Single-unit analysis of the human ventral thalamic nuclear group: Somatosensory responses

1. We have studied the functional and somatotopic properties of 531 single mechanoreceptive thalamic neurons in humans undergoing stereotactic surgery for the control of movement disorders and pain. The majority of these somatosensory cells had small receptive fields (RFs) and were activated in a reproducible manner by mechanical stimuli applied to the skin or deep tissues. These neurons, which we termed 'lemniscal,' could be further classified into those responding to stimulation of cutaneous (76% of lemniscal sensory cells) or deep (24%) structures. 2. The incidence of neurons having cutaneous or mucosal RFs in the perioral region, thumb, and fingers (66%) was much higher than that of neurons having RFs elsewhere on the body. Most of the deep cells were activated by movements of and/or mechanical stimuli delivered to muscles or tendons controlling the elbow, wrist, and fingers. 3. Sequences of cells spanning several millimeters in the parasagittal plane often exhibited overlapping RFs. However, RFs changed markedly for cells separated by the same distances in the mediolateral direction. This suggests that the cutaneous somatotopic representation of each region of the body is organized into relatively thin sheets of cells oriented in the parasagittal plane. 4. By comparing neuronal RFs in different parasagittal planes in thalamus of individual patients we have identified a mediolateral representation of body surface following the sequence from: intraoral structures, face, thumb through fifth finger to palm, with forearm and leg laterally. 5. Along many trajectories in the parasagittal plane the sequence of cells with overlapping RFs was interrupted by another sequence of cells with RFs corresponding to a different body region. The RFs of the intervening sequence characteristically represented body regions known to be located more medially in thalamus (see 3 above). These findings could be explained if the lamellae postulated above were laterally convex. 6. Cells responding to deep stimulation (deep cells) could be further classified into those responding to joint movement (63%), deep pressure (15%), or both (22%). Deep cells were found usually at the anterior-dorsal border and sometimes at the posterior border of the region containing cells responding to cutaneous stimuli. Athough there was some overlap in the RFs, deep cells representing wrist were found medial to those representing elbow, and both of these were found medial to cells representing leg. This suggested a mediolateral representation of deep structures progressing from jaw medially to fingers, wrist, elbow, shoulder, and leg. This representation of deep structures paralleled the mediolateral arrangement of cutaneous structures. 7. Cells responding to cutaneous stimulation were further categorized, on the basis of their response properties, as slowly and rapidly adapting and, on the basis of the minimum intensity required to activate them, as touch or pressure. Although these different types of cells were not located at specific sites within the thalamic representation of cutaneous structures, clustering of cells by type was observed to a statistically significant level. 8. Bipolar stimulation at sites where lemniscal cells could be recorded often evoked paraesthesiae over an area including the RFs of the cells recorded at that site. 9. These results obtained in human patients are similar to the results of recent detailed studies of the ventrobasal complex in nonhuman primates.