Cut false alarms before turning the volume up
Aliases: specificity over intensity · don't rescue with volume · PPV first
What it is
When true alarms are ignored, the intuitive patch is louder sound, redder color, a cue that cannot be muted. That is raising intensity. If ignoring comes from "it rang and nothing was wrong," intensity fixes detection, not credibility; people learn faster to treat the bigger sound as noise too. Raise positive predictive value first: fewer falses, then how heavy a true one should feel. Intensity is how a still-credible channel gets heard. It is not medicine for fatigue.
Why it happens
Detection and decision are two stages. Intensity hits the first: larger transients are harder to miss. Fatigue breaks the second: willingness to start handling. Turning the first stage up makes each failure of the second (handled, was false) more memorable, because the cost (startle, being overheard) is higher. The posterior then falls faster.
Specificity (fewer false positives) feeds the second stage directly: a ring is more likely real, so expected value of handling rises. When sensitivity (fewer misses) fights specificity, pushing the threshold toward "ring more" buys misses with fatigue. That trade is worth it only while true-event consequence is extreme and the channel is still trusted. Once the channel is tired, the trade loses on both sides.
Studying it
A 2×2 of high/low false alarms × normal/raised intensity, with a primary task of responding to true events. The outcome is hits on true events, not "did they hear it."
Independent variables: false-positive rate, level or visual salience, whether mute is forbidden. Dependent variables: true-event response rate and time, share of the channel muted or removed, affect or interruption cost when falses are handled.
If the experiment requires a response to every sound, the intensity advantage is inflated because the decision stage was taken away. Allow ignoring. Field contrasts are cleaner: a version that only changes intensity versus a version that only tightens rules, rather than changing both at once.
Where it stops holding
In extreme ambient noise (shop floor, cockpit), intensity is first about physical audibility, not fatigue treatment—yet the channel's rules should still be tightened, or audible noise becomes audible false alarms. Assistive-tech users may need stronger haptics or speech; that is channel adaptation, not a louder default for everyone. Where regulation forbids muting, intensity is locked and specificity is the only lever—more reason not to patch rules with volume.
Applying it
- When true-event response slows, audit false positives first: tighten triggers or split low-confidence detections onto a hint channel, and only then ask whether it was inaudible.
- Do not patch a false-alarm channel with "cannot be turned off" plus "bigger."
- Low-confidence detections go to status, logs, or a digest, not to an alarm sound.
- Verify by taking a high-false-positive rule for a week and doing only one of two things—raise volume one step, or tighten the trigger. If only the latter speeds true-event response, skip the intensity patch. Doing both mixes the books.