ResearchPod Summary
For decades, cognitive neuroscience has relied on group-level fMRI analyses, which average data across many participants to identify common brain features. While this approach has successfully mapped the general 'skeleton' of large-scale brain networks, it inherently blurs individual differences. Each human brain is unique, with variations in the size, shape, and precise location of functional regions. Precision fMRI, or precision functional mapping, addresses this by collecting hours of data from single individuals, allowing researchers to resolve the fine-scale, idiosyncratic topography of an individual's brain networks.
If every brain is unique, how can we identify shared organizational principles? The authors argue that we must move away from strict voxel-wise correspondence. Instead, they propose using 'cortical zones'—broad regions that account for the fact that functional areas do not always align with anatomical landmarks. By focusing on the relative positioning of networks and their functional responses during tasks, researchers can identify equivalent networks across individuals even when their precise coordinates differ. This approach has revealed that association cortex is organized into parallel, distributed networks that are remarkably consistent in their relative arrangement, despite individual differences in their exact location.
Precision fMRI has led to several fundamental discoveries, such as the identification of the Somato-Cognitive Action Network (SCAN) in the motor cortex and the resolution of the default mode network into distinct, parallel streams (DN-A and DN-B). Furthermore, these methods allow researchers to study how brain networks reorganize in response to development, perinatal damage, and experience. For example, individuals with early-life brain lesions often show significant reorganization of their functional networks, yet they frequently maintain typical cognitive performance, highlighting the brain's capacity for functional compensation.
[[RP_SECTION:limitations-of-group-averages|Limitations of group averages]]
Sam: [steady, matter-of-fact] The average brain is a statistical artifact — and a misleading one. It obscures the idiosyncratic network variants that actually define functional neuroanatomy. That's the central argument from a 2026 review by Caterina Gratton and Rodrigo Braga in the Annual Review of Psychology.
Alex: If the group average is just a blur, how do we actually map what's happening in a specific person? [[RP_SECTION:precision-functional-mapping|Precision functional mapping]]
Sam: [grounded, precise] Precision functional mapping. Instead of scanning thirty people for one hour each, you scan one person for many hours. The key move is bypassing group-template alignment entirely. When you force individual brains into a shared coordinate space, you lose high-frequency spatial information — particularly the idiosyncratic boundary transitions between functional networks. Longer single-subject acquisition preserves that signal.
Alex: It's like switching from a composite sketch to high-resolution satellite imagery. But does precision reveal something consistent across people, or is every brain just a collection of unique features?
Sam: [slower, teaching mode] That's the core tension. The answer is: both, and the framework matters. We identify commonalities by using cortical zones as relative anchors — defining network homology through spatial positioning and task-evoked responses rather than absolute voxel coordinates. What that reveals is that while exact topography varies substantially across individuals, the underlying functional skeleton is stable and heritable. The architecture is conserved; the implementation is not.
Alex: Heritable — so is this variation fixed at birth, or is neuroplasticity in play? [[RP_SECTION:genetic-and-plastic-variation|Genetic and plastic variation]]
Sam: [measured, building the case] Almost certainly both. The review distinguishes two forms of variation. Border shifts are changes in the size and shape of network regions — where one network ends and another begins. These are highly similar in monozygotic twins, which points to a strong genetic component. Ectopic variants are a different animal: intrusions of a network into a non-canonical location, somewhere it doesn't typically appear. Those show more evidence of experience-dependent plasticity, suggesting the environment can push the functional map around in ways that genetic factors alone don't predict.
Understanding the individual brain is critical for moving beyond the 'average' model of human cognition. Precision fMRI provides a more valid and reliable framework for linking brain organization to behavior and clinical outcomes. Although current studies show that individual differences in brain topography are associated with cognitive and personality traits, the effect sizes are often modest. The authors suggest that future progress requires not only more extensive imaging data but also more sensitive, reliable behavioral measures to fully capture how idiosyncratic brain organization influences human life.
AI-generated third-party summary by ResearchPod. Not official content or an endorsement by the paper authors or affiliated organizations.
Alex: [connecting the dots] So an ectopic variant might be the brain's way of adapting to something like a perinatal lesion?
Sam: [confident] Exactly. A patient with early brain damage can retain function because their motor network reorganizes into a non-canonical topography. Precision fMRI maps that reorganization. A group-average approach would simply flag it as misalignment noise — and discard it. The information you need to understand compensation is precisely the information the standard pipeline throws away.
Alex: So the limitation isn't just the data — it's the analytical framework. Forcing individual brains into a group template means you're systematically discarding the signal that explains how the brain routes around injury.
Sam: [sitting back, calm] That's right. And it has downstream consequences for effect sizes. When you ask whether network topography predicts behavior, the variance explained tends to be modest at the group level. Part of that is real — the brain has redundancy and degeneracy built in. Redundancy means multiple units can perform the same function; degeneracy means structurally different configurations can support the same behavioral output. So the same cognitive operation can run through different neural pathways in different people. Standard group-level analysis will always struggle to capture that, because it's looking for a fixed mapping that doesn't exist.
Alex: The brain is modular enough to route around damage, but that flexibility makes it harder to pin a specific behavior to a specific region across individuals.
Sam: [precise] Exactly. And that's why the field is moving from a nomothetic approach — seeking universal laws that apply across all brains — toward an idiographic one that treats the individual's functional map as the unit of analysis. The open question, and it's a real one, is determining which variants are behaviorally meaningful and which are just stochastic noise. Not every idiosyncratic feature is doing something important. Figuring out which ones are is the next hard problem. [[RP_SECTION:clinical-implications-of-mapping|Clinical implications of mapping]]
Alex: Which brings up the clinical angle. If interventions like TMS or DBS are targeted using a generic atlas, and the individual's actual functional boundaries are somewhere else entirely —
Sam: [quiet confidence] — then you're stimulating the wrong tissue. That's not hypothetical; it's a known source of variability in outcomes. The goal this review is pointing toward is using an individual's precision functional map to tailor targeting to their specific neuroanatomy. We're not there clinically at scale — the acquisition time alone is a real barrier — but the argument is that this is the only principled path forward. The average brain can't tell you where to intervene in a specific patient, because the average brain doesn't exist in any single skull. [[RP_SECTION:future-of-neuroimaging-research|Future of neuroimaging research]]
Alex: That's a clean summary of the problem. The group average was a useful scaffold for building the field, but it's now actively limiting what we can ask.
Sam: [measured] That's the Gratton and Braga argument in a sentence. The tools exist to do better — precision acquisition, individual-level parcellation, twin and longitudinal designs to separate genetic from plastic contributions. The bottleneck now is adoption and scale, and working out the right criteria for when individual variation is signal versus noise. Those are tractable problems, which is what makes this a useful moment to take stock of where the field stands. Thanks for listening to ResearchPod.