Audio can pull attention outside the field of view
Aliases: off-screen audio · spatialized notification · off-FOV cue
What it is
A headset’s visual cone is about a hundred degrees; objects behind, at the feet, or over the rear shoulder are invisible by default. A cue from behind the left shoulder turns the head and the target enters the view. That is auditory spatial cueing. Sound pulls attention outside the current frustum; it is not the same message read aloud. It fills a solid-angle hole in the field of view, not a hole in loudness.
Why it happens
Auditory space is nearly omnidirectional: interaural time and level differences plus pinna spectra give a heading in the horizontal plane, so a person can turn before seeing the source. Visual attention is tunnelled by the headset field of view; there are no pixels at the edge-behind. A directional sound opens a pointer outside that tunnel: the head turns along it until the target falls into central vision. A non-directional ding in the headphones can only say “something happened,” not “turn where,” so people sweep a circle and cueing collapses into an alarm.
Effective cueing needs the sound’s direction to match where the target will appear, and to arrive before unstructured visual search starts. A wrong direction sends the head into empty space; a late sound arrives after the wearer has already started turning at random, and becomes interference. Cueing is one orienting movement, not a continuous glue of attention to the ear.
Studying it
A visual search in which the target appears at a random heading outside the current view, compared across no sound, a central ding, and a cue spatialized at the target heading. Record time from onset until the target enters the view, wrong-direction turns, and total search time.
Independent variables: whether the cue has direction, angular error of the cue relative to the target, target offset from the initial heading. Dependent variables: latency to start the head turn, whether the first turn is in the correct half-field, time to capture.
If the lab only places targets within ±90°, the value of cueing the rear hemisphere is never measured. Outside-the-view must include behind. A loudspeaker array and HRTF headphones are not the same cue set; report the presentation.
Where it stops holding
Hearing loss, unilateral deafness, or a mixdown to mono narrows this channel; it cannot be assumed to turn everyone the right way. A noisy environment or several sources at once will swamp “turn where.” In a visually loaded task that already occupies central vision, a sudden spatial sound steals the current work and cueing becomes interruption. If the target will enter the view on its own (flying into the cone), the sound’s value drops. Users who are deaf need another channel; this leaf does not cover them.
Applying it
- Events that sit outside the current view and must be found get a short directional cue aimed at where the target will appear. Do not only play a central ding.
- Cue one direction at a time. Two spatial sounds at once send the head toward the louder one, not the more important one.
- After a timeout with no look toward the target, the sound may play once more. Do not call continuously.
- How to check: the target appears outside the initial view, including behind. With a spatial cue, the first head turn should enter the target’s half-field, and capture should be faster than with a non-directional cue. Turning into the opposite half-field, or sweeping in place first, means cueing has not yet been achieved.