The rotation of the Earth is variable in time. Variations in its amplitude are observed as changes in the length of day, whereas changes of the orientation of the rotation vector with respect to the mantle result in polar motion. The changes in the length of day that occur on a timescale of decades are caused by an exchange of angular momentum between the core and mantle. The core motions that participate in this exchange involve waves that are called torsional oscillations. These waves are predicted by theory and can be observed through their effect on the magnetic field. The connection between torsional oscillations and the changes in the length of day has thus confirmed a theoretical aspect of the geodynamo, and these waves provide us with a window through which we can study a part of the dynamics in the Earth's core. In the work that I will present, I attempt to use the observed variations in polar motion to achieve the same goal: to study the dynamics in the core through its manifestation on the orientation of the rotation vector. My effort is focused on one specific component of polar motion known as the Markowitz wobble, an irregular elliptical motion with a typical period of 30 years which remains unexplained. I explore the possibility that the Markowitz wobble is a consequence of torques on the inner core which lead to time-dependent changes in the orientation of its figure axis. These torques must be produced by fluid motions in the vicinity of the inner core. Though we have no direct information on such motions, they can be inferred based on flows near the core-mantle boundary and simple models of the dynamics in the core at decade timescales. By identifying the type of motion and the mechanism for the torque that can successfully reproduce the observed Markowitz wobble, we can discriminate between different models of the dynamics. In this way, decadal changes in polar motion allow us to study indirectly the dynamics taking place at a depth of more than 5000 km.