In a ScienceDaily item (here) there is a report on the nature of the fMRI signal by Finnish and Canadian researchers led be K. Kaila. They show that the signal depends on astrocytes rather than neurons.

Functional magnetic resonance imaging (fMRI) is a technique widely used in studying the human brain. However, it has long been unclear exactly how fMRI signals are generated at brain cell level. This information is crucially important to interpreting these imaging signals. … fMRI imaging does not directly measure the activity of nerve cells or neural networks, but local changes in cerebrovascular circulation during the execution of certain functions. … astrocytes in brain tissue play a key role in generating the fMRI signal. Astrocytes are not nerve cells, but neuronal support or glial cells that are present in the brain in greater abundance than nerve cells. Their signals change with changes in nerve cell activity in a manner that depends on the brain’s metabolic state, and the astrocyte signals thus created regulate the diameter of blood vessels in the brain thereby affecting local circulation…

“For example, it is generally believed that changes in fMRI signals associated with different brain diseases reflect changes in the function of nerve cells and neural networks, even though the explanation might lie in a pathological change in the characteristics and function of astrocytes. Likewise, the distinctive characteristics seen in fMRI signals measured from premature newborns is probably in large part based on the immaturity of astrocyte and blood vessel function,” Kaila explains.

Looking at astrocytes just shows how little we know about how the brain works. Recently Discover Magazine had an article by C. Zimmer on astrocytes (here).

All neurons have certain characteristic attributes: axons, synapses, and the ability to produce electric signals. As scientists peered at bits of brain under their microscopes, though, they encountered other cells that did not fit the profile. When impaled with electrodes, these cells did not produce a crackle of electric pulses. If electricity was the language of thought, then these cells were mute. … Astrocytes—named for their starlike rays, which reach out in all directions—are the most abundant of all glial cells and therefore the most abundant of all the cells in the brain. They are also the most mysterious. A single astrocyte can wrap its rays around more than a million synapses. Astrocytes also fuse to each other, building channels through which molecules can shuttle from cell to cell… All those connections put astrocytes in a great position to influence the goings-on in the brain. They also have receptors that can snag a variety of neurotransmitters, which means that they may be able to eavesdrop on the biochemical chatter going on around them. Yet for a long time, neuroscientists could not find any sign that astrocytes actually responded to signals from the outside. …It turned out that astrocytes, like neurons, can react to neurotransmitters—but instead of electricity, the cells produce waves of charged calcium atoms… wave of such activity that started in one astrocyte could spread to other astrocytes. Several research teams also discovered that astrocytes themselves release powerful neurotransmitters. They can produce glutamate (which excites neurons so that they are more likely to respond to a signal from another neuron) and adenosine (which can blunt a neuron’s sensitivity).

For some brain scientists, these discoveries are puzzle pieces that are slowly fitting together into an exciting new picture of the brain. Piece one: Astrocytes can sense incoming signals. Piece two: They can respond with calcium waves. Piece three: They can produce outputs—neurotransmitters and perhaps even calcium waves that spread to other astrocytes. In other words, they have at least some of the requirements for processing information the way neurons do. …a fourth piece. They find that two different stimulus signals can produce two different patterns of calcium waves (that is, two different responses) in an astrocyte. When they gave astrocytes both signals at once, the waves they produced in the cells was not just the sum of the two patterns. Instead, the astrocytes produced an entirely new pattern in response. That’s what neurons—and computers, for that matter—do.