Sharona Pearl

Deciding that people are a match who aren't because it's already got the image of you that you took in ideal conditions, matching it up to a captured image of you in ideal conditions, right?

what we see with cameras with embedded facial recognition technology either private or public our images the people captured not deliberately often and matched up against images that were not specifically necessarily taken for that purpose often scraped from social media in ways that can represent a pretty serious incursion upon our privacy and that has historically led to all kinds of false positives

people incorrectly being identified as a match for, say, somebody who has an arrest warrant.

Absolutely, and scientists are kind of split on this question.

Early face recognition researchers suspected that people who were at the extreme ends of these spectrums tended to focus on a particular feature, a nose, a mouth, ears, as opposed to doing what you're referencing, which is holistic processing.

looking at the face as a whole.

And you can see why from a super recognition standpoint, that makes sense.

If you only see an ear or a nose or an eye and you can still recognize that person, it's because you are actually taking in discrete points of information as opposed to making sense of the face as a whole.

Now, it's unclear whether or not that's actually as robust as we thought.

Maybe some super recognizers are also processing the face as a whole.

But I'll say this.

For me and for a lot of people, it actually becomes really important to see the whole face.

I don't know about you, but I certainly had more difficulty recognizing people when they were wearing masks that covered their nose and their mouth.

The other thing that emerged is that when people took their masks off, if you only met someone with a mask on, they looked really different because our faces were filling in information.

We were making mouths for them that didn't necessarily line up with how they looked at all.

That's a huge compliment from you.

Thank you so much.