Rapid encoding of task regularities in the human hippocampus guides sensorimotor timing

Ignacio Polti, Matthias Nau, Raphael Kaplan, Virginie van Wassenhove, Christian F. Doeller

Abstract

The brain encodes the statistical regularities of the environment in a task-specific yet flexible and generalizable format. Here, we seek to understand this process by converging two parallel lines of research, one centered on sensorimotor timing, and the other on cognitive mapping in the hippocampal system. By combining functional magnetic resonance imaging (fMRI) with a fast-paced time-to-contact (TTC) estimation task, we found that the hippocampus signaled behavioral feedback received in each trial as well as performance improvements across trials along with reward-processing regions. Critically, it signaled performance improvements independent from the tested intervals, and its activity accounted for the trial-wise regression-to-the-mean biases in TTC estimation. This suggests that the hippocampus supports the rapid encoding of temporal context even on short time scales in a behavior-dependent manner. Our results emphasize the central role of the hippocampus in statistical learning and position it at the core of a brain-wide network updating sensorimotor representations in real time for flexible behavior.

Introduction

When performing sensorimotor tasks, such as catching a ball, the brain must balance the specificity of the current sensory input with the statistical regularities of past experiences. This regularization often leads to a "regression-to-the-mean" effect, where estimates of extreme values (very fast or very slow) are biased toward the average. While timing functions are typically associated with the striatum and cerebellum, the source of the contextual information driving these statistical biases has remained unclear. This study investigates whether the hippocampal system—famed for building "cognitive maps"—provides the framework for updating sensorimotor representations in real-time.

Experimental Design and Methods

The researchers conducted an fMRI study with 34 healthy participants (ages 19–35) performing a Time-to-Contact (TTC) estimation task. In each trial, participants tracked a target moving at one of four speeds (4.2, 5.8, 7.5, or 9.1 °/s) in one of 24 directions for a 10° visual angle. The target disappeared before reaching a boundary, and participants had to estimate the moment of contact by pressing a button.

Figure 1: Task design and behavioral performance. (A) The TTC task sequence: fixation, visual tracking, TTC judgment, and feedback. (B) Behavioral results demonstrate a regression-to-the-mean bias, where participants overestimated short TTCs and underestimated long ones. (C) Distribution of feedback levels (Accurate, Small under-/overshoot, Large under-/overshoot) received by participants.

Immediate visual feedback was provided after each trial, categorized as high, medium, or low accuracy. The study analyzed brain activity using a mass-univariate General Linear Model (GLM) to identify regions tracking feedback and performance improvements, alongside Psychophysiological Interaction (PPI) analysis to examine functional connectivity changes during learning.

Key Findings

The study revealed that the hippocampus acts as a rapid learner, responding to feedback on a trial-by-trial basis. A functional division of labor was observed: while the caudate (striatum) tracked improvements for specific intervals, the hippocampus signaled improvements across the entire task structure, independent of the specific interval tested.

Crucially, hippocampal activity accounted for the behavioral "regression-to-the-mean" bias. This suggests that the hippocampus actively provides a statistical prior to help generalize behavior in noisy or novel environments. Furthermore, when performance was poor, hippocampal connectivity with the primary motor cortex and cerebellum increased, indicating its role in signaling the need to update sensorimotor maps when reality deviates from the brain's internal model.

Results and Behavioral Performance

Participants demonstrated high overall accuracy (Spearman’s $rho = 0.91$), yet exhibited a clear regression effect with slopes clustering around 0.5. Neural results showed:

  • Feedback Tracking: Anterior hippocampal activity was significantly predicted by the feedback from the previous trial ($p = 0.001$).
  • Improvement Signaling: The posterior hippocampus signaled performance improvements independent of the TTC interval ($p = 0.017$).
  • Neural Driver of Bias: Trial-wise activity in the anterior hippocampus reflected the magnitude of behavioral regression ($p < 0.0001$), positioning it as a driver of statistical regularization.

Discussion

These findings position the hippocampus at the center of a brain-wide network that updates sensorimotor representations in real time. While the study establishes a strong correlation between hippocampal activity and timing biases, further research using causal methods like Transcranial Magnetic Stimulation (TMS) or lesion studies is needed to confirm these roles. Future work may also explore how hippocampal "update" signals relate to an individual's ability to generalize to novel timing intervals, bridging the gap between sensorimotor control and high-level cognitive mapping.

Summary updated

Abstract

Abstract The brain encodes the statistical regularities of the environment in a task-specific yet flexible and generalizable format. Here, we seek to understand this process by converging two parallel lines of research, one centered on sensorimotor timing, and the other on cognitive mapping in the hippocampal system. By combining functional magnetic resonance imaging (fMRI) with a fastpaced time-to-contact (TTC) estimation task, we found that the hippocampus signaled behavioral feedback received in each trial as well as performance improvements across trials along with reward-processing regions. Critically, it signaled performance improvements independent from the tested intervals, and its activity accounted for the trial-wise regression-to-the-mean biases in TTC estimation. This suggests that the hippocampus supports the rapid encoding of temporal context even on short time scales in a behavior-dependent manner. Our results emphasize the central role of the hippocampus in statistical learning and position it at the core of a brain-wide network updating sensorimotor representations in real time for flexible behavior.
BibTeX
@inproceedings{polti2022rapid,
  author = {Ignacio Polti and Matthias Nau and Raphael Kaplan and Virginie van Wassenhove and Christian F. Doeller},
  title = {{Rapid encoding of task regularities in the human hippocampus guides sensorimotor timing}},
  booktitle = {eLife},
  year = {2022},
  doi = {10.7554/elife.79027},
}
Figure