by William Hogan
uring a night flight
home to California after five days in Washington, D.C., I discovered that I am
the last person you would want sitting next to you on a plane. Tired and unable
to bang on my laptop in the turbulence at 40,000 feet, I decided to watch Million Dollar Baby. I hadn’t seen it,
but I figured a Clint Eastwood–directed film that had won the Oscar for Best Picture
would be a deserved break for a hard week.
It is a wonderful film, and I became
deeply absorbed in it. The narrative is circumscribed by a father-daughter
story and concludes with an agonizing act. When the movie was over, the man
next to me said, “Sir, is there something I can do to help you?” I was crying. Well,
not really crying, more like heaving big sloppy sobs out of my eyes and nose
and mouth. Everyone around could hear me but I could not suppress my sadness.
After I recovered, I began to wonder
what had happened to me. I was cognitively intact, aware of my surroundings and
who I was. And yet the story was so engaging that it caused my brain to react
as if I were a character in the movie, as if one of my own daughters were the
one suffering. I experienced heartache as the movie ended, but then it was only
As a neuroscientist, I knew that
movies changed our brain activity in some way, but how?
I soon realized I had stumbled on a
potentially useful way to extend my studies of the social brain. My lab was the
first to discover that the neurochemical oxytocin is synthesized in the human
brain when one is trusted and that the molecule motivates reciprocation.
We found that the
human oxytocin response was similar to that found in social rodents,
another person (or rodent) is safe and familiar. Perhaps most surprising, we
found that in humans, this “you seem trustworthy” signal occurs even between
strangers without face-to-face interactions.
Oxytocin is an astonishingly interesting
molecule. It is a small peptide synthesized in the hypothalamus of mammal
brains. It is made of only nine amino acids and is fragile. Oxytocin is
classically associated with uterine contractions and milk-letdown for nursing.
Animal studies have shown that under physiologic stress oxytocin is released in
both brain and body.
This is unusual for a
brain-derived neurochemical, but it provides a powerful way to study oxytocin: After
a stimulus, changes in oxytocin in blood reflect changes in the brain’s
For more than a decade, I have run human
experiments measuring the endogenous release of oxytocin during social
interactions. My colleagues and I have studied oxytocin release in the
laboratory as well as in field studies spanning religious rituals, folk dances,
weddings, and a traditional war dance by indigenous people in the rainforest of
Papua New Guinea.
I also have demonstrated the causal
effect of oxytocin on prosocial behaviors by safely infusing synthetic oxytocin
into hundreds of people’s brains through their noses.
increases prosocial behaviors. It’s like turning on a garden hose and watching
the water spray out.
Provoking the Brain
Studies that only infuse oxytocin
into participants and then make claims about human behavior are suspect. This
approach does not identify what the brain itself is doing during social
interactions, including neurochemical promotion and inhibition of oxytocin
synthesis and dose-response relationships between oxytocin and behavior. The
key question is whether the brain produces its own oxytocin during the behavior
being studied; if so, the causal relationship between oxytocin and a particular
behavior can be demonstrated via an infusion study. But the reverse is not
true: Infusing oxytocin or any drug into the brain and observing a change in
behavior does not mean that this is how the brain works—it simply means that a
drug has changed behavior, as many drugs do. My studies complete the causal
circle by measuring what the brain does naturally and then intervening in this
system pharmacologically to show that the behavior can be provoked.
After years of experiments, I now
consider oxytocin the neurologic substrate for the Golden Rule: If you treat me
well, in most cases my brain will synthesize oxytocin and this will motivate me
to treat you well in return. This is how social creatures such as humans
maintain themselves as part of social groups: They play nice most of the time.
(Why people do not play nice is a fascinating story we also have studied; see
Zak, 2012 for evidence). But
I’m a skeptic at heart, so I always want to measure the
behavioral effects of oxytocin rather than simply ask people’s opinions about
how they feel.
The experience I had watching Million Dollar Baby caused me to wonder
if movies, in addition to direct personal interactions, would cause oxytocin
release. To test this, my colleague Jorge Barraza edited a set of a short video
clips that we obtained with permission from St. Jude Children's Research
Hospital. One version shows a father talking to the camera while his 2-year-old
son, Ben, who has terminal brain cancer, plays in the background. The story has
a classic dramatic arc in which the father is struggling to connect to and
enjoy his son, all the while knowing that the child has only a few months to
live. The clip concludes with the father finding the strength to stay
emotionally close to his son “until he takes his last breath.”
We also developed a video of the same
father and son spending a day at the zoo. This version does not mention cancer
or death, but the boy is bald (from his chemotherapy) and is called “miracle
boy” once during the clip. This video lacks the tension induced by the typical
story form but includes the same characters. This version was used as a control
story to see what the brain does when any video is being watched.
In our first study of narratives, we
took blood before and after participants watched one of the two versions of the
We found that the narrative with the dramatic arc
caused an increase in cortisol and oxytocin. Tellingly, the change in oxytocin had
a positive correlation with participants’ feeling of empathy for Ben and his
father. Heightened empathy motivated participants to offer money to a stranger
who was in the experiment. We connected a story to a feeling and then to a
prosocial behavior. The “flat” narrative of Ben and his father at the zoo did
not increase oxytocin or cortisol, and participants did not report empathy for
the story’s characters.
These findings suggest that
emotionally engaging narratives inspire post-narrative actions—in this case,
sending money to a stranger. But maybe this result only applied to videos of
dying children. Also, we did not know for sure that oxytocin was the reason participants
cared about the people in the video, just that oxytocin and empathy were
correlated. So we rolled up our sleeves and ran more experiments.
Our previous study pointed to
oxytocin as the biological instrument that puts people in thrall to a story. To
assess the causal impact of oxytocin on narrative immersion, we ran a study
using public service announcements (PSAs) in which participants received
intranasal infusions of synthetic oxytocin or a placebo. This time around, we
decided to test a larger set of video narratives. We wanted stories that most
people would not have seen before and ones that could elicit a prosocial action
at a cost (such as a donation). This would allow us to measure objectively whether
the story “got to you.”
We found a rich trove of public
service announcements from the United Kingdom that are well-produced and
engaging. The experiment used sixteen PSAs that ran for thirty or sixty seconds
on four topics: smoking, drinking to excess, speeding, and global warming. To
incentivize people to pay attention to the videos, each of the participants was
paid five dollars if they could correctly answer a factual question about the
ad immediately after watching it. For example, “Was there a car in the video?” Then,
our software asked participants if they would like to donate some of the five
dollars they had just earned to a charity associated with the cause shown in
the PSA. None of the PSAs solicited donations, they simply told stories about
social issues. Computer software presented all the videos and post-video
questions and we used random participant identifiers so that one’s donation
behavior was kept private.
Forty people received either 40 IU of
oxytocin or an equivalent amount of normal saline (placebo). Neither the
experimenters nor the participants knew what substance had been
administered. Participants started
watching the videos after an hour-long period during which the synthetic oxytocin
diffuses from the sinuses into the brain.
We found that those who received
oxytocin donated, on average, 56 percent more money to charity compared with participants
who received the placebo.
This confirmed the causal role of
oxytocin on post-narrative prosocial behavior. But why did this happen? We
discovered that participants who were given oxytocin showed substantially more
concern for the characters in the PSAs. This increased concern motivated them
to want to help by donating money to a charity that could alleviate the
suffering these stories depicted.
If you think about it, the donations
are quite odd. The narrative is over, but the effects linger. It is as if the
brain is lazy and is using a “monkey see, monkey do” approach to assess
appropriate social behaviors. (Indeed, the brain seeks to conserve energy by
using default pathways—a kind of “laziness.”) The PSAs seemed to persuade viewers
that (for example) nowadays the humans are very concerned about drinking too
much, so as a human, I, too, should be concerned. And I should demonstrate that
concern by donating money to charity. Such responses are what social creatures
with social brains do. And yet, participants understand that the stories are
fictional and are portrayed by professional actors. The money donated to
charity cannot help these actors out of their fictional binds. The money might
help prevent the harm depicted in the PSAs from happening to an unknown other
person, but this is a big “if.” Nevertheless, oxytocin makes people want to
help others in costly and tangible ways.
In another experiment,12
we sought to replicate our earlier study by taking blood samples before a group
of forty-two participants (who were not in the oxytocin infusion study) watched
one of the UK PSAs. We measured the change in oxytocin and in a fast-acting
arousal hormone with a long name that is abbreviated ACTH.
When the PSA elicited an increase in
both ACTH and oxytocin, donations were 261 percent higher than when one or both
of these biomarkers did not rise. The change in ACTH correlated with the amount
of attention people paid to the story. This finding makes sense: If we do not
attend to a story, it will not pull us into its narrative arc. Attention is a
scarce neural resource because it is metabolically costly to a brain that needs
to conserve resources. If a story does not sustain our attention, then the
brain will look for something else more interesting to do.
We also found that the change in
oxytocin was associated with concern for the characters in the story, replicating
our earlier finding. If you pay attention to the story and become emotionally
engaged with the story’s characters, then it is as if you have been transported
into the story’s world. This is why your palms sweat when James Bond dodges
bullets. And why you stifle a sniffle when Bambi's mother dies.
Narratives that cause us to pay
attention and also involve us emotionally are the stories that move us to
action. This is what a good documentary film does. More generally, stories with
a dramatic arc fit the requirements for high-impact narratives. This structure
sustains attention by building suspense while at the same time providing a
vehicle for character development. The climax of the story keeps us on the edge
of our neural seats until
the tension is relieved at the finish.
Theorists including Aristotle (Poetics, 335 BCE), Gustav Freytag (Die Technik des Dramas, 1863), and
Joseph Campbell (The Hero with a Thousand
Faces, 1949) have contended that the rising and falling tension of dramatic
performances facilitate the audience’s emotional connection to the characters.
Hollywood writers call this creating “surprising familiarity.” Every story is
different but somehow the same.
Now let’s get down to brass tacks:
Why are there so many dreadful movies? Humans have known about the three-act
structure and mythos, pathos, and ethos for 2,500 years. This is where the
neuroscience hits the flickering screen.
Like all experiments, we had to start
To answer these questions, we needed
to measure attention and oxytocin responses rapidly—second by second, or even
faster. Blood draws would not do. At the same time, the U.S. Department of
Defense wanted to know why narratives are persuasive and supported our research
and that of other labs as well. Attention is easy to measure rapidly, via a
quickened heartbeat or sweat coming from eccrine glands in the skin. But was
there a way to measure oxytocin rapidly? Nature provided a solution. While we were
mostly interested in oxytocin in the brain, the stimulus-induced co-release of
oxytocin in the brain and blood meant we could measure changing activity in
regions with densities of oxytocin receptors. The vagus nerve (the longest
cranial nerve, which innervates the heart and gut) is chock-full of oxytocin
receptors. With a bit of algorithmic fiddling, scientists can measure the
activity of the vagus using an electrocardiogram (ECG). We confirmed that the
change in oxytocin in blood correlates with changes in vagus nerve activity. Voilà,
we had a measurement technique. But would it predict behavior?
We returned to the story of the dying
child Ben because it is a reliable way to stimulate oxytocin release. This time
we measured cardiac activity using an ECG and sweat using an electrodermal
sensor on the fingers. Because we were developing a system that might be used
in a war zone, we built in redundancies. Attention was measured using both
heart rate and skin conductance changes from sweat on the fingers; emotional
resonance was quantified using two measures of changes in the brain’s
relaxation response driven by the vagus nerve. The exciting part was that we
could measure both effects up to one thousand times a second with off-the-shelf
But it is not so simple to isolate
the effects of a story from everything else the brain is doing to keep you
upright, breathing, and conscious. All neuroscience studies need to extract the
neurologic signal produced by a stimulus during an experiment from the
background noise of all other neural activity. To give you a sense of the scope
of this problem, for every thirty people we test for an hour each, we collect a
terabyte of peripheral neurologic data. Most of this data is not relevant to
understanding why people respond to stories, but the faint traces that are
relevant must be extracted and processed with extraordinary care. Once we did
all this, the data told us several interesting things.13
First among them is that the brain
does not work like the hypothetical story structure known as Freytag’s pyramid,
in which strictly rising action leads to a climax, and then strictly falling
action occurs as the story resolves. Even for the one-hundred-second “Ben” video,
one’s attention waxes and wanes. The brain is attending to the story and then
doing a quick search of the rest of the environment, and then refocusing on the
story as the tension rises. Nevertheless, the peak attentional response occurs
in the climax, when Ben’s father reveals that Ben is dying. That’s a bombshell to
which people pay attention.
The oxytocin response lags behind the
attentional spike as the story begins. After about thirty seconds, vagal
activity begins to increase as viewers get to know and then begin to empathize
with Ben and his father. Attention to the story provides a reason viewers
should care about the characters.
Not only were we able we track what
the brain is doing millisecond by millisecond during a story, we used the
neurologic data to build a predictive model of donations to a childhood cancer
charity—our measure of story impact. The statistical model we built predicts
whether a participant would donate money with 82 percent accuracy. That is, by
measuring how your peripheral nervous system responds to a story, we can almost
perfectly predict what you’ll do before you do it.
The participants who, for whatever
reason, either lost interest in the video or didn’t form an emotional
connection to Ben and his father almost never donated money to charity. But we
are still left with a mystery: Why donate money at all? The money will not save
Ben and it won’t offer relief to his father. It seems that once we are
attentive and emotionally engaged, our brains go into mimic mode and mirror the
behaviors that the characters in the story are doing, or might do. As social
creatures we are biased toward engaging with others, and effective stories
motivate us to help others.
Truth be told, Ben’s story is as near
to a perfect high-impact narrative as there is. We wondered if neurologic data
could identify bad stories, too. And what about stories that may be distasteful
but that are still desirable to watch? I watched Steven Spielberg’s Holocaust
movie Schindler’s List once. I’m glad
I did, but I don’t have much desire to watch it again. It was just overwhelming
Our next study tested stories about “hot-button”
issues to see how people reacted to potentially disagreeable topics. We used
first-person narratives from StoryCorps, a nonprofit that collects and
distributes personal stories. We choose six stories on racism, gun control, and
the terrorist attacks of September 11. Each anecdote lasted from two to four
minutes. For our “narrative impact” measure, we invited participants to donate
some of their earnings to a charity associated with the topic of the story.
These stories were challenging to
analyze because they varied substantially in structure and content. The
peripheral neurologic data we collected reflected these variations. Just as in
the “Ben” story, we confirmed that stories that sustain attention and generate
emotional resonance produce post-narrative donations—even stories on difficult topics.
To the brain, good stories are good stories, whether first-person or third-person,
on topics happy or sad, as long as they get us to care about their characters.
Psycholinguists have shown that
effective stories induce “transportation” into the narrative.14 Transportation happens when one loses
oneself in the flow of the story—just like I did while watching Million Dollar Baby. To understand the
psychological effects of stories, we included surveys of narrative
transportation and concern for story characters in the StoryCorps study. Both
narrative transportation and concern predicted post-story donations. This shows
why stories affect behavior after the story has ended: we have put ourselves
into the narrative. Even a week after the experiment, accurate story recall was
predicted by a single measure: narrative transportation.
Do We Know a Good Story When We See One?
You may be thinking that we have a
money-centric approach to assessing when people are moved by a story. Fair
enough. Let’s try a different approach: We’ll have thousands of people rate stories
instead. The stories we used were TV commercials. Conveniently, this is just
what USA Today asks readers to do on
Super Bowl Sunday: vote for the commercials they like the best. About five
thousand people voted for their favorite commercials in 2014, and the style and
content of these short narratives vary from the unusual to poignant to just
plain silly. This gave us a chance to further refine our algorithms and test
them against what people say they like.
not simply provide a ranking of commercials; it has its readers rate them on a
one to ten scale. Good idea! My group derived a quantification of narrative
engagement using neurologic data so we, too, could rate story quality. We
estimated the relative contribution of attention and emotional resonance on
story impact from our corpus of studied stories. We call this measure a story’s
ZEST (for Zak Engagement STatistic). By estimating each Super Bowl ad’s ZEST,
we could compare the USA Today
readers’ ad likability with the ZEST measure of brain activity.
Three days after the 2014 Super Bowl,
sixteen participants watched the top ten Super Bowl commercials in random order
in my lab while we measured their peripheral neurologic activity. The results
were astounding. There was no correlation at all between what USA Today readers said they liked and a
commercial’s ZEST. Either we had made a big mistake, or we had discovered
something important. So we ran another study using USA Today’s top ten 2013 Super Bowl commercials and found exactly
the same thing: zero correlation.
These findings suggest that people are
unable to articulate what they like and do not like. But their brains reveal
what is engaging for them to watch. Perhaps this should not surprise us. In a
classic study, psychologist and economics Nobel laureate Daniel Kahneman found
that people’s preferences for things they have not experienced are largely
Watching the Super Bowl commercials
myself, I sensed why it is hard to articulate what one likes. The best Super
Bowl commercial in 2014, according to USA
Today readers, was called “Puppy Love,” produced for Budweiser beer. In the
first ten seconds, one sees a puppy nuzzling the nose of a Clydesdale horse. One
immediately recognizes the Clydesdale as the Budweiser icon, and this tells
viewers what they can expect from the ad. The suspense is gone, and our
neurologic measures show that people’s attention wanders starting fifteen
seconds into the commercial. Without attention, the hoped-for emotional
resonance with the ad’s characters (and presumably the brand) fails to occur.
But ask people what they like and,
gosh, they see puppies and horses and wide open country and, well, of course we
love these images. But the brain does not lie. The commercial is dull.
In all our studies we ruled out
effects that might influence ZEST, including movement, cars, buildings,
attractive men and women, and many other factors. They don’t matter; it all
comes down to story.
The U.S. Department of Defense’s
funding of the emerging science of narrative jump-started the field.16 17 Storytellers have always known that attention and emotion are
important to develop during a narrative, but now we have ways to measure these
responses directly rather than rely on incohate impressions such as
“entertaining” or “fascinating.” Yet, even with millennia of practice, creating
a great story is difficult. The emerging science of narrative can guide the art,
but it cannot replace it. Humans are just too complex for an algorithm to
generate art. And this is where the artist comes in. The narrator in Million Dollar Baby
describes the heroine, Maggie’s, desire to be a boxer as “
magic of risking everything for a dream that nobody sees but you.
” Artists who create
worlds we cannot help but enter do the same.