tion, very much like DNA evidence in fo- terms of how we choose a specifc meth- Could you explain for our readers what this rensic analysis. Our answer to this is to put od, the primary tradeof is between cost brand personality framework is and how you people in an fMRI scanner. In the study you and portability on the one hand, and the mapped neural activity in certain regions of mentioned, we asked them to think about anatomical specifcity on the other. If the brain to specifc brand personalities? well-known brands like Gucci, Apple, Goo- you care about portability and are on a gle, or Ford. Afer the experiment, we asked limited budget, EEG is the likely choice : Brand personality is the idea that them to take the brand personality survey. because fMRI requires a 3-ton machine people can think about brands in We then showed that we could use their that costs about $1 million, so good luck MH anthropomorphic terms. For example, peo- brain activity to predict how they would wheeling it into a movie theater, for ex- ple rate Campbell’s Soup as high in terms describe brands in the subsequent survey. ample. Te downside of EEG is that it of being “wholesome,” or Google as being Because our participants didn’t know about doesn’t have very good spatial resolu- “imaginative.” It is intuitive but sometimes the survey when they were being scanned, tion, so you can’t easily tell which brain there was no risk that questions could bias their thoughts during the scanning session. Tis method is not perfect, but it provides some of the frst indications that we can use the brain to validate consumer self-reports, much like DNA evidence can be used to validate the account of a witness.

: In the same study, how were neu- BSJral responses to brands extracted and then mapped to one of the fve person- ality features?

: Tis is where we used ideas from MHmachine learning. We used 42 out of 44 brands as a training set to develop a predictive function for brand personali- ty based on a pattern of brain activity. We Figure 2: The technology used to study the consumer brain. While EEG is very learned how consumers respond to dif- popular because it is not very expensive, fMRI offers higher spatial precision.2 ferent brands and tried to explain this re-

FALL 2017 | Berkeley Scientifc Journal 33 Figure 3: This figure from Ming Hsu’s study3 maps how the scientists use hold-out and training brands to create a brain map of the five personality features from which consumer responses to unknown brands can be predicted. sponse in terms of fve personality traits: ex- keting to genetics to artifcial intelligence. REFERENCES citement, competence, sincerity, ruggedness, Tey can be used with destructive conse- and sophistication. We used two “hold-out” quences, but also for positive ends. Tese 1. Hsu, M. (2017). Neuromarketing: brands to check whether we could correctly technologies are simply tools, and the way Inside the mind of the consumer. predict brand personality based on the in- we use them ofen determines whether they California Management Review, 59(4), formation derived from the training sample. have destructive or positive consequences. 5-22. 2. Hsu, M., & Yoon, C. (2015). Te neu- : What are the ethical implications : If you were giving advice to roscience of consumer choice. Current BSJof studying the brains of consumers BSJa student who wants to do re- opinion in behavioral sciences, 5, with fMRI? search in the future, what would you say? 116-121. 3. Chen, Y. P., Nelson, L. D., & Hsu, : Let me give an example frst in- : Get involved early, whether as M. (2015). From “where” to “what”: MHvolving a collaboration of ours MHa research assistant in a lab or distributed representations of brand with a sports team. Tere, one question that an internship in a company! Start doing associations in the human brain. we could have asked was, “How can we use what you are passionate about. Also, try Journal of Marketing Research, 52(4), the brain to fgure out what prices people to select an area where you have a com- 453-466. are willing to pay for tickets?” Or, “How can parative advantage. For example, when I we fgure out what people would like to pay, started neuromarketing, this area of re- and what price would extract the maximum search was new. I was able to pick up the amount of proft?” Tat would have been a material even faster than my advisors be- really terrible study. Te fans would’ve hat- cause I had more time. Tere are so many ed it. Te team would hate it because they, exciting things that are happening! I’ve hopefully, care about what the fans think. had students start creative new research Instead, we asked, “How can we use neu- projects because they follow the most re- roscience to help the team fgure out how cent developments in AI, social media, to deliver a better fan experience?” Tat is and other technological advancements. a question that can beneft everyone! More generally, I think this is a challenge facing every new technology, from neuromar-

34 Berkeley Scientifc Journal | FALL 2017