Featured Research

Marmoset Murmurs: What We Can Learn from the Little Guys

Together with his team, CIN junior research group leader Steffen Hage wishes to understand the ‘Neurobiology of Vocal Communication’, i.e. the fundamentals of communication and vocal expression, which he is approaching from the level of behaviour all the way down to the single cell level. The goal is to gain insights into the way vocal utterances are produced in the brain, how they are perceived, learned and controlled. Moreover, Hage and his team seek to understand how these very basics of vocal motor and audio-vocal integration mechanisms might serve as precursors of human speech and language in the primate lineage (cf. Hage & Nieder, Trends Neurosci. 2016). In order to do so, they employ psychophysical, neuroethological and neurophysiological methodologies.

The ultimate goal of this work is to understand the mechanisms allowing the human brain to produce speech. This is important to better understand speech impediments and disorders and the role of speech in everyday social interactions, and their disturbance by afflictions such as autism. Since very little is yet known about the neuronal basis of vocalisation, about the evolution of speech and language, and its basis, cognitive vocal control, the research group’s line of investigation is intent on laying the groundworks. The group’s research requires a model that is evolutionarily very close to humans, necessitating the use of non-human primates in experiments.

Dr. Steffen Hage heads the CIN junior research group on "Neurobiology of Vocal Communication"

After his successful work on vocalising rhesus macaques, Hage decided to switch to the common marmoset (Callithrix jacchus) as his model organism of choice for the study of speech for several reasons:

  1. Marmosets are highly vocal animals – actually much more vocal than e.g. rhesus monkeys – with a rich repertoire. Moreover, they show a well-studied and well-defined counter-calling behaviour which is very useful to investigate audio-vocal integration mechanisms.

  2. Recent studies indicate that vocal development in marmoset monkeys is, at least partially, dependent on contingent parental auditory feedback (see also below).

  3. Recent studies indicate that common marmosets are capable to volitionally modulate the timing of their vocal output.

  4. Over the last years, marmosets have become an important tool to elucidate auditory processes in the primate brain, facilitating exchange with other scientists in neighbouring fields.

  5. As a New World monkey species, marmosets possess a so-called lissencephalic neocortex (a smooth brain surface). This characteristic will be of great advantage when studying the brain with large electrode arrays or neuro-imaging techniques.
The common marmoset (Callithrix jacchus) is an extremely promising animal model in neuroscience for many reasons

Establishing new model organisms in basic research can be a drawn-out and difficult procedure. Using common marmosets, however, has many advantages: they have short generational spans, are easily kept and bred and – being very small – do not require much space. For these and other reasons, the common marmoset has recently made the jump from a curiosity in just a few labs worldwide to one of the most promising model organisms in several branches of neuroscience. While much smaller than rhesus macaques, marmosets show comparably complex behaviour. These New World monkey species – just about the size of a squirrel – typically lives in family groups with 6-15 animals in the rainforests of northeastern Brazil. In these highly social groups, mothers typically give birth to twins twice a year, and both the father and older siblings carry the young as well. Marmosets communicate with each other visually and vocally; in fact, their vocal communication is much more varied and abundant than that of many other primate species, including macaques.

Hage’s lab is especially drawn by the varied vocalisation of marmosets, which manifests as a constant chatter of so-called whirrs, chirps, phees and tsiks. The animals are quite curious and might be willing to be trained to vocalise on cue: a capability they would then share with rhesus monkeys as previously shown by Hage in collaboration with CIN member Andreas Nieder (Hage & Nieder, Nat. Comm. 2013).

At the moment, the Hage lab is about to train marmoset monkeys to exhibit a similar behavioural response to affirm that both New and Old World monkey vocalisations are, to a specific degree, under cognitive control. Unlike songbirds (and humans), non-human primates do not seem to expand their vocal repertoire as they grow – the acoustic structures they can make are innate. However, monkeys are capable to control when and what to vocalise and when to be silent.

Monkeys, while not able to add to their congenital repertoire of utterances, can learn to control their vocalisation

While Hage’s research group aims to investigate links between behaviour and the underlying neuronal network employing electrophysiological methods, they have already made one unlooked-for and impressive discovery while working with the young animals and merely monitoring their behaviour (Gultekin and Hage, Nat. Comm. 2017). This is not altogether unusual: sometimes, a happy coincidence can play a significant role in science. From Alexander Fleming’s penicillin to Roy Plunkett’s teflon to William Kellogg’s Cornflakes – some of modernity’s most important inventions and discoveries have their root in chance making history. The researcher’s role in this case is to spot the remarkable coincidence and its extraordinary potential, to understand it, and to describe it scientifically.

Below you can find sonograms of some typical marmoset utterances. Click on the images to listen to  sound recordings.

A so-called double "Phee"
A "Twitter"
This is an example of a "Trill"
This is a "Tsik-Egg"

This is what happened for Hage’s research group (see our related press release here): one of the marmoset couples in their care had a litter of three infants, one of which the parents rejected. Since common marmosets usually give birth to twins, and parents are not able to fully care for three young in most cases, it is quite common that the third infant is rejected by its parents and would then die in their natural habitat. In this case, the third young animal was hand-fed by an animal caretaker at the CIN. However, marmosets are communal animals that need stable social groups. Therefore, the researchers wanted to reunify the rejected infant at least with its siblings as soon as possible.

After three months’ time, the young remaining with their parents had been weaned, and the three monkey siblings were brought together again. The resulting social group has proven stable. By now, the animals are fully grown up.

After a few weeks had passed, Yasemin Gültekin, a PhD student on the team, noticed an interesting peculiarity: the three siblings still produced typical baby monkey vocalisations in a sequence-like structure, so-called “babbling” behaviour. Normally, marmosets only show this vocal behaviour during the very first months after birth. At first the researchers assumed they would grow out of it, but even while the siblings started to add adult vocal behaviour, the rapidly growing monkeys retained the “babbling” behaviour as well. They employ both at subadult stage (at the age of 13 months), as Gültekin and Hage’s detailed records show. Hage concludes that “apparently, the animals need direct social or acoustic feedback from their parents for normal development of their vocal behaviour”.

Intrigued, the neuroscientists investigated further. When the parent monkeys had offspring a second time – twins this time –, they brought their microphones to bear. The second litter, which, as is the usual practice, were kept with the parents far longer than three months (three months being the age at which those animals of the first litter who had not been rejected were separated from their parents). The animals of this second litter developed vocalisations according to what the researchers had expected in marmosets.

At about seven months of age, they showed neither infant vocalisations nor infant ‘babbling’ anymore, only adult call behaviour.

Comparing the total of almost 14,000 individual calls on record for the five siblings, there is conclusive evidence: while marmosets may not learn twittering, pheeing and peeping from their parents, they apparently do learn what you can “say” when and where as an adult. It seems that developing the repertoire available to them heavily depends on direct feedback mechanisms much like our own. These conclusions are bolstered by recent work of Asif Ghazanfar’s work at Princeton, which shows that marmoset infants lose their infant vocalizations earlier when experiencing a high rate of contingent parental auditory feedback in comparison to those siblings perceiving a lower rate of such feedback (Takahashi et al., Curr. Biol. 2017).

This graphic may provide more insight into the call behaviour of adult marmosets raised by their parents, receiving normal amounts of parental feedback (top) and those that received limited parental feedback due to being raised apart (bottom). The normally raised marmosets exhibit no cries and next to no trills and peeps. Moreover, normally raised marmosets do not show "babbling" behaviour (here given as compound-cry). All these call types are typical of young marmosets and are still in use in adults raised with only limited parental feedback.

Differences in vocal behavior between normally-raised monkeys and monkeys with limited parental feedback after the third postnatal month. (A) Spectrograms of vocal sequences (B) Call type distributions related to call entropy and duration.
(click to enlarge)

What We Can Learn from the Little Guys – A Look to the Future

Like other nonhuman primates, marmosets live in groups. The important role of auditory communication notwithstanding, marmosets also use visual information – body posture, facial expressions, gaze direction and actions – to interact with others. Consequently, these animals are not only an ideal model for the investigation of vocal communication, but also for the brain machanisms involved in visually guided social interactions, as well as the question how auditory and visual communicative signals are integrated in the brain.

Previous work of Peter Thier´s team at the Hertie Institute for Clinical Brain Research has been able to identify key regions in the frontal and temporal lobes of the cortex in humans and rhesus monkeys which serve the processing of visual communicative signals. Thier´s team is now establishing the marmoset monkey as a complementary animal model to get a handle on the phylogeny of these brain regions and make use of the methodological advantages of the marmoset´s flat cortex. Together, Steffen Hage and Peter Thier are planning to decipher how the marmoset´s brain integrates the vocal and visual signals that guide their social interactions.

Humans may suffer from severe disturbances of normal social interactions such as in autism and schizophrenia. Many of these disorders have a genetic basis or are at least dependent of particular genetic profiles. Although many of the key genetic alternations have been established, how they lead to behavioural deficits remains nebulous, largely as a consequence of the emphasis on rodents as animal model. With the availability of the marmoset – a model system much closer to humans – and the advent of genetic modifications in marmosets we may hope that we will learn much more about the link between genes and social behaviour. In the long run, this promises new approaches to a number of diseases. The marmoset model suggests a promising pathway to a better understanding of social communication and its disorders.

As is usually the case in basic research, we do not know whether this path will lead to the desired goal. It may yet turn out to be much longer and less straight than expected from today’s perspective. But for the time being, it looks very promising and is quite exciting!


Gultekin Y. B., Hage S. R. (2017): Limiting Parental Feedback Disrupts Vocal Development in Marmoset Monkeys. Nature Communications 7: 14046.
doi: 10.1038/ncomms14046

Hage S. R., Nieder A (2013): Single Neurons in Monkey Prefrontal Cortex Encode Volitional Initiation of Vocalizations. Nature Communications 4: 2409.
doi: 10.1038/ncomms3409

Hage S. R., Nieder A. (2016): Dual Neural Network Model for the Evolution of Speech and Language. Trends in Neurosciences 39: pp. 813–829.
doi: 10.1016/j.tins.2016.10.006

Sasaki E., et al. (2009): Generation of Transgenic Non-Human Primates with Germline Transmission. Nature 459: pp. 523–527.
doi: 10.1038/nature08090

Takahashi D. Y., Liao D. A., Ghazanfar A. A. (2017): Vocal Learning via Social Reinforcement by Infant Marmoset Monkeys. Current Biology 27: pp. 1844–1852.
doi: 10.1016/j.cub.2017.05.004