Epilepsy patients walking a real maze reveal the right hippocampus as the brain’s mapmaker

Every time you take a shortcut through a park you have never crossed before, or recombine fragments of remembered streets into a brand-new route home, your brain is performing one of its most remarkable feats: reading and rewriting an internal map of the world. For decades, neuroscientists have suspected that the hippocampus, a seahorse-shaped structure buried deep in the temporal lobe, is the engine of that map. Now a study from LMU Munich, published in the Journal of Neurology, has put that idea to a strikingly concrete test, asking people with precisely localized brain dysfunction to physically walk their way through a complex building and watching, frame by frame, where their eyes and feet took them.

The research team, led by Denis Grabova and Florian Schöberl, recruited 47 adults with unifocal epilepsy from the tertiary epilepsy center at LMU University Hospital, together with 24 age-matched healthy controls. The patients fell into three carefully characterized groups: 16 with right-sided mesial temporal lobe epilepsy, in which the epileptic focus sits in or near the right hippocampus; 16 with left-sided mesial temporal lobe epilepsy; and 15 with frontal lobe epilepsy, whose seizures arise outside the temporal navigation network. Every participant had undergone extensive electrophysiological and imaging workup to pinpoint the epileptogenic zone, and only those with well-controlled seizures, no more than one per month on medication, were included. This clinical precision is what makes the cohort so scientifically valuable: each patient carries a naturally circumscribed lesion within a known brain network, allowing researchers to ask what happens to navigation when a specific node of that network goes offline.

The navigation challenge itself took place in a roughly 700-square-meter area of an outpatient clinic, an environment that was unfamiliar to all participants. In a first, examiner-guided exploration phase, each subject was walked along defined routes to five target items, pictures of a ball, a mushroom, a flower, a train and a house, placed at distinct locations. Immediately afterwards came the real test: over ten minutes, participants had to find 15 targets in a pseudo-randomized order dictated by the examiner. The first five routes duplicated the exploration paths, so they could be solved by simply retracing familiar steps, a task that can be managed with an egocentric, compass-like strategy in which locations are coded relative to one’s own body. The remaining ten targets, however, demanded recombined novel routes, requiring navigators to stitch together spatial knowledge into new trajectories and, ideally, to cut across previously unexplored territory using shortcuts. That flexible recombination is the behavioral signature of allocentric navigation, the map-like representation of space in which objects are related to each other independently of the navigator’s current position.

To capture not just whether people found their targets but how they tried, the researchers equipped participants with a gaze-controlled, head-fixed eye-tracking camera. Post hoc video analysis quantified saccades and fixations, including fixations directed at objects suitable as landmarks for reorientation, and reconstructed each walked trajectory. From these data the team derived a battery of quantitative measures: normalized error rates for familiar and novel routes separately, the percentage of available shortcuts actually used, the proportion of time spent at crossroads and in unexplored territory, and the ratio of traveled distance to the optimal path. Saccades were defined algorithmically as eye movements exceeding 240 degrees per second with acceleration above 3,000 degrees per second squared, while fixations were events lasting more than 100 milliseconds below those thresholds, a level of technical rigor that turns a walk down a corridor into a rich stream of measurable cognition.

The headline result was unambiguous. All three epilepsy groups performed significantly worse than healthy controls at finding navigation targets, but the deficits were not evenly distributed. Patients with right temporal lobe epilepsy posted a total normalized error rate of 71.7 percent, dramatically higher than the 45.3 percent of left temporal lobe patients, the 28.6 percent of frontal lobe patients, and the near-ceiling performance of controls. Crucially, the gap widened precisely where allocentric navigation matters most: on recombined novel routes, right temporal lobe patients erred on 84.2 percent of targets, far worse than every other group, while their performance on retracing familiar routes was statistically indistinguishable from the other patient cohorts. In other words, they could follow a memorized path but fell apart when asked to improvise a new one, the exact dissociation predicted if the right hippocampus is the hub for building and manipulating a cognitive map.

The trajectory data added a vivid behavioral texture to those numbers. Right temporal lobe patients spent significantly more time standing at crossroads, 28.1 percent of their navigation time compared with roughly 20 percent for the other groups and controls, and lingered longer in unexplored territory. They also used available shortcuts far less often than controls, managing only 13.9 percent compared with 65.0 percent in the healthy group. Eye tracking revealed a complementary pattern: all patient groups made fewer fixations on landmark-suitable objects than controls during the overall task, but on recombined novel routes it was specifically the temporal lobe patients who showed this landmark-attention deficit. Together, these findings sketch a coherent picture of a navigator whose internal map is metrically vague and landmark-poor, forcing hesitation at decision points and blindness to spatial shortcuts that a healthy map-holder would seize immediately.

The study also confronted a long-standing controversy in hippocampal research: whether spatial memory is lateralized to the right hemisphere at all. When the authors applied an analysis of covariance to correct for confounders, age, duration of epilepsy, and the presence of structural brain lesions, all of which correlated with worse performance, the raw quantitative differences between right and left temporal lobe patients largely dissolved. Yet the right-sided group remained clearly the worst performers overall, were the only subgroup impaired even on retracing familiar routes, and showed uniquely disorganized navigation strategies. The authors interpret this through a complementary-representation model in which the two hippocampi contribute different ingredients to navigation: the right hippocampus particularly important for encoding places, the left for temporal sequences, bridging spatial navigation to episodic memory. Lateralization, on this view, is real but graded and task-dependent rather than absolute.

The frontal lobe patients added their own lesson. Although spared the hippocampus, they too were significantly impaired on recombined novel routes relative to controls, consistent with the dorsolateral prefrontal cortex’s established role in goal-oriented planning and flexible wayfinding, and with the dense cortical interactions between prefrontal regions and the hippocampus that support path integration. Human navigation, the study underscores, is not the product of a single structure but of a distributed, multiregional network, and damage at different nodes produces characteristically different failures: hippocampal damage corrupts the map itself, while frontal damage undermines the planning that exploits it.

Beyond its theoretical weight, the work carries practical implications. Spatial navigation deficits are among the earliest functional markers of Alzheimer’s disease, and this same real-space paradigm has previously differentiated amyloid-positive from amyloid-negative patients with mild cognitive impairment. Establishing how the hippocampal network fails in focal epilepsy refines the interpretive toolkit for such studies and highlights how age, disease duration, and lesion status must be disentangled before navigation scores can be read as network signatures. The authors also acknowledge limitations, including age differences between subgroups, the clinical heterogeneity of prior epilepsy surgery, and psychometric tests that may have missed subtle executive or visual deficits, and they call for future work with more sensitive batteries for visual scene perception and executive function.

What lingers most from the study is its methodological honesty: rather than simulating space on a screen, it asked broken brains to navigate real space, with all its demands on locomotion, gaze, memory and planning intertwined. The verdict from those walked corridors is that the hippocampus remains the master cartographer of human goal-oriented navigation, with the right hippocampus holding a particularly prominent pen when the map must be redrawn on the fly, and that clinically well-characterized epilepsy patients offer one of the sharpest available windows into how the human brain charts the world.

Subject of Research: The role of the right and left hippocampi in allocentric real-space navigation in patients with focal epilepsy

Article Title: Real-space navigation in patients with focal epilepsy underpins the prominent role of the hippocampus for allocentric navigation in humans

Article References: Grabova, D., Kaufmann, E., Pradhan, C., Rémi, J., Vollmar, C., Masouris, I., Wlasich, E., Kustermann, J., Zwergal, A., & Schöberl, F. (2026). Real-space navigation in patients with focal epilepsy underpins the prominent role of the hippocampus for allocentric navigation in humans. Journal of Neurology, 273(10), Article 626. https://doi.org/10.1007/s00415-026-14164-7

Image Credits: AI Generated

DOI: 10.1007/s00415-026-14164-7

Keywords: hippocampus, allocentric navigation, spatial navigation, focal epilepsy, temporal lobe epilepsy, cognitive map, eye tracking, neuropsychology, frontal lobe epilepsy, real-space navigation, lateralization, Journal of Neurology