Despite over 40 years of research on Bell-type inequalities and the question of non-locality, new technical results that have general foundational relevance can still be obtained. In this talk will present a number of new results that deal with the question of how to discern local, quantum and no-signaling correlations. • 1) I will present a non-trivial no-signaling inequality that discerns no-signaling correlations from general correlations - the first to our knowledge. This inequality has a striking similarity with the CHSH inequality, yet it is crucially different. • 2) I will next discuss interesting relationships that can be inferred between some well-known conditions at different hidden- variable levels (such as the assumptions of outcome and parameter independence). The upshot of the analysis will be that which conditions are to be obeyed by different kinds of correlations and which are not, depends on the level of consideration. A conclusive picture therefore depends on which hidden-variable level is considered to be fundamental. • 3) I will further comment on interesting relationships that exist between inferences on the surface and subsurface level. Here the surface level deals with experimentally accessible probabilities (e.g., via relative frequencies) and the sub-surface level deals with probabilities that are conditioned on a hidden-variable (or the quantum state). The most interesting such a relationship is the following: any deterministic hidden-variable theory that obeys no- signaling and gives non-local correlations must show randomness at the surface, i.e., the surface probabilities cannot be deterministic. This is the case in Bohmian mechanics but this result shows it to be generic. Throughout the talk I will show how these three topics are related, and comment on the foundational impact of the results obtained.