Brain Behav Evol 2014;84:93–102 Published online: September 20, 2014 DOI: 10.1159/000365346 Convergent Evolution of Vocal Cooperation without Convergent Evolution of Brain Size a, b a, b, c Jeremy I. Borjon Asif A. Ghazanfar a b c Princeton Neuroscience Institute and Departments of Psychology and Ecology and Evolutionary Biology, Princeton University, Princeton, N.J. , USA Key Words complex cognitive machinery is not needed for vocal coop- Vocal communication · Turn taking · Vocalizations · eration to occur. Consistent with this idea, the temporal Common marmosets · Humans structure of marmoset vocal exchanges can be described in terms of coupled oscillator dynamics, similar to quantitative descriptions of human conversations. We propose a simple Abstract neural circuit mechanism that may account for these dynam- One pragmatic underlying successful vocal communication ics and, at its core, involves vocalization-induced reductions is the ability to take turns. Taking turns – a form of coopera- of arousal. Such a mechanism may underlie the evolution of tion – facilitates the transmission of signals by reducing the vocal turn taking in both marmoset monkeys and humans. amount of their overlap. This allows vocalizations to be better © 2014 S. Karger AG, Basel heard. Until recently, non-human primates were not thought of as particularly cooperative, especially in the vocal domain. We recently demonstrated that common marmosets (Calli- Introduction thrix jacchus) , a small New World primate species, take turns when they exchange vocalizations with both related and un- Humans orchestrate behavior through spoken lan- related conspecifics. As the common marmoset is distantly guage, gestures and, more often than not, a combination related to humans (and there is no documented evidence of speech and gestures. The evolutionary origins for this that Old World primates exhibit vocal turn taking), we argue uniquely human form of communication are mysterious that this ability arose as an instance of convergent evolution, for many reasons. Primarily, there are few ways to recon- and is part of a suite of prosocial behavioral tendencies. Such struct what ancestral vocalizations sounded like and how behaviors seem to be, at least in part, the outcome of the co- they were used. Moreover, soft tissues such as the brain operative breeding strategy adopted by both humans and and parts of the vocal apparatus do not fossilize. Relatedly, marmosets. Importantly, this suite of shared behaviors oc- the dynamic social behavior that may have driven the evo- curs without correspondence in encephalization. Marmoset lution of more sophisticated communication is also diffi- vocal turn taking demonstrates that a large brain size and cult to reconstruct from the archaeological record. That © 2014 S. Karger AG, Basel Asif A. Ghazanfar 0006–8977/14/0842–0093$39.50/0 Princeton Neuroscience Institute, Princeton University Princeton, NJ 08544 (USA) E-Mail [email protected] E-Mail asifg @ princeton.edu www.karger.com/bbe Downloaded by: Princeton University Library 198.143.38.65 - 3/22/2016 8:45:42 PM said, we can infer from indirect evidence that early homi- Given its central importance in everyday human social nids were quite social. For example, evidence of hominid interactions, it is natural to ask how vocal turn taking, a fire use dates to at least one million years ago [Berna et al., form of cooperation, evolved. It has been argued that hu- 2012] and evidence for a group of humans using fire in the man cooperative vocal communication is unique and, es- form of a hearth has recently been pegged to at least sentially, evolved in three steps (put forth most cogently 300,000 years ago [Shahack-Gross et al., 2014]. Whether by Tomasello [2008], but see also Hewes [1973] and Riz- our hominid ancestors gestured or spoke to one another zolatti and Arbib [1998] for similar scenarios). First, ape- (or did both) in order to light the hearth is an open ques- like ancestors used manual gestures to point and direct tion. Yet necessary and foundational to any instance of the attention of others. Second, later ancestors with pro- successful cooperative communication – no matter the social tendencies used manual gestures to mediate shared modality or sophistication – is the capacity to take turns. intentionality. Finally, and most mysteriously, a transi- tion from primarily gestural to primarily vocal forms of cooperative communication came about, perhaps in or- Cooperative Communication: Turn Taking in der to express shared intentionality more efficiently. Typ- Humans and Marmoset Monkeys ically, no primates other than humans are thought to ex- hibit cooperative vocal communication, the implication One way to enhance signal quality during communi- being that communication via turn taking requires a big cation is to prevent interference through taking turns. By brain and complex cognitive mechanisms. pausing after transmission, a sender allows signals from Conversely, we hypothesize that vocal turn taking other individuals to transpire and be heard before anoth- could have evolved through a voluble and prosocial an- er signal is emitted. The elimination of overlap via turn cestor without the prior scaffolding of manual gestures or taking increases the likelihood of the signal being heard big brains. To test this hypothesis, we studied the vocal accurately. As a consequence, an exchange of signals be- exchanges of a small New World primate found in South tween two or more individuals has a structure. A success- America: the common marmoset (Callithrix jacchus) ful instance of human vocal turn taking, for example, [Takahashi et al., 2013]. Marmosets are part of the Cal- would involve person 1 speaking while person 2 attends, latrichinae subfamily of the Cebidae family of New World followed by a response from person 2, be it a statement or primates. The common marmoset is approximately 20 an indication for person 1 to continue speaking. The im- cm tall, weighs an average of 400 g, and exhibits no sexu- position of such structure during communication is a key al dimorphism in body size ( fig. 1 a). Marmosets display feature of human social development [Jasnow and Feld- little evidence of shared intentionality and they do not stein, 1986; Jaffe et al., 2001]. Mothers will use pauses to produce manual gestures. Like humans, they are coop- establish rhythmicity in vocal, gestural and gaze-driven erative breeders and voluble. Marmosets are among the interactions with their infant [Jaffe et al., 2001]. The es- very few primate species that form pair bonds and exhib- tablishment of this rhythmicity has been demonstrated to it biparental and alloparental care of infants [Zahed et al., result in measurable gains in word learning [Yu and 2008]. These cooperative care behaviors are thought to Smith, 2012]. Thus, we can envision the imposition of scaffold prosocial motivational and cognitive processes, structure as a scaffold upon which a developing child can such as attentional biases toward monitoring others, the build mature communication strategies. Of course, not all ability to coordinate actions, increased social tolerance conversations in humans adhere to a strict turn taking and increased responsiveness to others’ signals [Snowdon rule. Face-to-face conversations between familiar indi- and Cronin, 2007; Burkart and van Schaik, 2010]. Besides viduals tend to be rapid speech exchanges with a substan- humans, and perhaps to some extent in bonobos [Hare et tial amount of overlap [O’Conaill et al., 1993]. However, al., 2007], this suite of prosocial behaviors is not typically between unfamiliar individuals or in unfamiliar contexts, seen in other primate species. or when individuals are out of view of one another, con- When out of visual contact, marmoset monkeys and versational exchanges solely via speech signals become other callitrichid primates will participate in vocal ex- more regulated, falling back onto basic turn taking behav- changes with conspecifics [Ghazanfar et al., 2001, 2002; ior [O’Conaill et al., 1993; Sellen, 1995]. Thus, in general Miller and Wang, 2006; Chen et al., 2009]. In the labora- and throughout human development, turn taking is a tory and in the wild, marmosets typically use ‘phees’, a foundational principle upon which more flexible com- high-pitched vocalization that can be monosyllabic or munication can occur. multisyllabic, as their contact call ( fig. 1 b, c) [Bezerra and 94 Brain Behav Evol 2014;84:93–102 Borjon /Ghazanfar DOI: 10.1159/000365346 Downloaded by: Princeton University Library 198.143.38.65 - 3/22/2016 8:45:42 PM Souto, 2008]. A phee call contains information about sex, identity and social group [Norcross and Newman, 1993; Miller et al., 2010]. As for humans, marmoset conversa- tions occur spontaneously with another conspecific re- gardless of pair-bonding status or relatedness. Marmoset vocal exchanges can last as long as 40 min and have a temporal structure that is strikingly similar to the turn taking rules used by humans [Stivers et al., 2009; Taka- hashi et al., 2013]. First, there are rarely if ever overlap- ping calls (i.e. no interruptions and thus, no interference). a Second, there is a consistent silent interval between utter- ances across two individuals ( fig. 1 b, c). The similarities 1 Marmoset 1 do not stop there. Marmoset 2 In humans, dynamical system models incorporating coupled oscillator-like mechanisms are thought to ac- 0 count for the temporal structure of conversational turn Amplitude taking and other social interactions [Chapple 1970; Oul- –1 lier et al., 2008; Schmidt and Morr, 2010]. In the vocal b 0 102030 domain, such mechanisms have two basic features: (1) Time (s) periodic coupling in the timing of utterances across two interacting individuals ( fig. 1 d) and (2) entrainment, 20 where if the timing of one individual’s vocal output quick- 15 ens or slows, the other’s follows suit. The vocal exchanges 10 of marmoset monkeys share both of these features [Taka- 5 hashi et al., 2013]. Thus, marmoset vocal communica- (kHz) Frequency 0 tion, like human speech communication, can be modeled 0102030 as loosely coupled oscillators.