New Advances in Hippocampal Spatiotemporal Coding—Multiplexed Theta Phase Coding and the Regulation by External vs. Self-Motion Cues

—Review of the latest Nature Neuroscience paper: “allothetic and idiothetic spatial cues control the multiplexed theta phase coding of place cells”

1. Academic Background & Motivations

Spatial navigation and memory have long been focal topics in neuroscience, with the hippocampus serving as the brain’s central structure for generating and maintaining the cognitive map, and playing a pivotal role in the encoding and retrieval of spatial information. For decades, how the hippocampus integrates external and internal spatial cues to generate stable yet flexible spatial representations has remained a key theoretical and experimental challenge.

This study focuses on the temporal coding mechanism of principal hippocampal cells—place cells—under the theta (θ) rhythm. The theta rhythm (an ~8 Hz local field potential [LFP] oscillation) provides an important “time window” for the hippocampus, acting as a temporal scaffold for spatial computations. As animals traverse their place fields, the firing pattern of place cells exhibits systematic phase shifts (phase precession), allowing individual cell spikes to encode spatial positions and compress spatial sequences (theta sequences), thereby predicting future trajectories and guiding spatial decisions. Meanwhile, evidence has shown that the early phases of theta may support retrospective spatial coding and the encoding of new associations, with some literature suggesting this phase mediates “phase procession” and synaptic plasticity mechanisms. However, functional differentiation between early and late theta phases and their neural mechanisms lack thorough experimental validation, especially when animals must integrate external landmark (allothetic cues) and self-motion (idiothetic cues) information.

To uncover the intrinsic multiplexed nature of theta phase coding and the regulatory roles of external and self-motion cues, it is necessary to construct scenarios with conflicting spatial cues and apply novel experimental systems—such as the virtual reality “dome” (planetarium-style dome VR)—in conjunction with electrophysiological recording and computational modeling. This study arises from precisely such an academic context and challenge.

2. Source and Author Team

The paper was published in top-tier journal Nature Neuroscience (nature neuroscience | volume 28 | october 2025 | 2106–2117). The corresponding author is James J. Knierim, with main contributors Yotaro Sueoka, Ravikrishnan P. Jayakumar, Manu S. Madhav, Francesco Savelli, Noah J. Cowan, et al. The research team is primarily based at Johns Hopkins University (Solomon H. Snyder Department of Neuroscience, Zanvyl Krieger Mind/Brain Institute, Department of Mechanical Engineering, Laboratory for Computational Sensing and Robotics, Kavli Neuroscience Discovery Institute), as well as University of British Columbia and University of Texas San Antonio. Portions of the data were previously reported in related studies but analyzed from independent or novel perspectives.

3. Research Procedure in Detail

1. Design and Implementation of the Virtual Reality Spatial Dissociation System

Researchers employed a unique “dome” virtual reality platform, consisting of a hemispherical projection shell and an annular treadmill for animal navigation. The annular track provides a one-dimensional spatial trajectory, with the dome’s inner surface projecting static or rotatable visual landmarks. Animals are tethered to the center via an arm, equipped with reward mechanisms and optical encoders for real-time movement recording. White noise masks other environmental cues. During the experiment, the projected dome landmarks are controlled as follows:

• Experimental Gain g Mechanism: The ratio of landmark movement speed to the animal’s actual running speed. g=1 indicates static landmarks; g>1, landmarks move counter to the animal, perceived as “speeding up”; g, landmarks move in the same direction, perceived as “slowing down”.
• Multi-Frame Synchronous Recording: All spatial data are registered not only in conventional lab frames, but also the landmark frame and the hippocampal cognitive map frame (hippocampal gain, h).

2. Animal Subjects & Recording

Five male Long–Evans rats (5–8 months old), singly housed, and fully compliant with animal ethics protocols. CA1 place cells were recorded with microelectrode arrays, yielding on average 4.74 ± 3.54 effective cell units per session.

3. Experimental Stages and Core Design

• Epoch 1 (Baseline, g=1): Landmarks remain static, establishing standard spatial representation.
• Epoch 2 (Gain ramping): The experimental gain g is linearly changed, and landmarks begin rotating, generating persistent conflict between self-motion and landmark spatial cues.
• Epoch 3 (Sustained conflict, g=g_final): Landmarks rotate steadily at the target speed, forming sustained spatial cue conflict, requiring rats to constantly learn new landmark/self-motion associations.
• Epoch 4 (Landmarks removed): All salient visual landmarks are switched off, leaving only an illumination ring, and the hippocampal representation is driven primarily by self-motion information—probing intrinsic encoding mechanisms.

4. Data Analysis & Algorithmic Innovation

• Spatial Field Normalization & θ Phase Analysis: Each cell’s place field is normalized in different reference frames; circular-linear regression is used to analyze the relationship between θ phase and spatial position, especially leveraging single traversal analysis to expose heterogeneity in phase coding.
• Spike Phase Spectrum & Compression Factor Analysis: Measures the frequency association between cell firing phase and θ rhythm, and the degree of sequence compression within the θ cycle.
• Continuous Attractor Network (CAN) Model: Models external and self-motion cues as Gaussian inputs with feedback inhibition, simulating the dynamics of two distinct θ phase coding modes.

4. Major Findings in Detail

1. Place Cells Are Firmly Anchored by External Landmarks; Landmarks Dominate Place Fields Under Conflict

In the dome experiment, most sessions (⁴⁰⁄₅₁) showed highly stable place fields in the rotating landmark reference frame; as landmarks rotated, place fields co-rotated. This means the field size distribution remained almost constant in the landmark frame, while physical running distance changed proportionally to g. The results reveal that under spatial cue conflict, external landmarks overwhelmingly dominate the hippocampal map.

2. θ Phase Precession Is Robustly Preserved in the Landmark Frame, Unaffected by Spatial Cue Conflict

Regardless of gain value, all cells maintained classic phase precession in the landmark frame: as animals traversed the place field, spike phase advanced systematically. Parameters such as precession slope, correlation coefficient, and phase offset were invariant. This structure scaled with the landmark reference frame, demonstrating that hippocampal coding locks preferentially to the most salient spatial cue.

3. Theta-Modulated Bursting Frequency Is Adaptively Regulated to Maintain Coding Stability

When g ≠ 1, to preserve stable phase precession, rats had to adjust their physical speed to traverse the landmark-frame-defined field, and place cells actively adjusted their theta-related bursting frequencies (as shown by spike phase spectrum). The normalized precession rate (npr) varied linearly with gain, ensuring spatial sequences were consistently and accurately coded within compressed theta cycles.

4. Single Traversal Analysis Reveals Multimodal Structure of θ Phase Coding: Two Modes Coexist, Linked to Prospective and Retrospective Spatial Coding

Finer single traversal analysis revealed some laps with negative coding slope (dominant phase precession, predicting future locations), others with positive slope (dominant phase procession, reflecting retrospective paths or possibly new association formation). The overall distribution was bimodal, with positive slope traversals biased toward early theta phases, coinciding with regions previously implicated in reverse theta sequences and synaptic plasticity.

5. Persistent Cue Conflict and Landmark Removal Selectively Impair Phase Procession (Early θ Phase Coding), While Predictive Coding Is Preserved

When g deviates from 1 and when landmarks are absent, positive slope (phase procession) in single traversals significantly decreased; related statistics (like Second Lobe Index, SLI) fell, while negative-slope (phase precession) structure persisted. This indicates that constant demands for new association encoding and spatial prediction errors markedly suppress early theta phase firing—likely inhibiting retrospective or associative encoding but preserving predictive spatial coding.

6. Neural Circuit Level: CA3 and Entorhinal Cortex Inputs Cooperatively Modulate θ Phase Coding of External/Self-Motion Information

Combining spike firing and gamma rhythm coupling analysis revealed that gain conflict affects not only medium gamma, but potentially slow gamma coupling as well. This indicates that both CA3 and EC III input streams encode both allothetic and idiothetic information, with their cooperative regulation effectively captured by the multiplexed coding mechanism.

7. CAN Model Reveals Selective Regulation of Coding Modes by Relative Cue Strength & Feedback Inhibition

The modified CAN model precisely replicated the experimental observations: when landmark vs. self-motion input strengths were unbalanced or misaligned, feedback inhibition intensified, selectively impairing phase procession (processing lobe vanishing) while preserving precession lobe. The former relies more on balanced cue activation, providing mechanistic support for the multimodal θ phase coding dynamics.

5. Conclusions & Scientific Significance

1. Theta Phase Coding Exhibits Functional Multiplexing Properties

This study systematically demonstrates, for the first time, that θ phase coding of CA1 place cells encompasses both prospective (phase precession) and retrospective/new associative (phase procession) modules, and these modes can be independently regulated according to spatial cue context. Prospective coding is robust, driven by the strongest spatial cue, ensuring continuous navigation; early theta phase coding weakens selectively with environmental association demands or cue conflict, possibly providing a substrate for new associative memory formation or retrospective evaluation.

2. Encoding Switching Mechanisms Are Universal & Evolutionarily Significant

The study reveals that when new environmental associations are being formed, the hippocampus preferentially safeguards navigation and spatial prediction functions, delegating new information encoding to more plastic states (e.g., subtle dendritic spiking in early θ phase). This division of labor is meaningful not just for spatial navigation but also occurs in non-spatial cognition and other brain regions, potentially underpinning theoretical advances in flexible neural network coding.

3. Technical & Methodological Innovations

The dome VR system and spatial cue conflict experiment design, combined with multi-frame dynamic analysis and continuous attractor modeling, have furnished spatial cognition neuroscience with a novel toolkit, broadly applicable for probing neural coding frame selection, memory formation, interference, and related complex phenomena.

6. Research Highlights & Application Value

• Pioneeringly demonstrates the dual-pathway multiplexing mechanism of hippocampal θ phase coding and its precise regulation by external/self-motion cues.
• Empirically tests multiplexed coding’s dynamic changes under continuous new learning demands, bridging a prior gap between theory and experiment.
• Presents a novel theoretical framework directly linking spatial prediction, retrospective sequence coding, and new memory formation to neural temporal activity structures.
• The combination of dome VR technology and continuous attractor modeling opens new avenues for studies of neural circuit dynamics and intervention for cognitive dysfunction.
• Offers insights for cross-disciplinary fields like AI spatial navigation and robotic cognitive mapping.

7. Other Valuable Content

The end of the paper contains detailed methodological descriptions, offering a template for future replication or extension. The literature review systematically traces the history of hippocampal spatial coding, serving as an essential reference for understanding development in the field. The authors outline future prospects for theta-gamma coupling states, reference frame selection, and multi-level path dependency mechanisms, broadening new research directions.

8. Summary

Through virtual reality conflict experiments and innovative algorithmic analysis, this study reveals the multiplexed essence of hippocampal place cell θ rhythm phase coding, clearly disentangling the complex relation between external landmarks and self-motion cues. The division of labor between prospective and retrospective coding provides a new paradigm for understanding cognitive map flexibility and memory formation mechanisms, injecting powerful momentum into both neuroscience and interdisciplinary cognitive engineering innovation.