Symmetry of Broca’s area homolog in chimpanzee and a surface probabilistic atlas

Through a semisupervised pipeline (summarized in Fig 1A), we precisely reconstructed the cortical surface of 9 chimpanzee brains from their structural MRI data. Having their surface representation, we projected both BA44 and BA45 volumetric histological segmentations [29] to each individual surface. We examined all individuals for evidence of surface area asymmetry of both histologically defined regions (BA44 and BA45) using a Wilcoxon signed-ranks test. Although there was considerable asymmetry in some individuals (see Fig 1B), both BA44 and BA45 showed no asymmetry at the population level (BA44 T = 16, p = 0.49; BA45 T = 10, p = 0.16).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1.

(A) Reconstruction pipeline for the cytoarchitectonic surface maps. First, the raw MRI data were cleaned using noise reduction and contrast inversion. Next, the individual surfaces were reconstructed in FreeSurfer. The individual maps of BA44 and BA45 are displayed in black and yellow, respectively. Finally, the individual surfaces and cytoarchitectural maps were registered to the JUNA template surface (B) Probabilistic atlas of regions BA44 and BA45 in the chimpanzee brain, derived from the individual maps, alongside the lateralization index for each individual brain. The underlying data can be found in the GitHub/Zotero repository, under the results/chimpanzee-atlas folder.

https://doi.org/10.1371/journal.pbio.3002266.g001

To enable comparison across subjects, we coregistered the individual brain surfaces to the surface reconstruction of the JUNA [34] chimpanzee template (see Fig 1A). On the JUNA surface, we averaged all the individual segmentations, deriving a high-quality probabilistic atlas of BA44 and BA45 homologs in the chimpanzee brain (Fig 1B). The resulting atlas is open access and available for direct download (see Data and Code Availability Statement).

Comparison between human Broca’s area and its chimpanzee homolog

Leveraging advanced surface registration [31,33], we coregistered the JUNA surface to the surface reconstruction of the MNI-2009c human template (Fig 2A) [35]. This enabled us to compare the human BA44 and BA45 histological atlases derived by Amunts and colleagues [21], with our probabilistic atlas of the chimpanzee homolog (Fig 2B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

(A) Two-step surface registration; in the first step, we align gross anatomical landmarks. This first alignment is then used to start a more granular one, based on sulcal depth. (B) Side-by-side comparison of our chimpanzee probabilistic atlas with the human population overlap of Amunts and colleagues [21] in the human brain template. Left BA44 is the area that grew the most and shows a large anterior expansion, which is not present in right BA44. The underlying data and scripts used can be found in the GitHub/Zotero repository, under the scripts/ and results/human-comparison folders.

https://doi.org/10.1371/journal.pbio.3002266.g002

After projecting the chimpanzee segmentations to the human brain, we computed their volumes using the MNI template’s cortical thickness. We found the projected chimpanzee BA44 to have an average size of 2,331 mm³ (SD: 789) in the left hemisphere, and 1,955 mm³ (SD: 935) in the right hemisphere. In contrast, Amunts and colleagues [21] reported the human BA44 to have an average size of 3,839 mm³ (SD: 2,277) in the left hemisphere, and 2,527 mm³ (SD: 1,597) in the right hemisphere. This means that, when scaled and projected to the same surface template, the human BA44 is 1.64 times larger in the left hemisphere than in the chimpanzee, and 1.29 times larger in the right hemisphere. Moreover, Fig 2B shows that such enlargement is likely the result of a substantial anterior expansion, not present in the right BA44.

For the chimpanzee BA45, the average size after projecting to the human brain (Fig 2B) was 3,187 mm³ (SD: 1,002) and 2,329 mm³ (SD: 1,308) for the left and right hemispheres, respectively. For the same region in humans, Amunts and colleagues [21] reported an average size of 3,242 mm³ (SD: 1,149) and 3,173 mm³ (SD: 1,637) for the left and right hemispheres, respectively. The human BA45 was only 1.02 times larger in the left hemisphere than the chimpanzee’s homolog area, while being 1.36 times larger in the right hemisphere.

Comparing the projection of chimpanzee BA44 with human functional maps

To better understand the behavioral role of the observed expansion, we computed the overlap of the projected chimpanzee BA44 with functional subdivisions of human BA44 related to action and syntax [17,36–38]. To compare only with the core chimpanzee BA44, we thresholded the projected atlas at the 0.5 level. We found that the chimpanzee BA44 overlapped most with the regions involved in action [17,36] (Figs 3, left, and S1 and Table 1). The highest overlap was found with the area Clos 4, associated with action imagination [36], of which 34% was contained by the chimpanzee’s BA44. Following this were the regions Clos 1 (26% contained, associated with phonology and overt speech tasks [36]), Clos 5 (20%, associated with phonology and semantics [36]), Papitto’s region (18%, associated with action execution/imitation [17]), and Clos 2 (7%, associated with semantics, orthography, and covert speech [36]). In contrast, the region Clos 3, associated with basic syntactic operations [36–38], had only a 3% overlap with the chimpanzee BA44. Similar results were obtained when comparing across different levels of thresholding (see S1 Fig). Finally, when visually compared, the projected chimpanzee BA44 shared spatial location and extension with the receptorarchitectural division of the human BA44 [22,23] (see S2 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Percentage of overlap between the chimpanzee BA44 and functional subdivisions of the human BA44 [17,36–38].

Action-related regions present the highest overlap with action-related areas and virtually no overlap with the syntax area. The chimpanzee BA44 atlas was thresholded at 0.5 to maintain only its core area. The functions being reported are those with the highest P (Activation | Domain) as reported by Clos and colleagues [36] and Papitto and colleagues [17], except Clos 1, which was originally reported to be a syntax area, but further studies did not find to be involved in basic syntactic operations [36,37]. The underlying data and scripts used can be found in the GitHub/Zotero repository, under the scripts/ and results/human-comparison folders.

https://doi.org/10.1371/journal.pbio.3002266.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Anatomical overlap between functional areas [17,36] with the projected BA44 chimpanzee thresholded at 0.5.

The functional role of each area, as reported by their authors, is stated in the third column. For Clos regions, we report the functions with the highest P (Activation | Domain).

https://doi.org/10.1371/journal.pbio.3002266.t001

BA44 became increasingly left lateralized in evolution

We tested for asymmetry in surface area for both BA44 and BA45 in chimpanzee brains. Consistent with Schenker and colleagues’ [29] volumetric analysis, we found both regions to show no statistical difference in size, thus being symmetric at the population level. This result is in clear contrast with the strongly left lateralized BA44 in humans [21]. Since both chimpanzee and human segmentation were obtained through similar histological procedures, the results support the conclusion that the asymmetry of BA44 developed after humans diverged from our last common ancestor with chimpanzees.

BA44 expanded the most in humans relative to chimpanzees, extending anteriorly

We projected the BA44 and BA45 of each chimpanzee to the surface of a human cortical template and computed an average volume using the template’s cortical thickness. Assuming the MNI template is representative of the Amunts and colleagues [21] population, their reported volumes for humans are directly comparable with our scaled-up volumes for chimpanzee. Our comparison revealed that BA44 enlarged by a factor of 1.64 and 1.29 in the left and right hemispheres beyond the amount of overall cortical expansion, respectively, in this cross-species comparison. Meanwhile, BA45 enlarged only in the right hemisphere, by a factor of 1.36. Our results show that Broca’s area enlargement is remarkable in context of the evolution of human prefrontal cortex size [26,29]. Moreover, our findings suggest that BA44 became left lateralized thanks to a large anterior expansion in the left hemisphere.

The chimpanzee BA44 projects to human areas related to action, and not syntax

We compared the projected chimpanzee atlas with functional subdivisions of human Broca’s area. Our results show that the core chimpanzee BA44 overlapped solely with action-related areas, with the greatest overlap found with regions functionally associated with (in descending order) action imagination, phonology, and action execution/imitation [17,36]. Indeed, we found almost no overlap between the core BA44 chimpanzee homolog and the Broca’s subdivision involved in basic syntax operations [36–38]. Our findings were consistent across multiple levels of thresholding for the chimpanzee BA44 probabilistic map. These results indicate that a simple anatomical scale and shift of the chimpanzee Broca’s area does not explain the existence of the syntax subregion of Broca’s area in the human brain.

Cross-species differences in BA44 support a segregation of action and syntax in humans

Recent functional imaging studies found that both language and action recruit nonoverlapping subdivisions of Broca’s area in the human brain, with language being processed more anteriorly than action [17–19,36–38]. Moreover, it has been found that left BA44 segregates action and syntactic processes of language in 2 distinct subregions, with syntax recruiting its anterior part and action the posterior one [17,36–38]. This functional subdivision of human BA 44 mirrors the underlying receptorarchitectonic organization, which is a powerful indicator of functional diversity [22].

In this study, we have provided further evidence for this action/language segregation, finding that it is likely the result of an evolutionary process involving Broca’s area. By coregistering chimpanzee and human brains, we found that BA44 underwent a large expansion and left lateralization. Moreover, we found the chimpanzee BA44 maps anatomically to the posterior section of human BA44, functionally associated to action. Indeed, virtually no overlap was found between the chimpanzee BA44 and the human syntax regions. Furthermore, when visually compared, the projected chimpanzee BA44 shared spatial location and extension with the receptorarchitectonic division of human BA44. Taken together, this suggests that the left human BA44 evolved to accommodate syntax through an anterior expansion in the inferior frontal cortex.

Along with contributing to the debate on the relationship between action and language, our findings add to the broader topic of language origins. Although there is an evolutionary continuity in auditory-vocal processing and their underlying neurobiological substrate in temporo-frontal networks [39–42], functional neuroanatomical changes appear to play a crucial role during the evolution of prefrontal cortex. Functional studies indicate that when macaques process simple grammatical sequences activation of both Broca’s area and the frontal operculum is observed [43]. In contrast, human brains solely recruit the frontal operculum for simple grammars, whereas BA44 comes into play when processing complex grammatical sequences that nonhuman primates cannot process [44,45]. Thus, despite the observed similarity in the organizational principles across primates, it may have been the expansion of cytoarchitectonically defined BA44 throughout evolution that paved the way for the representation of language in the human brain.

When comparing human and chimpanzee Broca’s area, an implicit assumption is that the chimpanzee homolog can serve as a referential model to that of our shared last common ancestor. It is important to acknowledge that this is not entirely accurate, as chimpanzee brains have also certainly changed along their evolutionary lineage. Nevertheless, the limited fossil evidence from endocranial morphology suggests similarities between extant great apes and early hominins in the region of the inferior frontal gyrus, until a more pronounced “Broca’s cap” in the left hemisphere becomes evident from certain crania of the genus Homo starting at approximately 1.8 million years ago [46]. Of additional note, the anatomical coregistration method we employed cannot offer a complete answer to the exact underlying evolutionary process that Broca’s area underwent (e.g., recycling, neural reuse or cultural reuse [47–49]); for a thorough discussion, please refer to Amunts and colleagues [22]. Finally, our sample size is relatively small, meaning further cross-species comparative studies will be needed on the cytoarchitectonic and receptorarchitectonic organization of BA44 to test our conclusions.

Through a cross-species comparison, our study contributes key insights to Broca’s area reorganization and the ongoing debate on the relationship between language and action. Our findings support the interpretation that BA44 was modified from an action area, as found in nonhuman primates, to a bipartite system serving syntax anteriorly and action posteriorly. In this way, our results underline distinct neural bases for action and syntactic processes in the human brain, and thus, an independence of both cognitive domains.

Ethics statement

All subjects in this study were housed in accordance with federal and state laws governing the welfare and care of nonhuman primates in the United States. All procedures were approved by the Emory University Institutional Animal Care and Use Committee (protocol #YER2000673012513). We emphasize that the collection of postmortem brains used in this study were obtained opportunistically when individuals died from natural causes or were euthanized for quality-of-life reasons.

Cytoarchitecture segmentation of Broca’s area in human brains

We downloaded the publicly available data from Amunts and colleagues [21], in which the left and right BA44 and BA45 were manually segmented on 10 subjects following histological procedures. While the individual maps are not available, the Julich institute has released the probabilistic cytoarchitectural map of both areas derived from the Amunts’ dataset.

Cytoarchitecture segmentation of Broca’s area homolog in chimpanzee brains

Our chimpanzee cytoarchitectural data come from a previous study [29], in which both BA44 and BA45 were bilaterally delineated and guided by the same cytoarchitectonic criteria used in the human maps [21]. Whole-brain MRI data were acquired ex vivo for all chimpanzees (see S1 Supplementary Methods). From the population of 12 chimpanzees, we discarded 3 based on inadequate corresponding MRI data quality, retaining 9 subjects (Pan troglodytes, 5/4 males/females, age = 32.8 ± 11.8 years, age range = 12 to 44.5 years). The demographics of the included chimpanzees are summarized in S1 Table. For additional information on the data acquisition, please refer to the original publication [29].

Broca’s homolog in chimpanzees: Deriving a probabilistic atlas and studying population symmetry

We derived a probabilistic atlas of Broca’s area homolog from the individual cytoarchitectonic segmentations of BA44 and BA45 and their associated ex vivo MRI scans. We performed all the analysis on the reconstructed cortical surfaces, as surface analysis better captures and aligns brains based on their gyrification, thus being more robust than volumetric analysis [50].

The procedure can be summarized in 5 steps: (I) reconstruct 3D brain surfaces from the ex vivo MRI scans using FreeSurfer; (II) project each chimpanzee’s BA44 and BA45 volumetric segmentation to their corresponding surfaces; (III) register all surfaces to a common template, namely, the JUNA chimpanzee brain template [34]; (IV) map the individual cytoarchitectural regions to the JUNA template; and (V) aggregate them to derive a high-quality probabilistic atlas. See Fig 1A for a graphical explanation and S1 Supplementary Methods for a detailed explanation of each step. The processing scripts for the computation of the open access atlas are readily available for download (see Data and Code Availability Statement).

We further leveraged the individual reconstructions to study the surface-area asymmetry of Broca’s homolog in chimpanzee brains. For this, we computed the areas of BA44 and BA45 on each individual chimpanzee and tested their bilateral symmetry through a Wilcoxon signed-ranks test.

Mapping chimpanzee cytoarchitectural maps to the human brain

To enable cross-species comparison, we aligned the cortical reconstruction of JUNA template [34] to that of the human MNI template (ICBM152 9c Asymmetric) [35]. Given the differences in brain shape and volume, we opted to use surface-based registration algorithms, which have been proven successful in aligning the brains of chimpanzees and humans [33].

Based on the work of Eichert and colleagues [33], we performed the surface-based registration in 2 stages. In the first stage, we performed a first alignment of the brain templates using gross anatomical regions. Specifically, we aligned the brains based on their inferior frontal gyrus, as defined by the Desikan atlas (Fig 2A) [51]. Starting from that rough alignment, we then carried a more granular registration based on the sulcal patterns. For a detailed explanation of each stage, please refer to S1 Supplementary Methods as well as the open access processing script (see Data and Code Availability Statement).

Expansion of BA44 and BA45 in humans relative to chimpanzees

In their histological study, Amunts and colleagues [21] report the average gray matter volume for human BA44 and BA45. Since our chimpanzee regions stem from similar histological procedures, we can study how much BA44 and BA45 expanded through evolution by mapping them to a common space and by comparing their size across species.

Having morphed chimpanzee BA44 and BA45 to the human template, we computed their individual volumes using the MNI template cortical thickness. In this way, we obtained volumes for chimpanzee Broca’s area subregions that are scaled up and projected onto the template human cortical surface. Assuming the MNI template is representative of the Amunts and colleagues [21] population, their reported volumes for humans are directly comparable with our scaled-up volumes for chimpanzee.

Functional aspects of the BA44 homolog in the human brain

We aimed to understand the relation between function and the location of the projected chimpanzee Broca’s area homolog—with a particular interest in language and action. For this, we projected the functional subdivisions of human BA44 defined by Papitto and colleagues [17] and Clos and colleagues [36] to the MNI cortical surfaces. There, we compared them to the core chimpanzee BA44, obtained by thresholding the atlas at the 0.5 level, i.e., the points in the surface where the majority of the chimpanzee population had their BA44 located. Particularly, for each functional region, we computed their overlap with the chimpanzee BA44, defined as how much of the functional area was contained by the chimpanzee BA44. We further visually compared our projected chimpanzee BA44 with an existing receptorarchitectonic division of BA44 [23]. The comparison had to be carried out visually since no volumetric nor surface data are publicly available.