What Research Rhythm Tests Actually Measure

A plain-language tour of the standard timing batteries, and of what each score is and is not evidence for.

Short answer: research timing tests are not one test. They are small families of tasks, each isolating one thing: can you hear that two durations differ, can you hear that one event in a regular sequence was displaced, can you hear whether a click track sits on the beat of a piece of music, can you place your own taps on an external pulse, and can you hold a tempo once the sound has gone. The best known battery, BAASTA, covers all five. Separate instruments such as the H-BAT, the CA-BAT and the Beat-Drop Alignment Test attack narrower slices with more psychometric machinery.

Why a battery instead of one test

The reason is dissociation. Beat perception and beat production can come apart in the same person, which is the central finding running through the beat deafness literature. A single composite "rhythm score" would average that dissociation away and tell you nothing about which half is weak. Batteries exist so that a profile can be drawn rather than a rank assigned.

The second reason is that timing has more than one time scale. Judging whether a 200 ms tone is longer than a 220 ms tone is a different job from tracking a pulse across thirty seconds of music, and the brain appears to handle those with partly different machinery, as our guide to the neural mechanisms of rhythm describes. A battery samples several time scales on purpose, because a person can be excellent at one and weak at another.

BAASTA, task by task

BAASTA, the Battery for the Assessment of Auditory Sensorimotor and Timing Abilities, was published by Dalla Bella, Farrugia, Benoit, Bégel, Verga, Harding and Kotz in Behavior Research Methods in 2017. It is the most widely used tool of its kind, and its structure is worth knowing because a large share of popular claims about "rhythm ability" trace back to one of its subtests.

Duration discrimination. Two intervals, one standard and one comparison. Which is longer? This is the purest measure of interval sensitivity, with no beat and no movement involved at all. It is the task behind the thresholds discussed in tempo just-noticeable difference and Weber's law.

Anisochrony detection. A regular sequence plays, one interval is shortened or lengthened, and you say whether anything was wrong. It is run with plain tones and again with real music, because those are very different problems: music brings its own accents, and a displaced event can be masked or exposed depending on where it falls in the bar.

Beat alignment. Music plays with a click track over it, aligned or misaligned, and you judge which. No tapping. This is the standard way to measure whether the beat is being extracted at all, independently of whether the body can act on it.

Paced and unpaced tapping. Tap with a metronome, tap with music, and tap with nothing. Unpaced tapping reveals the rate the internal timekeeper drifts toward, which is the phenomenon our guide to your internal clock tempo covers.

Synchronization-continuation. The pacing sound plays, then stops, and you keep going. This is the cleanest available split between "following" and "holding," and it is the reason the game Chronecho exists in the form it does.

Adaptive tapping. The sequence changes tempo and you have to follow it. This measures error correction, the continuous small adjustment that keeps a tapper attached to a pulse that is not perfectly stable, and it is the ability our guide to tempo changes describes from the musical side.

BAASTA has been examined for test-retest reliability by Bégel and colleagues in Annals of Physical and Rehabilitation Medicine in 2018, and a mobile implementation with adult norms was published in Behavior Research Methods in 2024. That progression matters: a battery without repeat-measurement data and without norms is a set of tasks, not an instrument. Reliability tells you whether the same person scores similarly twice; norms tell you what a score means relative to anyone else.

The narrower instruments

Three other tests appear constantly in the literature, each solving a specific problem with the general battery.

The Harvard Beat Assessment Test (H-BAT), published by Fujii and Schlaug in Frontiers in Human Neuroscience in 2013, was built specifically to test whether beat perception and beat production dissociate. It has four subtests: a music tapping test, a beat saliency test, a beat interval test, and a beat finding and interval test. Three of them use adaptive staircase methods, meaning the difficulty tracks your performance rather than following a fixed list, so a threshold is estimated efficiently in relatively few trials.

The Computerised Adaptive Beat Alignment Test (CA-BAT), published by Harrison and Müllensiefen in Scientific Reports in 2018, took the beat alignment idea and rebuilt it with item response theory, adaptive testing and automatic item generation, then validated it across four studies. The authors' own conclusion is worth repeating because it is unusually candid: the test performed well in the laboratory but was not well suited to online testing, because it depends on fine perceptual discrimination that consumer setups degrade.

The Beat-Drop Alignment Test (BDAT), published in Attention, Perception, & Psychophysics in 2022, addresses a different weakness. In a conventional beat alignment task the superimposed click is itself a sensory cue to the beat, which means a listener can partly solve the task by local acoustic comparison rather than by tracking a pulse. The BDAT tests beat perception without giving such a cue, by asking listeners to judge alignment at a point where the beat has to be internally maintained.

What each number is evidence for

Read the last column carefully. The games in it are analogues, not implementations. They borrow the shape of a task, not its calibration, its normative sample, or its controlled administration.

Why your phone is the weak link

Every one of these tasks assumes that the sound reaches the ear when the software thinks it does, and that a tap is registered when the finger lands. Consumer devices break both assumptions. Output latency, input latency, wireless audio buffering and browser scheduling can each add tens of milliseconds, and the total is often larger than the effect being measured. That is precisely why the CA-BAT authors flagged online administration as a limitation, and it is why we published a separate audio latency and timing offset test. If you want your own numbers to mean anything across sessions, measure your offset first and keep the hardware constant.

To be explicit about scope: none of Chronkle's games are diagnostic instruments. They are not screening tools, not assessments and not therapy. They are daily perception puzzles. A score tells you how a particular round went on a particular device, and any concern about timing, hearing, speech or movement belongs with a qualified clinician rather than with a browser game.

Keep reading

• Tempo JND and Weber's law: what a discrimination threshold is and why it scales.
• Beat deafness: what these batteries were built to detect.
• Audio latency and timing offset test: measure your own device before trusting any score.

Sources: Dalla Bella, Farrugia, Benoit, Bégel, Verga, Harding & Kotz, BAASTA: Battery for the Assessment of Auditory Sensorimotor and Timing Abilities (Behavior Research Methods, 2017); Bégel et al., Test-retest reliability of the Battery for the Assessment of Auditory Sensorimotor and Timing Abilities (Annals of Physical and Rehabilitation Medicine, 2018); Dalla Bella et al., Mobile version of the BAASTA: implementation and adult norms (Behavior Research Methods, 2024); Fujii & Schlaug, The Harvard Beat Assessment Test (H-BAT): a battery for assessing beat perception and production and their dissociation (Frontiers in Human Neuroscience, 2013); Harrison & Müllensiefen, Development and validation of the Computerised Adaptive Beat Alignment Test (CA-BAT) (Scientific Reports, 2018); Testing beat perception without sensory cues to the beat: the Beat-Drop Alignment Test (BDAT) (Attention, Perception, & Psychophysics, 2022).