To study the coupling of the fMRI-BOLD signal and the underlying physiological activity, paradigms are ideal that allow manipulating one of these signals while affecting the others only little. One such example is provided by the motion sensitive area MT of the primate. Neurons in this region respond well to patterns of moving random dots, but responses are much diminished when each dot is paired with a neighbor moving in the opposite direction. This motion opponency is supposedly generated in MT and not present in the synaptic input. As a result, motion opponency should reduce the spiking activity, while leaving the local field potential and BOLD signal much unaltered. We performed simultaneous electrophysiological recordings and fMRI measurements in anaesthetized macaque monkeys and a high-field (7Tesla) environment. We used counterphase sinusoidal gratings as well as locally paired and unpaired dots as stimuli along with their unidirectional counterparts. For both conditions we found dissociations of spiking and fMRI activity during motion opponency, compared to stimulation with unidirectional patterns. All of our stimuli consistently generated positive BOLD responses in the vicinity of the electrode but some of them led to a suppression of spiking activity while others increased firing rates. Primarily grating stimuli were able to reduce the spike rate below baseline while still eliciting positive BOLD responses. This leads us to conclude that in macaque neocortex spike rate changes do not necessarily alter the BOLD response, implying that spikes are not the main cause of this signal. Further our results are compatible with the hypothesis that inhibitory signaling can cause positive BOLD responses.