Objectives

To illustrate fundamental principles of long-range connectivity in the cerebral cortex of monkeys and mice as revealed by anatomical tracer experiments and to discuss how they inform efforts to estimate brain connectivity using noninvasive neuroimaging in humans and nonhuman primates.

Methods and Results

Recent studies of cortical organization have provided evidence for ~41 distinct cortical areas in the mouse, ~117 areas in the marmoset, ~140 areas in the macaque, and ~180 areas in humans (see Van Essen and Glasser, 2016). Retrograde tracer studies in macaque monkeys, marmosets, and mice reveal evidence for highly distributed inter-areal connectivity and a wide range of connection weights (many orders of magnitude) for different area-to-area pathways (Markov et al., 2014; Majka et al., 2016; Gamanut et al., 2018). An additional complexity is that the axonal trajectories of individual cortical neurons can be remarkably complex. Comparisons between connectivity as measured by invasive tracer injections and as inferred from noninvasive imaging have been profitably explored in the macaque monkey, A recent parcellated analysis reported a correlation of 0.59 between tractography-based connectivity (using postmortem diffusion imaging data) and tracer-based connectivity (Donahue et al., 2016); the correlation is weaker for areas that are more widely separated, which highlights the challenges of inferring direct connectivity using tractography. To compare tracers with fMRI-based connectivity, in recent unpublished findings from a collaboration with the Henry Kennedy and Takuya Hayashi labs, we have (i) generated ‘dense’ (parcellation-free) maps of retrogradely labeled neurons registered to a macaque surface-based atlas from 28 tracer injections; (ii) mapped resting-state functional connectivity from a separate group of 1 alert and 30 anesthetized macaques; and (iii) compared anatomical connectivity (both dense and parcellated) with seed-based functional connectivity, revealing correlations that differ according to the region injected (higher for sensorimotor areas, lower for prefrontal and visual areas) and according to the method of functional connectivity analysis (full correlation outperforms partial correlation). This approach holds promise for developing approaches that more accurately predict anatomical connectivity using functional connectivity and tractography. Translating the lessons learned from monkey connectivity analyses to humans is challenging for several reasons: (i) there are no methods available in humans for directly measuring ‘ground truth’ long-distance anatomical connectivity; (ii) cerebral cortex in humans is vastly larger than in monkeys and is highly nonuniform in its evolutionary expansion; and (iii) human cortex likely contains many areas that are absent in the monkey lineage. Nonetheless, it is possible to carefully compare both areal organization and connectivity across species using interspecies surface-based registration.

Discussion and Conclusions

Elucidating how the wiring of the human brain gives rise to our uniquely human characteristics and behaviors is one of the great challenges of modern systems neuroscience. This is a daunting endeavor, but also an exciting one in which progress is accelerated using modern surface-based approaches and ‘HCP-style’ analyses promulgated by the Human Connectome Project (HCP).

Acknowledgements

No acknowledgement found.