The brain science behind storytelling – part II (and sort of a use-case)

Pedro Almeida | December 4, 2017

A while ago, I wrote about my presentation at Unilever about the brain and storytelling. Here’s the second part to the story…

If you recall the post, I ended up defending the view that we like narratives because building stories allow us to make sense of the world and ourselves, and that this is an evolved ability that confers humans a specific evolutionary advantage.

I then went on to apply this principle to ad success, by illustrating with a nice study by Quesenberry and Coolsen (2014), where they showed that Superbowl ad success was related to the number of dramatic acts present in the ad. Notably, dramatically richer Superbowl ads, are better rated by viewers. Paul Zak further showed that good stories, with engaging and clear narrative acts, are intimately related to emotional activation processes (with physiological correlates such as ACTH and oxytocin release). And, in fact, a beautiful study by Reagan et al, using text sentiment analysis showed that all books from Project Gutenberg mostly follow 6 types of emotional arches intimately connected to their narrative acts.

This tells us that in fact, the drama in the story is linked to an emotional voyage that we want the viewer to embark on, and that successful ads are those that can take the viewers on that ride.
Now, the issue is, how do we measure this emotional arch effectively? We could, of course, ask the user, but that would interrupt the process and introduce a less than desirable bias in the emotional process itself. We would also lose temporal granularity, as it is hard to collect declarative measures for each second of the ad. We could also use facial coding, but we’re finding that in fact, it is not the norm (especially if viewers are alone) to produce facial expressions when one is looking at ads (do you?). In fact, a generally accepted view is that facial expressions evolved as a set of tools for effective social communication among humans, and signal situations and themes that are emotionally charged and recurrent in our evolutionary past. How charged are you when looking at ads? And who are you communicating with?

So, we resorted to our preferred methods: peripheral physiological indexes (electrodermal activity and heart rate) of emotional activation. We showed them that this was a really cool way to measure the emotional trip of the spectator and link it to emotional events, as the following case-study from the Superbowl Kia ad 2016 illustrates.


The reason why we like the people from Unilever PT so much was that they were the first to trust in MindProber, and in fact, these results are from the first live study we ever did.

What we did back then, was to register the physiological activation of a relatively small group (55) of participants in a room (still not in their homes), to Unilever brands vs. direct competitors and ask them about ad likability and memory impact. We also asked participants to signal, through our app, the moments they especially liked or dislike.
In addition to getting a useful benchmark of emotional impact, clients could see the emotional arch of their content, understand whether it fits what they intended on the drawing board, and relate it to memory associations by participants. Here are biometric the results (sorry, can’t tell you which is what…).


Our favorite way of looking at the data is:

1 – understand how impactful the initial seconds are by looking at activation rise time (this is especially important if you’re on digital media and want to avoid the skip button),

2 – look at the moments where you’re supposed to produce impact and analyze whether that happened across segments,

3 – search other points of interest (moment with unexpected peaks, where participants express significant liking or disliking) and try to figure our its drivers;

4 – compare overall activation with benchmarks and

5 – analyze post-visualization surveys and relate them to the activation arch.

Doing this provided some important insights, which unfortunately I can’t tell you much about. One thing we can disclose is that we were able to understand that although ads where producing impact where intended (especially via visual load), a couple would produce unforeseen memory associations in high arousal moments that granted some caution (stuff we didn’t want people to associate with the ad). We were also able to show that that long format ads are much more effective (especially when the narrative is good and impactful) in involving customers.

An important point, of course, is that thanks to our automated pipeline we’re the only company out there that can run these studies in hundreds of people in a few days and provide maximum autonomy to clients. Our mission is to teach them how to use the platform and get out of their way.

share on

About the author
Pedro Almeida

You may also like…