Abstract:

Organisms use two spatial reference frames to represent spatial information: the egocentric frame (i.e., self-centered, defined relative to the subject) and allocentric frames (i.e., a world-centered). The transformation between these two frames is essential for forming and applying cognitive maps. One widely recognized model, proposed by Bicanski and Burgess (BB model), suggests that the coordinated activity of the parietal cortex, retrosplenial cortex, and the medial temporal lobe networks underpins spatial reference frame transformation. Specifically, during spatial learning, the processed sensory information forms the egocentric representations in the parietal cortex. The retrosplenial cortex integrating this egocentric information and the head direction signal from the limbic regions (i.e., the anterior nucleus of the thalamus), which are then projected to the medial temporal lobe, where allocentric representations are formed. The reverse process facilitates the transformation from allocentric to egocentric frames, which is crucial for memory retrieval, imagination, and action execution.

Despite a substantial body of empirical research has been conducted surrounding this model, several unresolved issues remain: 1) the topological organization of egocentric and allocentric representations in the parietal-retrosplenial-temporal lobe loop is still debated; 2) it is unclear whether the human retrosplenial cortex supports joint representations of egocentric and allocentric information as seen in rodents, or if it is the sole brain region responsible for combining these spatial variables; 3) the hypothesized flow of information during spatial reference frame transformation still lacks experimental support. 4) there is a lack of causal evidence regarding the neural mechanisms underlying spatial reference frame transformation.

To address these questions, we propose propose 3 studies. In Studies 1 and 2, we will investigate the dynamic representation of egocentric, allocentric, and joint spatial information, as well as cross-brain communication within the parietal-retrosplenial-temporal lobe loop during spatial reference frame transformation, using high spatial-temporal resolution intracranial EEG. In Study 3, we will causally test the role of this neural loop in reference frame transformation and explore electrical stimulation protocols to enhance spatial reference frame transformation abilities using deep brain stimulation (DBS).

Studies 1 and 2 will recruit refractory epilepsy patients with electrodes implanted in the parietal cortex, retrosplenial cortex, and medial temporal lobe, who will perform a spatial memory task. Study 1 will focus on the direction learning phase (egocentric to allocentric transformation), where participants learn the directions of target buildings from a first-person perspective with different head direction, later identifying these directions on a global map. EEG time-frequency signals will be used to decode egocentric and allocentric target directions and head direction signals, identifying which brain regions within the parietal-retrosplenial-temporal lobe loop represent these spatial variables. Further, using Granger causality analysis and cross-frequency coupling analysis, we will investigate the information flow between brain areas which represent allocentric and egocentric information during time periods that can significantly decode these variables, and correlate the information flow with learning performance. Representational similarity analysis will be used to explore joint encoding of egocentric target directions and head directions by the retrosplenial cortex. Study 2 will focus on the direction testing phase (allocentric to egocentric transformation), where participants will convert memorized allocentric target directions into egocentric ones. The analysis will follow the same methodology as Study 1. In Study 3, we will investigate how electrical stimulation affects the conversion process from egocentric to allocentric reference frames. Refractory epilepsy patients with electrodes implanted in the retrosplenial cortex and medial temporal lobe will participate in a path integration task. Participants will use either an allocentric or an egocentric strategy during the task. Different stimulation protocols (50 Hz, theta-burst, or no stimulation) will be applied to the retrosplenial cortex to assess how these patterns affect behavioral performance with allocentric strategies.

This research combines computational models and multimodal neurotechnology to investigate the neural mechanisms of spatial reference frame transformation. It aims to uncover the topographical distribution of egocentric and allocentric representations in the brain, explore the role of low-frequency neural oscillations in cross-brain communication during reference frame transformation, and validate the causal role of the retrosplenial cortex in this process. These findings will not only provide evidence for existing computational models but also contribute to translating spatial navigation theory into clinical interventions.

Key words: spatial reference frame, egocentric spatial representation, allocentric spatial representation, cognitive map, intracranial EEG