Blood oxygenation level-dependent (BOLD) fMRI signals often exhibit pronoun
ced over- or undershoot upon changes in stimulation state. Current models p
ostulate that this is due to the delayed onset or decay of perfusion-depend
ent attenuating responses such as increased cerebral blood volume or oxygen
consumption, which are presumed to lag behind the rapid adjustment of bloo
d flow rate to a new steady-state level. If this view is correct, then BOLD
overshoot amplitudes in a specific tissue volume should be correlated with
steady-state increases in perfusion, independent of stimulus type. To test
this prediction, we simultaneously recorded BOLD and relative perfusion si
gnals in primary visual cortex while inducing graded perfusion increases wi
th three types of visual stimulus. Two of these, a diffuse chromatic stimul
us with no luminance variation and a very high spatial frequency luminance
grating, did not produce detectable BOLD overshoot (or undershoot) when an
equal mean luminance baseline was used. Radial checkerboard stimuli, howeve
r, caused pronounced over/undershoot of both BOLD and perfusion signals eve
n when temporal mean luminance was held constant and stimulus contrast was
adjusted to produce the same steady-state blood flow increases evoked by th
e other stimuli. Transient amplitudes were relatively invariant in spite of
large changes in steady-state response, demonstrating nonlinear BOLD and p
erfusion step responses in human V1, These findings suggest that, rather th
an a purely tissue-specific biomechanical or metabolic phenomenon, BOLD ove
rshoot and undershoot represent transient features in the perfusion signal
whose effects may be amplified by slowly evolving blood volume changes. (C)
1999 Academic Press.