Categorization and discrimination of human and non-human primate affective vocalizations: Investigation of frontal cortex activity through fNIRS

Abstract

Previous research has highlighted the involvement of frontal regions in human participants while they engaged in the explicit decoding, such as categorization (AvsB) and discrimination (Avsnon-A), of affective signals. Given its adaptive value and deep evolutionary history, this human capacity to recognize the affective content in human calls is likely to extend to the vocalizations of other closely related species, such as non-human primates. However, few comparative studies have thus far investigated this process at both the behavioral and neural levels. Here, we aimed to study the role of frontal regions in human participants while they engaged in the explicit affective content decoding of primate calls using functional Near Infrared Spectroscopy (fNIRS). Specifically, we recorded frontal regions of participants while they categorized or discriminated positive and negatively valenced vocal signals produced by four different primates: humans, chimpanzee and bonobo (both great apes species), and rhesus macaques (a more distant species). We also analyzed whether behavioral responses correlated with recorded frontal activations. fNIRS data revealed more activations within the inferior frontal cortexpars triangularis(IFC_tri), the frontopolar (FPC), and middle frontal cortices (MFC) during discrimination compared with categorization. Activity in these regions was modulated by both the species and the type of task, with greater activity during the discrimination of agonistic chimpanzee calls compared with categorization. Categorization was itself characterized by a decrease of frontal activity during the correct recognition of all chimpanzee calls, and of affiliative rhesus macaque and agonistic bonobo vocalizations. Our results also highlighted behavioral differences related to the type of task. Participants discriminated almost all affective cues of all four species vocalizations above chance level. In comparison, they correctly categorized the affective content of most human and great ape vocalizations above chance level, but not those of rhesus macaque calls, highlighting an effect of both phylogenetic relatedness and the type of task. Overall, these findings support the hypothesis of an evolutionary ancient affective recognition processing system situated in the frontal cortex, inherited from our last common ancestor with other great apes.

Keywords: NIRS; affect; frontal; primate; recognition; vocalization.

Fig. 1.

Representative waveforms of 750-ms-long angry/threatening vocalizations expressed by human (in blue), chimpanzee (in green), bonobo (in orange), and rhesus macaque (in pink) species. These graphical representations were extracted using the PhonTools package (Barreda, 2015) in Rstudio (Rstudio Team, 2020).

Fig. 2.

Probe locations into the MNI space by using SPM12 software implemented in MatLab R2018b (www.fil.ion.ucl.ac.uk/spm/). Red and blue dots indicate transmitters and receivers’ positions, respectively. Yellow dots indicate the channel numbers.

Fig. 3.

Structure of the experiment, with each of the 6 blocks made of 12 mini-blocks, which in turn comprised 12 individual trials.

Fig. 4.

Mean and SE of human recognition of human (in blue), chimpanzee (in green), bonobo (in orange), and rhesus macaque (in pink) affective vocalizations for categorization (CAT) and discrimination (DIS) tasks and the different kinds of affective states. All contrasts were significant within each condition after Bonferroni correction with P_corrected= .05/24 = .002, excluding the following contrasts: chimpanzeevsrhesus macaque and bonobovsrhesus macaque for affiliative cues and bonobovsrhesus macaque for threatening contents in discrimination task (seeSupplementary Material Table 2).

Fig. 5.

Mean and SE of concentration changes of O₂Hb (µM) in right and left FTC/MFC and IFC_triduring the categorization and the discrimination tasks by human participants of primate affective vocalizations. N = 20. ***p < .001, *p < .05.

Fig. 6.

Interaction between participants’ accuracy and O₂Hb concentration changes in IFC_triwithin each affect and species for (A) categorization and (B) discrimination. Confidence interval at 0.95. Figures were made on Rstudio using the package Visreg (Breheny & Burchett, 2017).