We propose to deploy the (almost unique) machinery centered around the adjoint model of the MITgcm in conjunction with the ocean state estimates developed within the Estimating the Circulation and Climate of the Ocean (ECCO) consortium in a dedicated effort to elucidate annual to decadal variability of the Atlantic circulation over the period covering the satellite altimetric record (1992 to present). The adjoint model furnishes the complete set of the system’s dual space variables or Lagrange multipliers for any scalar-valued model diagnostic (climate index) considered, and whose time evolution are the transient sensitivities of that target diagnostic with respect to the model state. The availability of state estimates as baseline trajectories for the linearization which have been fit to most of the available observations serves as a strong link between inferred sensitivities and observations. Our aim is to expose dominant mechanisms, time scales, and regions which influence key indices of oceanic variability. The approach is centered around the Atlantic meridional overturning circulation (MOC), but considered more broadly as a complex three-dimensional system of horizontal and vertical flows and mixing processes, with its associated meridional transports of heat and freshwater. A hierarchy of proposed work will include (1) dynamical interpretation of the transient dual variables in terms of anomaly propagation (Kelvin and Rossby waves, advection, conversion processes at the boundaries, etc); (2) through conversion from sensitivities to mormalized responses, establish a hierarchy of variables that dominate the indices’ variability, with emphasis on surface forcing (responses to wind stress vs. heat flux), interior processes (anomaly propagation at depth), and parameterized mixing variability; (3) in conjunction with (2), establish dominant time scales and spatial patterns (main regions of influence) affecting the variability; (4) correlation of adjoint sensitivities for different indices (e.g. transient structure of mass transport vs. heat transport vs. heat content sensitivities). The dual space analyses will be complemented with time-varying budget diagnostics of the state estimates to understand the link between heat and freshwater content variability and their lateral transports, as well as air-sea fluxes. The exact dynamic and kinematic consistency render the ECCO state estimates particularly suitable for this task. Special emphasis will be put on the role of vertical processes such as convection, downwelling, subduction, and vertical mixing. The proposed work items, put together, should improve our understanding of the nature and causes of low frequency memory residing in Atlantic circulation variability as a result of ocean dynamics, with consequences for potential predictability.