What is Motor Imagery?

The mental rehearsal of a movement without physical execution, producing measurable changes in sensorimotor cortex oscillations that can be detected by EEG or intracortical electrodes and used for BCI control.

What category does Motor Imagery belong to in BCI?

Motor Imagery is a Signal Types concept in brain-computer interface science.

Motor Imagery — BCI Glossary

Motor imagery (MI) is the cognitive process of imagining a movement — clenching a fist, moving a foot, rotating a wrist — without physically performing it. This mental rehearsal activates many of the same motor cortex circuits involved in actual movement, producing detectable changes in neural oscillations. Motor imagery is one of the foundational paradigms for non-invasive EEG-based BCIs and also plays a role in intracortical BCI control.

Neural Basis

When a person imagines moving their right hand, the left sensorimotor cortex (contralateral to the imagined movement) shows a decrease in mu rhythm (8-12 Hz) and beta rhythm (13-30 Hz) power — a phenomenon called event-related desynchronization (ERD). Simultaneously, the ipsilateral sensorimotor cortex and other non-involved regions may show event-related synchronization (ERS). This lateralized pattern can be detected by EEG electrodes over the motor strip and classified to determine the imagined movement type.

BCI Applications

Motor imagery BCIs typically classify 2-4 movement types:

Left hand vs. right hand: Binary control based on lateralization of mu/beta ERD
Feet: Midline ERD pattern, distinct from hand imagery
Tongue: Lateral and inferior ERD, usable as a third or fourth class

Classification accuracy for 2-class MI typically ranges from 70-90% in healthy subjects, though a significant minority of users (15-30%) are "BCI illiterate" — unable to produce reliably classifiable MI patterns. Four-class MI is more challenging, with typical accuracies of 60-75%.

Training and Adaptation

Unlike ERP-based BCIs (which exploit involuntary brain responses), MI BCIs require the user to learn a cognitive strategy that produces consistent, classifiable neural patterns. This training process can take hours to weeks. Modern approaches use neurofeedback — real-time visualization of the user's oscillatory patterns — to guide learning. Co-adaptive systems simultaneously train the classifier to the user and the user to the classifier, improving convergence.

Intracortical Motor Imagery

In intracortical BCI studies (BrainGate, Neuralink), participants often use attempted or imagined movements to generate neural commands. A participant with tetraplegia who imagines moving their paralyzed hand produces motor cortex firing patterns that can be decoded into cursor velocity or robotic arm commands. The higher signal quality of intracortical recordings allows much finer-grained decoding than EEG-based MI, enabling continuous 2D or 3D control rather than discrete class selection.