Eye Tracking — The Best Lie Detector

You are welcome to comment this article on our blog on Medium .


A person can learn to control facial expressions, at least partially, and learn to “dodge” with the help of body language. However, eye movements are more difficult to control сonsciously. In an experiment participants were asked to reproduce their gaze trajectory with the pointer on the screen — and even in such a controlled setting, subjects were not conscious of 42–55% of their eye movements (Marti et al., 2014).

Our eyes can also say a lot when we are trying to lie. That is why eye tracking can be successfully used for lie detection.

Lying is strongly related to cognitive load. According to the theory of cognitive load on deception, a false answer would require more resources (attention and memory) than the truth, since a lie would lead to additional cognitive operations (self-control, more information to be held in working memory, etc.) (Spence et al., 2001; Vrij et al., 2008).

In a concealed information test (CIT), a classic experiment on lie detection, subjects are asked to hide information about some crime that innocent people would not know (Matsuda et al., 2012). For example, to recognize some crime-relevant item (a knife) among other irrelevant items. A polygraph (a lie detector) records physiological responses while the person is giving the answers about whether or not he or she recognizes the items. One general problem though with all polygraph methods is that they detect increases in measures that reflect increased arousal, which are typically interpreted as reflecting guilt or fear, which is not always true (Ganis et al., 2003).

In eye tracking, the subject would be presented with a set of images associated with the “crime”, and neutral images. At the same time, the eye movements will be observed and recorded.

Several experiments have shown that liars’ eye movements differ from those of innocent people. And this is how:

- Dilated pupils, which are associated with the state of increased anxiety and working memory load (Proudfoot et al., 2016).

- The strategy of avoidance, when the subject is trying not to look at the “crime object”, and pays more attention to the neutral objects instead (Proudfoot et al., 2016, Kim et al., 2016).

- Change in the blink rate, when at the of moment when a person is telling a lie, the number of blinks is fewer than in the neural state, but after the person has lied — the number of blinks increases dramatically (Vrij et al., 2008).

- All of these changes can be successfully traced via video-oculographic equipment in the laboratories. The creation of a software eye-tracker, capable of fine tracking of eye movements and blinking, will allow to take these technologies to a whole new level. We at Neurodata Lab are working to create a software eye tracker to detect eye movements with high accuracy and without specific equipment.


- Ganis, G., Kosslyn, S., Stose, S., Thompson, W., Yurgelun-Todd, D. Neural correlates of different types of deception: An fMRI investigation. Cerebral Cortex, vol. 13, issue 8 (2003), pp. 830–836.

- Kim, K., Kim, J., Lee, J. Guilt, Lying, and Attentional Avoidance of Concealed Information. Social Behavior and Personality: an international journal, vol. 44, issue 9 (2016), pp. 1467–1475.

- Marti, S., Bayet, L., Dehaene, S. Subjective report of eye fixations during serial search. Consciousness and Cognition, vol. 33 (2015), pp. 1–15. Published by Academic Press Inc.

- Matsuda, I., Nittono, H., Allen, J. The current and future status of the concealed information test for field use. Frontiers in Psychology, vol. 3 (2012), p. 532.

- Proudfoot, J., Jenkins, J., Burgoon, J., Nunamaker, J. More Than Meets the Eye: How Oculometric Behaviors Evolve Over the Course of Automated Deception Detection Interactions. Journal of Management Information Systems, vol. 33, issue 2 (2016), pp. 332–360. Published by Routledge.

- Spence, S., Farrow, T., Herford, A., Wilkinson, I., Zheng, Y., Woodruff, P. Behavioural and functional anatomical correlates of deception in humans. NeuroReport, vol. 12, issue 13 (2001), pp. 2849–2853. Published by Lippincott Williams and Wilkins.

- Vrij, A., Fisher, R., Mann, S., Leal, S. A cognitive load approach to lie detection. Journal of Investigative Psychology and Offender Profiling, vol. 5, issue 1–2 (2008). pp. 39–43.