Amorphous solids -- solids that lack long-range order of their constituent particles -- are common in both nature and industry. Window glass, dense polymers, and food grains are three examples of amorphous solids familiar to us. In many amorphous solids, shear banding -- plastic deformation in which strain is accumulated in a thin band of the material -- is common. Consequently, many amorphous solids are brittle, a trait which has limited the technological adoption of otherwise promising materials such as metallic glasses. Therefore, a fundamental understanding of shear banding -- i.e., the progression from particle level plastic events to a macroscopic shear band, identification of the sites in the material from which shear banding is most likely to originate, the effect of structural modifications on shear banding, and mechanisms that arrest shear band operation before failure -- is crucial for predicting failure and engineering ductility in amorphous materials. This dissertation describes efforts to illuminate elements of plasticity in amorphous solids using model systems of colloidal particles. The bulk of the results focuses on colloidal pillars subjected to uniaxial compression. Results from instrumented compression experiments reveal that the pillars exhibit a scaling of strength with stiffness that is similar to the scaling found in metallic glasses, which we interpret in the context of the energetics and kinematics of a critical shear band nucleus. In 4D \emph{in-situ} compression experiments we are able to observe the microscopic evolution of a shear band and the associated mechanical response in and around the shear band. The results from this experiment lend credence to the interpretation of shear banding as localized, anisotropic glass transition. In addition to the pillar geometry, we perform confined compression experiments on a confined colloidal glass to investigate the structural fingerprints of the particles that are most likely to rearrange in an amorphous solid. The results from these experiments are interpreted in the context of a recently introduced machine-learning based approach to the identification of particles most susceptible to rearrangement termed "softness". We report preliminary application of softness to the shear banding pillars.