Small steps. Big impact.

THE FITBIT BLOG

Fitbit Asks: Who “Won” Last Night’s Debate?

| 8 Comments

After last night’s presidential debate, the nagging question seems to be “who won: President Obama or Governor Romney?”  It’s a subjective, squishy question that can only be made somewhat quantitative by polling a lot of people on how they felt about the debate.  But what if, instead, you could get some objective measurements on the candidates themselves?  What if you could get insight into their mental state while they debated?  That would be cool… and it’s the object of today’s Fitbit blog post.

You can tell a lot from a person’s face.  A person smiles, they’re happy (usually); a person frowns, they’re not.  That’s the easy stuff.  Believe it or not, you can go much deeper.  Did you know that you can actually measure a person’s heart rate from his or her face?  Really. And from someone’s heart rate, there are various inferences that we can make: such as if they’re stressed, or even lying.

Once we figured out that this was possible, our first thought was, “Wouldn’t it be cool if we could use this to measure the stress levels of our nation’s leaders? Say, for example, during the next presidential debate?” Last night, under the bright lights of the presidential debate stage, that’s exactly what Fitbit Labs did. Just by looking at their faces.

In the above images, a faint blue box shows the area we’re focusing on for our results. It’s pretty cool: using a regular video camera and some advanced computer algorithms, it’s possible to observe small color changes on someone’s skin as his or her heart pumps.  It’s very much like checking for someone’s pulse by touching his wrist, except in this case the “pulses” are color fluctuations in the face coming from blood volume changes under the skin. In order for this technique to work, the person shouldn’t move too much or the motion ends up interfering with the very subtle colors that you’re trying to measure.  And forget about fancy camera panning like you see used in movies for dramatic effect.

Of course, “normal” heart rates vary from person to person a lot as well. Luckily, in the US we actually publish our leaders’ physicals for the public to see, so we know what each candidate’s heart rate was starting off.

Table 1: Resting heart rate for Pres. Obama and former Gov. Romney from their most recent medical reports

Table 1: Resting heart rate for Pres. Obama and former Gov. Romney from their most recent medical reports

During Gov. Romney’s opening remarks, he stayed still for a few seconds and we saw that his heart rate was 73 bpm, or a 83% elevation above his resting heart rate of 40 bpm.  Not bad.  President Obama’s rebuttal a few minutes later came up at 96 bpm, or a 43% elevation above his resting heart rate of 67 bpm. Even though Romney’s heart rate was lower in the beginning, compared to his resting heart rate he was more worked up right from the start.

Now here’s a breakdown of the results across the debate:

  • Romney’s opening argument: 73 bpm (+83%)
  • Obama’s rebuttal: 96 bpm (+43%)
  • Obama’s closing speech: 59 bpm (-12%)
  • Romney’s closing speech: 75 bpm (+88%)
Or, maybe it’s easier to see as a timeline across the debate from opening to close:
Figure 2: Heart rate elevation for Pres. Obama and former Gov. Romney through the debate

Figure 2: Heart rate elevation for Pres. Obama and former Gov. Romney through the debate

What does this mean? Let’s walk through the debate.

There’s actually a number of different inferences you can make from the data we found. A person’s heart rate naturally accelerates during times of stress — such as a crucial debate being televised to millions of people that decide who will run the entire country. We could look at an elevated heart rate as a measure of nervousness. In that case, we would infer that President Obama was more nervous towards the beginning of the debate, but proved to be one cool cat as the pressure mounted towards the end. In this line of reasoning, Romney, was steadily nervous throughout – and to a higher degree than Obama was. But in this case, we’re making the assumption that the elevated heart rate is related to stress, when that’s not necessarily the case. What if, instead, Romney’s heart rate is a reflection of how amped up he is about his ideas? This could just mean that Romney was really “in the zone” for the debate, while Obama failed to bring a certain excitement for the task at hand. This was, in fact, a popular criticism of Obama’s performance.

So who DID win the debate? Well, if you want the candidate who kept his cool, it was Obama. If you want the one who was most excited, it was Romney. Maybe before the next debate Obama should drink some caffeine to get that energy up. It will be interesting to see how the next debate compares, now that the pressure is really on to be more energetic.

This also makes us here at Fitbit wonder if this sort of technology might have a place in debates in the future. Would a candidate ever agree to be hooked up to an EKG during the debate? Or at the very least, could such data be used to train for debates? We can’t wait to find out.

Disclaimers:

This is just our best guess at the candidates’ heart rates during the debate.  There are a lot of reasons to take these with a grain of salt, but we’d be thrilled to see if others could reproduce or refute our analysis.  Of the things to consider:
1. The heart rate measurements were taken during small clips when there’s no movement. No movement is necessary for accuracy, but ideally we’d have used longer clips (~15 seconds, at least).
2. There’s some natural variation in heart rates as well, especially when a person is calm, and “heart rate” is usually defined as the average heart rate over some time. Since we’re using short clips, we might not have seen enough heart beats to make an accurate measurement.
3. Lighting can change on stage and that would affect the results.
4. In all likelihood, both candidates are wearing makeup for the debate. Since we’re measuring changes in skin coloring, makeup can mask some of this.
5. The videos we used were compressed, and of lower quality. A higher definition video would likely be more accurate.
6. Heart rate alone is not the best measure of stress.  Measures of heart rate variability, skin conductance, etc. used together are much more reliable.