TL;DR
This study uses fMRI to compare how human and monkey brains process vocalizations, revealing shared early auditory processing but distinct higher-level specialization in humans for speech. It highlights the evolutionary expansion of the human superior temporal sulcus (STS) for language, a region less involved in vocalization processing in monkeys.
What Problem Does This Paper Address?
The evolutionary origins of human language are a profound mystery. While vocalizations are crucial for social interaction in many species, it's unclear how much of our brain's capacity for processing speech is shared with our primate relatives. This paper aims to bridge this gap by directly comparing the neural mechanisms underlying vocalization processing in humans and macaque monkeys using fMRI, seeking to understand which brain regions are conserved and which have evolved specifically for human language.
Methods
The researchers employed functional Magnetic Resonance Imaging (fMRI) to scan both human and macaque monkey participants while they listened to various auditory stimuli. These stimuli included:
- Monkey vocalizations (Mvoc): Natural calls from rhesus monkeys.
- Human emotional vocalizations (Hemo): Non-linguistic sounds like laughter and cries.
- Intelligible speech (French): Spoken French, the native language for human participants.
- Unintelligible speech (Arabic): Spoken Arabic, unintelligible to the human participants.
- Bird songs (Bs): Included for human participants.
- Scrambled controls: These were created by processing the natural sounds through a filterbank and shuffling segments to preserve spectral characteristics but destroy intelligibility and naturalness. These served as a baseline to isolate the processing of natural vocalizations.
The fMRI data were analyzed using SPM software, with species-specific preprocessing and normalization to anatomical templates. A key methodological aspect was the attempt to keep experimental conditions as similar as possible across species, including the use of a visual fixation task to minimize movement. For monkeys, a contrast agent (MION) was used to enhance the MR signal, reflecting cerebral blood volume (CBV) changes, while for humans, the standard Blood Oxygenation-Level Dependent (BOLD) signal was measured.
Datasets
The study utilized existing datasets of monkey vocalizations (Rhesus Monkey Repertoire) and bird songs. Human and monkey vocalizations were recorded specifically for this study. No new datasets or benchmarks were introduced.
Key Insights and Ideas
- Shared Early Auditory Processing: Both humans and monkeys show similar patterns of activation in early auditory areas (like Heschl's gyrus) when processing scrambled, non-meaningful sounds. This suggests a conserved fundamental auditory processing system across species.
- Higher-Level Specialization in Humans: While both species show activations in the superior temporal gyrus (STG) for vocalizations, humans exhibit a distinct specialization in the superior temporal sulcus (STS) for processing human speech, particularly intelligible speech. This region appears to be heavily recruited for language in humans.
- Monkey Vocalization Processing is Less Specialized: Monkeys do not show a clear "species-specific" response in the STS to their own vocalizations. Their processing of monkey calls appears to engage regions similar to those processing unintelligible speech or emotional sounds in humans, suggesting a less complex or specialized neural architecture for vocal communication compared to humans.
- The STS as a Language Hub: The study strongly suggests that the human STS has been significantly repurposed or expanded for language processing, a stark contrast to its role in monkeys. This highlights a key evolutionary divergence in the neural basis of communication.
- Intelligibility Matters: In humans, the intelligibility of speech (French vs. Arabic) elicits stronger responses in the STS, underscoring the importance of semantic and syntactic processing for language comprehension.
Key Results
- Early Auditory Cortex: Both species showed widespread activations in response to scrambled stimuli, extending beyond primary auditory areas into frontal and parietal regions. Small regions near primary auditory cortex were sensitive to spectral differences within scrambled stimuli.
- Higher-Order Auditory Processing: When comparing intact vocalizations to scrambled controls, both species showed activations in the STG. However, in humans, speech stimuli (French) elicited significantly stronger responses in the STG/STS compared to non-speech vocalizations (monkey calls, emotional sounds).
- Species Specificity: Monkeys did not show significant species-specific responses in the STS to monkey calls. In contrast, humans showed a clear preference for human vocalizations over monkey calls in the STG and STS.
- Speech Processing: Intelligible speech (French) in humans activated the STS more strongly than unintelligible speech (Arabic), and both activated the STS more than non-linguistic vocalizations. The left inferior frontal gyrus and left precentral gyrus were also recruited for intelligible speech.
Limitations and Future Work
- Scanning Differences: The authors acknowledge differences in scanning procedures and analysis parameters between species (e.g., voxel size, MR signal type, smoothing kernel), which could affect direct comparisons, though they argue the statistical power was comparable.
- Stimulus Presentation: While efforts were made to match stimuli, the number of segments concatenated differed between species, and the monkey calls were from unfamiliar individuals.
- Passive Listening Task: The study used a passive fixation task, meaning participants were not actively engaged in discriminating or understanding the sounds. The authors suggest that future work could explore how active tasks might influence auditory processing.
- Future Directions: The study opens avenues for further research into the evolutionary trajectory of vocalization processing, the specific roles of different sub-regions within the STS, and how these neural circuits might interact with other cognitive functions like social cognition and action processing. The authors suggest that the observed differences reflect a genuine species divergence in the neural architecture for vocal communication.
Summary updated