The study involved 130 participants, each of whom performed a different mental task, ranging from reading, to memorizing a list, to making complex decisions about whether to take monetary risks, while being scanned using fMRI. The researches were able to identify which of eight tasks participants were involved in with more than 80-percent accuracy by analyzing the participants' fMRI data against classifications developed from the fMRIs of other individuals. The researchers also were able to identify what class of objects (faces, houses, animals, etc.) a person was viewing before he or she could report that information by analyzing the pattern of brain activity at the back of the brain where information is processed and then conveyed towards the frontal regions associated with awareness.
"It's the same principle experienced during a car accident. The car accident actually happens tens of a milliseconds before you are aware you have actually been hit," explains Hanson. "By looking at the back of the brain, we can 'read out,' for example, that a person is looking at dogs and cats before they actually know they are looking at a dog or a cat."
Unlike most research that has focused on specific areas of the brain, Hanson and his team looked at the pattern of activity across a half million points in the brain. Interestingly, the patterns of neural networks involved in each of the eight tasks on the surface appear very similar. The reason, Hanson explains, is that various mental functions tend to draw on many of the same processes. For example, memorizing a list of words that include the word dog is likely to draw up a memory of a pet, the same as reading a story about a dog would. Using machine learning techniques (a support vector machine), capable of analyzing and categorizing large amounts of data, the researchers were able to identify those slight differen
'/>"/>
| Contact: Helen Paxton paxton@andromeda.rutgers.edu 973-353-5262 Rutgers University Source:Eurekalert |