The upcoming Laser Interferometer Space Antenna (LISA) mission will detect and prove low-frequency gravitational-waves. The milliHertz frequency band to which LISA is sensitive will witness a rich symphony of gravitational-wave signals, with sources ranging from galactic white-dwarf binaries to extreme mass ratio inspirals. Many of the weaker signals from these sources, being individually unresolvable, will overlap and incoherently add with each other producing a stochastic confusion noise. While the frequency content of this noise is governed by the dynamics of the binary inspirals, its angular dependence will be modulated by the distribution of the sources across the sky. Measuring the distribution of the confusion noise on the sky can give us new tools to probe its astrophysical or cosmological history. Besides, a cosmological stochastic gravitational-wave background is also possible whose angular power spectrum, if detected, will be a unique probe of early universe physics. In this talk, I will give a brief overview of the LISA mission and describe its sensitivity to the stochastic gravitational-wave confusion noise. I will then describe a Bayesian technique that we developed to detect and probe the anisotropies of the confusion noise using LISA. I describe a novel decomposition based on Clebsch-Gordan coefficients to show how a spherical harmonic basis can be used to study the anisotropies while constraining the power in all spatial directions to be non-negative. I demonstrate this with an end-to-end pipeline for analyses with LISA, using a series of data simulations and analyses rooted in Bayesian inference.