Time Course of Attentional Modulation in the Frontal Eye Field During Curve Tracing

Paul S. Khayat, Arezoo Pooresmaeili, Pieter R. RoelfsemaJournal of Neurophysiology 2009 · DOI

TL;DR

This study investigates how the brain selects relevant visual information for action, focusing on the timing of neural activity in the frontal eye field (FEF) and visual cortex (V1) during a curve-tracing task. The key finding is that the neural "selection signal" emerges at remarkably similar times in both early visual areas (V1) and higher-level frontal areas (FEF), suggesting a distributed and interactive process rather than a strictly hierarchical one.

What Problem Does This Paper Address?

Understanding how our brains filter out irrelevant visual information and focus on what's important for guiding our actions is a fundamental question in neuroscience. Previous research had shown that neurons in both early visual areas (like V1) and higher-level areas (like FEF) become more active when a stimulus is behaviorally relevant. However, the precise timing of this "attentional modulation" and how it unfolds across different brain regions remained unclear. Specifically, it was debated whether early visual areas simply process information and pass it on, or if higher-level areas actively influence processing in earlier stages. This paper aims to clarify the temporal dynamics of attentional selection by comparing neural activity in V1 and FEF during a task that requires sustained attention and a subsequent motor response.

Methods

The researchers used a curve-tracing task where monkeys had to mentally trace a target curve connected to a fixation point, while ignoring a distractor curve. This task required shifting attention along the target curve. Single-unit recordings were performed in the frontal eye field (FEF) of two macaque monkeys. To analyze the timing of neural responses, they employed sophisticated curve-fitting methods to estimate the latency of both the initial visual response and the subsequent attentional modulation (the point at which neural activity starts to differentiate between the target and distractor). They also leveraged previous data from their lab on V1 activity during the same task, allowing for a direct comparison. Control trials were included to rule out confounds from stimulus features near the fixation point. Neurons were classified into visual, visuomovement, and movement types based on their response properties.

Datasets

Not discussed in the paper. The study involved recordings from macaque monkeys performing a specific behavioral task.

Key Insights and Ideas

  • Synchronized Selection Signals: The most striking finding is the near-identical timing of the attentional selection signal in V1 and FEF. This challenges models that propose a strict hierarchy where early areas passively receive information and higher areas initiate selection. Instead, it suggests a more integrated process.
  • Distributed Attention: The similar timing implies that attentional selection is not confined to a single brain region but likely involves interactions between widely separated cortical areas. This supports the idea of a "frontoparietal network" working in concert.
  • Visual Response Precedes Selection: As expected, the initial visual response to the stimulus occurred earlier in both V1 and FEF than the selection signal, indicating that the brain first registers the presence of stimuli before it starts to prioritize them.
  • Task-Specific Control: The study highlights that the control of attention can be dynamic and task-dependent. In tasks requiring detailed visual analysis (like curve tracing), visual cortex might play a more dominant role in selection, while in tasks requiring strategic planning, frontal areas might exert more top-down control.

Key Results

  • FEF Visual Response Latency: The initial visual response in FEF neurons occurred at a latency of approximately 50 ms after stimulus onset.
  • FEF Selection Signal Latency: The attentional selection signal, where FEF neurons began to discriminate between the target and distractor curves, emerged at a latency of approximately 147 ms after stimulus onset.
  • V1 Visual Response Latency: In comparison, V1 visual responses occurred earlier, around 41 ms.
  • V1 Selection Signal Latency: Crucially, the attentional selection signal in V1 occurred at approximately 144 ms, very similar to the FEF latency.
  • Modulation Strength: While the timing was similar, the strength of attentional modulation (measured by a modulation index) was significantly stronger in FEF neurons compared to V1 neurons.

Limitations and Future Work

  • Animal Model: The study was conducted in macaque monkeys, and while highly informative, direct translation to human cognition requires caution.
  • Task Specificity: The findings are specific to the curve-tracing task. Whether this temporal synchrony holds for other attentional tasks (e.g., visual search) needs further investigation.
  • Simultaneous Recordings: While they used data from the same animal for V1 and FEF comparisons, simultaneous recordings would provide even stronger evidence by eliminating potential confounds from different recording sessions or animals.
  • Future Directions: The authors suggest exploring how different types of attentional control (goal-directed vs. stimulus-driven) might influence these temporal dynamics. They also propose investigating the role of feedback connections in shaping these selection signals and how these processes might be implemented in attractor network models.

Summary updated

Abstract

Neurons in the frontal eye fields (FEFs) register incoming visual information and select visual stimuli that are relevant for behavior. Here we investigated the timing of the visual response and the timing of selection by recording from single FEF neurons in a curve-tracing task that requires shifts of attention followed by an oculomotor response. We found that the behavioral selection signal in area FEF had a latency of 147 ms and that it was delayed substantially relative to the visual response, which occurred 50 ms after stimulus presentation. We compared the FEF responses to activity previously recorded in the primary visual cortex (area V1) during the same task. Visual responses in area V1 preceded the FEF responses, but the latencies of selection signals in areas V1 and FEF were similar. The similarity of timing of selection signals in structures at opposite ends of the visual cortical processing hierarchy supports the view that stimulus selection occurs in an interaction between widely separated cortical regions.
BibTeX
@article{khayat2009time,
  author = {Paul S. Khayat and Arezoo Pooresmaeili and Pieter R. Roelfsema},
  title = {{Time Course of Attentional Modulation in the Frontal Eye Field During Curve Tracing}},
  journal = {Journal of Neurophysiology},
  year = {2009},
  doi = {10.1152/jn.91050.2008},
}
Figure