Donna Rose Addis: Employing brain scans for lie detection just fuzzy logic

Professor Robin Palmer, University of Canterbury's director of Clinical Legal Studies. Photo / Supplied
Professor Robin Palmer, University of Canterbury's director of Clinical Legal Studies. Photo / Supplied

• Donna Rose Addis is a professor of psychology at the University of Auckland.

Can we use brain scanning for lie detection? According to a recent article in the Herald, we can.

We can use scans of brain waves to determine if information pertinent to a legal case is in a suspect's head. If the suspect says otherwise, it can be assumed they are lying.

Professor Robin Palmer of the University of Canterbury calls this "knowledge detection".

In the Brain does not Lie Project, his team will assess how reliable this technology is for gathering information from the brain.

The problem is, this proposal goes beyond the limits of brain scanning technology.

In fact, neuroscientists the world over have rejected the idea that this technology could be used for lie " or knowledge " detection.

Brain scanning cannot tell us what someone knows or doesn't know for one simple reason: a single pattern of brain activity can indicate many different types of thoughts.

Called a "one-to-many mapping", this means a brain scan can be interpreted in a number of ways and is therefore not definitive.

It is also often impossible to get reliable brain scan results from an individual person.

Most of our understanding about brain function is derived from studies on groups of people, yielding averaged patterns of brain activity. In contrast, lie detection focuses on brain activity in one person: the defendant.

Because every brain is different, it is impossible to conclude definitively that an individual's brain patterns reflect the presence or absence of knowledge; perhaps their brain just works differently from the norm.

There are, of course, many contexts in which we want to interpret the brain scans of individual people.

Healthcare is an obvious one, but there our question is different. We are using brain scans to ask why someone has some cognitive or psychological problem. Is a part of their brain not functioning normally? In the health arena, we are not asking brain scans to reveal what someone knows.

There is a very real concern that judges and juries will view brain scans as hard evidence.

With any form of lie/knowledge detection, there exists the possibility that those using the technology for this purpose may be duped. A suspect undergoing this type of brain scanning could evade a positive result by simply "scrambling" their thoughts. They just have to think random thoughts instead of what they are required to. In doing so, they will cloud the results, if not render them useless.

Brain scans can, at best, provide an approximate idea of what someone may be thinking, and then only if the person is compliant. This isn't definitive evidence.

There is a very real concern, however, that judges and juries would view brain scans as hard evidence. It is not obvious to most people that the quality of brain scans are affected by multiple factors, including processing and analysis steps that each affect the final images.

This, combined with the individual characteristics of the person being scanned, means brain scans are open to misinterpretation.

Why do neuroscientists feel so strongly about the issue of using brain scanning for lie/knowledge detection?

Because it represents a misuse of the technology. It pushes brain scanning beyond the boundaries of what it can safely tell us to make conclusions which, if wrong, could have dire consequences.

An innocent person could spend years incarcerated or a criminal could get away scot-free. It is therefore critical that people understand brain scans cannot be used to read minds.

- NZ Herald

Get the news delivered straight to your inbox

Receive the day’s news, sport and entertainment in our daily email newsletter


© Copyright 2016, NZME. Publishing Limited

Assembled by: (static) on production bpcf03 at 27 Oct 2016 12:49:23 Processing Time: 510ms