Neuromarketing Proof? UCLA Brain Scans Predict Ad Success
For years neuromarketing firms have been selling their services to help advertisers optimize TV commercials, product packaging, and other media. While these companies all claim success in helping their clients boost sales, there’s been little in the way of published academic research that demonstrates measuring consumer brain activity can reliably predict subsequent behavior. A new study published in Psychological Science brings us closer to that point: scientists using a UCLA fMRI facility analyzed anti-smoking ads by recording subject brain activity. They also asked subjects about the commercials and whether the ads were likely to change their behavior. The researchers found that activity in one specific area of the brain predicted the effectiveness of the ads in the larger population, while the self-reports didn’t.
The methodology involved comparing brain activity in subjects who viewed ads from three campaigns to actual performance of the campaigns in increasing call volumes. The researchers focused on a subregion of the medial prefrontal cortex (MPFC) but also compared activity in other brain regions for control purposes. They found that the ad campaign which created the greatest activity in the MPFC region generated significantly more calls to a stop-smoking hotline. Here’s a sample ad (sorry, not sure which campaign it’s from):
Experts Failed, Too
The subjects failed to identify which ads would change their behavior; in fact, the most effective campaign, “C,” was the one judged to be least likely to work. The researchers also asked a group of industry experts to predict which campaign would work best. Like the experimental subjects, the so-called experts also predicted that “C” would be the least effective.
Before we get carried away…
A few cautions are in order before we break out the champagne and claim that neuromarketing studies have been validated and the field has officially come of age:
- This was a single study that tested 31 total subjects. (That’s actually a good number for an fMRI study.) These subjects were smokers who wanted to quit, a group likely to be receptive to the messaging.
- FMRI is rarely used by commercial neuromarketing firms due to equipment cost, lack of portability, subject immobility, and small sample sizes. Most neuromarketing companies use EEG, biometrics, and/or other techniques not capable of the 3D brain region localization of fMRI.
- The ads weren’t commercial campaigns and the measured behavior wasn’t a product purchase or even a brand preference. The same technique might not work for the commercial campaigns of interest to most marketers.
Even if this single, small study of smoker behavior can’t be readily extrapolated to campaigns for BMW or Pepsi, it’s still of great significance in proving neuromarketing studies can actually work. As the authors note, “The approach described here is novel because it directly links neural responses with behavioral responses to the ads at the population level.” Simply put, the brain scan data correctly predicted how the ads would perform in the real world – not just how the subjects would behave, but the broader public audience. That’s a major milestone.
It’s interesting that the authors never use the term “neuromarketing” in the text of their paper, but rather refer to their study as a “neural focus group.” That’s actually a rather nice term, as it positions the idea of using brain scans in a context familiar to traditional marketers. Solid wordsmithing AND exciting data!
Abstract & Full Text: From Neural Responses to Population Behavior Neural Focus Group Predicts Population-Level Media Effects by Emily B. Falk, Elliot T. Berkman, and Matthew D. Lieberman.
I used to market to people living with schizophrenia and I met with a few players in this space. Unfortunately the researchers were not confident in their ability to read this audience. I was intrigued with the possibility of seeing a lymbic response to our creative, and learning which activated the memory (although it wouldn’t be clear whether storage or retrieval was being observed).
It’s going to be fascinating to watch as more studies like this unfold. Really goes to show how little we understand about our own brains and what makes us tick.
I know that a lot of companies would pay big bucks for this kind of research if neuromarketing claims could be effectively and adequately substantiated. Very interesting stuff here!
Interesting finding !!
“….They found that the ad campaign which created the greatest activity in the MPFC region generated significantly more calls to a stop-smoking hotline. …”
The ultimate question : how to develop an ad that is able to create the greatest activity in the MPFC region.
The problem, Eric, is that the area in question isn’t a magic bullet. It worked for anti-smoking ads, but might be totally irrelevant for beer or detergent. Still, this work is one significant step in the right direction.
I like that you are cautious about this result and I agree that one swallow a sample does not make. However, if they were right in one instance they might be on the way to be right in more instances.
Interesting that the experts were also wrong. Could be part of the reason why most anti-smoking advertising also does not work.
I think I have pointed you to the site http://antismokingscam.wordpress.com/ which argues this.
Some of the anomaly between experts and the result might be that the respondents were smokers that wanted to stop while the experts would not have been.
Unfortunately this is therefor also not a good anti-smoking campaign, but only one that directs potential buyers of the service to the service.
You may be right about why the experts missed, Erik. I’d propose an even simpler explanation: experts simply aren’t all that good at predicting what works in advertising. They do a lot better at explaining why a campaign did well, or flopped, after the results are in. My years in direct marketing (where we had constant quantitative data on what worked) taught me that even smart, experienced people could be confounded by actual consumer behavior.
Sometimes sleep makes one realize that you are missing an important point(s).
1. Thomas Ramsoy taught me that when doing an fMRI it is not just an area in the brain that lights up. Many areas light up. The analyst then deduct one picture from the rest to see whether the area of interest lit up more under the test condition. Therefor there has to be a apriori hypothesis about which areas activities one will inspect. So your point that other areas could have been activated, they were just not looked at is right – and important.
2. Much more important for neuromarketing (and deserving a separate post by you): According to Lindstrom these ads should have also activated the Nucleus Accumbens – and thereby created a desire to smoke.
i) the activation among the researched respondents might have just been to make them smoke? (What Lindstrom would have concluded), or
ii) Lindstroms interpretation of his results, that anti-smoking warnings make people want to smoke, is not right. It made people want to stop as the experiment above showed – because it related to calls to the help-line (external verification).
To your first point, Erik, I suppose the combination of varying activities in different areas (as well as differences between individual subjects) means researchers need to be careful they don’t find the pattern they are looking for while potentially ignoring others. By the time research gets distilled to the level of the general press (and marketing blogs like this one!), the findings sound quite uncomplicated: “Stimulus A lit up region X, so we concluded…”
I’ll have to look and see what Lindstrom actually found. I recall that the main thrust was that peripheral items that were associated with smoking (packaging, even with ugly lung images, and red Formula 1 cars, for example) became part of the “trade dress” and could trigger cravings. An anti-smoking ad that didn’t include either the product itself (cigarettes, people smoking) or imagery associated with smoking or cigarette brands, might be less likely to trigger craving than one that had an equally powerful message but included smoking-related imagery. All speculation, of course.
Roger, exactly my point: It all depends on the POV of the speculator!
Of course – some might be right.
Amazing how little we really know about how our brains work. I love these stories. It’ll be interesting to see how future studies shape up.
As a physician I would like to see this research extended to the idea being actively promoted by health policy makers and care managers that patient satisfaction is linked to improved outcomes. My gut tells me that the curb appeal of a health message doesn’t matter so much in some situations when difficult behavior change is needed, and thus the patient “doesn’t hear what they want to hear.” It would be helpful to know from a more objective source whether appropriate confrontation is sometimes effective or necessary.
I hadn´t read this interesting article of yours and the original paper. I´ve been working in marketing research for many years and have focused specially in the importance of PFC and vmPFC for persuasion (not leaving other cerebral areas behind). My hipothesis has been that all this cerebral areas are responsible for making tv ads persuasive. I have developed a complete theory (to be published next year) on what “mental contents” can explain and predict the activation of these relevant areas generating persuasion. This study seems to be helpful to prove some of my hypotheses: The mPFC is central (though not exclusive) to understanding persuasion. The “mental contents” and “mental processes” taking place in this area (and other “sub” areas related to persuasion) are the central aspect of marketing research.
On the other hand, cognitive linguists have developed strong theories on how most of our actions and behaviors are shaped and are represented in the brain (image schemas / cognitive metaphors). Some cognitive liguists are developing theories on what brain areas are invoved in processing these two kinds of “mental activities”. We have developed qualitative research techniques (for the last 15 years) using cognitive linguistics to “find” motivations and persuasive strategies for our clients with relatively good rates of success. So we hypotesize that these cognitive processes are responsible for activating mPFC and other brain areas involved in persuasion.
Most of our clients test the tv ads that we recommend with quantitative techniques (specially MB TV links) and the correlation of persuasion scores are mostly high.
Hope you find this interestig!.
I’m a believer in neuromarketing but am curious Roger about your opinion of what Vaughan Bell (clinical and neuropsychologist at the Universidad de Antioquia, Colombia and visiting senior research fellow at the Institute of Psychiatry, King’s College London) wrote about in The Guardian on “The Trouble With Brain Scans”?
Any technique can be overused or even abused, Michael. I wrote about the dead salmon study here: Are Brain Scan Findings Fishy?. Any findings need to be reproducible by other labs to be truly accepted. As we get more research going, I expect that we’ll develop a body of research that is recognized as high quality work. At the moment, at least from a neuromarketing standpoint, we aren’t quite there. Studies like the UCLA work are pointing in the right direction.
I see that you are presenting in Sao Paulo at the NMSB conference!
Will your talk be available to poor people that cannot afford the trip across continents?
Good question, Erik, and one to which I don’t know the answer. Some conferences tape the speeches and make them public, some sell video access for a modest fee to people who couldn’t attend live, and some don’t tape at all. One conference I spoke at did a free live video feed for anyone who cared to tune in. It was sold out in person, so the video availability didn’t seem to hurt demand for registrations.
If it looks like videos will be available for the Sao Paulo event, I’ll put up a post. Thanks for raising the question!
This is early study no need to be too tough to it. Holes can always be picked in methods. Every study is a hint of further things that need more testing.
It does appear that subjective experience-self-reports-conscious expression is actually a pretty good contrary indicator.
Brain research says that behavior is “decided” in 140ms or less so there is no way any verbal statement can ever know, let alone, what is happening or going to happen.