Neural coding of visual objects rapidly reconfigures to reflect sub-trial shifts in attentional focus

Lydia Barnes, Erin Goddard, Alexandra WoolgarbioRxiv 2021 · DOI

TL;DR

This study uses MEG and advanced decoding techniques to show that the brain can rapidly reconfigure how it represents visual information within milliseconds, specifically when task demands shift mid-trial. This rapid adaptation is crucial for flexible behavior and appears to be more pronounced in challenging tasks, suggesting the brain only engages this costly process when necessary.

What Problem Does This Paper Address?

Humans are remarkably adept at flexibly shifting their attention to focus on what's important in a dynamic environment. This ability is thought to be supported by "adaptive coding" in brain regions like the frontoparietal cortex, where neural representations change to prioritize task-relevant information. While fMRI studies have shown that these representations change over trials, the temporal resolution of fMRI is too coarse to observe how quickly these changes occur within a single trial, especially when attention needs to shift mid-task. This paper tackles the question of whether the brain can reconfigure its neural codes on a rapid, sub-trial timescale to support these momentary shifts in attentional focus, a key aspect of how we break down complex tasks into manageable steps.

Methods

The researchers employed Magnetoencephalography (MEG), a neuroimaging technique with excellent temporal resolution, to record brain activity. They used Multivariate Pattern Analysis (MVPA), specifically Linear Discriminant Analysis (LDA), to decode information about task-relevant features (color and shape) from the MEG data. This involved training classifiers to distinguish between different task rules and then testing how well these classifiers could predict which feature was currently relevant. The key innovation here is applying these time-resolved decoding methods to investigate rapid, within-trial shifts in attention, a feat not possible with fMRI.

Datasets

Not discussed in the paper. The study used novel stimuli created specifically for the experiments.

Key Insights and Ideas

  • Sub-millisecond Reconfiguration: The brain can indeed reconfigure its neural codes for visual features on a very fast, sub-trial timescale (within hundreds of milliseconds) to match shifting attentional demands.
  • Task Difficulty Matters: This rapid adaptive coding is not always engaged. It was only reliably observed in a more difficult version of the task, suggesting the brain conserves this resource and only deploys it when processing demands are high.
  • Momentary Focus is Supported: This rapid reconfiguration provides a neural basis for how we can maintain focus on specific aspects of a task for brief periods, even as the task evolves.
  • Dissociation in Simpler Tasks: In a simpler task with highly distinguishable stimuli, the brain showed robust coding of all features but no preferential coding for the task-relevant ones, highlighting the role of task complexity.

Key Results

  • Experiment 1 (Simple Task): While rule information was decodable, there was no reliable evidence of preferential coding for task-relevant features (color or shape) over irrelevant ones.
  • Experiment 2 (Difficult Task): Robust and sustained preferential coding for the relevant feature (color or shape) was observed, starting around 500 ms after stimulus presentation and lasting throughout the trial. This effect was significantly stronger for relevant features compared to irrelevant ones.
  • Epoch Comparison: The temporal dynamics of preferential coding were comparable between the first and second stimulus epochs, even though attention had to shift mid-trial. This suggests the reconfiguration process is remarkably fast.
  • Behavioral Performance: Participants performed well above chance in both experiments, but overall performance was lower in the more difficult Experiment 2.

Limitations and Future Work

  • Task Differences: The authors acknowledge that several factors differed between the two experiments (stimulus complexity, difficulty, number of objects), making it challenging to pinpoint the exact reason for the differential findings. Future work should systematically manipulate these factors.
  • Optimal Engagement: It's unclear what specific conditions push the brain to engage in preferential coding. Future research should investigate the "cost" of reconfiguring attention and identify when it's an optimal strategy.
  • Broader Implications: The authors suggest exploring how these findings relate to fluid intelligence and using more advanced neuroimaging techniques (like MEG-fMRI fusion) to better understand the interaction between domain-general and task-specific brain regions. They also propose investigating how rapid stimulus streams or self-directed attention shifting might further probe the limits of neural reconfiguration.

Summary updated

Abstract

Abstract Every day, we respond to the dynamic world around us by flexibly choosing actions to meet our goals. This constant problem solving, in familiar settings and in novel tasks, is a defining feature of human behaviour. Flexible neural populations are thought to support this process by adapting to prioritise task-relevant information, driving coding in specialised brain regions toward stimuli and actions that are important for our goal. Accordingly, human fMRI shows that activity patterns in frontoparietal cortex contain more information about visual features when they are task-relevant. However, if this preferential coding drives momentary focus, for example to solve each part of a task, it must reconfigure more quickly than we can observe with fMRI. Here we used MVPA with MEG to test for rapid reconfiguration of stimulus information when a new feature becomes relevant within a trial. Participants saw two displays on each trial. They attended to the shape of a first target then the colour of a second, or vice versa, and reported the attended features at a choice display. We found evidence of preferential coding for the relevant features in both trial phases, even as participants shifted attention mid-trial, commensurate with fast sub-trial reconfiguration. However, we only found this pattern of results when the task was difficult, and the stimulus displays contained multiple objects, and not in a simpler task with the same structure. The data suggest that adaptive coding in humans can operate on a fast, sub-trial timescale, suitable for supporting periods of momentary focus when complex tasks are broken down into simpler ones, but may not always do so.
BibTeX
@misc{barnes2021neural,
  author = {Lydia Barnes and Erin Goddard and Alexandra Woolgar},
  title = {{Neural coding of visual objects rapidly reconfigures to reflect sub-trial shifts in attentional focus}},
  year = {2021},
  doi = {10.1101/2021.05.25.445712},
  howpublished = {bioRxiv},
}
Figure