A new method for estimating population receptive field topography in visual cortex

We introduce a new method for measuring visual population receptive fields (pRF) with functional magnetic resonance imaging (fMRI). The pRF structure is modeled as a set of weights that can be estimated by solving a linear model that predicts the Blood Oxygen Level-Dependent (BOLD) signal using the stimulus protocol and the canonical hemodynamic response function. This method does not make a priori assumptions about the specific pRF shape and is therefore a useful tool for uncovering the underlying pRF structure at different spatial locations in an unbiased way. We show that our method is more accurate than a previously described method (Dumoulin and Wandell, 2008) which directly fits a 2-dimensional isotropic Gaussian pRF model to predict the fMRI time-series. We demonstrate that direct-fit models do not fully capture the actual pRF shape, and can be prone to pRF center mislocalization when the pRF is located near the border of the stimulus space. A quantitative comparison demonstrates that our method outperforms the direct-fit methods in the pRF center modeling by achieving higher explained variance of the BOLD signal. This was true for direct-fit isotropic Gaussian, anisotropic Gaussian, and difference of isotropic Gaussians model. Importantly, our model is also capable of exploring a variety of pRF properties such as surround suppression, receptive field center elongation, orientation, location and size. Additionally, the proposed method is particularly attractive for monitoring pRF properties in the visual areas of subjects with lesions of the visual pathways, where it is difficult to anticipate what shape the reorganized pRF might take. Finally, the method proposed here is more efficient in computation time than direct-fit methods, which need to search for a set of parameters in an extremely large searching space. Instead, this method uses the pRF topography to constrain the space that needs to be searched for the subsequent modeling. (C) 2013 Elsevier Inc. All rights reserved.