Signal Detection Theory
- Crystal Kwan
- Sep 27, 2019
- 2 min read
This week's topic signal detection theory is a whole new concept to me so I searched for many real-life examples to strengthen my understanding of it. Firstly, this theory is designed to assess decisions made under uncertainty.
"Noise" means anything (not only hearing) that can interfere with detection, which is in fact everywhere. There is no perfectly quiet environment. For example, even though you are ready to take a SAT test in a new and completely quiet environment, you can still get distracted by heart beating and breathing sound, which belongs to the noise you don't usually pay attention to in a noisy environment. All these different types of noise may interfere with your ability to detect a stimulus.
In relation to signal, there are two possible errors, a false alarm and a miss. A false alarm is when the signal is not present, but you say you have noticed it. And a miss is you don't notice the signal while it is present. The response criteria can be liberal or conservative when you lean towards yes or no response. This uncertain answer can vary from person to person when doing the same test. This problem urges us to compare the degree of drawbacks of two approaches, while this causes us a lot of hesitations when answering some people's question in the real world. For example, when you are in college and need to choose classes for next semester, you have to balance your choice in the face of overlapping time period of different classes you may want to take. This kind of realistic scenario when spending time deciding between two choices dominates most of my life to be honest. I am a person not good at making quick decisions at all, always investigating two types of errors produced by each choice. There are definitely many negative sides of this type of behaviour. However, after learning response criteria, I become more uncertain about either changing this "bad" habit or not since sometimes to think about potential outcome of your choice may be a good thing.
Crystal, I really like the way that you gave background information on this topic to make it informative for not only subject matter experts, but also an average person that could be new to the topic.
I always think SDT is an interesting topic in terms of preventing human machine relationships that are over-reliant on the machine. At airports, extra alarms are built into the baggage scanning detection system in order to keep the TSA agents in check and remind them to err on the side of caution. It is better in this scenario to have more alarms or false alarms than to miss the dangerous bag.
Hi Crystal! It was refreshing to read about how someone might thing about SDT for the first time! I like to think about it in a few parts, the first being the decision criterion or threshold for making decisions and the second being the possible results in a given search or vigilance task (where the misses and false alarms come into play). Sometimes defining your criterion can be really helpful in order to establish the reason for making a decision, then you can feel more certain about it!