A brief silent gap is enough for the auditory system to register as an event boundary

What it is

Insert a sufficiently short silent gap into a sound, and the auditory system doesn't just "notice a gap" — it treats what comes before and after as two separate events, splitting what was a continuous sound into perceptual units that can be counted and processed independently. The gap functions as an event boundary, not merely a detectable physical stimulus.

This is a different question from "can the gap be detected at all": that's a matter of pure detection ability, while this entry is about what happens once the gap is detected — it gets used as the dividing line between "this is one thing" and "that is another."

Why it happens

In the real world, a continuously producing physical process (friction, resonance, airflow) rarely goes completely silent for no reason and then restarts, unless something discrete really did change — contact was broken, a new impact occurred. The auditory system builds an empirical inference rule on this basis: a genuine interruption in energy is strong evidence that "a discrete event transition happened here," so it assigns the sound before and after the interruption to different perceptual units rather than continuing to treat it as the ongoing same process.

Because the auditory system already has extremely fine gap-detection ability (see the sibling entry), this inference rule can be triggered at very short physical durations — it doesn't take a long silence to "count" this as two things; a gap on the order of tens of milliseconds is often enough to trigger segmentation. This explains why the auditory system is much sharper than vision, under comparable conditions, at chopping a continuous stream into countable, separately-processable units.

Studying it

A common paradigm constructs two short tones separated by a silent gap of varying duration, and asks participants to judge whether they heard "one continuous sound" or "two separate sounds," finding the shortest gap duration that reliably triggers a "two events" judgment — this threshold is not identical to the raw gap-detection threshold, since it's usually longer, because the segmentation judgment involves a higher-level categorization step rather than pure physical detection.

Common independent variables: gap duration, whether the pitch and timbre before and after the gap match (a bigger difference before/after makes the same gap more likely to be judged as two events). Common dependent variables: proportion of "two events" reports, minimum gap duration required for the judgment.

Where it stops holding

• The similarity between the sounds before and after the gap strongly affects this threshold: when pitch and timbre match closely on both sides, a longer gap is needed before it's judged as two separate events; when the sounds differ noticeably, a much shorter gap suffices. This boundary interacts with pitch- and timbre-based grouping cues and cannot be considered in isolation from gap duration alone.
• In a fast, repeating rhythmic sequence, a short gap that would otherwise trigger segmentation can instead be reinterpreted as part of the beat pattern itself (a regular pause) rather than an event boundary — this interference from regularity is explored further in the sibling entry.
• This rule describes segmentation of "one thing versus two things" within a single sound stream; it does not address which stream multiple simultaneous or alternating sound sources should be attributed to — that is a different-level question.

Applying it

• To make a series of continuously played voice prompts, sound effects, or data announcements perceived as several countable, separately processable units, inserting even a silent gap of only tens of milliseconds between units is usually enough to establish a clear boundary sense, without needing to change timbre or pitch.
• Conversely, to make content perceived as an uninterrupted, continuous whole (ambient background sound, a synthesized speech output meant to sound smooth), strictly avoid even a brief silent gap at splice points, or it risks being heard as an unintended break.
• Verification: play the designed sequence to listeners and simply ask "how many sounds/segments did you count," checking whether the reported number matches the design intent — an overcount suggests an unintended silent gap exists; an undercount suggests the gap wasn't long enough or was masked by similarity between the surrounding sounds.

Related

• Same group: A3.13.1 Auditory temporal resolution is finer than visual temporal resolution · A3.13.3 A steady rhythm induces a tendency to synchronize movement or attention · A3.13.4 Rhythmic pattern can carry an information channel independent of pitch and loudness
• Nearby: A3.15 Auditory stream segregation
• Search terms: event boundary · auditory segmentation · temporal grouping · gap detection