Precise Motor Control Enables Rapid Flexibility in Vocal Behavior of Marmoset Monkeys

Abstract

Investigating the evolution of human speech is difficult and controversial because human speech surpasses nonhuman primate vocal communication in scope and flexibility [1-3]. Monkey vocalizations have been assumed to be largely innate, highly affective, and stereotyped for over 50 years [4, 5]. Recently, this perception has dramatically changed. Current studies have revealed distinct learning mechanisms during vocal development [6-8] and vocal flexibility, allowing monkeys to cognitively control when [9, 10], where [11], and what to vocalize [10, 12, 13]. However, specific call features (e.g., duration, frequency) remain surprisingly robust and stable in adult monkeys, resulting in rather stereotyped and discrete call patterns [14]. Additionally, monkeys seem to be unable to modulate their acoustic call structure under reinforced conditions beyond natural constraints [15, 16]. Behavioral experiments have shown that monkeys can stop sequences of calls immediately after acoustic perturbation but cannot interrupt ongoing vocalizations, suggesting that calls consist of single impartible pulses [17, 18]. Using acoustic perturbation triggered by the vocal behavior itself and quantitative measures of resulting vocal adjustments, we show that marmoset monkeys are capable of producing calls with durations beyond the natural boundaries of their repertoire by interrupting ongoing vocalizations rapidly after perturbation onset. Our results indicate that marmosets are capable of interrupting vocalizations only at periodic time points throughout calls, further supported by the occurrence of periodically segmented phees. These ideas overturn decades-old concepts on primate vocal pattern generation, indicating that vocalizations do not consist of one discrete call pattern but are built of many sequentially uttered units, like human speech.

Compensatory mechanisms affect sensorimotor integration during ongoing vocal-motor acts in marmoset monkeys

“…In a 235 previous study, we demonstrated that marmosets interrupt their vocalizations shortly after 236 noise onset when perturbation starts after vocal onset [18], supporting the idea that these 237 animals tend to avoid calling in ambient noise. Such call interruptions, however, were rare 238 (2.6% of all calls), indicating stark neuronal and/or anatomical constraints in exhibiting such 239 behavior [18] and resulting in a large fraction of phee calls being perturbed by noise bursts. 240…”

“…Based on the current work and earlier studies [14,18], we propose a hypothetical neuronal 251 model suggesting various audio-vocal control mechanism involving cortical, subcortical, and 252 corticofugal connections capable of modulating vocal behavior in a noisy environment (Fig. 253 4).…”

“…These findings suggest that 49 while these animals generally seem to prefer avoiding calling in a noisy environment, they do 50 exhibit the involuntary audio-vocal effects discussed above when doing so. This idea is 51 supported by a recent study showing that marmoset tend to produce single calls instead of 52 call sequences in response to perturbing noise stimuli [18]. Interestingly, marmoset monkeys 53 are also capable of interrupting ongoing vocalizations rapidly after noise perturbation onset 54 [18], overturning decades-old concepts regarding vocal pattern generation [19][20][21], 55 indicating that vocalizations do not consist of one discrete call pattern but are built of many 56 sequentially uttered units that might be modulated and initiated independently of each other.…”

“…This idea is 51 supported by a recent study showing that marmoset tend to produce single calls instead of 52 call sequences in response to perturbing noise stimuli [18]. Interestingly, marmoset monkeys 53 are also capable of interrupting ongoing vocalizations rapidly after noise perturbation onset 54 [18], overturning decades-old concepts regarding vocal pattern generation [19][20][21], 55 indicating that vocalizations do not consist of one discrete call pattern but are built of many 56 sequentially uttered units that might be modulated and initiated independently of each other. 57 However, it is yet unclear whether audio-vocal mechanisms, such as the Lombard effect and 58 its accompanied changes in call frequency, can be rapidly elicited in cases where the 59 perturbing noise starts after call onset or whether such effects only occur if noise 60 perturbation starts prior to call onset.…”

“…Current studies have control when [14], where [38], and what to vocalize [39]. In addition, recent studies revealed 232 that marmosets are able to modulate distinct call parameters in response to acoustic 233 feedback [18,40]. This vocal flexibility allows marmosets to avoid calling in the presence of 234 environmental noise and predominantly initiate their vocalizations in silent periods [14].…”