David Linhardt¹ (), Siddharth Mittal¹, Michael Woletz¹, Samuel Schwarzkopf², Christian Windischberger¹; ¹Medical University of Vienna, ²University of Auckland

Population receptive field (pRF) mapping is widely used to quantify visual field representations, yet the degree to which reproducibility depends on specific stimulus properties remains unclear. Here we systematically tested the impact of stimulus sequence design and temporal sampling on pRF test–retest reliability. We re-analyzed the CHN retinotopy dataset (N=12; two sessions, three runs per session; Chang et al., 2025; openneuro.org/datasets/ds004698). pRF parameter estimations were obtained based on the 2D Gaussian model in GEM-pRF (Mittal et al., 2025), using two modelling pipelines. In the standard pipeline, the stimulus was downsampled to the TR (1.2 s) before HRF convolution. In the high-resolution pipeline, HRF convolution was performed at the stimulus-native temporal resolution, and only the final model time course was downsampled. Each run used an otherwise identical, smoothly moving bar stimulus that differed only in the pseudo-randomized starting angles for each crossing. Order of the three-bar presentations was identical across sessions and participants. Reliability was assessed via within-subject correlations between session-matched runs. High-resolution modelling produced a statistically reliable improvement in eccentricity and size estimates (p < 0.001). The absolute effect was small (Delta r ~ 0.01), given the already high baseline correlations. In contrast, the choice of starting angle had a much larger impact. Despite matched duration, number of peaks and stimulus energy, median reliability varied markedly across stimulus configurations, by almost an order of magnitude more than the effect of high temporal resolution modelling. Together, these results show that while high-resolution modelling yields a measurable benefit, stimulus sequence design is a stronger factor for pRF reliability. Careful control of stimulus conditions offers a simple way to improve reproducibility and is straightforward to implement in studies aiming for clinical use.

Acknowledgements: Research was supported by the Austrian Science Fund (FWF) [10.55776/P35583, 10.55776/PAT8722623]