Abstract
The linearity of BOLD responses is a fundamental presumption in most analysis procedures for BOLD fMRI studies. Previous studies have examined the linearity of BOLD signal increments, but less is known about the linearity of BOLD signal decrements. The present study assessed the linearity of both BOLD signal increments and decrements in the human primary visual cortex using a contrast adaptation paradigm. Results showed that both BOLD signal increments and decrements kept linearity to long stimuli (e.g., 3 s, 6 s), yet, deviated from linearity to transient stimuli (e.g., 1 s). Furthermore, a voxel-wise analysis showed that the deviation patterns were different for BOLD signal increments and decrements: while the BOLD signal increments demonstrated a consistent overestimation pattern, the patterns for BOLD signal decrements varied from overestimation to underestimation. Our results suggested that corrections to deviations from linearity of transient responses should consider the different effects of BOLD signal increments and decrements.
Introduction: The Linearity Assumption in fMRI
Most functional MRI (fMRI) studies rely on the assumption that the Blood Oxygen Level Dependent (BOLD) signal behaves as a Linear Time-Invariant (LTI) system. This "superposition principle" implies that the brain's response to a long stimulus can be predicted by summing the responses to shorter constituent pulses. However, previous research has shown that for very short (transient) stimuli, this linear model often fails, typically resulting in an "overestimation" error where the predicted response is larger than the actual measured signal. While this phenomenon is well-documented for signal increments (stimulus onset), significantly less is known about signal decrements (stimulus offset or reduction). If decrements do not follow the same linear rules as increments, current mapping of brain activity—particularly for fast-paced or complex tasks—may be systematically biased.
Methods: Contrast Adaptation and Voxel-Wise Analysis
To investigate these dynamics, researchers employed a contrast adaptation paradigm with 12 healthy participants in a 3T MRI scanner. Unlike traditional designs that compare a stimulus to a blank screen, this study used a steady 25% contrast checkerboard as a continuous baseline.
The experimental setup involved briefly increasing contrast to 100% (increment) or decreasing it to 6.25% (decrement) for durations of 1, 3, or 6 seconds. The researchers performed a high-resolution voxel-wise analysis of 223 individual voxels within the primary visual cortex (V1). Two primary metrics were used: the Dice Index, to measure the similarity between predicted linear responses and actual measurements, and a Contrast Index, to determine whether the model was overestimating or underestimating the real-world signal.
Key Results: Asymmetric Non-linearity in Transient Responses
The study found that BOLD linearity is highly dependent on stimulus duration. For "long" stimuli (3s and 6s), the linear model performed accurately for both increments and decrements. However, for 1-second transient stimuli, significant deviations from linearity emerged.
The most striking finding was the asymmetry between "on" and "off" signals:
- Increment Pattern: For 1s increments, the distribution of the Contrast Index was heavily shifted toward positive values, indicating a consistent overestimation pattern across nearly all voxels.
- Decrement Pattern: For 1s decrements, the Contrast Index distribution was much wider, spanning both negative and positive values. This indicates a mixed pattern of overestimation and underestimation at the voxel level.
These results suggest that the "underestimation" of decrements reported in some previous studies was likely an artifact of averaging across voxels; in reality, different parts of the visual cortex respond to "off" signals in fundamentally different ways.
Discussion and Limitations
The findings demonstrate that standard hemodynamic models need significant updates to account for the complex behavior of signal decrements. The voxel-wise variation in decrement responses may be due to local differences in vascular architecture or specific populations of "off-center" neurons.
While robust, the study is limited by its exclusive focus on the primary visual cortex (V1) and a narrow range of contrast levels. Future research should investigate whether these non-linearities are consistent across higher-order cognitive areas and sensory cortices, and further explore the physiological basis for the observed voxel-wise heterogeneity.
Data and stimulus code for this study are available via the OSF repository.
Summary updated