Masking extends forward and backward in time

What it is

Masking does not only happen when two sounds are strictly simultaneous. For a short interval after a loud sound ends, a weaker sound that follows can still have its detection threshold raised enough to become inaudible — this is forward masking. More counterintuitively, a weak sound played a short interval before a loud sound can also be affected, retroactively, by the loud sound that comes after it, becoming similarly hard to detect — this is backward masking. Together these are called temporal masking, extending simultaneous masking along the time axis.

Why it happens

Forward masking reflects the fact that the neural response triggered by a loud sound has a decay process — it does not drop to zero the instant the sound physically ends. This lingering neural activity continues to occupy the channel that would otherwise be used to detect a subsequent weak sound; the weak signal's contrast against this lingering activity is too small, the same logic as "background activity swamps the weak signal" in simultaneous masking, except here the background activity comes from a sound that already ended rather than one playing at the same instant. The lingering activity decays over time, so the later the weak sound appears, the less it is affected.

Backward masking is more involved: it reflects that the neural response to the weak sound has not finished being processed before the much larger response to the following loud sound arrives and interrupts or overrides that processing. Because this depends on a window during which the weak sound's processing is still incomplete, and that window is much shorter than the decay process behind forward masking, backward masking operates over a noticeably shorter time range than forward masking — once the gap grows, the weak sound's processing is already complete and backward masking disappears, while forward masking's lingering activity may not yet have fully decayed. This forward-backward asymmetry is the defining feature that sets temporal masking apart from simultaneous masking.

Studying it

The standard paradigm holds the masker's and target's frequency and level fixed while systematically varying the time interval between them (testing both orders — masker first, target first), measuring the target's threshold elevation at each interval to trace out separate decay curves for forward and backward masking.

Typical independent variables: the interval between masker and target, and the masker's own level and duration. Typical dependent variables: the target's detection-threshold elevation relative to its unmasked baseline.

This paradigm is mainly used to characterize the auditory system's temporal resolution limits, providing basic data for understanding whether two closely spaced auditory events can be heard as distinct. It also supplies parameters frequently cited in the psychoacoustic models behind audio coding and compression algorithms.

Where it stops holding

• Forward masking's effective window is typically much longer than backward masking's; the two are not symmetric time windows, so it should not be assumed that "leaving an equal gap on each side" is equally safe on both sides.
• Masking strength decays as the interval grows — a longer gap means less impact — but the decay rate and absolute window length shift with the masker's level and duration, so no single millisecond figure applies universally.
• The temporal masking discussed here builds on the same frequency-proximity relationship as simultaneous masking, just extending its reach from "the same instant" to "a short interval before or after." The two sounds still need to be close enough in frequency for a clear effect; when frequencies are far apart, temporal masking is weak too.
• Interference effects that can still appear over longer intervals are usually no longer this energetic-level mechanism, but belong to a different category involving memory or attention, and should not be explained using this entry's account.

Related

• Same group: A3.04.1 A loud sound masks a weaker sound close to it in frequency · A3.04.3 Cue tones need a spectrum offset from ambient noise
• Nearby: A3.13 Temporal resolution and rhythm perception (the auditory system's limit for telling adjacent events apart — a related but different question from temporal masking)
• Search terms: temporal masking · forward masking · backward masking · psychoacoustics