[Neuroscientists: feel free to correct anything here with better information for non-scientists to learn. Others: please read this as a primer.]
Some people soundly reject the premise that eye-tracking is somehow better (today) than other forms of technological mind-reading. I can understand this reaction. Eye tracking (ET) is not literally reading minds — neurons to be precise — where the best science fiction imagines lifting our thoughts directly.
We could argue that our eyes are only the place where our central nervous systems (CNS) are directly exposed to the world. The eyes are at the very least a window into something much more complex and much closer to where our thoughts really happen. Curiously, as humans, we somehow read other people’s emotions and intentions, in no small way, through their eyes.
Most other common BCI approaches, like EEG and fMRI, aren’t reading our thoughts directly either. It’s more about finding useful proxies for localized brain activity in blood oxygenation and electrical activity. EEG usually measures varying electrical potentials near the surface of the skull, but not individual neurons.
“Neural lace,” on the other hand, ideally interfaces directly with neurons (reading and potentially writing data as electrical signals). But “neural lace” also requires surgery, adaptation and significant individual training to be able to reliably interpret the results. It’s likely to be deployed only in high-value cases for now, such as for people who are paralyzed.
Eye tracking w/pupillometry and scene understanding (via world-facing cameras) represents a readily-available low-dimensional signal that’s much easier to parse than tracking billions of individual neurons or waves of electrical activity. In this signal, our brains have already filtered billions of neurons down to a few outputs and presented them visually for us to read.
It’s as if we had a ticker tape reading out what we’re focusing on and how we’re feeling about it and we’re just reading the tape. Without world-facing cameras, this information would be only partially useful. Cameras and some AI processing gives us the missing context of what we’re looking at, how long, how often, how emotionally, and with how much concentration.
With that, we could determine what interests you, what you pay attention to, how attracted you are to people or things, how you feel in…