Model 2: Processing of Correctly Categorized Species trials

Figure 58: Wholebrain results when selectively contrasting processing of bonobo and macaque vocalizations against human voices. Enhanced brain activity for bonobo (ab) and macaque (cd) compared to human voice (dark blue to green). ac: left hemisphere; bd: right hemisphere. Brain activations (a-d) are independent of low-level acoustic parameters for all species (fundamental frequency ‘F0’ and mean energy of vocalizations). Data corrected for multiple comparisons using FDR at a threshold of p<.05. Hum: human; Chimp: chimpanzee; Bon: bonobo; Mac: macaque.

5.5.1.2. Model 2: Processing of Correctly Categorized Species trials

We computed the same contrasts of interest as in model 1 and observed the enhancement of similar brain regions as above for chimpanzee vocalizations. Thus, brain activity specific to

the correct categorization of chimpanzee vocalizations ([chimpanzee hits > human hits, bonobo hits, macaque hits]) was enhanced in IFG bilaterally (see Figure 59 ab). When directly contrasting hits for chimpanzee vs human vocalizations ([chimpanzee hits > human hits]), activity was enhanced in similar regions as mentioned above in bilateral IFG (see Figure 59 cd). Following this, to the exception of the left pars orbitalis, the recognition of macaque calls [macaque hits > human hits]) also led to an increase of hemodynamic responses as described above in bilateral IFG (see Figure 60 cd). On the contrary, the correct categorization of bonobo vocalizations [bonobo hits > human hits] enhanced bilaterally activity in the pars triangularis only (see Figure 60 ab). As in model 1, we found no significant activity in IFG for the correct recognition of bonobo and macaque vocalizations compared to other species’ vocalizations.

Figure 59: Wholebrain results when selectively contrasting correct categorization of chimpanzee against other species’ vocalizations. Enhanced brain activity for chimpanzee compared to other species’ vocalizations (ab) and chimpanzee calls compared to human voices (cd) (dark red to yellow). ac: left hemisphere; bd: right hemisphere. Brain activations (a-d) are independent of low-level acoustic parameters for all species (fundamental frequency ‘F0’ and mean energy of vocalizations). Data corrected for multiple comparisons FDR at a threshold of p<.05. Hum: human; Chimp: chimpanzee; Bon: bonobo; Mac: macaque.

Figure 60: Wholebrain results when selectively contrasting correct categorization of bonobo and macaque against other species’ vocalizations. Enhanced brain activity for bonobo (ab) and macaque (cd) vocalizations compared to human voices (dark red to yellow). ac: left hemisphere;

bd: right hemisphere. Brain activations (a-d) are independent of low-level acoustic parameters for all species (fundamental frequency ‘F0’ and mean energy of vocalizations). Data corrected for multiple comparisons using voxel-wise false discovery rate (FDR) at a threshold of p<.05. Hum:

human; Chimp: chimpanzee; Bon: bonobo; Mac: macaque.

5.5.2. Accuracy

As described in Chapter 4, we found significant differences regarding the recognition of the four species by human participants (F(3) = 69.097 p<.001). Participants categorized human

voices more accurately compared to chimpanzee (χ²(1) = 73.337, p<.001), bonobo (χ²(1) = 196.54, p<.001), and macaque vocalizations (χ²(1) = 95.433, p<.001). Participants were also better at categorizing chimpanzee vocalizations compared to bonobo calls (χ²(1) = 29.761, p<.001). Note that the contrast [chimpanzee vs macaque] did not reach significance after Bonferroni correction (see Figure 61).

Figure 61: Rate (%) and SD of human recognition of primate species through their vocalizations.

The dotted line represents the 25% chance level.

5.6. Interim Discussion

The present study emphasizes for the first time the role of IFG in the human categorization of heterospecific vocalizations expressed by phylogenetically close species, namely other primates.

Wholebrain analyses indeed revealed an enhancement of activity in bilateral IFG^tri, IFG^ope and IFG^orb for the human categorization of chimpanzee calls compared to human, bonobo and macaque vocalizations. Moreover, to the exception IFG^orb for bonobos, similar fMRI activity maps were found within the bilateral IFG for the categorization of chimpanzee, bonobo and macaque screams when contrasting to human voices. These findings extend previous fMRI results showing a gradient of activations in the left IFG (uncorrected threshold of p < .001) for the human discrimination of emotions in human voice compared to chimpanzee vocalizations and then compared to macaque calls (Fritz et al., 2018).

Interestingly, while the correct recognition of chimpanzee vocalisations also led to an

involvement of bilateral IFG, the correct categorization of bonobo and macaque calls enhanced activity in specific subparts of the frontal regions with respectively IFG^tri and IFGtri/ope.

Extending the existing literature on human auditory processing of human voices, and in light of our findings, we can surmise that the categorization of NHP vocalizations require the involvement of i) the right IFG for the processing of low auditory variations (Frühholz &

Grandjean, 2013b; Schirmer & Kotz, 2006); ii) the left IFG for the decoding of short scale information such as the call roughness (Frühholz & Grandjean, 2013b; Grandjean, 2020); iii) the bilateral IFGtri and IFGope for the general processing of decision making in heterospecific vocalizations (Dricu et al., 2017); and iv) the bilateral IFG^orb for the integration of prosodic modulation (Belyk et al., 2017) in chimpanzee calls only via its connection to the posterior superior temporal sulcus (Frühholz et al., 2015).

In line with our fMRI results and previous comparative studies (Kelly et al., 2017), the behavioural data also highlighted the difference of recognition between the two great ape species. Human participants were indeed capable of identifying human voices, chimpanzee vocalizations as well as macaque calls but were unable to do so for bonobo calls.

Consequently, the peculiar evolutionary pathway of bonobos as well as the infrequent acoustic features involved in their calls (Grawunder et al., 2018; Hare et al., 2012) seem to prevent human participants from accurately recognizing the bonobo species, despite their close phylogenetic proximity with Homo sapiens and their affiliation to the great ape clade.

Finally, regarding the correct categorization of macaque calls, previous comparative studies interestingly demonstrated the relationship between the capacity of humans to accurately recognize the emotional contents in vocalizations expressed by macaques (Macaca sylvanus and Macaca arctoides) and the modulation of acoustic features such as fundamental frequency (Filippi et al., 2017) and energy (Linnankoski et al., 1994).

Overall, humans would be able to identify most of the NHP species, enhancing activity in bilateral IFG. However, the involvement of all subparts of IFG requires vocalizations expressed by evolutionary, acoustically and behaviourally close species to humans.

To conclude, the present study revealed for the first time the sensitivity of bilateral IFG to NHP vocalizations, highlighting the evolutionary function of these specific frontal regions.

Furthermore, the absence of activations in IFGope and IFGorb as well as the failure of human

divergence in such processes. Overall, our results support the hypothesis of a continuum in the primate brain evolution and will hopefully contribute to a better understanding of IFG functions in the human brain. Future studies focused on emotional judgement may be particularly informative in this matter. Most NHP vocalizations are indeed expressed in a motivational or emotional context that modulates the prosodic feature, which in turn have been found to influence the categorization processing at behavioural and cerebral level in human voices (Belin, 2006; Filippi et al., 2017; Fritz et al., 2018; Kelly et al., 2017; Scheumann et al., 2014, 2017). We expect a similar pattern to emerge in the listening of closely related vocalizations, underlying once again the close continuity at the brain level amongst all primates including humans.

5.7. Supplementary Material

Reaction time

ANOVA analysis revealed significant differences in participants’ reaction time between the species recognition (F(3) = 87.264, p<.001) . All contrasts were found significant after Bonferroni correction excepted the following contrasts chimpanzee vs macaque and bonobo vs macaque.

Therefore, participants showed a faster reaction time in categorizing human voices compared to chimpanzee (χ2(1) = 113.2, p<.001), bonobo (χ2(1) = 126.95, p<.001), and macaque vocalizations (χ2(1) = 218.74, p<.001). Participants were also faster to recognize human and then chimpanzee and bonobo vocalizations compared to macaque calls (χ2(1) = 265.56, p<.001).

Figure 62: Reaction time (ms) and SD of human recognition of primate species through their vocalizations. All contrasts were significant after Bonferroni correction with Pcorrected = .01.

Chapter 6.

Brain Activation Lateralization in Monkeys (Papio anubis) following Asymmetric Motor and Auditory stimulations through functional Near Infrared Spectroscopy

Coralie Debracque*, Thibaud Gruber*, Romain Lacoste, Didier Grandjean‡ and Adrien Meguerditchian‡

Submitted

6.1. Abstract

Hemispheric asymmetries have long been seen as characterizing the human brain; yet, an increasing number of reports suggest the presence of such brain asymmetries in our closest primate relatives. However, most available data in non-human primates have so far been acquired as part of neuro-structural approaches such as MRI, while comparative data in humans are often dynamically acquired as part of neuro-functional studies. In the present exploratory study in baboons (Papio anubis), we tested whether brain lateralization could be recorded non-invasively using fNIRS device in two contexts: motor and auditory passive stimulations. Under light propofol anaesthesia monitoring, 3 adult female baboons were exposed to a series of i) left- versus right-arm passive movement stimulation; and ii) left- versus right-ear versus stereo auditory stimulations while recording fNIRS signals in the brain related areas (i.e., motor central sulcus and superior temporal cortices respectively).

For the motor condition our results show that left-arm versus right-arm stimulations induced typical contralateral difference in hemispheric activation asymmetries in the 3 subjects for all 3 channels. For the auditory condition, we also revealed typical human-like patterns of hemispheric asymmetries in 1 subject for all three channels, namely i) typical contralateral differences in hemispheric asymmetry between left-ear versus right-ear stimulations, and ii) a rightward asymmetry for stereo stimulations. Overall, our findings support the use of fNIRS to investigate brain processing in non-human primates from a functional perspective, opening the way for the development of non-invasive procedures in non-human primate brain research.

Keywords: fNIRS, hemispheric lateralization, primate, motor perception, auditory perception

6.2. Introduction

Lateralization is often presented as a key characteristic of the human brain, which separates it from other animal brains (Eichert et al., 2019); yet, an increasing number of studies, particularly in non-human primates (from here onward, primates), dispute this claim in a broad array of topics ranging from object manipulation, gestural communication to producing or listening to species-specific vocalizations (Fernández-Carriba et al., 2002;

Hook-Costigan & Rogers, 1998; Lindell, 2013; Margiotoudi et al., 2019; Meguerditchian et al., 2012, 2013). For instance, several primate studies present behavioral evidence of manual lateralization (Fitch & Braccini, 2013; Meguerditchian et al., 2013) which have been shown to be associated with contralateral hemispheric correlates at the neuro-structural level (Margiotoudi et al., 2019; Meguerditchian et al., 2012). Other examples show orofacial asymmetries during vocal production, as evidenced by more pronounced grimaces on the left side of the mouth, which is suggestive of right hemisphere dominance in monkeys and great apes (Fernández-Carriba et al., 2002; Hook-Costigan & Rogers, 1998), as has been documented in humans (Moreno et al., 1990). In addition, comparative structural neuroimaging has shown that particular areas known to be leftwardly asymmetric in humans such as the Planum Temporale in the temporal cortex presented also leftward asymmetry in both monkeys and great apes (Gannon et al., 1998; Hopkins et al., 2015; Marie et al., 2018; Pilcher et al., 2001), although the bias at the individual level seems to be more pronounced in humans (Rilling, 2014; Yeni-Komshian & Benson, 1976).

At the neural functional level using fMRI or Positron Emission Tomography (PET) scan, most available studies in nonhuman primates focused on lateralization of perception of synthesized sinusoidal or more complex vocal signals and reported inconsistent results. For instance, in rhesus macaques, the processing of species-specific and/or heterospecific calls as well as non-vocal sounds, elicited various patterns of lateralized activations within STG such as in the left lateral parabelt, either toward the right hemisphere or the left depending on the study (Gil-da-Costa et al., 2006; Joly et al., 2012; Petkov et al., 2008; Poremba et al., 2004). In chimpanzees, a similar PET study reported a rightward activation within STG for processing conspecific calls (Taglialatela et al., 2009). In general, such a variability of

direction of hemispheric lateralization for processing calls appears similar to hemispheric lateralization’s variability described in humans for language processing depending of the type of auditory information and of language functions that are processed (Belin et al., 2000;

Schirmer & Kotz, 2006; Zatorre & Belin, 2001).

Compared to the leftward bias suggested for language, research investigating emotion perception in primates has strengthened the idea of a right bias in lateralization specific to emotion processing (Lindell, 2013). For example, Parr and Hopkins found that right ear temperature increased in captive chimpanzees when they were watching emotional videos, consistent with a greater right hemisphere involvement (Parr & Hopkins, 2000). The rightward hemisphere bias documented in chimpanzees is also found in other primate species such as olive baboons during natural interactions, as evidenced by studies investigating the perception of visual emotional stimuli (Baraud et al., 2009; Casperd &

Dunbar, 1996; Wallez & Vauclair, 2011). Yet, while the right hemisphere has understandably received much focused, the left hemisphere is also involved for emotion processing. For example, Schirmer and Kotz have suggested that the left hemisphere is particularly involved in the processing of short segmental information during emotional prosody decoding (Schirmer & Kotz, 2006). Whether this functional differentiation, essential for speech perception in humans (Grandjean, 2020), is also present in non-humans is unclear. Baboons appear in this respect a particularly interesting animal model to study for lateralization, with several recent studies underlying the similarities in manual and brain asymmetries with humans (Margiotoudi et al., 2019; Marie et al., 2018; Meguerditchian & Vauclair, 2006).

Furthermore, the baboon brain is on average twice as large as the macaque brain (Leigh, 2004), which may facilitate the specific investigation of sensory regions. Finally, this species has all the primary cortical structures found in humans (Kochunov et al., 2010).

However, a major drawback in current studies lies in the complexity with which brain asymmetry can be investigated comparatively in primates. Here, we used fNIRS in baboon brains to test whether the blood oxygen level dependent (BOLD) response differed accordingly between the two hemispheres following left- versus right-asymmetric auditory and motor stimulations. fNIRS is a non-invasive optical imaging technique that has been developed to investigate brain processes in potentially at-risk populations such as human premature new-borns, but which is now widely used with adult human participants. fNIRS

research(Boas et al., 2014). Considering its portability and its lessened sensitivity to motion artefacts (Balardin et al., 2017) compared to other non-invasive techniques, it might be an excellent methodology to study brain activations in primates under more ecologically relevant testing conditions, for example with a wireless and wearable device. As a first step, the present study tested the fNIRS in baboons immobilized under light anesthesia monitoring. In relation with each of the stimulation types, we targeted relevant corresponding brain regions of interest – the motor cortex within the central sulcus and the auditory cortex regions in the temporal lobe respectively - in both hemispheres by positioning the two sets of fNIRS channels (one by hemisphere for a given region). We predicted that, if fNIRS was suitable to record brain signal in baboons, it would reflect contralateral hemispheric asymmetries in signals for each stimulation type within their corresponding brain region of interest, namely the motor cortex, associated with right- versus left-arm movements, and the temporal cortex, associated with the right- versus left- versus stereo ear auditory presentations. Our latter prediction was modulated by the knowledge that auditory regions are less lateralized, with about fifty percent of fibers projecting in the bilateral regions (Robinson & Burton, 1980; Smiley & Falchier, 2009), compared to cortical motor regions.

6.3. Materials and methods

6.3.1. Subjects

We tested 3 healthy female baboons (Talma, Rubis and Chet, mean age = 14.6 years, SD ± 3.5 years). The subjects had normal hearing abilities and did not present a neurological impairment. All animal procedures were approved by the “C2EA -71 Ethical Committee of neurosciences” (INT Marseille), and were conducted at the Station de Primatologie CNRS (UPS 846, Rousset-Sur-Arc, France) within the agreement number C130877 for conducting experiments on vertebrate animals. All methods were performed in accordance with the relevant French law, CNRS guidelines and the European Union regulations (Directive 2010/63/EU). All monkeys were born in captivity from 1 (F1) or 2 generations (F2) and are housed in social groups at the Station de Primatologie in which they have free access to both outdoor and indoor areas. All enclosures are enriched by wooden and metallic climbing structures as well as substrate on the group to favour foraging behaviours. Water

is available ad libitum and monkey pellets, seeds, fresh fruits and vegetables were given every day.

6.3.2. Subject’s hand Preference in Communicative Gesture and Bi-Manual task Impacts of subject’s handedness on cerebral lateralization of language, motor and visual functions are well known in human neuroscience (35). For that purpose, we report here the hand preference of individual baboons during visual communicative gesture (CG slapping one hand repetitively on the ground in direction to a conspecific to threaten him) and a bi-manual tube task (BM - holding a PVC tube with one hand when removing the food inside the tube with the fingers of the other hand). Hence, in both contexts, Talma was left-handed (CG: n=27, HI=-0.56, z-score=-2.89; BM: n=31, HI=-0.42, z-score=-2.33) whereas Rubis showed a preference toward the right hand (CG: n=16, HI=0.25, z-score = 1; BM: n=79, HI= 1, z-score=8.88). On the other hand, Chet was left-handed in communicative gesture (n=25, HI

= -0.44, z-score = -2.2) but right-handed in the bi-manual tube task (n=11, HI = 0.45, z=score = 1.51).

6.3.3. Recordings

We selected one of the most wearable, wireless and light fNIRS device available on the market (Portalite, Artinis Medical Systems B.V., Elst, The Netherlands) to measure the brain activations in baboons during the motor and auditory stimulations. The data were obtained at 50 Hz using six channels (three by hemisphere), three inter-distance probes (3 – 3.5 – 4 cm) and two wavelengths (760 and 850 nm). To localize our regions of interests (ROIs), the motor and auditory cortices, the fNIRS probes were placed using T1 MRI scanner images previously acquired by the LPC group on baboons (see Figure 63).

Each fNIRS session was planned during a routine health inspection undergone by the baboons at the Station de Primatologie. As part of the health check, subjects were isolated from their social group and anesthetized with an intramuscular injection of ketamine (5 mg/kg - Ketamine 1000®) and medetomidine (50µg/kg - Domitor®). Then Sevoflurane (Sevotek®) at 3 to 5% and atipamezole (250 µg/kg - Antisedan®) were administered before recordings. The area of interest on the scalp was shaved. Each baboon was placed in ventral decubitus position on the table and the head of the individual was maintained using foam positioners, cushions and Velcro strips to remain straight and to reduce potential motion

a drip of NaCl was put in place during the entire anaesthesia. Just before recording brain activations, sevoflurane inhalation was stopped and the focal subject was further sedated with a minimal amount of intravenous injection of Propofol (Propovet®) with a bolus of around 2mg/kg every 10 to 15 minutes or by infusion rate of 0.1 – 0.4 mg/kg/min. After the recovery period, baboons were put back in their social group at the Station de Primatologie and monitored by the veterinary staff.

Figure 63: Schematic representation of fNIRS channel locations on ROIs according to T1 MRI template from 89 baboons (Love et al., 2016) for (a) the motor and (b) the auditory stimulations.

Red and blue dots indicate receivers and transmitters’ positions respectively. Yellow dots indicate the channel numbers.

6.3.4. Motor stimulations

The motor stimulations consisted of 20 successive extensions of the same arm, alternatively right and left repeated three times according to the same set plan (L-R-R-L-L-R) for all baboons, resulting in a total of 120 arm movements. One experimenter on each side of the baboon extended slowly their respective arm while stimulating the interior side of the hand (gentle rhythmic tapping) with their fingers throughout the duration of the extension (about

5s) upon a brief vocal command triggered by another experimenter. Between each block, there was a 10s lag.

6.3.5. Auditory stimulations

The auditory stimuli consisted of 20s long series of agonistic vocalizations of baboons and of chimpanzees recorded in social settings (in captivity in an outside enclosure for baboons;

and in the wild for chimpanzees). Equivalent white noise stimuli matched for the energy dynamics (i.e. the sound envelopes) were produced and used for comparison to control for the sound energy dynamic differences. In the present study and analysis, we only examine the effect of the lateralization of auditory stimulations (i.e., left ear versus right ear versus stereo) as a whole on hemispheric asymmetry and thus do not distinguish between auditory signal types or species (e.g. white noise and vocalizations). The auditory stimuli were

and in the wild for chimpanzees). Equivalent white noise stimuli matched for the energy dynamics (i.e. the sound envelopes) were produced and used for comparison to control for the sound energy dynamic differences. In the present study and analysis, we only examine the effect of the lateralization of auditory stimulations (i.e., left ear versus right ear versus stereo) as a whole on hemispheric asymmetry and thus do not distinguish between auditory signal types or species (e.g. white noise and vocalizations). The auditory stimuli were