Being in the visual field is not the same as being seen
Aliases: LBFS · gaze-awareness dissociation
What it is
Gaze landing directly on an object, even resting there for a while, does not guarantee the object was "seen" — where seeing means forming a percept that can be reported and remembered. The stronger version of this claim is called looked-but-failed-to-see (LBFS): the classic example from traffic accident analysis is a dashcam or eye-tracking record showing a driver's gaze clearly swept across a motorcycle's location, while the driver later insists they never saw it at all.
This goes a step further than "attention occupied elsewhere causes obvious objects to be missed": it shows that even when gaze itself — where the eyes are pointed — is aimed directly at the target, that alone is not sufficient proof the target reached awareness. Eye-tracking data can tell you light fell on the fovea; it cannot tell you whether that signal was processed to the level where it could be reported.
Why it happens
Where the eyes point (overt attention) only determines where high-resolution sensory input lands. What actually determines whether that input gets processed into a reportable percept is the allocation of attentional resources (covert attention) — and the two can come apart. Under normal conditions eye movements track attention shifts, but when cognitive resources are consumed elsewhere, the eyes can "pass over" an object — the retinal signal is generated as usual, foveal resolution is as high as ever — while processing resources are never allocated to it. That high-quality sensory input stalls at the pre-attentive stage and never reaches the layer that can be recalled and reported.
Inattentional blindness studies combined with eye tracking confirm this directly: a participant's fixation point genuinely lands on the unexpected object, sometimes for a non-trivial duration, yet when asked afterward, they still report not seeing it. This shows that "seeing" carries two entirely distinct meanings — the optical sense (the retina receiving a signal) and the perceptual sense (the signal being promoted to reportable conscious content) — and the two can come apart completely.
Studying it
The main approach combines eye tracking with the inattentional blindness paradigm: record participants' fixation location and duration, confirm the target object genuinely fell within the foveal region for a sufficient duration, then check that against the post-hoc detection report to see whether the two agree.
In traffic safety research, the evidence for this comes mainly from accident reconstruction: combining dashcam footage with reconstructed gaze paths to analyze "looked-but-failed-to-see" accidents — cases where the driver's gaze trajectory proves their eyes passed over the location of the hazard, yet the person had no awareness of having seen it at all afterward.
Dependent variables are accordingly split into two groups: gaze metrics (first-fixation latency, fixation duration, fixation count) and detection metrics (post-hoc report, content recall) — the core evidence for this entry is precisely that these two groups of measures dissociate.
Methodological caveat: fixation duration and detection rate are genuinely positively correlated — longer fixations generally raise detection probability. This entry does not deny that; it only points out that fixation alone is not sufficient proof of being seen, and the two need to be measured separately rather than one standing in for the other.
Where it stops holding
- This does not deny that longer fixations raise detection probability; a brief, saccade-embedded glance (lasting tens of milliseconds) is more likely to be missed than a sustained dwell fixation (several hundred milliseconds or more) — the two should not be treated the same.
- This dissociation is less pronounced under low perceptual load, where a fixation is more often accompanied by sufficient processing resources, strengthening the correlation between gaze and detection.
- The evidence base combines experimental gaze-plus-report studies in the lab with observational accident reconstructions in the field; the two carry different causal strength — the former is experimental evidence, the latter observational, which can only suggest correlation.
Applying it
- Do not treat "the user's gaze swept across this region" on an eye-tracking heatmap as evidence that "the user saw this element," especially when evaluating whether a warning, legal notice, or critical action button was noticed.
- Verifying that an element was truly seen requires behavioral or verbal confirmation — can the user recall the content, did they act on it correctly — rather than relying purely on eye-tracking metrics like fixation duration or count.
- For critical information carrying compliance or liability implications, if the only evidence is eye-tracking data showing "the user's gaze covered that region," that evidence is not strong enough to support a conclusion like "the user was informed."
- Verification: collect gaze data and post-hoc report or task-correctness data together; only conclude "the user noticed it" when the two agree. When they disagree, trust the report, not the gaze data.