This is the implementation of a social science experiment testing the participants ability to detect liars. The experiment is presented through a web interface and participants are asked if the presented subjects are lying or not.
This experiment is past of a research lead by a group of researchers at EPFL and UNIL.
This version of the experiment presents subjects using images. For the same experiment but using videos instead of images, please refer to the videos branch.
This code provides the interface for several variations of the experiment. Many settings are available such as: including the dots challenge, set a timeout for each question, include one or more breaks, etc... For all the details on the experiment settings, see the [documentation](/Documentation/Constants Definitions.md).
All the users information and results are stored in a database.
The interface is available in English and French.
This experiment is part of a research trying to define the nature of our ability to detect lies. In fact, by observing the participants accuracy and, more importantly, their response time, this experiment tries to define whether our natural ability to detect lies is a central (ie. active, trained, slow) process or a modular (ie. passive, innate, fast) process.
Learn more about the experiment and results on the following report.
To setup the experiment on a server, follow these instructions.
- All content (images or videos) are preloaded and the start of the experiment, ensuring no delay through the actual survey.
- Every subject is presented several times to measure accuracy (typically, 4 repetitions).
- The subjects are presented in a pseudo random order. More specifically, the order is random but yet ensures that no two same subjects follow each other.
- Session tracking: This prevents the users tries to access a page in the wrong order or tries to restart the survey by changing the URL or refreshing the page.
- Add readme for code
- Test images version