Developing a Lie Detector Using Experimental Economics
Author Information
Author(s): Bershadskyy Dmitri, Dinges Laslo, Fiedler Marc-André, Al-Hamadi Ayoub, Ostermaier Nina, Weimann Joachim
Primary Institution: Otto-von-Guericke University Magdeburg
Hypothesis
Can experimental economics contribute to the development of machine learning algorithms for lie detection?
Conclusion
The study demonstrates that experimental economics can improve the quality of datasets used for training lie detection algorithms, achieving an accuracy rate of 67%.
Supporting Evidence
- The study achieved a lie detection algorithm accuracy of 67%.
- Monitoring did not change the overall lying behavior compared to the original experiment.
- Participants were incentivized to lie, reflecting real-world conditions.
Takeaway
This study shows how scientists can use experiments to help computers learn to tell when someone is lying, making it easier to catch lies in real life.
Methodology
The study modified a classic experiment on lying to collect video data, allowing for individual-level analysis of lying behavior.
Potential Biases
Potential biases in the dataset due to the specific participant selection and the controlled experimental environment.
Limitations
The study's findings may not generalize beyond the specific experimental conditions and participant demographics.
Participant Demographics
148 students from Otto-von-Guericke University Magdeburg, with a focus on excluding psychology students to avoid bias.
Statistical Information
P-Value
p<0.0001
Statistical Significance
p<0.0001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website