Multiple timescales of sensory-evidence accumulation across the dorsal cortex

Lucas Pinto, David W. Tank, Carlos D. BrodyeLife 2022 · DOI

TL;DR

This study investigates how different brain regions contribute to decision-making that relies on accumulating sensory evidence over time. By selectively inactivating parts of the mouse cortex during a virtual reality task, researchers found that frontal areas are crucial for integrating older evidence, while posterior areas handle more recent information, mirroring a known hierarchy of neural processing speeds.

What Problem Does This Paper Address?

Making decisions often involves integrating information over time. For example, deciding whether to trust a news source might involve considering recent articles and older patterns of reporting. This process, known as evidence accumulation, is thought to involve widespread brain networks. However, it's unclear why so many brain areas are involved and how their specific roles relate to the inherent processing speeds (timescales) of different neural circuits. This paper aims to bridge this gap by testing the hypothesis that the known hierarchy of intrinsic timescales in the cortex—where some areas process information faster than others—is fundamental to how we accumulate evidence for decision-making.

Methods

The researchers trained mice to perform a virtual reality task where they had to accumulate visual evidence (towers) over several seconds to make a navigational decision. To pinpoint the role of different brain regions, they used optogenetics to temporarily silence specific cortical areas during different phases of the task. This allowed them to observe how inactivating these areas affected the mice's ability to weigh sensory evidence from different points in time. They also analyzed large-scale calcium imaging data from the mouse cortex during the task to measure the intrinsic temporal integration windows of various brain regions.

Datasets

The study utilized data from mice trained on a virtual reality evidence accumulation task. For the calcium imaging analysis, they re-analyzed previously collected wide-field calcium imaging data from mice expressing the GCaMP6f calcium indicator. No new benchmark datasets were introduced.

Key Insights and Ideas

  • Frontal vs. Posterior Roles: The study reveals a functional division in evidence accumulation: frontal cortical areas are more critical for integrating older sensory evidence, while posterior areas are more involved with recent evidence.
  • Timescale Hierarchy Matters: The findings strongly suggest that the inherent differences in how quickly different brain regions process information (their intrinsic timescales) are not just incidental but are a key organizational principle for evidence accumulation.
  • Beyond Simple Accumulation: Some frontal areas showed effects on behavior even after the evidence was presented, hinting at roles in decision commitment or post-accumulation processing.
  • Cortical Dynamics Reflect Hierarchy: The calcium imaging data confirmed that cortical areas exhibit a gradient of temporal integration windows, with frontal areas having longer timescales, aligning with the inactivation results.

Key Results

  • Optogenetic inactivation of various dorsal cortical areas consistently impaired evidence accumulation, though the degree of impairment varied.
  • Inactivating frontal cortical areas led to more pronounced deficits in using older sensory evidence compared to inactivating posterior areas.
  • Calcium imaging data showed that the decay time constants of neural activity (a measure of temporal integration) systematically increased from visual to frontal areas.
  • While most inactivations primarily affected evidence processing and memory, some frontal areas also showed effects on processes beyond immediate accumulation, suggesting roles in decision commitment.

Limitations and Future Work

The authors acknowledge that the optogenetic inactivation method has limitations, including potential spread to neighboring regions and the possibility of complex interactions between inactivated areas. They suggest future work should explore these interactions more deeply, perhaps using simultaneous inactivation and large-scale recordings. Additionally, it remains an open question whether evidence is accumulated in parallel or sequentially, and how information from different timescales is ultimately combined to form a decision. Future studies could probe these questions with tasks that explicitly manipulate temporal demands.

Summary updated

Abstract

Cortical areas seem to form a hierarchy of intrinsic timescales, but the relevance of this organization for cognitive behavior remains unknown. In particular, decisions requiring the gradual accrual of sensory evidence over time recruit widespread areas across this hierarchy. Here, we tested the hypothesis that this recruitment is related to the intrinsic integration timescales of these widespread areas. We trained mice to accumulate evidence over seconds while navigating in virtual reality and optogenetically silenced the activity of many cortical areas during different brief trial epochs. We found that the inactivation of all tested areas affected the evidence-accumulation computation. Specifically, we observed distinct changes in the weighting of sensory evidence occurring during and before silencing, such that frontal inactivations led to stronger deficits on long timescales than posterior cortical ones. Inactivation of a subset of frontal areas also led to moderate effects on behavioral processes beyond evidence accumulation. Moreover, large-scale cortical Ca<sup>2+</sup> activity during task performance displayed different temporal integration windows. Our findings suggest that the intrinsic timescale hierarchy of distributed cortical areas is an important component of evidence-accumulation mechanisms.
BibTeX
@inproceedings{pinto2022multiple,
  author = {Lucas Pinto and David W. Tank and Carlos D. Brody},
  title = {{Multiple timescales of sensory-evidence accumulation across the dorsal cortex}},
  booktitle = {eLife},
  year = {2022},
  doi = {10.7554/elife.70263},
}
Figure