Marion Forano
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Marion Forano has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- David W. Franklin
Technical University of Munich, Germany
6 shared publications Alin Albu-Schäffer
Affiliations Sensor Based Robotic Systems and Intelligent Assistance Systems, TUM School of Computation, Information and Technology, Technical University of Munich (TUM), Garching, Germany, Institute of Robotics and Mechatronics, German Aerospace Center (DLR), Wessling, Germany, Munich Institute of Robotics and Machine Intelligence (MIRMI), Technical University of Munich (TUM), Munich, Germany
2 shared publications- Annika Schmidt
Affiliations Sensor Based Robotic Systems and Intelligent Assistance Systems, TUM School of Computation, Information and Technology, Technical University of Munich (TUM), Garching, Germany, Institute of Robotics and Mechatronics, German Aerospace Center (DLR), Wessling, Germany, Munich Institute of Robotics and Machine Intelligence (MIRMI), Technical University of Munich (TUM), Munich, Germany
2 shared publications - Arne Sachtler
Technical University of Munich (TUM) Munich Germany
2 shared publications Davide Calzolari
Affiliations Sensor Based Robotic Systems and Intelligent Assistance Systems, TUM School of Computation, Information and Technology, Technical University of Munich (TUM), Garching, Germany, Institute of Robotics and Mechatronics, German Aerospace Center (DLR), Wessling, Germany
2 shared publications- Jordan A. Taylor
Department of Psychology, Princeton University, Princeton, NJ, USA
2 shared publications
Latest Publications
- Open Access
- 3 September 2024
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 20 (9) , e1011478
Finding the rhythm: Humans exploit nonlinear intrinsic dynamics of compliant systems in periodic interaction tasks
Activities like ball bouncing and trampoline jumping showcase the human ability to intuitively tune to system dynamics and excite motions that the system prefers intrinsically. This human sensitivity to resonance has been experimentally supported for interactions with simple linear systems but remains a challenge to validate in more complex scenarios where nonlinear dynamics cannot be predicted analytically. However, it has been found that many nonlinear systems exhibit periodic orbits similar to the eigenmodes of linear systems. These nonlinear normal modes (NNM) are computable with a recently developed numerical mode tool. Using this tool, the present resarch compared the motions that humans excite in nonlinear systems with the predicted NNM of the energy-conservative systems. In a user study consisting of three experiment parts, participants commanded differently configured virtual double pendula with joint compliance through a haptic joystick. The task was to alternately hit two targets, which were either aligned with the NNM (Experiments 1 and 2) or purposefully arranged offset (Experiment 3). In all tested experiment variations, participants intuitively applied a control strategy that excited the resonance and stabilized an orbit close to the ideal NNM of the conservative systems. Even for increased task accuracy (Experiment 2) and targets located away from the NNM (Experiment 3), participants could successfully accomplish the task, likely by adjusting their arm stiffness to alter the system dynamics to better align the resonant motions to the task. Consequently, our experiments extend the existing research on human resonance sensitivity with data-based evidence to nonlinear systems. Our findings emphasize the human capabilities to apply control strategies to excite and exploit resonant motions in dynamic object interactions, including possibly shaping the dynamics through changes in muscle stiffness.
- 3 September 2024
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 20 (9) , e1011478
Open AccessFinding the rhythm: Humans exploit nonlinear intrinsic dynamics of compliant systems in periodic interaction tasks
Activities like ball bouncing and trampoline jumping showcase the human ability to intuitively tune to system dynamics and excite motions that the system prefers intrinsically. This human sensitivity to resonance has been experimentally supported for interactions with simple linear systems but remains a challenge to validate in more complex scenarios where nonlinear dynamics cannot be predicted analytically. However, it has been found that many nonlinear systems exhibit periodic orbits similar to the eigenmodes of linear systems. These nonlinear normal modes (NNM) are computable with a recently developed numerical mode tool. Using this tool, the present resarch compared the motions that humans excite in nonlinear systems with the predicted NNM of the energy-conservative systems. In a user study consisting of three experiment parts, participants commanded differently configured virtual double pendula with joint compliance through a haptic joystick. The task was to alternately hit two targets, which were either aligned with the NNM (Experiments 1 and 2) or purposefully arranged offset (Experiment 3). In all tested experiment variations, participants intuitively applied a control strategy that excited the resonance and stabilized an orbit close to the ideal NNM of the conservative systems. Even for increased task accuracy (Experiment 2) and targets located away from the NNM (Experiment 3), participants could successfully accomplish the task, likely by adjusting their arm stiffness to alter the system dynamics to better align the resonant motions to the task. Consequently, our experiments extend the existing research on human resonance sensitivity with data-based evidence to nonlinear systems. Our findings emphasize the human capabilities to apply control strategies to excite and exploit resonant motions in dynamic object interactions, including possibly shaping the dynamics through changes in muscle stiffness.
- 1 July 2024
- Published by American Physiological Society in Journal of Neurophysiology
- Vol. 132 (1) , 1-22
Reward actively engages both implicit and explicit components in dual force field adaptation
Here we show that rewarding participants’ performance during dual force field adaptation primarily affects the initial rate of learning and the early timescales of adaptation, with little effect on the final adaptation level. However, reward affects both explicit and implicit components of adaptation. Whereas the learning rate of the slow process is increased implicitly, the fast learning and retention rates are increased through both implicit components and the use of explicit strategies.
- 1 July 2024
- Published by American Physiological Society in Journal of Neurophysiology
- Vol. 132 (1) , 1-22
Reward actively engages both implicit and explicit components in dual force field adaptation
Here we show that rewarding participants’ performance during dual force field adaptation primarily affects the initial rate of learning and the early timescales of adaptation, with little effect on the final adaptation level. However, reward affects both explicit and implicit components of adaptation. Whereas the learning rate of the slow process is increased implicitly, the fast learning and retention rates are increased through both implicit components and the use of explicit strategies.
- Open Access
Finding the rhythm: Humans exploit nonlinear intrinsic dynamics of compliant systems in periodic interaction tasks
Everyday activities, like jumping on a trampoline or using a swing-stick, show that humans seemingly effortless support systems in their intrinsically preferred motions. Although this observation seems obvious, data-based evidence proving that humans indeed match system dynamics has been lacking, since everyday objects usually exhibit complex, nonlinear dynamics, which are in general not analytically solvable. Recent insights in the field of nonlinear mode theory and the development of a tool to compute modes for nonlinear systems enabled us to investigate human strategies to excite periodic motions in the interaction with nonlinear systems. In the setup of a high score game, participants interacted with differently configured virtual compliant double pendulum systems through a haptic joystick. Through the joystick, the user could command positions to a motor link connected to the pendulum by a spring and received resulting spring forces in return to convey the feeling of holding a flexible stick. The participants were asked to alternately hit two targets located on the computed nonlinear mode of the system as often as possible. All participants intuitively exploited the elasticity of the system by choosing a holding strategy of the motor link and only compensate for energy losses with small motions. In this way, the intrinsic dynamics of the double pendulum system were exploited leading to the predicted fast motions along the nonlinear modes. The human strategy stayed consistent when decreasing the target size or increasing the mass of the lower pendulum link, i.e., changing the dynamics. Consequently, the presented research provides data-based evidence that humans can indeed estimate the nonlinear dynamics of system and intuitively exploit these. Additionally, the introduction to nonlinear modes and ways to compute them could be a powerful tool for further investigations on human capabilities and strategies in periodic interactions with nonlinear systems. Author summary: Without thinking about it, humans interact with a wide variety of objects in everyday life. This includes objects with very complex nonlinear dynamics such as flexible rods or ropes. Since it is not trivial to enforce trajectories far away from the system’s intrinsic motions and frequencies, it is likely that humans explore and, whenever possible, exploit the natural dynamics of the system. By using a tool to predict the trajectories of systems with nonlinear dynamics, we collected human user data to validate this hypothesis for repetitive tasks with a virtual double pendulum. Indeed our research showed that humans supported mechanical systems in their respective intrinsic motions and were able to intuitively match the systems’ eigenfrequencies. In doing so, only little control effort and motion was needed from the users, which could aids to save energy and mental resources. Since both these aspects are limiting factors in continuous tasks, it seems to have an evolutionary benefit that humans are very capable in estimating and exploiting the natural dynamics of compliant systems and tune their own control strategy to be be synchronized to the controlled system.
Open AccessFinding the rhythm: Humans exploit nonlinear intrinsic dynamics of compliant systems in periodic interaction tasks
Everyday activities, like jumping on a trampoline or using a swing-stick, show that humans seemingly effortless support systems in their intrinsically preferred motions. Although this observation seems obvious, data-based evidence proving that humans indeed match system dynamics has been lacking, since everyday objects usually exhibit complex, nonlinear dynamics, which are in general not analytically solvable. Recent insights in the field of nonlinear mode theory and the development of a tool to compute modes for nonlinear systems enabled us to investigate human strategies to excite periodic motions in the interaction with nonlinear systems. In the setup of a high score game, participants interacted with differently configured virtual compliant double pendulum systems through a haptic joystick. Through the joystick, the user could command positions to a motor link connected to the pendulum by a spring and received resulting spring forces in return to convey the feeling of holding a flexible stick. The participants were asked to alternately hit two targets located on the computed nonlinear mode of the system as often as possible. All participants intuitively exploited the elasticity of the system by choosing a holding strategy of the motor link and only compensate for energy losses with small motions. In this way, the intrinsic dynamics of the double pendulum system were exploited leading to the predicted fast motions along the nonlinear modes. The human strategy stayed consistent when decreasing the target size or increasing the mass of the lower pendulum link, i.e., changing the dynamics. Consequently, the presented research provides data-based evidence that humans can indeed estimate the nonlinear dynamics of system and intuitively exploit these. Additionally, the introduction to nonlinear modes and ways to compute them could be a powerful tool for further investigations on human capabilities and strategies in periodic interactions with nonlinear systems. Author summary: Without thinking about it, humans interact with a wide variety of objects in everyday life. This includes objects with very complex nonlinear dynamics such as flexible rods or ropes. Since it is not trivial to enforce trajectories far away from the system’s intrinsic motions and frequencies, it is likely that humans explore and, whenever possible, exploit the natural dynamics of the system. By using a tool to predict the trajectories of systems with nonlinear dynamics, we collected human user data to validate this hypothesis for repetitive tasks with a virtual double pendulum. Indeed our research showed that humans supported mechanical systems in their respective intrinsic motions and were able to intuitively match the systems’ eigenfrequencies. In doing so, only little control effort and motion was needed from the users, which could aids to save energy and mental resources. Since both these aspects are limiting factors in continuous tasks, it seems to have an evolutionary benefit that humans are very capable in estimating and exploiting the natural dynamics of compliant systems and tune their own control strategy to be be synchronized to the controlled system.
- Open Access
Reward actively engages both implicit and explicit components in dual force field adaptation
Motor learning occurs through multiple mechanisms, including unsupervised, supervised (error-based) and reinforcement (reward-based) learning. Although studies have shown that reward leads to an overall better motor adaptation, the specific processes by which reward influences adaptation are still unclear. Here, we examine how the presence of reward affects dual-adaptation to novel dynamics, and distinguish its influence on implicit and explicit learning. Participants adapted to two opposing force fields in an adaptation/de-adaptation/error-clamp paradigm, where five levels of reward (a score and a digital face) were provided as participants reduced their lateral error. Both reward and control (no reward provided) groups simultaneously adapted to both opposing force fields, exhibiting a similar final level of adaptation, which was primarily implicit. Triple-rate models fit to the adaptation process found higher learning rates in the fast and slow processes, and a slightly increased fast retention rate for the reward group. While differences in the slow learning rate were only driven by implicit learning, the large difference in the fast learning rate was mainly explicit. Overall, we confirm previous work showing that reward increases learning rates, extending this to dual-adaptation experiments, and demonstrating that reward influences both implicit and explicit adaptation. Specifically, we show that reward acts primarily explicitly on the fast learning rate and implicitly on the slow learning rates. New and Noteworthy: Here we show that rewarding participants’ performance during dual force field adaptation primarily affects the initial rate of learning and the early timescales of adaptation, with little effect on the final adaptation level. However, reward affects both explicit and implicit components of adaptation. While the learning rate of the slow process is increased implicitly, the fast learning and retention rates are increased through both implicit components and the use of explicit strategies.
Open AccessReward actively engages both implicit and explicit components in dual force field adaptation
Motor learning occurs through multiple mechanisms, including unsupervised, supervised (error-based) and reinforcement (reward-based) learning. Although studies have shown that reward leads to an overall better motor adaptation, the specific processes by which reward influences adaptation are still unclear. Here, we examine how the presence of reward affects dual-adaptation to novel dynamics, and distinguish its influence on implicit and explicit learning. Participants adapted to two opposing force fields in an adaptation/de-adaptation/error-clamp paradigm, where five levels of reward (a score and a digital face) were provided as participants reduced their lateral error. Both reward and control (no reward provided) groups simultaneously adapted to both opposing force fields, exhibiting a similar final level of adaptation, which was primarily implicit. Triple-rate models fit to the adaptation process found higher learning rates in the fast and slow processes, and a slightly increased fast retention rate for the reward group. While differences in the slow learning rate were only driven by implicit learning, the large difference in the fast learning rate was mainly explicit. Overall, we confirm previous work showing that reward increases learning rates, extending this to dual-adaptation experiments, and demonstrating that reward influences both implicit and explicit adaptation. Specifically, we show that reward acts primarily explicitly on the fast learning rate and implicitly on the slow learning rates. New and Noteworthy: Here we show that rewarding participants’ performance during dual force field adaptation primarily affects the initial rate of learning and the early timescales of adaptation, with little effect on the final adaptation level. However, reward affects both explicit and implicit components of adaptation. While the learning rate of the slow process is increased implicitly, the fast learning and retention rates are increased through both implicit components and the use of explicit strategies.
- 1 November 2021
- Published by American Physiological Society in Journal of Neurophysiology
- Vol. 126 (5) , 1490-1506
Direct and indirect cues can enable dual adaptation, but through different learning processes
We present evidence that learning of context-dependent dynamics proceeds via different processes depending on the type of sensory cue used to signal the context. Visual workspace location enabled learning different dynamics implicitly, presumably because it directly enters the dynamic state estimate. In contrast, a color cue was only successful where learners were apparently able to leverage explicit strategies to account for changed dynamics. This suggests a unification for the previously inconclusive role of color cues.
- 1 November 2021
- Published by American Physiological Society in Journal of Neurophysiology
- Vol. 126 (5) , 1490-1506
Direct and indirect cues can enable dual adaptation, but through different learning processes
We present evidence that learning of context-dependent dynamics proceeds via different processes depending on the type of sensory cue used to signal the context. Visual workspace location enabled learning different dynamics implicitly, presumably because it directly enters the dynamic state estimate. In contrast, a color cue was only successful where learners were apparently able to leverage explicit strategies to account for changed dynamics. This suggests a unification for the previously inconclusive role of color cues.
- Open Access
Direct and indirect cues can enable dual-adaptation, but through different learning processes
Switching between motor tasks requires accurate adjustments for changes in dynamics (grasping a cup) or sensorimotor transformations (moving a computer mouse). Dual-adaptation studies have investigated how learning of context-dependent dynamics or transformations is enabled by sensory cues. However, certain cues, such as color, have shown mixed results. We propose that these mixed results may arise from two major classes of cues: “direct” cues, which are part of the dynamic state and “indirect” cues, which are not. We hypothesized that explicit strategies would primarily account for adaptation for an indirect color cue but would be limited to simple tasks while a direct visual separation cue would allow implicit adaptation regardless of task complexity. To test this idea, we investigated the relative contribution of implicit and explicit learning in relation to contextual cue type (colored or visually shifted workspace) and task complexity (one or eight targets) in a dual-adaptation task. We found that the visual workspace location cue enabled adaptation across conditions primarily through implicit adaptation. In contrast, we found that the color cue was largely ineffective for dual adaptation, except in a small subset of participants who appeared to use explicit strategies. Our study suggests that the previously inconclusive role of color cues in dual-adaptation may be explained by differential contribution of explicit strategies across conditions. New & Noteworthy We present evidence that learning of context-dependent dynamics proceeds via different processes depending on the type of sensory cue used to signal the context. Visual workspace location enabled learning different dynamics implicitly, presumably because it directly enters the dynamic state estimate. In contrast, a color cue was only successful where learners were apparently able to leverage explicit strategies to account for changed dynamics. This suggests a unification for the previously inconclusive role of color cues.
Open AccessDirect and indirect cues can enable dual-adaptation, but through different learning processes
Switching between motor tasks requires accurate adjustments for changes in dynamics (grasping a cup) or sensorimotor transformations (moving a computer mouse). Dual-adaptation studies have investigated how learning of context-dependent dynamics or transformations is enabled by sensory cues. However, certain cues, such as color, have shown mixed results. We propose that these mixed results may arise from two major classes of cues: “direct” cues, which are part of the dynamic state and “indirect” cues, which are not. We hypothesized that explicit strategies would primarily account for adaptation for an indirect color cue but would be limited to simple tasks while a direct visual separation cue would allow implicit adaptation regardless of task complexity. To test this idea, we investigated the relative contribution of implicit and explicit learning in relation to contextual cue type (colored or visually shifted workspace) and task complexity (one or eight targets) in a dual-adaptation task. We found that the visual workspace location cue enabled adaptation across conditions primarily through implicit adaptation. In contrast, we found that the color cue was largely ineffective for dual adaptation, except in a small subset of participants who appeared to use explicit strategies. Our study suggests that the previously inconclusive role of color cues in dual-adaptation may be explained by differential contribution of explicit strategies across conditions. New & Noteworthy We present evidence that learning of context-dependent dynamics proceeds via different processes depending on the type of sensory cue used to signal the context. Visual workspace location enabled learning different dynamics implicitly, presumably because it directly enters the dynamic state estimate. In contrast, a color cue was only successful where learners were apparently able to leverage explicit strategies to account for changed dynamics. This suggests a unification for the previously inconclusive role of color cues.
- Open Access
- 19 October 2020
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 16 (10) , e1008373
Timescales of motor memory formation in dual-adaptation
The timescales of adaptation to novel dynamics are well explained by a dual-rate model with slow and fast states. This model can predict interference, savings and spontaneous recovery, but cannot account for adaptation to multiple tasks, as each new task drives unlearning of the previously learned task. Nevertheless, in the presence of appropriate contextual cues, humans are able to adapt simultaneously to opposing dynamics. Consequently this model was expanded, suggesting that dual-adaptation occurs through a single fast process and multiple slow processes. However, such a model does not predict spontaneous recovery within dual-adaptation. Here we assess the existence of multiple fast processes by examining the presence of spontaneous recovery in two experimental variations of an adaptation-de-adaptation-error-clamp paradigm within dual-task adaptation in humans. In both experiments, evidence for spontaneous recovery towards the initially learned dynamics (A) was found in the error-clamp phase, invalidating the one-fast-two-slow dual-rate model. However, as adaptation is not only constrained to two timescales, we fit twelve multi-rate models to the experimental data. BIC model comparison again supported the existence of two fast processes, but extended the timescales to include a third rate: the ultraslow process. Even within our single day experiment, we found little evidence for decay of the learned memory over several hundred error-clamp trials. Overall, we show that dual-adaptation can be best explained by a two-fast-triple-rate model over the timescales of adaptation studied here. Longer term learning may require even slower timescales, explaining why we never forget how to ride a bicycle. Retaining motor skills is crucial to perform basic daily life tasks. However we still have limited understanding of the computational structure of these motor memories, an understanding that is critical for designing rehabilitation. Here we demonstrate that learning any task involves adaptation of independent fast, slow and ultraslow processes to build a motor memory. The selection of the appropriate motor memory is gated through a contextual cue. Together this work extends our understanding of the architecture of motor memories, by merging disparate computational theories to propose a new model.
- 19 October 2020
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 16 (10) , e1008373
Open AccessTimescales of motor memory formation in dual-adaptation
The timescales of adaptation to novel dynamics are well explained by a dual-rate model with slow and fast states. This model can predict interference, savings and spontaneous recovery, but cannot account for adaptation to multiple tasks, as each new task drives unlearning of the previously learned task. Nevertheless, in the presence of appropriate contextual cues, humans are able to adapt simultaneously to opposing dynamics. Consequently this model was expanded, suggesting that dual-adaptation occurs through a single fast process and multiple slow processes. However, such a model does not predict spontaneous recovery within dual-adaptation. Here we assess the existence of multiple fast processes by examining the presence of spontaneous recovery in two experimental variations of an adaptation-de-adaptation-error-clamp paradigm within dual-task adaptation in humans. In both experiments, evidence for spontaneous recovery towards the initially learned dynamics (A) was found in the error-clamp phase, invalidating the one-fast-two-slow dual-rate model. However, as adaptation is not only constrained to two timescales, we fit twelve multi-rate models to the experimental data. BIC model comparison again supported the existence of two fast processes, but extended the timescales to include a third rate: the ultraslow process. Even within our single day experiment, we found little evidence for decay of the learned memory over several hundred error-clamp trials. Overall, we show that dual-adaptation can be best explained by a two-fast-triple-rate model over the timescales of adaptation studied here. Longer term learning may require even slower timescales, explaining why we never forget how to ride a bicycle. Retaining motor skills is crucial to perform basic daily life tasks. However we still have limited understanding of the computational structure of these motor memories, an understanding that is critical for designing rehabilitation. Here we demonstrate that learning any task involves adaptation of independent fast, slow and ultraslow processes to build a motor memory. The selection of the appropriate motor memory is gated through a contextual cue. Together this work extends our understanding of the architecture of motor memories, by merging disparate computational theories to propose a new model.
- Open Access
Timescales of motor memory formation in dual-adaptation
The timescales of adaptation to novel dynamics are well explained by a dual-rate model with slow and fast states. This model can predict interference, savings and spontaneous recovery, but cannot account for adaptation to multiple tasks, as each new task drives unlearning of the previously learned task. Nevertheless, in the presence of appropriate contextual cues, humans are able to adapt simultaneously to opposing dynamics. Consequently this model was expanded, suggesting that dual-adaptation occurs through a single fast process and multiple slow processes. However, such a model does not predict spontaneous recovery within dual-adaptation. Here we assess the existence of multiple fast processes by examining the presence of spontaneous recovery in two experimental variations of an adaptation-de-adaptation-error-clamp paradigm within dual-task adaptation in humans. In both experiments, evidence for spontaneous recovery towards the initially learned dynamics (A) was found in the error-clamp phase, invalidating the one-fast-two-slow dual-rate model. However, as adaptation is not only constrained to two timescales, we fit twelve multi-rate models to the experimental data. BIC model comparison again supported the existence of two fast processes, but extended the timescales to include a third rate: the ultraslow process. Even within our single day experiment, we found little evidence for decay of the learned memory over several hundred error-clamp trials. Overall, we show that dual-adaptation can be best explained by a two-fast-triple-rate model over the timescales of adaptation studied here. Longer term learning may require even slower timescales, explaining why we never forget how to ride a bicycle. Author Summary: Retaining motor skills is crucial to perform basic daily life tasks. However we still have limited understanding of the computational structure of these motor memories, an understanding that is critical for designing rehabilitation. Here we demonstrate that learning any task involves adaptation of independent fast, slow and ultraslow processes to build a motor memory. The selection of the appropriate motor memory is gated through a contextual cue. Together this work extends our understanding of the architecture of motor memories, by merging disparate computational theories to propose a new model.
Open AccessTimescales of motor memory formation in dual-adaptation
The timescales of adaptation to novel dynamics are well explained by a dual-rate model with slow and fast states. This model can predict interference, savings and spontaneous recovery, but cannot account for adaptation to multiple tasks, as each new task drives unlearning of the previously learned task. Nevertheless, in the presence of appropriate contextual cues, humans are able to adapt simultaneously to opposing dynamics. Consequently this model was expanded, suggesting that dual-adaptation occurs through a single fast process and multiple slow processes. However, such a model does not predict spontaneous recovery within dual-adaptation. Here we assess the existence of multiple fast processes by examining the presence of spontaneous recovery in two experimental variations of an adaptation-de-adaptation-error-clamp paradigm within dual-task adaptation in humans. In both experiments, evidence for spontaneous recovery towards the initially learned dynamics (A) was found in the error-clamp phase, invalidating the one-fast-two-slow dual-rate model. However, as adaptation is not only constrained to two timescales, we fit twelve multi-rate models to the experimental data. BIC model comparison again supported the existence of two fast processes, but extended the timescales to include a third rate: the ultraslow process. Even within our single day experiment, we found little evidence for decay of the learned memory over several hundred error-clamp trials. Overall, we show that dual-adaptation can be best explained by a two-fast-triple-rate model over the timescales of adaptation studied here. Longer term learning may require even slower timescales, explaining why we never forget how to ride a bicycle. Author Summary: Retaining motor skills is crucial to perform basic daily life tasks. However we still have limited understanding of the computational structure of these motor memories, an understanding that is critical for designing rehabilitation. Here we demonstrate that learning any task involves adaptation of independent fast, slow and ultraslow processes to build a motor memory. The selection of the appropriate motor memory is gated through a contextual cue. Together this work extends our understanding of the architecture of motor memories, by merging disparate computational theories to propose a new model.