Joint access point (AP) association and physical carrier sensing (PCS) threshold selection has the potential to improve the performance in high density wireless LANs (WLANs) under high contention, interference and self-interference (SI) limited transmissions. Using tools from stochastic geometry, user and AP locations are independent realizations of spatial point processes. Considering the inherent effects of the channel access protocol, the spatial density of throughput (SDT), which depends on channel access probability and coverage rate, is derived as the performance objective. Leveraging spatial statistics of the network, a throughput-utility maximization problem is formulated to seek AP association and PCS threshold selection policies that jointly maximize SDT. The AP association and the PCS threshold selection policies are derived analytically while an algorithm is proposed for numerical solution. Under simulated scenarios involving full-duplex (FD) nodes, optimizing AP association yields performance gains for low to high node density in large-scale wireless networks. Considering PCS threshold selection optimization jointly with AP association is shown to improve performance by effectively separating concurrent transmissions in space. It is shown that AP association in FD WLANs groups users into minimal contention domains and PCS threshold optimization reduces the interference domain of user groups for additional performance gains.