Where does efference copy collide?

An efference copy is a duplicate of a motor command that the nervous system keeps when it moves, using it to anticipate the sensory feedback the movement will cause (the reafference) and to cancel that self-caused signal from what the senses report. Von Holst and Mittelstaedt's 1950 Das Reafferenzprinzip introduced the idea to explain why the world looks stable when we move our own eyes but appears to swing when the eyeball is pushed passively: only a self-generated movement carries a matching copy to cancel. The referent is the stored prediction of a movement's sensory consequences, and it is the same construct that Sperry independently named corollary discharge and that control theory later called a forward model.[1]

Unit

The predicted sensory feedback of a self-generated movement

Recovery means

Not applicable

Measured by

Inferred from the cancellation of self-generated sensory signals and from reafference errors when movement is perturbed

Corollary discharge is Sperry's name, coined in the same year, for a signal that accompanies a motor command and forewarns sensory systems about the movement the animal is about to make. Sperry's 1950 Neural basis of the spontaneous optokinetic response produced by visual inversion inferred it from fish whose eyes had been surgically rotated: the animals circled endlessly because the expected and the actual visual motion no longer matched. It is not a different construct from efference copy but the identical idea discovered on a separate track: one latent thing travelling under two names, the textbook case of a jangle.[2]

Unit

The advance signal a motor command sends to sensory systems

Recovery means

Not applicable

Measured by

Inferred from disrupted perception of stability when predicted and actual sensory motion are made to disagree

In control theory and computational motor neuroscience the same construct is called a forward model: an internal model that takes a copy of the outgoing command and predicts the system's next state and the sensory signals it will produce, so the controller need not wait for slow feedback. Wolpert, Ghahramani, and Jordan's 1995 An internal model for sensorimotor integration gave behavioural and computational evidence that the brain runs such a predictor when estimating the position of the hand. It is the engineering-flavoured name for what neuroscience calls efference copy and corollary discharge, the same prediction of a movement's consequences, formalised as a model.[3]

Unit

The model-predicted next state and sensory outcome of a command

Recovery means

Not applicable

Measured by

Estimated by comparing predicted and actual state during reaching, and by state-estimation models fit to behaviour