This explains the failure of the subjects to completely compensate for the target shift when it occurred late in the movement because the velocity feedback gain prevented complete adaptation of the endpoint

position. Finally, if the brain utilizes some kind of OFC, then the reflex responses Gemcitabine should exhibit the same kind of responses as seen in voluntary control because the same neural structures must be responsible for both (Scott, 2004). This means that not only will the responses vary according to the physical demands of the task being performed but that these responses approximate the later “voluntary” responses (Pruszynski et al., 2009). Although the short-latency (monosynaptic) stretch reflex responds only to muscle stretch, the long-latency response has long been known to respond to other factors (e.g., Lacquaniti and Soechting, 1986). However, more recently, it has been shown that the long-latency stretch reflex responses actually reflect the internal model of the limb, corresponding to the required joint torques to offset the overall disturbance of the limb (Kurtzer et al., 2009 and Kurtzer et al., 2008). Both time delays and noise in the sensorimotor system

impede our ability to make accurate estimates of relevant features of movement, such as the state of our limbs. Motor prediction, as instantiated by a forward Epigenetic inhibitor nmr model, is a key computational component that can alleviate this problem (Desmurget and Grafton, 2000, Miall et al., 1993 and Wolpert and Kawato, 1998). We have touched upon this issue previously in our description to of the Kalman filter, in which a combination of motor output and sensory input is used to estimate the current state. A forward model is a putative computational element within the nervous system that predicts the causal relation between actions and their consequences (Wolpert and Kawato, 1998). The forward model instantiates a model of the neuromuscular system and external world, thereby acting as a neural simulator that makes predictions of the effect of motor commands. A necessary

input to the forward model is a copy of the motor output (termed efference copy) that will act on the neuromuscular system. The output of the model can then be used for state estimation, prediction of sensory feedback, or for predictive control. Forward models are not only useful to counteract the effects of delays and noise but also can help in situations where identical stimuli can give rise to different afferent signals depending on the state of the system. For example by modulating the γ static and γ dynamic drive to the muscle spindles, the sensorimotor system will receive different sensory responses for the identical physical input (Matthews, 1972). To infer state in such situations, the sensorimotor system needs to take into account the motor output to interpret the sensory input.