So far there is no instrument to measure the “thoughts” of the biological brain. Neuroscientists are interested in studying how the brain thinks, and in taking simulated neuronal judgments about conditions to a new level.
Researchers at Baylor College of Medicine and Rice University have used artificial intelligence models to open up a new way of thinking about brain activity.
“For centuries, neuroscientists have studied the way the brain works by relating brain activity to information about inputs and outputs. To study the neuroscience of movement, for example, scientists first measure muscle movement and neuronal activity in the brain, and then link the two to find patterns.”
Xaq Pitkow, an assistant professor of neuroscience at Baylor College of Medicine, one of the lead authors of this study, said, “But to study the brain’s consciousness, we couldn’t find anything to compare with neural activity.”
To study the process by which the brain generates ideas, the researchers thought they would first have to measure an idea. But how do you measure an idea when you can’t see or touch it? They developed a system of backward reasoning, inferring backward from the behavior to the idea that most likely led to that behavior.
For the other part of this study, the researchers developed another trained artificial intelligence model that then took the brain activity that would be displayed by the idea obtained from the previous part of reverse inference and compared it to the observed brain activity.
“We took the ideas obtained from the inference and compared them to the brain activity that showed those ideas, and if they matched, we had a handle on the inference process by which the brain arrived at those ideas.”
In one experiment, researchers asked an animal to tell whether a fruit inside a box was raw or ripe. In fact, all of the fruit was immature, but this model of the system inferred from the animal’s behavior that the animal’s thoughts were that some of these fruits were ripe and some were immature.
Pitko gave another example that previously, scientists thought that the animal’s brain always takes the best solution to accomplish the task. But this study, through this new way of studying, found that it does not always have to be the case.
“Sometimes animals make judgments about their surroundings that are ‘wrong,’ and based on that judgment, they go about thinking about the best course of action. Maybe that’s why we see animals sometimes behaving in a suboptimal way.” Pitko said.
For example, Pitko said, animals are important in hunting when the judgment of the surrounding sound information. If they judge that these sounds are coming from the same prey, then the best course of action decided by the brain should be to focus all actions on the source of this sound.
If this predator incorrectly believes that these sounds come from different prey, this leads them to adopt a different behavioral scheme, a suboptimal scheme – such as constantly scanning around to identify one of the prey.
The study therefore concluded that the animals were making “rational” but seemingly “suboptimal” behavioral plans based on their judgment. The researchers are confident that their model has indeed reasoned out the “thoughts” of the animal’s brain.
The study was recently published in the Proceedings of the National Academy of Sciences of the United States of America (PNAS).
Recent Comments