For my final thesis in electronic engineering I created a portable stress recorder that checked a patient's galvanic skin response, breath rate and heart rate variability.

After a while I realized that my gadget could double as a polygraph (lie detectors rely mostly on the fact that telling a lie will cause some stress, such as fear of being caught).

I'm rather sure that one of the algorithms used in professional polygraphs involves the heart rate variability. You measure the time between two QRS complexes (that's the time between two heartbeats), apply some windowing to the data and then feed the result to a Fourier transform. High-frequency peaks in the resulting graph are related to parasympathetic activity, while low-frequency peaks are linked to the sympathetic system. (VERY LOOSELY speaking, this detects small irregularities in the rhythm. The more stressed you are, the more irregularly your heart beats, creating a high-frequency peak).

I was rather surprised to see that the circuit could tell if I was standing, sitting or lying down - the high frequency peak showed up quite clearly in the first case, and was nonexistant in the third. Another way to test the system was to hook somebody up and make him count down from 1000 in steps of 13, as quickly as possible (a psychology text suggested that, halfway through the count, the examiner should check his watch and comment that all the other test subjects were faster - just to make the poor victim more nervous!)

I'm not saying that lie detectors work; quite the opposite, it's so easy to induce a stress response in your body that the examiner will have a hard time distinguishing them from your lies. I wholly agree with narzo's writeup.


Amusing anecdote: my gadget had a 4-digits hexadecimal display to show its internal status; each digit checked a different subsystem. In my very first test I glued the electrodes to my chest and wrist, pushed the ON button and received the shock of my life when the display cheerfully informed me that I was "DEAD"...

Next time I'll stick to the decimal system.