TL;DR
We experience a continuous visual world despite frequent blinks, a phenomenon often attributed to the brain "filling in" missing information. This study, using intracranial recordings in epilepsy patients, reveals that instead of filling in, the brain actively suppresses visual activity transients during blinks. This suppression, rather than filling-in, explains why blinks are less noticeable than external visual interruptions.
What Problem Does This Paper Address?
We seamlessly perceive a continuous visual stream, even though our eyes blink about 100-150 times per minute, momentarily obscuring our vision. This raises a fundamental question: how does the brain maintain this sense of continuity? Two main hypotheses exist: 1) the brain "fills in" the missing visual information during blinks, maintaining a continuous neural representation, or 2) the brain actively "edits out" or suppresses the neural signals associated with blinks, making them less noticeable than external visual interruptions (like a sudden darkening of the room). This paper aims to disentangle these hypotheses by investigating the neural correlates of visual perception during blinks versus external visual interruptions.
Methods
The researchers used electrocorticography (ECoG) to record brain activity directly from the surface of the visual cortex in 14 patients undergoing evaluation for epilepsy. This provides high spatial and temporal resolution, crucial for distinguishing rapid neural events. They presented participants with images and introduced three types of visual interruptions:
- Spontaneous blinks: Natural, unprompted eye closures.
- Voluntary blinks: Blinks performed on cue.
- Gaps: External, brief periods where the visual stimulus was replaced by a blank screen (black or gray).
They analyzed the high-frequency broadband (HFB) power, an indicator of neural activity, in response to these events. A key methodological innovation was the use of time-domain deconvolution to disentangle the responses to the visual stimuli from the responses to the interruptions (blinks and gaps). This is a more sophisticated approach than standard event-related averaging, allowing for accurate estimation of each event's contribution even when they overlap.
Datasets
The study used data from 14 patients undergoing intracranial monitoring for epilepsy. No new datasets were introduced or used for training.
Key Insights and Ideas
- Early visual cortex is "blind" to perception: In early visual areas (V1 and V2), the brain's response to a blink and an external gap was remarkably similar. This suggests that at this basic level, the brain is primarily responding to the physical interruption of light, not its perceived significance.
- Higher-level areas differentiate: As the signal travels up the visual hierarchy, higher-level areas begin to distinguish between blinks and gaps. Specifically, the reappearance of the stimulus after a gap elicits a sustained neural "overshoot" (increased activity), while the reappearance after a blink does not.
- Suppression, not filling-in: The data strongly argue against the "filling-in" hypothesis. Instead of maintaining activity during blinks, the brain appears to actively suppress the neural transients associated with blink-induced visual disappearances. This suppression is more pronounced for blinks than for gaps.
- "Filling-in" is a misnomer: The perceived continuity during blinks isn't due to sustained neural activity representing the missing image, but rather due to the absence of a noticeable neural signal indicating a disruption.
Key Results
- In early visual cortex (V1, V2), both blinks and gaps caused a dip in activity followed by an overshoot upon stimulus reappearance.
- In higher-level visual cortex, the reappearance-related overshoot was significantly larger for gaps than for blinks. This difference increased as the signal moved anteriorly.
- The disappearance-related dip was actually slightly more pronounced for blinks than for gaps, contradicting the idea of a suppressed dip for blinks.
- Statistical analyses confirmed these findings, showing significant differences in reappearance-related overshoots between gaps and blinks in higher-level visual areas.
Limitations and Future Work
- Simulated gaps: The study acknowledges that the "gaps" used were not a perfect physical replication of blinks' retinal impact. While controls were implemented to mitigate potential confounds (duration, latency, luminance), the inherent differences between these events remain.
- Incomplete coverage: The study's recording sites did not cover all brain regions, particularly sub-cortical areas that might be involved in generating the blink-suppressing signal. Future work could explore these areas.
- Mechanism of suppression: The exact neural mechanism for blink suppression remains to be fully elucidated. The authors suggest possibilities like "corollary discharge" (an efference copy of the motor command) or a more general prediction error minimization mechanism.
- Role of dorsal stream: The paper hints at a potential role for dorsal stream areas in signaling perceived discontinuities (like gaps), which might be actively inhibited during blinks. Further investigation into these regions is warranted.
Summary updated