The human brain is constantly processing data for statistical evaluations. According to a study by US neuroscientists, human sense of confidence, seen as subjective, it is based on similar calculations that a computer makes.
US – The self confidence is based on statistical calculations performed by the brain just as it is done by a computer, says a study conducted at Cold Spring Harbor Laboratory (New York) and published in the journal Neuron.
Adam Kepecs, research professor of neuroscience at the center and lead author has developed a model of confidence, but its future goal is to find out where this internal statistical system is in the brain and how it makes data processing.
The feeling of self-confidence often related to important issues such as choosing a career or financial investment decisions. And also guide the stock options every day, for example, to turn while driving.
“Every time we make a decision, we need confidence. Without a precise mechanism that would allow us trust our acts, it would be very difficult to choose or evaluate the different choices. Human trust is a feeling, but there is also a scientific concept which is based on statistical methods that calculate the certainty of a hypothesis,” he adds.
To calculate statistical confidence involves studying a set of data and reach a conclusion. How does our brain do it?
Previous studies on this topic had concluded that largely the sense of confidence comes from the approach and trial and error, and that is very objective and prone to errors. But if confidence tended to err, simple tasks such as the decision to make a turn while driving would be difficult.
In order to determine whether the human feeling of trust is built by calculations, Kepecs, in collaboration with his colleague Joshua Sanders, created a series of video games to compare the human brain and computer performance.
They played sounds of clicks to volunteers and asked them to determine which ones were faster. Then participants rated the reliability of each option on a scale of one to five. The team found that human responses were similar to computer capable of extracting patterns from statistical data calculations.
The model Kepecs on human trust maintained its consistency in tracking another experiment in which participants responded to questions comparing populations from different countries. Unlike perception test, this had the added complexity of the individual knowledge base of each volunteer.
Even human weaknesses, like being overconfident in difficult decisions with limited data, or be undecided easy questions, were consistent with the model. “This subjective feeling of confidence is based on statistical calculations, it is not something heuristic or a shortcut,” says the author.
Kepecs plans to use the model as a starting point to search the headquarters of confidence in the brain and understand their system of neural circuits. “To have a theory about this feeling is a need to find out how the brain manages first step and the way nerve cells perform this process.”
The work could also have implications in fields such as statistics and, in particular, in machine learning. “Humans are even better than computers to solve complex problems,” concluded the neuroscientist.