U.S. researchers say they've developed a computational model of how our brains learn to classify and categorize incoming sensory stimuli, a basic neural process that allow us to make important judgments on a daily basis.

The scientists at New York University have reported their work and its outcome in the journal Nature Communications.

"Categorization is vital for survival, such as distinguishing food from inedible things, as well as for formation of concepts, for instance 'dog vs. cat,' and relationship between concepts, such as hierarchical classification of animals," says study leader Xiao-Jing Wang, a professor of mathematics, neural science and physics.

The initial model was limited to showing how our neural circuits learn and categorize simple, basic visual stimuli, Wang explains.

"Future research is needed to explore if the general principles extracted from this model are applicable to more complex categorizations," he says.

In the researchers' neural-circuit computer model, low-level neural circuits in the brain's cortex passed information from visual stimuli on to higher-level neural circuits where a stimulus feature -- for example, a pattern of moving dots -- gets classified into simple binary categories, dubbed A or B, in this case, which direction the dots are moving -- up or down or left or right.

"Stable category representations develop in neurons intermediate to sensory and decision layers if they exhibit choice-correlated activity fluctuations," the researchers wrote.

The researchers' category-learning computer model was able to create specific predictions of category creation that were confirmed by analysis of single-neuron electrical activity observed in an experiment, they said.

Surprisingly, learning a correct category boundary -- the dividing of a continuous stimulus into A and B -- was seen to require feedback of a top-down nature from higher-level category-selective neurons to the lower-level initial feature-coding neurons, the researchers determined.

Previously, most scientists had believed a choice of category was influenced by more-or-less random activity in initial sensory neurons, traveling over a "bottom-up" pathway, sensory to category.

The new computer model suggests a different hypothesis, that "choice probability" in the brain is a matter of "top-down" category-to-sensory signaling, the researchers say.

It addresses the subject of feedback projections in the brain, the functional significance of which have been a long-standing puzzle, they say.