Paradigm-dependent modulation of event-related fMRI activity evoked by the oddball task

Abstract

We have previously shown that event-related functional magnetic resonance imaging (ER-fMRI) may be used to record responses to the rapid, interleaved presentation of stimuli in the three-stimulus oddball task. The present study examined the sensitivity of ER-fMRI responses to variations in the range of inter-stimulus intervals (ISIs, calculated as the time from the offset of one stimulus to the onset of the next stimulus) and the type of behavioral response task used. ISIs were varied between a wide ISI range (550-2,050 msec) and a narrow ISI range (800-1,200 msec), while maintaining a similar mean ISI (approximately 1 stimulus per sec) between experiments. The response task was varied between button press and subvocal target counting. Gradient echo, echo planar images were acquired for each of three experiments (wide ISI with button press, narrow ISI with button press, and wide-ISI with counting) in five subjects. Target stimuli generated increased fMRI signal in a wide range of brain regions. The use of a narrow ISI range generated a greater volume of subcortical activity and a reduced volume of cortical activity relative to a wide ISI range. The counting task generated a larger amplitude and longer lasting evoked response in brain regions that responded during all three experiments. Rare distractor stimuli evoked fMRI signal change primarily in orbitofrontal, ventral-medial prefrontal and superior parietal cortex. These results illustrate that although ER-fMRI is relatively insensitive as a technique to small variations in the timing of stimulus-evoked responses, it is remarkably sensitive to consequences such variations have for the topographic location and amplitude of neural responses to stimuli.

Figure 1

Example regressors used in this experiment. (a) Modeled hemodynamic response time series (black lines) shown for the target stimuli for one representative run. Approximate time of presentation of each target stimulus in this run shown as the letter “T” placed below the abscissa. Main effect regressor (dotted line) modeled average stimulus response magnitude. Linear trend regressor (solid line) modeled response amplitude as dependent upon the number of prior stimuli presented in this run, as shown. Responses overlapping in time were modeled as adding in a linear manner, explaining the large early response in both regressors. Similar regressors were used for distractor stimuli, following the time course of distractor stimulus presentation. (b) Two examples of weighted linear sums of main effect and linear trend regressors. Dotted line shows predominant main effect regressor with some diminution of response modeled by the addition of a negative linear trend regressor. This combination models a predominant main effect with a small reduction over stimulus repetitions due to habituation. Solid line shows linear trend regressor inverted, with no contribution from main effect regressor. This negative to positive pattern of response was found to predominate for the distractor stimulus in orbitofrontal cortex.

Figure 2

Regions responding to target stimuli in all experiments. (a) Regions that responded to target stimuli with a significant main effect of increased evoked signal intensity in all three experiments. Legend shows relative activation levels compared across the three tasks for this and all following figures. As illustrated in the legend in lower right of figure, regions that responded more during Experiment 1 (button press, wide‐ISI range) shown in red, regions that responded more during Experiment 2 (counting, wide‐ISI range) shown in green, regions responding more in Experiment 3 (button press, narrow ISI range) shown in blue. Regions responding to multiple tasks shown as a linear combination of red, green or blue depending on relative Z score value obtained for each. Spatially normalized statistical data averaged over all 5 subjects were plotted onto the normalized structural scan of one subject. Slices moving from left to right going from more inferior to more superior in the brain. (b) Percentage change in signal intensity in response to target stimuli relative to baseline, defined as the mean of the −2.15 sec and 0 sec time points. Data averaged across all voxels that responded with significant signal increases to all three tasks. Response for Experiment 1 (button press, wide‐ISI range) shown in thick red, Experiment 2 (counting, wide‐ISI range) shown in thin green, and Experiment 3 (button press, narrow ISI range) shown in dashed blue. (c) Regions that responded to target stimuli with a significant decrease in evoked signal intensity in all three tasks. Legend as in (a). (d) Percentage change in signal intensity in response to target stimuli averaged across all voxels that responded to target stimuli with significant signal decrease in all three tasks. (e) Regions with significant negative linear trend response across repetitions of target stimuli for all three experiments.

Figure 3

Comparisons among experiments. (a) Regions that responded to target stimuli with a significant main effect in both wide‐ISI experiments (1 and 2) but not in the narrow‐ISI experiment (3). (b) Regions that responded to target stimuli during the narrow‐ISI experiment, but not during the wide‐ISI experiments. (c) Regions that responded during the button press experiments (1 and 3), but not during the counting task experiment (2). The somato‐motor strip was primarily activated. (d) Regions that responded during the counting task, but not during the button press task. (e) Regions that responded to rare distractor stimuli with a significant main effect in both wide‐ISI experiments (1 and 2) but not in the narrow‐ISI Experiment (3). (f) Regions that responded to target stimuli during the narrow‐ISI experiment, but not during the wide‐ISI experiments. (g) Shows regions that responded to the distractor stimuli with a significant linear trend in the counting experiment (2) alone. (h) Regions that responded to the distractor stimuli with a significant linear trend in both button press experiments (1 and 3). (i) Percentage change for each distractor stimulus in sequence at 6.45 sec latency relative to mean of pre‐stimulus baseline for medial prefrontal ROI shown in 3g. Solid line shows best‐fit linear regression of signal amplitude across repetitions.