The descending pathways that mediate the periaqueductal gray (PAG)-evoked coordination of respiratory, laryngeal, and orofacial activity for vocalization have yet to be delineated. Two hypotheses have been offered. One theory is that this activity is mediated by a diffuse descending projection to parvocellular reticular interneurons, adjacent to the relevant laryngeal and orofacial motoneuronal pools. The second hypothesis is that the motor activity for vocalization is integrated via a projection from the PAG to a caudal medullary column of neurons, the nucleus retroambigualis (NRA). These hypotheses were tested with the use of a series of medullary transections combined with PAG stimulation. Transections that eliminated, in a series of caudal-to-rostral steps, the NRA, also eliminated the PAG-evoked cricothyroid and most of the thyroarytenoid laryngeal motor activity. These results indicate that the final common pathway for much of the laryngeal activity in PAG-evoked vocalization includes un initial synapse in the caudal medulla, presumably in the NRA. 2. The electromyographic changes evoked by microinjection of D,L-homocysteic acid (DLH) in the NRA of the unanesthetized, precollicular decerebrate cat were analyzed in order to delineate the NRA contribution to the coordinated respiratory, laryngeal, and oral muscle changes in vocalization. A total of 415 DLH injection sites were located at or caudal to the level of the obex. Vocalization was evoked at 46 of these sites, which were all confined to a restricted region of the ventrolateral medulla 1-3 mm caudal to the obex. This region corresponded to the rostral half of the NRA and the immediately adjacent medullary tegmentum. 3. In all experiments evidence was obtained that variable muscle activation, rather than functional and integrated muscle patterns, was represented within the NRA. Vocalization evoked by DLH microinjection in the NRA was usually associated with excitation of the cricothyroid, thyroarytenoid, external oblique, internal oblique, internal intercostal, and diaphragm muscles that occurred in a different manner from site to site. That is, injection at sites separated by 0.3-0.5 mm evoked quite different responses. 4. NRA-evoked vocalization was compared with PAG-evoked vocalization using small injections (1.5-4.5 nl) into each region. As well, larger microinjections (15-120 nl) into NRA were made for comparison with previous results from the PAG using similar doses. Within the PAG, stereotyped and relatively "fixed" patterns of muscle activity are represented, whereas within the NRA there was no representation of specific muscle patterns, but rather a partial topographic separation of "premotor neurons" regulating different muscles. Correspondingly, stereotyped vocalizations were never evoked from the NRA. Further, most NRA-evoked vocalizations were unusual in quality and would not be identified generally as feline. 5. Evidence was obtained for a separation of pathways from the PAG regulating sound production and orofacial modulation of that sound. In contrast to the results from the PAG, excitation of NRA neurons rarely evoked activity in the oral muscles (genioglossus or anterior belly of digastric) or orofacial modulation of sound production. 6. Our finding suggests that the NRA serves as an important substrate for the generation of respiratory pressure and larynges adduction, which are two essential aspects of not only vocalization but also several behaviors involving Valsava maneuvers such as coughing, vomiting, and defecation.
