If you're seeing this message, it means we're having trouble loading external resources on our website.

If you're behind a web filter, please make sure that the domains *.kastatic.org and *.kasandbox.org are unblocked.

Main content

Signal detection theory - part 1

Created by Ronald Sahyouni.

Want to join the conversation?

Video transcript

Voiceover: In this video, I'm going to be talking about something known as Signal Detection Theory. Signal Detection Theory, basically, looks to see how we make decisions, so decision making, under conditions of uncertainty, so with uncertainty. Let me give you an example of what Signal Detection Theory is trying to do. I want you to look at the screen and tell me if there is any change. It was pretty obvious that I put up this bright green circle. Okay? Keep looking at the screen and think to yourself and try and notice any changes. Okay? This time around, I put up this fainter green dot as compared to this bright green dot. The bright green dot is a fairly strong signal and this faint green dot is a fairly weak signal. Signal Detection Theory is, basically, trying to figure out at what point is a signal strong enough that we are able to notice it, in the first place, and also in order to… Signal Detection Theory is, basically, trying to decide at what point are we able to detect a signal, and it had its origins in radar. Back when radar was being developed, they had to figure out a way to determine whether a strong signal is a ship or a large whale or a school of fish, and that's where it had its origins. Signal Detection Theory also plays a role in psychology and, in psychology, imagine that we show a list of words to an individual, and then we show them a second list of words, and we ask them to recall which words from the second list were on the first list. The decision that they have to make is to decide which word on the second list was also present on the first list, and the uncertainty is their ability to memorize all the words on the first list, so they're not sure, 100%, whether a word is exactly the same as the one on the first list, or very similar, and I can give you a real-world example of Signal Detection Theory. Imagine that you're driving to work or school and you're waiting at a traffic light. It's a foggy day, and you have to decide when to start driving, so you have to decide when the light turns green and you have to start driving. Now it's really hard to see the green light, so it might be, kind of, faint, kind of, like this green light, and you have to decide, at what point in time, how strong does the signal have to be in order for you to say, "Yes, "the light is definitely green. Let me start driving." In that case, there are a few different options. Let me just draw a quick table here. There are a few different options. Either the signal is present, so the light is green, so signal can be present, or the signal can be absent, so the light is red, or it's not green. You can either say, "Yes, the light is green," or you can say, "No, the light is not green." There are a few different possibilities. If you say, "Yes, the light is definitely green," you're 100% sure, maybe it's something like this, then that would be a hit. However, if the light is present, or maybe it's really faint, and it's present, but you're not 100% sure whether it's green or not, you might say, "No," and, since the signal is present, and you're saying, "No," that's a miss, so it's incorrect. Another possibility would be the signal being absent, so maybe the signal's absent, but you say, "Yes," and that would be a false alarm, so false alarm. The final possibility is that the signal is absent and you say, "No," and that's correct, so that would be a correct rejection, a correct rejection. If the signal is really, really strong, so if it's this, you might always get it right. Whenever the signal is present, you'll always say, "Yes," and when the signal's gone, you'll always say, "No." In that case, it's pretty easy to decide whether a signal is present or not. On the other hand, if it's a fairly weak signal, maybe this faint green dot, you might get some false alarms. You might say, "Yes," when the signal's actually present, or you might say, "No," because you don't see this faint green dot, and you might get some misses. An easy signal, like the really bright green dot, would create more hits than misses, whereas a weak signal would create less hits than misses. Now, the strength of a signal, which is what we were just talking about, is a variable known as D Prime, so this is the strength of a signal. Another variable is C, and that is strategy. Let me just talk about this. Strategy would be… Let's look at the example of you driving to work and you're waiting at the traffic light. One strategy could be, if you see any light, you're going to say, "Yes," and start driving. Another strategy would be, if you see any green light, you're going to say, "Yes," and start driving, or, third strategy would be, if it's a green light and it's elevated up off the floor, and it was presented immediately after a red light, then you're going to start driving, so you have different strategies. There are two big strategies. You could either have a conservative strategy, so conservative, or you can have a liberal strategy. If you have a conservative strategy, you would always say, "No," unless you're 100% sure that the signal is present. You're always going to be saying, "No," unless you're 100% sure the signal is present, and the bad thing about that is that, even though you'll get all the correct rejections, you might also get some misses. On the other hand, you can have a liberal strategy, where you always say, "Yes," and, in that case, you'll always get all the hits; however, you might get a few false alarms. These are the two different strategies that you can use, so this would be strategy and this would be the strength, so D Prime and C.