Functional brain imaging — most commonly, positron emission tomography (PET) or functional magnetic resonance imaging (fMRI) — has become an important tool for many neuroscientists. But, despite recent advances, it is still far from clear exactly how the changes in cerebral blood flow and oxygen metabolism that these techniques measure are related to neuronal activity. Caesar et al. have developed a new model in which to investigate this relationship, and they find that it depends largely on context.

The authors used laser Doppler flowmetry to measure local cerebral blood flow while they stimulated one or both of two inputs to cerebellar Purkinje cells. One input, carried by the climbing fibres, is excitatory, whereas the other, carried by the parallel fibres, is inhibitory. When either of these inputs was stimulated, the authors measured a frequency-dependent increase in cerebral blood flow that seemed to be coupled with an increase in the postsynaptic element of the local field potential. This is consistent with previous work by Logothetis and colleagues, in which simultaneous fMRI and neurophysiological recordings showed that the fMRI signal correlated more closely with the local field potential than with spiking activity.

On the basis of a theoretical model, Caesar et al. hypothesized that imposing an inhibitory input onto an existing excitatory input would reduce the ongoing increase in cerebral blood flow, compared with that produced by the excitatory signal alone. However, when they carried out this experiment — stimulating the climbing fibres and then the parallel fibres — they found that the inhibitory input caused the cerebral blood flow to increase further. The same thing happened when they reversed the order of stimulation so that an excitatory input was added to an existing inhibitory one. However, in both cases, the combined stimulation produced an increase in cerebral blood flow that was less than the sum of the increase produced by the two inputs when each was stimulated alone.

So although the results did not agree with the authors' hypothesis, they did show that the change in cerebral blood flow caused by stimulation of these inputs depends on the context of that stimulation. It is possible that this effect results from a refractory period that restricts the Purkinje cells' response to the additional input and is related to the electrical response properties of these neurons.

As Kim points out in a related article, this work helps to answer some unresolved questions in functional imaging. For example, the careful analysis demonstrates that it is possible to infer changes in postsynaptic field potential from haemodynamic signals, even though the latter are much slower than the former. Step by step, we are coming closer to a full understanding of just what functional imaging can tell us about brain activity.