Dr. Arthur Lee

I tried to get in contact multiple times.

I can give you a lie detector, let's say, that has 99% accuracy in the lab, but unless I can tell you that we are actually measuring lying with full validity, there is always going to be a way for somebody to cheat that.

So I can give you a lie detector, let's say, that has 99% accuracy in the lab, but unless I can tell you that we are actually measuring lying with full validity, there is always going to be a way for somebody to cheat that and try to invoke other related processes that we are using in our lie detector to try and seem as if they're not lying when they are.

So I first wanted to bring up this idea of accuracy versus validity.

It's not only important that a test is very accurate, but it also needs to be the case that the test is measuring what it purports to measure.

So the question then became for us was,

okay, if we use the brain to say, classify when somebody is lying or not, how can we be sure that we are actually measuring lying?

And if we find that we are not measuring lying, but some related construct like selfishness or conflict or nervousness,

How do we go about changing that?

How do we go about improving our validity so that as science progresses one study after another, we keep on improving our validity of our estimates of lying and maybe one day we'll get there.

So this changes the paradigm from let's just increase the accuracy as high as we can make it into let's try to get a hold of the construct of lying as closely as we can as possible.

And then maybe we can talk about accuracy after that.

So the way we did it was we first designed an experiment that is akin to that movie called The Wolf of Wall Street, where there is a financial broker who recommends stocks to people.

And in the movie, it's not a very honest broker who recommends very bad stocks to clients, but the clients don't know.

They don't have the information.

They go by what the broker says, but the broker makes a lot of money in the process.

So we put the experimental participants in a game like that, where they are told, hey, you're playing the role of a financial advisor, and you're recommending one of two financial options to another person that you don't see.

One of that option is more beneficial for your client that you're trying to help.

The other option is actually more beneficial for you yourself.

It has more kickback for you, but not so much for the other client.