It has proved difficult to attribute precise respiratory function to localized anatomical substrates in man because lesions are rarely isolated and, even with newer imaging techniques, ante-mortem localization is imprecise. Furthermore accurate diagnosis of respiratory insufficiency has led to earlier therapeutic intervention with controlled ventilation. Finally there is probably considerable redundancy and plasticity of neural function where progressive and destructive mass lesions can have little or no functional consequence whilst acute discrete lesions in a similar distribution may lead to profound respiratory impairment.

It has been conventional to consider neural control of respiration to depend on two anatomically and functionally independent pathways (Plum, 1970). Metabolic (automatic) respiration is the homeostatic pathway by which ventilation may be mediated to maintain acid–base status and oxygenation to the metabolic requirements. The behavioural (voluntary) system operates during wakefulness and allows voluntary modulation of respiration in response, for example, to speaking, singing, breath-holding and straining. This has proved a valuable and durable model to explain and predict patterns of respiratory insufficiency; however, there is increasing evidence for interactions between the two systems (Orem & Netick, 1986; Murphy et al., 1990).

Techniques of central motor stimulation and functional imaging have confirmed that the respiratory muscles are activated behaviourally with phrenic motoneurones being controlled by rapidly conducting, oligosynaptic pathways from the contralateral motor cortex. Inspiratory muscles have a direct representation in the primary motor cortex, premotor cortex, supplementary motor area and thalamus, whilst for active expiration the areas are more extensive and involve limbic cortex (Gandevia & Rothwell, 1987; Macefield & Gandevia, 1991; Colebach et al., 1991; Maskill et al., 1991; Ramsay et al., 1993).