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

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

TL;DR

The hippocampus is traditionally viewed as a hub for long-term memory and spatial navigation, but this study reveals it also acts as a real-time engine for statistical learning. By tracking trial-by-trial feedback, the hippocampus rapidly encodes task regularities to guide sensorimotor timing, helping the brain balance specific task demands with general environmental "rules."

What Problem Does This Paper Address?

When you try to catch a ball, your brain performs a complex trade-off. It must focus on the specificity of the current moment (how fast is this ball moving?) while also regularizing based on past experience (how fast do balls usually move?). This regularization often manifests as a "regression-to-the-mean" effect, where our estimates of extreme values (very fast or very slow) are biased toward the average.

While sensorimotor timing is often attributed to the striatum, cerebellum, and motor cortex, the source of the "contextual" information that drives these biases has remained elusive. The authors investigate whether the hippocampal system—famed for building "cognitive maps"—provides the statistical framework necessary to update sensorimotor representations in real-time.

Methods

The researchers combined functional magnetic resonance imaging (fMRI) with concurrent eye tracking in 34 participants.

  • The Task: A fast-paced "Time-to-Contact" (TTC) estimation task. Participants tracked a moving target that disappeared before hitting a boundary. They had to press a button at the exact moment they estimated the target would have made contact.
  • Variables: The target moved at four different speeds and in 24 different directions.
  • Feedback: Participants received immediate visual feedback on their accuracy (high, medium, or low).
  • Analysis: The team used a mass-univariate General Linear Model (GLM) to identify brain regions tracking feedback and performance improvements. They also employed Psychophysiological Interaction (PPI) analysis to look at functional connectivity—how the hippocampus "talked" to other regions like the motor cortex and cerebellum during learning.

Datasets

The study utilized a primary dataset of 34 healthy human participants (ages 19–35). No external pre-existing datasets or benchmarks were used, as this was an original experimental study.

Key Insights and Ideas

  • The Hippocampus as a Rapid Learner: Contrary to the idea that the hippocampus only handles memories over days or weeks, this study shows it responds to feedback on a trial-by-trial basis (seconds).
  • A Functional Division of Labor: The brain uses a "divide and conquer" strategy for timing. The caudate (striatum) tracks improvements for specific intervals (specificity), while the hippocampus tracks improvements across the entire task structure, independent of the specific interval (regularization).
  • The Source of Behavioral Bias: Hippocampal activity accounted for the "regression-to-the-mean" bias. This suggests the hippocampus isn't just "making mistakes" when it biases our timing; it is actively providing a statistical prior to help us generalize to noisy or novel situations.
  • Real-Time Network Updating: When participants performed poorly, hippocampal connectivity with the primary motor cortex and cerebellum increased. This suggests the hippocampus signals the need to "update" the sensorimotor map when reality doesn't match the brain's internal model.

Key Results

  • Behavioral Accuracy: Participants performed well overall (Spearman’s $rho = 0.91$), but showed a clear regression effect, with regression slopes clustering at 0.5 (where 1 is perfect accuracy and 0 is total regression to the mean).
  • Feedback Tracking: Activity in the anterior hippocampus was significantly predicted by the feedback received in the previous trial ($p = 0.001$).
  • Improvement Signaling: The posterior hippocampus signaled performance improvements that were independent of the specific TTC interval tested ($p = 0.017$).
  • Neural Correlation with Bias: Trial-wise hippocampal activity reflected the magnitude of the behavioral regression effect ($p < 0.0001$ in the anterior HPC), positioning it as the neural driver of statistical regularization.

Limitations and Future Work

The authors acknowledge that while they found a strong correlation between hippocampal activity and timing biases, fMRI cannot prove causality. Future research using Transcranial Magnetic Stimulation (TMS) or studying patients with hippocampal lesions could confirm if the "regression effect" disappears when the hippocampus is offline.

Additionally, the authors predict that individuals with stronger hippocampal "update" signals should be better at generalizing to entirely new timing intervals. Testing this prediction directly would be a logical next step to bridge the gap between sensorimotor control and high-level cognitive mapping.

Summary updated

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 bridging 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 is in line with the idea 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