When a brain region becomes active, a flood of blood arrives within a few hundred milliseconds to service local neurons with the oxygen and glucose they need for energy. Scientists exploit this flow when they use functional magnetic resonance imaging to determine what parts of the brain respond to different stimuli. Recent estimates, however, peg the rush of blood to be nearly 10 times the amount neurons need for metabolism.

Now neuroscientist Christopher I. Moore of the Massachusetts Institute of Technology has proposed a new theory behind the excess flow—the blood, he says, may actually be involved in information processing in the brain. Moore’s “hemo-neural hypothesis” posits several mechanisms for how blood might modulate neuron activity. Molecules in the blood might diffuse into the brain and affect neurotransmitter release, or changes in the volume, pressure or temperature of blood vessels may stress neuronal membranes to regulate transmission. Or there may be a middleman—astrocytes, the nonneuronal supporting cells that surround capillaries in the brain, could secrete chemical signals to neurons in response to a change in blood flow.

Previous research supports Moore’s idea, such as the recent work on Alzheimer’s disease suggesting that vascular decline may precede, and facilitate, neurodegeneration. Further, if blood were to play a tempering role, disruptions in its flow could explain the mechanism behind epilepsy, which can result from overexcited neurons.

Although some in the neuroscience community are dismissive, many believe that a true model of brain processing must include some role for blood. If his hypothesis proved true, Moore says, cerebral blood flow would no longer be thought of simply as a means to investigate brain function. “It would be a Heisenberg sort of thing,” he suggests, referring to the way observing a quantum state changes it, “where what you’re looking at is actually a part of the computation going on.”