Early work on motor cortex function

Back in the 1940s, Canadian neurosurgeon Wilder Penfield wanted to know which bits of epileptic's brains he could suck out without them noticing. To work out which were the really important bits of cortex he electrically stimulated the cortical surface and observed the results. He found that stimulation of Brodmann's area 4 readily elicited localised muscle twitches. Furthermore there appeared to be a “motor map” of the body surface along the gyrus that comprises area 4. Area 4 is therefore now known as the primary motor cortex. Following this discovery, he found that stimulation of regions rostral (in front) of M1 caused more complicated movements; however, more current was required to initiate movements from these areas. These 'premotor' cortical areas are located in Brodmann's area 6.

The motor cortical areas are now typically divided into three regions which have different functional roles:

1. primary motor cortex (M1)
2. pre-motor area (PMA)
3. supplementary motor area (SMA)

Penfield's experiments make everything seem pretty straightforward: the purpose of M1 is to connect the brain to the lower motor neurons via the spinal cord in order to tell them which particular muscles need to contract. These upper motor neurons are found in layer 5 of the motor cortex and contain some of the largest cells in the brain (Betz cells whose cell bodies can be up to 100 micrometres in diameter. For comparison, rod photoreceptors are about 3 micrometres in across). The descending axons of these layer 5 cells form the CST. However, a single layer 5 forms synapses with many lower motor neurons which inervate different muscles. Furthermore, the same muscle is often represented over quite large regions of the brain's surface and there is overlap in the representation of different regions of the body. These facts mean that M1 neurons do not form simple connections with lower motor neurons. The activity of single M1 neuron could cause contraction of more than one muscle. This suggests that M1 may not simply be coding the degree of contraction of individual muscles. To understand what is going on we need to record from the motor cortex of unaesthetised monkeys whilst they are performing a task:

More recent work: coding of reach direction in M1

Much of the seminal work in this field has been done by Georgopolous, who recorded from neurons in M1 whilst a monkey used its arm to point to targets at various locations in space. He found that single neurons would respond to a broad but restricted range of reach directions. This made him think that cells in M1 are encoding the direction of reach rather than activity of single muscles. However, a single cell can't accurately encode reach direction because its resolution is too coarse. He recorded the preferred reach direction of hundreds of cells in M1 and found there to be a broad range of preferred directions. The question now arises: how does the make accurate reaching motions when individual neurons code direction fairly coarsely? Georgopolous found that when, for a particular reach direction, the preferred direction and firing of all the neurons are added together as a vector, the sum accurately predicts the final reach direction of the monkey. Furthermore, he found evidence of a mental rotation of the population vector when a monkey is asked to point at an angle to a target: the population vector initially points directly at the target and then “rotates” rapidly until it points at the desired angle to the target.

Is M1 really coding reach direction?

The occurrence of the population vector in M1 is undisputed but it is uncertain what it means. It turns out that single neuron responses in M1 do not correlate only with reach direction but also with a number of other co-variables such as initial arm position, acceleration, movement preparation, target position, distance to target, joint configuration, muscular force, etc. Recent work has suggested that the population vector in M1 codes motion around single joints and that the apparent reach-direction code is an epiphenomenon due partly to the physical restrictions of the musculoskeletal system.

What is needed for the execution of a correctly coordinated movement?

Whatever M1 is really coding, it would appear to be relatively low level. Looking at M1 responses doesn't tell us how complex movements are planned or how they are they are guided. For this we have to look at the premotor areas and how they interact with the basal ganglia to initiate movement and the cerebellum for the visual guidance of movement. A key feature in both of these tasks is sensory feedback. Without sensory feedback normal initiation and (particularly) guidance of movement is not possible. Initiating a movement at the right time is usually based upon sensory information. Correct completion of a movement (such as reaching) requires you to know a lot about your environment, such as the initial length of the muscle, muscle velocity, load on the muscle, and posture. You also need to know how the movement is progressing in the form of visual and proprioceptive feedback in order to correct the movement as it happens or improve it on subsequent occasions through trial and error learning.

The role of the premotor areas, PMA and SMA

Stimulation of SMA and PMA causes initiation of complex, often bilateral, movements. Initiation of these movements requires more current to be delivered than that required for initiation of movements from M1. This is because the premotor areas are only indirectly connected with lower motor neurons.

PMA receives input from cerebellum and is involved in planning movements based on external(especially visual) cues. Involved mostly in control of postural and proximal limb muscles. Lesions of PMA disrupt learned responses to visual cues. SMA has strong reciprocal connections with the basal ganglia and is involved in planning learned sequences of movements. Is site of readiness potential which begins 1s before a movement is initiated. Stimulation of SMA creates an urge to move. Bilateral SMA lesions blocks all movement and causes flaccid paralysis. More minor, unilateral, lesions disrupt learning of movement sequences which don't depend on external cues.

PMA and SMA are the routes through which the cerebellum and basal ganglia form parallel processing loops with the motor cortical areas. It is through these loops that movements are initiated and guided. Damage to structures in these loops causes specific movement impairments.