Thursday, October 7, 2010

Can We Trust Confidence?

As I sink into the couch and duly turn on the TV to watch re-runs of Law and Order (the original, and still the best), I’m never sure whether to root for the prosecution or the defense. I tend to believe the witnesses when choosing whether or not to support the prosecutor’s case. More often than not in these episodes, witnesses have a strong impact on the viewers, and more importantly, on the jury. Even in real life courts, witness testimony is powerful.


But just because one witness is more confident, does that mean he or she is more accurate? In the September 17th, 2010 issue of Science, Fleming et al. report on the possible correlation between confidence and accuracy. Using magnetic resonance imaging (MRI) and controlling for response bias, Fleming et al. were able to characterize how well a subject’s confidence can distinguish between incorrect and correct responses, called a “type 2 performance.” Put another way, a high type 2 performance means that there was a close relationship between confidence and accuracy when subjects had to identify a stimulus.


A type 2 performance, in theory at least, is different from a type 1 performance, which is how accurately a subject identifies a stimulus. A type 2 performance is supposed to correlate with higher level thinking and metacognition, the ability to think about thinking. Type 1 performance represents lower level thinking, such as stimulus processing. Researchers faced the problem that type 2 cognition can encompass type 1 performance. For example, a subject could just be bad at making accurate judgments about stimuli, so correct judgments may result purely from chance. To control for type 1 performance, Fleming et al. programmed a computer to give harder tasks to the better observers and easier tasks to the poorer ones. Structural MRI results show greater signal intensity in the gray matter regions of the prefrontal cortex for high type 2 performers, leading researchers to believe that confidence (remember that this experiment controls for response bias, such as brash or reticent personalities) may result from introspection of the perception process (other cognitive processes).


Results from Fleming et al. bring up controversial ideas about self awareness and uncertainty, in not only humans but higher level cognition mammals, such as monkeys and dolphins. It is still a bit unclear if the confidence results from the strength of the external stimulus, or self reevaluation. Nevertheless, Fleming et al. take us one step closer to understanding our perception of the environment and how we perceive the environment. We should be more careful in how we judge confidence. Although the experiment controlled for personality times, it’s entirely possible that the confident witness in a court or an episode of Law and Order may simply be a bombastic and arrogant person.


Check out the articles here:

http://www.sciencemag.org/cgi/content/full/329/5998/1478

http://www.sciencemag.org/cgi/content/full/329/5998/1541


No comments:

Post a Comment