Reconstructing the Neanderthal brain using computational anatomy

Abstract

The present study attempted to reconstruct 3D brain shape of Neanderthals and early Homo sapiens based on computational neuroanatomy. We found that early Homo sapiens had relatively larger cerebellar hemispheres but a smaller occipital region in the cerebrum than Neanderthals long before the time that Neanderthals disappeared. Further, using behavioural and structural imaging data of living humans, the abilities such as cognitive flexibility, attention, the language processing, episodic and working memory capacity were positively correlated with size-adjusted cerebellar volume. As the cerebellar hemispheres are structured as a large array of uniform neural modules, a larger cerebellum may possess a larger capacity for cognitive information processing. Such a neuroanatomical difference in the cerebellum may have caused important differences in cognitive and social abilities between the two species and might have contributed to the replacement of Neanderthals by early Homo sapiens.

Figure 1

Reconstructed Neanderthal brains. (a) Population-average. (b) Representative modern human subject. (c) The reconstructed brains with the neuroanatomical labels.

Figure 2

Comparisons of the brain surface morphology among Neanderthal, early Homo sapiens and modern Homo sapiens. (a) Surface statistical map shows the surface area where the differences are statistically significant (p < 0.05 with family-wise error (FWE) correction). See Extended Data Fig. 1 for more details. (b) Surface displacement maps show the morphological difference in the direction perpendicular to the tangential surface. The displacement maps were calculated by subtracting modern Homo sapiens (MH) from Neanderthal (NT), early Homo sapiens (EH) from NT, and MH from EH, respectively. See Extended Data Fig. 2 for more details.

Figure 3

Comparisons of the relative volumes of the parcellated brain regions among NT, EH and MH. (a) Each parcellated volume was normalized to the mean MH volume to calculate a ratio (i.e. relative volume unit). The regionally specific volume differences were evaluated after removing the effects of ICV by analysis of variance (ANOVA) across 13 parcellated regions. We employed Bonferroni correction for multiple comparisons, so that the threshold of p < 0.003 (=0.05/13) is set to statistically significant. The relative volume differences were found in Pa SI (F_2,1190 = 9.31, p = 0.0001), Oc SM (F_2,1190 = 8.15, p = 0.0003), Ce V (F_2,1190 = 7.34, p = 0.0007), and Ce P (F_2,1190 = 6.70, p = 0.0013) (Extended Data Table 3). The mean (±standard deviation) MH volumes of the parcellated brain regions are 161.61 ± 5.22 cc for Fr SM, 41.57 ± 1.00 cc for Fr I, 65.46 ± 3.45 cc for Fr O, 96.93 ± 2.82 cc for Sm, 88.30 ± 1.98 cc for Pa SI, 37.38 ± 1.41 cc for Pa TP, 91.27 ± 3.92 cc for Te SM, 82.50 ± 2.60 cc for Te I, 93.70 ± 4.17 cc for Oc SM, 39.99 ± 1.32 cc for Oc I, 12.38 ± 0.23 cc for Ce V, 13.86 ± 0.21 cc for Ce A, and 114.41 ± 3.76 cc for Ce P. (b) As the ANOVA results indicated a significant group-by-laterality difference in the size-adjusted volume of the cerebellar hemisphere (F_2,1190 = 14.28, p < 0.001 for Ce A, F_2,1190 = 12.73, p < 0.001 for Ce P), we tested if there is a significant difference between the size-adjusted volume of the left and right cerebellar regions within each group and between groups based on the symmetrized volume analysis. Fr, frontal lobe; Pa, parietal lobe; Te, temporal lobe; Oc, occipital lobe; Ce, cerebellum; Sm, sensorimotor cortex; SM, superior and middle region; I, inferior region; O, orbitofrontal region; SI, superior and inferior region; TP, temporo-parietal junction; A, anterior region; P, posterior region; V, vermis. *p < 0.05 corrected for multiple comparisons. Data are means ± s.d. See Extended Data Table 1 for correspondence between the automated anatomical labelling (AAL) atlas and the parcellated brain regions.