Your Brain Might Build Mental Images By Turning Neurons Off, Not On

Close your eyes and picture a red apple. Most people can do this without effort, conjuring a vivid mental snapshot of something that isn’t actually in front of them. For decades, brain scientists assumed this trick worked by running the visual system in reverse, with higher brain regions firing up the same neurons that would activate if you were actually seeing the apple. A provocative new paper argues that this long-held assumption is probably wrong, and that the brain may create mental images through a far stranger process: not by switching neurons on, but by selectively silencing them.

Researchers at the University of New South Wales, the University of Technology Sydney, and the University of the Sunshine Coast propose that mental imagery works by reshaping the brain’s background “noise.” Rather than feedback signals from executive brain regions causing neurons in the visual processing area to fire, the authors suggest imagery suppresses the neurons irrelevant to the imagined object while leaving the relevant ones relatively untouched. It’s less like painting a picture on a blank canvas and more like sculpting a figure by chipping away marble.

Published in Psychological Review, the paper pulls together decades of research on brain connectivity, neural recordings, brain imaging, and behavioral experiments rather than presenting new data of its own.

What Happens When the Brain Processes Mental Images

Normal vision works in a forward direction: light hits the eye, signals travel through a chain of brain regions, and early areas process simple features like edges before higher areas handle complex tasks like recognizing a face. Mental imagery runs in reverse. Brain scanning studies show that imagining something starts activity in executive areas near the front of the brain, which then flows backward toward early visual regions through “feedback” connections, signals traveling top-down rather than bottom-up.

Feedback connections outnumber their forward-moving counterparts by about two to one, according to primate studies cited in the paper. Yet despite being more numerous, they are generally thought to be weaker and slower, mainly acting as adjusters rather than initiators. Authors Roger Koenig-Robert, Thomas Pace, and Joel Pearson argue that forward connections cause neurons to fire, while feedback connections adjust the rate of neurons already firing, often by turning the volume down rather than up.

Why Brain Scans May Have Misled Scientists About Mental Imagery

Much of what scientists know about mental imagery comes from fMRI, a brain scanning technique that measures blood flow as a rough stand-in for neural activity. Numerous fMRI studies have shown that when people imagine visual scenes, early visual areas light up, widely interpreted as evidence that imagination fires up the same neurons as real seeing.

According to the paper, this interpretation has a critical flaw. Blood flow signals don’t specifically measure neurons firing; they reflect a mix of electrical activity, including signals that fall below the threshold needed for a neuron to fire at all. Some studies combining fMRI with direct electrode recordings in animals have shown it’s possible for strong blood flow responses to occur even when local neuron firing decreases, given adjusting signals arriving from other brain regions. In short, fMRI shows that something is happening in early visual areas during imagination, but can’t confirm whether neurons are being switched on or tuned down.

What Animal Studies and Human Experiments Suggest

Since scientists can’t ethically implant electrodes in healthy human visual cortex, much of the direct evidence comes from animal experiments. In higher brain regions, feedback can clearly drive neuron firing. Recordings from human epilepsy patients revealed that 88% of neurons in the medial temporal lobe, a memory-related region rather than early visual cortex, maintained the same selectivity whether their preferred image was seen or merely imagined.

In early visual areas, the picture looks different. When forward-moving input is knocked out in animal studies, activity in the next area up the chain drops sharply. Feedback removal, by contrast, tends to have far more limited effects, suggesting its role there is to refine rather than generate. In one attention experiment in monkeys, stimulating a frontal brain area boosted neuron firing in a mid-level visual area only when a stimulus was already present. When no stimulus was there, feedback had little or no effect, consistent with adjustment rather than direct driving.

Behavioral results in humans are mixed. Imagining motion appears to produce adaptation effects in motion-detecting brain areas, suggesting, but not proving, that feedback drives neurons in those mid-level regions. For basic features processed in early visual areas, results pointed more toward suppression than activation.

Brain May Use Suppression, Not Activation, to Build Mental Images

Pulling these threads together, the authors propose that mental imagery works differently depending on where in the brain’s visual chain one looks. In higher regions, feedback likely drives neurons to fire. In early visual areas, it may instead reshape ongoing background activity by suppressing neurons coding for irrelevant features, carving out a representation rather than lighting one up.

That could explain why mental images are almost always less vivid than real perception, and why they feel unstable: background neural activity is inherently variable, so a representation sculpted from it would naturally be more fragile than one driven by direct activation. Whether the brain builds pictures in the mind’s eye by turning lights on or knowing which ones to switch off remains unresolved, but the answer may change how scientists think about imagination, memory, and the brain’s inner life altogether.

---