Abstract The neurodegeneration in the substantia nigra of the basal ganglia that leads to Parkinson's disease (PD) produces a progressive decline in walking function. A major barrier limiting effective walking interventions in people with PD is the occurrence of insufficient motor learning. In the absence of disease modifying options, dopaminergic medications and deep brain stimulation are often used as the disease progresses. Although effective at improving gait, these solutions are temporary, concealing the concurrent degeneration of dopaminergic neurons. As a result, these solutions become less effective at improving gait as the disease progresses and can wear off later in the day. Physical therapy has the potential to increase walking capacity as well as create long-term improvements through intensive training that focuses on motor learning. In particular, the use of rhythmic auditory cues is a well-established intervention that attempts to overcome the damage to internal cueing mechanisms. Despite consistent benefits observed immediately after training, the long-term (retention) effects are more mixed. We contend that reduced retention is a result of inadequate motor learning. Conventional rhythmic auditory cueing employs strategic (explicit) learning with use-dependent learning. To add implicit learning mechanisms to rhythmic auditory cueing, a subtle error in the tempo can be introduced that requires recalibration (i.e., error-based learning). We propose that the addition of frequent movement recalibrations via subconscious temporal distortions of rhythmic auditory cues will repeatedly engage the early phase of implicit learning for individuals with PD to involve intact neural pathways (i.e., hippocampal driven encoding) for improved motor learning and longer term retention of gait improvements. As this training has the potential to alter neural circuitry, we will assess the neural substrates induced by this intervention through multimodal markers of hippocampal and striatal structure and connectivity. Specifically, we will use a randomized controlled design to perform four weeks of gait training using either no metronome, a metronome with a fixed tempo (strategic learning), or a metronome with a slowly changing tempo (strategic learning + error-based learning). In Aim 1, we will assess the gait behavioral changes post-training and at a 3-month follow-up to determine retention. In Aim 2, we will explore structural and functional neural changes induced by the gait training interventions. Our team is exceptionally well prepared to perform these Aims, consisting of experts in gait neurorehabilitation and biomechanics (Lewek), motor learning and neuroimaging (Dayan), and Parkinson's disease clinical care (Browner). At the conclusion of this project, we will have determined the available mechanisms of motor learning for people with PD, and will have examined the neural substrates that can be targeted to maximize remaining intact neural circuitr...