People often recognize a blurry object after seeing it clearly just once, a process called one-shot learning, where a single experience changes perception forever. This ability helped early humans spot dangers quickly. A recent study explores this by focusing on perceptual learning, which is the brain's way of improving recognition after one exposure. The work identifies the high-level visual cortex, or HLVC - a part of the brain that processes complex visual information - as the key area where priors, which are memories of past images, are stored and used.
The study used Mooney images, which are faded pictures of animals or objects. Participants viewed blurred versions first, then clear ones, and afterward recognized the blurry ones much better. This showed how priors help fill in missing details. Researchers noted that people with schizophrenia or Parkinson's disease struggle with this, as their priors can overpower current views, causing hallucinations - false perceptions like seeing things that aren't there.
Sharper insights from brain scans
To map this, scientists combined several methods. They used functional magnetic resonance imaging, or fMRI, a scan that tracks blood flow to active brain areas. This revealed activity patterns in the HLVC matching the priors. Behavioral tests changed image sizes, positions, or angles to see what affects learning, finding that priors capture specific patterns but not broader ideas. Electroencephalography, or EEG, recorded brain waves for timing, while intracranial EEG, done during surgery, caught fast signals confirming HLVC's early role.
An artificial intelligence (AI) model, was built to mimic this. It stored priors in one part and used them to recognize images in another, achieving human-like one-shot learning better than other models. This suggests future AI could learn new tasks with little training. The findings offer a theory for testing hallucinations and may connect to "aha" moments in understanding ideas. Overall, it advances knowledge of how the brain blends memory and sight for quick perception.
This research is published in Nature Communications.