I hypothesize that lying triggers a natural stress response, which can be used to identify deception. I estimate that, by observing these stress signals, I will correctly detect approximately 70% of the participants' lies.

What is a polygraph?

A polygraph is a device that measures stress when someone lies. They can do this with many different methods, such as heart rate, sweat, and blood pressure. However, the negatives of a polygraph is that they can show false positives, due to a unrelated increase in stress, and can also have false negatives, as a person can purposefully attempt to fool the polygraph, keeping their perceived stress levels low.

Are polygraphs used in law?

It is rare for a police officer or lawyer to conduct a lie detector test, as lie detector results are inadmissible in a court in most places, due to the possibility of false negatives or false positives as mentioned above. Some US states do allow polygraph results to be used as evidence, but only if both parties agree. A newer method of lie detection is called fMRI. Unlike traditional MRI, which shows a model of the brain, a fMRI shows how the brain is functioning, and which areas are active. If fMRI can be proved to be impossible to fool, it may become a method for lie detection that would be admissible in court, assisting lawyers and police officers with telling if a criminal is lying about something.

How can I measure stress?

Some of the methods of measuring stress are easier to do than others. fMRI is not something a normal person can do, as they would need a highly expensive machine, and need to be a trained neurologist. However, sweat response, heart rate and blood pressure are much easier, although measuring blood pressure does require a small machine, as pressure is needed to be put into the sleeve on the participants arm. Sweat can be measured by using a GSR sensor, or Galvanic Skin Response.

What is GSR?

GSR, or Galvanic skin response, is a sensor that measures skin conductivity, as sweat is salty water, which is conductive. The more sweat produced, the more the reading will increase. These increases and decreases of sweat production are very small amounts, and not visible amounts of sweat.

How can I monitor heart rate?

Heart rate monitoring is very simple, as many watches use them, and require no machinery and can cost $10. Most commonly they use an optical LED, or light emitting diode. A sensor then reads how much light is bounced back from the skin, and this shows when the heart beats, as arteries will expand during a heartbeat.

Previous projects from others

I have seen other people do non electrical lie detectors such as having a participant hold their arms out, and when they are stressed, their arms will feel different when the participant places hands on the testers hands. However, there is no real number proven by science, but by the interpretation of the pressure felt, which could be different depending on the person.

In terms of doing it electrically, it has been done by having participants be asked 5 questions truthfully, to be a low or truth band, and then 5 questions which are told false, and this is the high band or lie band, and then 10 questions that are randomly asked, and whichever band they are in is the result. I found that a game of two truths and a lie would be optimal, as people will be stressed that their lie will be found.

Credits: Science buddies Cleveland clinic American Psychological Association

In this experiment, the variables are as follows:

Controlled

A controlled variable is a variable that never changes throughout the experiment. In my experiment, here are some examples from my project.

• The lie detector itself
• Type of measurement, in this case, two truths and a lie
• Amount of questions

Manipulated

A manipulated variable is the variable that changes, which is the basis of your experiment. There can only be one, because if there is more than one, you cannot know which change caused the outcome. In this case, the manipulated variable is the statements that the participants choose to say.

Responding

A responding variable is a variable that is the result of the experiment. In this experiment, the responding variable is whether I was able to find out the statement the participants lied on.

My procedure goes as follows.

Baseline

I ask all participants three baseline questions. These are to ensure the sensor is working correctly, and to allow me to know how their body reacts when asked a normal question. I observe the graph during these questions and record what I observe: whether it was nominal, dropped, or climbed. I chose questions that would not cause stress, for example, asking for someone's age can be stressful or embarrassing. I would ask what the participant's name was, what school they attended, or currently attend, and what their favourite colour is. I believe these questions will not stress the participants out at all.

2 Truths & a Lie

I then tell the participant they may say their first statement. As they say their statement, I watch my graph, watching for anything abnormal. The graph will continuously move up and down in a sine-pattern wave; however, when the participant is stressed, the line will drop farther than normal. If I observe nothing out of the ordinary, I will record in my spreadsheet that the statement one was an OK response. However, if there is a large drop, I will record a Drop. The same is repeated for statements 2 and 3, and I allow the graph to go back to normal before asking my participant to say their next statement.

Result

After all three statements have been said and recorded, I will look at my spreadsheet. If I marked a statement as a drop, that question is most likely a lie, because lying causes stress. I will record the statement I think is a lie, and then ask my participant which of their statements was a lie. I will record whether I was correct, and if not, what the correct answer was. No personal information is recorded in the spreadsheet; only the graph response to each baseline and question is recorded. The responses to the baseline questions and the statement the participant says are also not recorded in any form.

In this project, I observed a few different things. First of all, I observed that some participants chose to say statements that were embarrassing to say. Because of this, on the graph, this showed up as stress. People were surprised when they saw this, and it shows stress that you may not have thought would show up on the graph. Second, I noticed that when a participant was unsure whether to say that statement or not, this showed up as stress too, throwing me off on whether it was a lie or not.

My analysis is that lie detectors are not fully accurate for most people; however, they can detect people's lies 41 percent of the time. I noticed that some statements people say can cause stress, even though they are not a lie. Embarrassing or statements they aren't sure whether to say are some reasons for false lying. These are a few reasons why lie detectors still shouldn't be used for serious situations. Another reason for such a low score is that some of the tests were performed in a pool viewing area, being a loud area. This may have caused questions that were not lies to be shown as lies, due to the perceived stress, as a lot is happening, with people yelling and making noise.

In conclusion, I have found that a lower percentage than what I initially hypothesized was correct. I found that a noisy environment may negatively affect my ability to perceive lies, and someone who is embarrassed to say their statement may also affect it. Overall, it may have been easier if my environment were better, but in suboptimal situations, it was mediocre.

The applications of a Lie detector are many, although because of the uncertainty of whether the lie detector is lying, they aren't used in certain places, such as law enforcement. A few US states do allow polygraph evidence to be used as evidence in a court, this is only if both parties agree, which means that the accused may choose not allow it, as it proves they are guilty.

Credit: American Psychological Association

I think some of my sources of error are in my choice of environment. I found it was much more reliable when testing in a closed office. However, to be able to get more participants, I had to perform some of my tests in a noisy environment. The noise most likely created more stimulations, meaning even though the person was not lying, they were stressed, and my sensor measures stress. I also think maybe telling my participants not to choose statements that may be embarrassing would help make it more reliable, and performing all experiments in a quiet room.

Science buddies: Simple lie detector test Cleveland clinic: Heart rate monitor American Psychological Association: Do "lie detectors" work?

I acknowledge the following people for their assistance in my project.

• My dad, for helping me with designing my lie detector itself
• My sisters, for helping me with my trifold and being my first testers for the proof of concept
• My mother, for helping with the trifold, and helping me with writing and grammar.
• My teacher, Mr.Bringloe, for hosting and making sure I format my project correctly.