The ability to break symmetry and polarize through self-organization is a fundamental feature of cellular systems. A prevailing theory in yeast posits that symmetry breaking occurs via a positive feedback loop, wherein the adaptor protein Bem1 promotes local activation and accumulation of Cdc42 by directly tethering Cdc42GTP with its guanine nucleotide exchange factor (GEF) Cdc24. In this paper, we find that neither Bem1 nor the ability of Bem1 to bind Cdc42GTP is required for cell polarization. Instead, Bem1 functions primarily by boosting GEF activity, a role critical for polarization without actin filaments. In the absence of actin-based transport, polarization of Cdc42 is accomplished through Rdi1, the Cdc42 guanine nucleotide dissociation inhibitor. A mathematical model is constructed describing cell polarization as a product of distinct pathways controlling Cdc42 activation and protein localization. The model predicts a nonmonotonic dependence of cell polarization on the concentration of Rdi1 relative to that of Cdc42.