Estimation of the temporal information encoded in the observed hemodynamic response in functional MRI (fMRI) is of great interest. One assumption that most of the current studies make is that the mean waveform observed is consistently locked with the stimulus and variably distributed about its mean. Because the noise is high any violations of this assumption are not easily observed. We have developed a method for filtering and tracking the hemodynamic response using a recursive least squares (RLS) algorithm and probabilistic shift maps (psm). Our initial results clearly demonstrate an overall change in the amplitude from greater to lesser and a latency shift from shorter to longer in primary visual cortex. These changes may indicate fatigue or adaptation of the neuronal patterns or blood supply. The fact that we observe a change in primary visual cortex is notable since this suggests that changes in brain regions which are specialized for higher congnitive functions may exhibit even larger changes. This suggests the importance of determining the degree to which the measured brain region adapts to the paradigm presented in any fMRI experiment.