In the lab, a broad picture window thats actually a one-way mirror looks into the testing room, where subjects are wired up to sensors as they watch videos. The perimeter of the testing room itself is lined with small carrels, each with its own computer workstation. A Webcam is perched atop each monitor. I had already gone through a demonstration of SPARK Neuros system myself, watching a couple of old Super Bowl ads and a grainy clip of the movie Crash, and had seen, precisely, the peaks and valleys of my attention. According to what Ive seen so far, Gerrol said, it seems like you feel empathy and emotion in strong ways, but you do not seem like someone who has massive highs and lows. It was an anodyne analysis that didnt tell me anything I didnt know, but it wasnt supposed to. The more interesting test would happen when the dozen volunteers began to react to the candidates onstage in Des Moines.
To read emotions and gauge attention, a tight plastic cap outfitted with an array of highly conductive metal electrodes was placed over the participants heads as they concentratedor noton a relaxing, intentionally monotonous beach scene playing on the screen in front of them. They could watch waves forming in the middle distance and hear surf lap the shore. Once the electrodes were in place, the electroencephalogram (EEG) cap began to broadcast electrical signalsbrain wavesevery four milliseconds, establishing each participants baseline: essentially, the brain at rest. This would give the researchers a way to measure the intensity of a participants reactions once that tranquil scene was replaced by the rough-and-tumble of the debate. Our brains share a lot in common with how we process information, but were not all coming into it at the same starting point, Gerrol explained. Maybe youre having a good day. Maybe youre having a bad day. Maybe you just had way too much coffee. And so its really important to understand each persons starting point, so the change in the difference from that baseline becomes much more meaningful than some absolute scale.
The researchers also clipped a heart-rate monitor to each persons left earlobe to keep track of vascular changes and attached a galvanic-skin-response (GSR) sensor to their middle and index fingers to measure sweat production, the way a lie detector does. (Sweaty hands typically occur when someone is anxious. Sweat conducts electricity; the greater the signal coming from the GSR, the greater the arousal.) As the participants looked at the computer monitor, the researchers used an eye tracker to see, instantly, exactly where on the screen they were looking, while a Webcam observed tiny, nearly imperceptible changes in their facial expressions. According to Gerrol, EEG makes up the majority of the signal, GSR contributes to converging evidence around emotional arousal, and facial expressions occasionally help.
How SPARK Neuros proprietary algorithm weights all of this data, Gerrol said, is the secret sauce. Its the way that we combine different sensors, from brain activity to peripheral-nervous-system activity, which can provide insights about the nature of emotions, he told me. Sometimes that is clear joy or clear fear or clear anger. However, keep in mind we are measuring emotions, not like versus dislike. Sometimes a politicianwill purposefully evoke fear as part of their strategy, perhaps scaring people, for example, about war and its consequences or about terrorism. Or they will use anger as a strategic rhetorical device, perhaps evoking anger at a lack of action on climate change. In other words, even if we read anger, that does not mean that the person is angry at the candidate; it very well may mean that the candidate did a good job of riling them up on a topic they care about.
Around midnight, when the debate was over, a couple of the volunteers stayed behind to talk with Gerrol and me. We had already seen a graph of the participants reactions; they had not. Michael Bradley Cohen, a thirty-three-year-old actor and licensed New York City tour guide, who is white, had snapped to attention when Bernie Sanders spoke and flatlined when it was Bidens turn. I came here trying to hold my top three candidatesButtigieg, Sanders, and Warren, in that orderin the same place in my head, Cohen told us. But the data appeared to show something else. His strongest reactions were to Warren, Sanders, and then Buttigieg. When Warren, in an advertisement, spoke about a wealth tax, she really got Cohens attention, even though he thought he had been more attuned to a Buttigieg ad that, comparatively, had not elicited much of an emotional response. During the debate, Cohens reaction to Warrens answer to a question about her fitness to serve as Commander-in-Chief showed a sustained set of spikes, each one climbing higher as she continued to talk, like notes moving up a musical staff. When Gerrol pointed this out, Cohen was, as Gerrol predicted, suddenly introspective. I think it was actually really poignant to me, hearing a womans voice on the stage. I know she spoke after Senator Klobuchar, but to hear her speak about this in a way that I trusted felt really good at the time. Yeah, it showed up. Thats what it was. And probably what were also witnessing is that I had a bit of a glow, like, Oh, theres the girl in class that I like, and that might also be what was picked up on.
Faradia Kernizan, who is twenty-nine and black, recently finished a masters degree in public health. She arrived that evening already a Warren supporter, and it showed in the data. When Warren spoke, Kernizans attention moved from her baseline of two all the way up to eight and a half, a jump that Gerrol found especially telling. I mean, thats what I might expect to see if you were watching a horror movie, he said, not something like a debate, with its dry content. But Kernizan didnt find it dry at all. This is our future, she said. I mean, this is all so exciting. Were at a point where, I mean, hopefully, things will change and maybe we get a different leader, and I think were at a point where we can get somebody that we really believe in.
She was paying special attention to Warrens views on foreign policy, she said, so she would be able to repeat them to friends and family members she hoped to convince to vote for Warren. But Kernizan was also attuned to some of the other candidates as well, like Biden, since there was a chance that he would be the nominee, and Sanders, whom she supported in 2016. That Cohen and Kernizan were not especially moved by the political ads didnt surprise Gerrol. My impression, having studied thousands and thousands of ads, is that political ads are especially boring. The Bloomberg ad, Cohen said, looked like he was running for President in 1988. As Cohen watched it, his attention dropped to his baseline.
A few days after we had gone through this exercise, I got an e-mail from Gerrol telling me about a third participant, whose attention graph looked similar to Cohens and Kernizans, with significant surges of attention when Sanders and Warren were speaking. The difference was that, unlike Cohen and Kernizan, that participants attention was an expression of his antipathy: Sanders and Warren were his least favorite candidates. It was a perfect illustration of the danger of conflating attention with affection. We had seen something similar, too, when Cohens graph showed his attention picking up every time one of the candidates mentioned Trump. Yeah, when we hear the name Voldemort, Cohen said, we all have a response.
Go here to see the original:
The Neuroscience of Picking a Presidential Candidate - The New Yorker