Decibels are logarithmic. The decibel scale uses a base-10 logarithm to express the ratio between two power-like quantities, such as sound intensity or electrical signal strength. The confusion with “exponential” is understandable, though, because the relationship works both ways depending on which direction you read it: the decibel scale is logarithmic, but the physical intensity it represents grows exponentially as the decibel number climbs. That dual nature is exactly why people argue about it online, and both camps are, in a sense, talking about the same mathematical relationship from opposite ends.
What It Means to Say the Scale Is Logarithmic
A logarithmic scale takes a massive range of values and squashes it into something human-sized. Sound is a perfect case: the quietest thing your ears can detect and the roar of a jet engine differ in intensity by a factor of about a trillion. Writing out numbers that span twelve orders of magnitude every time you want to compare two sounds would be absurd, so the decibel scale applies a logarithm to the ratio. The decibel, technically one-tenth of a bel, uses a base-10 (decadic) logarithm to measure the ratio of two power-like quantities such as mean-square sound pressure in acoustics.1Metrologia. On logarithmic ratio quantities and their units The result is a compact number, usually somewhere between 0 and about 140 for everyday sounds, instead of a number with a dozen zeros.
When someone says “exponential,” they are typically looking at the same relationship in reverse. If you start with decibels and want to recover the actual physical intensity, you raise 10 to a power. Going from 70 dB to 80 dB does not mean the sound got 14% louder in raw energy; it means the intensity multiplied by 10. Going from 70 dB to 90 dB means it multiplied by 100. That explosive growth is exponential behavior, and it is baked into the scale precisely because the scale itself is logarithmic. Logarithms and exponentials are inverse operations, two sides of one coin. The scale is defined logarithmically; the underlying quantity it maps behaves exponentially.
Why “Every 10 dB Equals 10 Times the Power” Matters More Than the Math
You do not need to remember the formula to use decibels usefully. The single most practical thing to internalize is the 10-dB rule: an increase of 10 dB corresponds to a tenfold increase in sound power. A 20 dB increase is a hundredfold increase. A 3 dB increase roughly doubles the power. These relationships hold because of the logarithmic definition, but knowing the rule of thumb is far more helpful than knowing the equation behind it.
This is where real-world intuition goes sideways. Most people hear “80 versus 90 decibels” and picture something modestly louder, the way 80 versus 90 miles per hour feels modestly faster. But 90 dB carries ten times the acoustic power of 80 dB. A jackhammer at 100 dB is not twice as intense as a vacuum cleaner at 70 dB; it is a thousand times as intense. The linear-thinking habit that serves us well for speed, temperature, and weight is exactly the wrong habit for decibels, and it trips up almost everyone who has not had a reason to think carefully about it.
Perceived loudness does not even track the intensity multiplier cleanly. Roughly speaking, a 10 dB increase sounds about twice as loud to most listeners, even though the physical power has gone up tenfold. So the logarithmic scale is already partially correcting for human perception, but not perfectly. Your ears have their own compression going on.
Why Sound Measurement Uses a Logarithmic Scale in the First Place
The choice was not arbitrary. Three properties of sound and hearing made a logarithmic scale almost inevitable.
First, the raw dynamic range is enormous. The threshold of human hearing sits near 0.00002 pascals of pressure. The threshold of pain is around 20 pascals. That is a factor of a million in pressure and a factor of a trillion in intensity (because intensity scales with the square of pressure). No linear ruler handles that range gracefully.
Second, human hearing itself is roughly logarithmic. Doubling the physical intensity of a sound does not feel like double the loudness. It takes roughly ten times the intensity to sound twice as loud. A scale that compresses intensity logarithmically maps much more closely to how loud things feel than a linear scale would. This is the same basic principle behind the Richter scale for earthquakes and the stellar magnitude scale in astronomy: when a phenomenon spans many orders of magnitude and perception is approximately logarithmic, a log scale keeps the numbers useful.
Third, engineers in telecommunications needed to add and subtract signal gains and losses. Logarithms convert multiplication into addition. If a signal passes through an amplifier that multiplies power by 100 (a 20 dB gain) and then through a cable that cuts it by half (a 3 dB loss), the net result is 20 minus 3, or 17 dB. Without the log scale, you would be multiplying and dividing large numbers at every stage of a signal chain. The decibel made bookkeeping manageable, and that engineering convenience is a big part of why the unit stuck.
The 3 dB Versus 5 dB Exchange Rate Debate
One place where the logarithmic nature of decibels has real health consequences is in occupational noise regulation. The question regulators face is: if a worker is exposed to louder noise for a shorter time, how do you calculate the equivalent daily dose? The “exchange rate” is the number of decibels by which noise must increase to halve the allowed exposure time.
A strict energy-equivalent model uses a 3 dB exchange rate. This follows directly from the physics: because 3 dB represents a doubling of acoustic power, halving the duration at 3 dB higher delivers the same total energy to the ear. Most international standards and the U.S. National Institute for Occupational Safety and Health (NIOSH) use this 3 dB rate. The Occupational Safety and Health Administration (OSHA), however, uses a 5 dB exchange rate. That more lenient rate assumes the ear recovers somewhat during quieter intervals between noise bursts.
The difference is not academic. In a large study of monitored U.S. workers, half had full-shift exposures at or above 85 dB when measured using OSHA’s 5 dB rate, but nearly three-quarters exceeded 85 dB under the stricter 3 dB rate. Only about 14% exceeded 90 dB under the 5 dB calculation, yet 42% exceeded 90 dB under the 3 dB calculation.2Occupational and Environmental Medicine. Noise exposure and hearing loss prevention programmes after 20 years of regulations in the United States The choice of exchange rate dramatically shifts how many workers appear to be at risk and how aggressively employers must intervene.
A systematic review of studies on permanent hearing threshold shifts found that the 3 dB exchange rate systematically overestimates the risk of noise-induced hearing loss for intermittent or fluctuating noise. The 5 dB rate appeared more accurate for predicting actual hearing outcomes, though it also overestimated risk at exposures above 100 dB.3Ear and Hearing. Exchange Rates for Intermittent and Fluctuating Occupational Noise: A Systematic Review of Studies of Human Permanent Threshold Shift This suggests that the ear does get some recovery benefit from intermittent quiet periods, making the pure energy-equivalent 3 dB model overly conservative for real-world fluctuating environments. The debate remains unresolved, but it illustrates a broader point: the logarithmic math is clean, but biology does not always follow the math perfectly.
Equal-Loudness Contours and the Frequency Problem
Decibels measure physical intensity, but loudness as you experience it depends heavily on frequency. Your ears are not equally sensitive across the pitch spectrum. A 40 dB tone at 1,000 Hz sounds much louder than a 40 dB tone at 50 Hz. To hear a very low-frequency sound as equally loud, the physical level often needs to be tens of decibels higher.
This frequency dependence is mapped by equal-loudness-level contours, which trace the dB level needed at each frequency to sound as loud as a reference tone. These contours were originally standardized in the mid-twentieth century, but significant errors were reported in the older standard by 1985. After 18 years of revision work, a fully updated international standard (ISO 226:2003) was published. That standard was revised again in 2023, incorporating an updated power exponent relating loudness perception to physical intensity and refining the mathematical expressions for better precision.4J-STAGE / Acoustical Science and Technology. Revision of ISO 226 “Normal Equal-Loudness-Level Contours” from 2003 to 2023 edition: The background and results
One revision was prompted by a seemingly tiny change: the threshold of hearing at 20 Hz was lowered by just 0.4 dB in a related standard. On a linear scale, 0.4 units would be trivial. But at very low frequencies where the ear is already straining, that fraction of a decibel reflects a meaningful shift in what we know about the boundary of audibility. It is another reminder that the compressed nature of the log scale can make small-looking numbers deceptively important.
The practical upshot is weighting filters. When you see a noise measurement reported in “dBA,” the A-weighting filter has already adjusted the raw decibel reading to approximate human frequency sensitivity. A reading in dBA de-emphasizes very low and very high frequencies where the ear is less sensitive, giving a single number that better predicts how loud something sounds. Without that filter, a measurement in flat (unweighted) decibels could be misleading. Two environments with the same unweighted dB reading can feel dramatically different in loudness if one is dominated by low-frequency rumble and the other by mid-frequency speech.
Common Misconceptions About Decibel Readings
One widespread mistake is treating decibel values as additive in the intuitive sense. If one lawnmower produces 90 dB, two identical lawnmowers running side by side do not produce 180 dB. They produce about 93 dB, because combining two equal power sources doubles the total power, and doubling power adds 3 dB. People who expect the numbers to stack linearly are often surprised that a second noise source barely moves the meter.
A related error is assuming that halving the distance to a sound source doubles the decibel reading. In a free field (outdoors, no reflections), moving half as far away increases the level by about 6 dB, not by doubling it. The inverse-square law governs intensity, and the logarithmic scale compresses that relationship. Indoors, reflections from walls and ceilings complicate things further, which is why a dishwasher that seemed quiet in the showroom can feel relentless in a small kitchen with hard surfaces.
Another misconception involves “zero decibels.” People sometimes think 0 dB means silence or the absence of sound. It does not. Zero dB SPL (sound pressure level) is defined as the reference pressure of 20 micropascals, which roughly corresponds to the quietest sound a healthy young human can detect at 1,000 Hz. Sounds can and do register below 0 dB SPL; they are just quieter than that reference threshold. Anechoic chambers routinely measure negative decibel values. Similarly, 0 dB in an electronic signal chain simply means the signal is at whatever reference level was defined for that system, not that nothing is happening.
How Other Species Interact with the Same Scale
Decibels describe physics, not biology, so the scale applies equally to sounds made by any source. But different species process those intensities in strikingly different ways. Research on bats illustrates this nicely. Bats that echolocate need to judge the loudness of returning echoes to estimate target distance and size. A study of multiple bat species found that for frequencies used in echolocation, brainstem auditory response functions were significantly shallower than for lower frequencies used in social calls.5PubMed Central. Hearing sensitivity and amplitude coding in bats are differentially shaped by echolocation calls and social calls In practical terms, the bat’s auditory system compresses loudness coding differently depending on the frequency band. For echolocation frequencies, the neural response grows more gradually with intensity, which gives the bat finer resolution for distinguishing small differences in echo strength, a critical ability when tracking prey in the dark at high speed.
This is a biological parallel to the engineering logic behind the decibel itself. Bats evolved a kind of internal gain control that is tuned to the task: broad, slow-growing response curves for frequencies where precise loudness discrimination matters, and steeper response curves for social-call frequencies where simply detecting the call is enough. The decibel scale was designed to serve a similar purpose for human engineers, compressing a huge range into workable numbers so that meaningful differences stand out. Evolution arrived at the same strategy independently, which says something about how universal the problem of handling wide dynamic ranges really is.
Decibels Outside of Acoustics
If you only encounter decibels in the context of sound, it is easy to forget that the unit was born in electrical engineering and lives a busy life far beyond acoustics. Signal-to-noise ratio in radio, Wi-Fi signal strength, fiber-optic link budgets, antenna gain, seismology, and even some medical imaging measurements all use decibels. The underlying logic is always the same: you have a ratio of two power-like quantities, and you express it logarithmically to make the numbers compact and the arithmetic easy.
Wi-Fi signal strength is one place you may have encountered decibels without realizing what was going on. Your router’s signal is often reported in dBm, which references one milliwatt. A typical home Wi-Fi signal might be around −40 dBm close to the router and −70 dBm in a distant room. That 30 dB drop means the signal power at the far end is a thousand times weaker, even though “−40 to −70” does not look like a thousand-fold difference at a glance. Understanding that the gap is logarithmic explains why your video call works fine in the living room and stutters in the garage: you are not dealing with a modest reduction in signal but an enormous one.
In fiber-optic networks, engineers add up gains and losses along a link in decibels, exactly the way telecommunications engineers did when the unit was first formalized in 1929. A splice might lose 0.1 dB, a connector might lose 0.5 dB, and an amplifier might add 20 dB. The entire link budget is a simple addition problem, which is vastly easier than multiplying and dividing the raw power ratios at each point. This additive convenience is arguably the single biggest reason decibels persisted across so many fields: once you are working in log space, cascaded gains and losses become grade-school arithmetic.
When “Logarithmic” Becomes Misleading
Calling the decibel scale logarithmic is accurate, but it can create a false sense of linearity in perception. People hear “logarithmic” and sometimes assume it means the scale perfectly tracks how humans perceive loudness. It does not. The scale was designed to track power ratios, not subjective experience. The fact that roughly 10 dB corresponds to a doubling of perceived loudness is a convenient coincidence for mid-range levels, but the relationship bends at extremes. At very low levels near the hearing threshold, small decibel changes can feel dramatic. At very high levels, the ear’s own compression and protective reflexes alter the loudness-perception curve further.
Loudness perception also depends on duration. A 1-millisecond click at 80 dB does not sound as loud as a sustained 80 dB tone, even though the peak level is the same. This temporal integration means that a sound-level meter reading, which captures instantaneous or short-term-averaged decibel values, does not fully predict how loud something will feel to a listener. Standards for environmental noise often use time-averaged metrics like the equivalent continuous sound level for this reason, folding duration into the picture. The decibel captures the physics faithfully, but translating physics into human experience always requires an additional layer of interpretation that the scale alone cannot provide.