Abstract
Rapid event-related designs are an increasingly popular method in fMRI studies of perception. These designs typically produce smaller fMRI response amplitudes than spaced and blocked designs, but whether they preserve the relative amplitudes of responses is unknown. It is also unclear whether the signal reductions arise from nonlinearities in the hemodynamics or in the neural responses. To address these issues, we compared contrast response functions in visual cortex measured with spaced and rapid event-related designs. During scanning, subjects viewed 18 degree contrast reversing checkerboard patterns presented for 500 msec on a gray mean field. Condition ordering was counterbalanced using an m-sequence to optimize efficiency. To control attention, subjects performed a demanding RSVP task at fixation throughout all scans. fMRI signal (BOLD, 3T) was averaged from early visual areas, and linear response kernels were estimated for each condition from the average timeseries. Estimated conditional response amplitudes were then averaged across subjects. In Experiment 1, patterns at three different contrast levels were displayed every 16 seconds (spaced scans), or 3 seconds (rapid scans). For all contrast levels, average responses measured in the 3-second design fell close to those measured in the 16-second design. In Experiment 2, seven patterns of different contrast were displayed every three seconds, or every one second. For all contrasts, average responses measured with the one-second design were smaller than those measured with the three-second design. The difference between the one- and three-second contrast response functions was best characterized by a horizontal shift along the log contrast axis. This result suggests that reducing the time between stimulus presentations from three to one second reduces the contrast gain of those neurons producing the fMRI response, perhaps due to increased masking or contrast adaptation.
Supported by NIH EY 11862, G. Heckman supported by NSF Graduate Research Fellowship