<- Back to laws
MOTOR CONTROL / SPEED-ACCURACY / TARGET ACQUISITION

Fitts's
Law

Aimed movements take longer when the target is farther away or narrower along the approach axis. The law turns a speed-accuracy trade-off into a testable quantitative model.

START
DISTANCE D
WIDTH W
MT = a + b log2(D/W + 1)FARTHER + NARROWER = MORE DIFFICULT
TypePsychomotor performance model
DomainMotor control / HCI / ergonomics
Scientific statusReplicated quantitative regularity
Primary predictionMovement time
Central trade-offSpeed versus accuracy
LIVE TARGET LAB

Make the target harder to acquire.

Adjust the target width and center-to-center distance, then complete ten alternating selections. The predicted time uses an illustrative calibration; your measured time also includes browser, device, reaction, and strategy effects. This is a teaching demo, not an ISO test.

SHANNON FORMULATIONID = log2(D / W + 1)Index of difficulty, measured in bits
DIFFICULTY2.58 bits
MODEL387 ms
YOUR MEAN--
MISSES0
280 px

Adjust the task, then begin.

This condition is moderately difficult. Increase distance or reduce width to raise the index of difficulty.

01 / MEANING

Distance matters relative to tolerance.

Fitts's Law models the time required for a rapid aimed movement. A farther target requires a larger movement amplitude. A narrower target allows less endpoint variability. Together they define an index of difficulty that often relates approximately linearly to average movement time within a given task, device, population, and experimental range.

The width is not simply the target's largest visible dimension. In the one-dimensional model it is the tolerance along the movement axis. A wide horizontal button approached vertically may offer less relevant tolerance than its visual area suggests. For two-dimensional targets, researchers must choose an appropriate geometric or effective-width formulation.

DOUBLE D, KEEP W
Harder

The movement spans more distance for the same terminal tolerance.

KEEP D, DOUBLE W
Easier

The endpoint can vary more while still counting as a successful selection.

DOUBLE BOTH
Similar ratio

The ratio D/W is preserved, so the model predicts similar difficulty.

02 / MODEL & VARIANTS

The equation is a fitted model, not a universal stopwatch.

COMMON HCI FORMID = log2(D/W + 1)

The Shannon formulation always gives a nonnegative difficulty and is widely used in HCI.

MOVEMENT MODELMT = a + b(ID)

a is the fitted intercept; b is the fitted slope in time per bit for one condition set.

ORIGINAL FORMID = log2(2A/W)

Fitts's 1954 reciprocal-tapping work used movement amplitude A and target tolerance W.

THROUGHPUTTP = IDe / MT

Bits per second, normally calculated over a sequence using effective difficulty.

DWD/WIDINTERPRETATION
1608021.58near + forgiving
3204083.17far + precise
48020244.64very demanding
Do not transfer a and b blindly.

They absorb device latency, movement mode, participant population, posture, feedback, selection method, and task procedure. Refit them when the experimental system changes.

03 / WHY THE RELATION EMERGES

Fast movement creates endpoint uncertainty.

Fitts framed the result using information theory: greater amplitude relative to tolerance requires more information to control the movement. Later accounts emphasize variability in motor commands, feedback-guided corrections, and optimized submovements. These interpretations are related but should not be collapsed into one settled neural mechanism.

01 / PLAN

Initial impulse

The system selects direction, force, and duration based on target geometry and prior experience.

02 / TRANSPORT

Rapid approach

Much of the distance is covered quickly; higher speed can increase endpoint variability.

03 / CORRECT

Feedback corrections

As the pointer nears the target, smaller adjustments trade time for accuracy.

04 / SELECT

Commitment

Click, lift, dwell, or touch-up rules add device- and technique-specific demands.

FASTERMORE ENDPOINT VARIABILITYMORE ACCURATE
04 / MEASUREMENT PROTOCOL

A serious Fitts study measures behavior, not button pixels.

01Define the task

Specify pointing, tapping, dragging, touch, gaze, stylus, or another movement and its selection rule.

02Sample difficulty

Use multiple distance-width combinations across a meaningful range rather than one easy condition.

03Collect sequences

Use repeated trials, counterbalanced conditions, practice, and enough participants for stable estimates.

04Record endpoints

Time alone cannot reveal the speed-accuracy strategy or effective target width.

05Compute effective width

Project selection coordinates onto the task axis and estimate We from endpoint dispersion.

06Report the model

Give errors, MT, a, b, fit diagnostics, throughput method, exclusions, and uncertainty.

NOMINAL TARGET W
OBSERVED ENDPOINT SPREAD
We = 4.133 SDx

Under the standard one-dimensional method, effective width reflects the variability participants actually produced. It adjusts difficulty for speed-accuracy bias.[7]

STANDARD CONTEXT

ISO/TS 9241-411:2012 specifies evaluation methods for physical input devices including mice, touchpads, touchscreens, tablets, joysticks, trackballs, and styli. Its procedure is much more controlled than this page's live demo.[5]

05 / HISTORY

From metal plates to graphical interfaces.

1954Paul M. Fitts

Reciprocal tapping experiments connect movement amplitude and tolerance to an information-based index of difficulty.[1]

1964Fitts & Peterson

Discrete target reaches extend the analysis and distinguish movement time effects from reaction time effects.[2]

1980s-90sHCI adoption

Researchers apply the model to mice, joysticks, trackballs, menus, and graphical target selection.

1992MacKenzie synthesis

A major review clarifies model variants, theoretical issues, and the law's value as an HCI research tool.[3]

2000s-nowStandardized evaluation

Effective width, multidirectional tasks, and throughput support comparative input-device testing across desktop, touch, gaze, and immersive systems.

06 / DESIGN APPLICATIONS

Design with the model, but validate the interface.

DESKTOP

Menus and toolbars

Frequent controls benefit from adequate motor tolerance, stable placement, and reduced travel from likely starting positions.

TOUCH

Finger selection

Occlusion, contact area, posture, handedness, and absolute touch precision complicate a simple mouse-derived width rule.

ACCESSIBILITY

Motor variability

Larger activation regions, spacing, alternatives, and personalization can reduce precision demands without visually overwhelming the layout.

SCREEN EDGES

Boundary assistance

For pointer devices that stop at a screen boundary, an edge target can gain effective tolerance in the blocked direction.

VR / AR

Distant and expanding targets

Depth, ray instability, head motion, selection confirmation, and dynamic target expansion change the acquisition problem.

SAFETY-CRITICAL

Separate destructive actions

Distance and tolerance should work with confirmation, undo, labeling, and mode awareness - not replace them.

FLOATING TARGET

The pointer can overshoot on every side.

EDGE TARGET

The physical screen boundary can stop one direction of travel.

BOUNDARIES CHANGE EFFECTIVE TOLERANCE
07 / CASE STUDIES

Geometry is only one layer of usability.

CASE A / TINY ICON

A frequent action sits far from the work.

CURSOR

Increasing its hit region or moving it nearer reduces predicted pointing difficulty. But discoverability and labeling still need separate evaluation.

CHANGE: geometry + placement
CASE B / TOUCH KEYBOARD

Visual keys and touch distributions differ.

Model actual touch endpoints, adjacent-key errors, hand posture, and language prediction rather than treating nominal key width as the whole task.

CHANGE: effective width + error model
CASE C / DESTRUCTIVE CONTROL

Delete is large and easy to hit.

DELETEUNDO?

Fitts-friendly acquisition can increase accidental activation. Safety requires separation, confirmation proportional to risk, and recoverability.

CHANGE: motor ease + error recovery
CASE D / NARROW PATH

The task is steering, not pointing.

Moving through a constrained tunnel is better described by the Steering Law, which integrates path length and width along the trajectory.

CHANGE: choose the correct model
08 / LIMITS & MISUSE

Where the shortcut fails.

WRONG

"Big buttons are always better."

Layout density, scanning, occlusion, semantics, frequency, and neighboring actions also matter.

WRONG

"Visual size equals target width."

The relevant tolerance depends on movement direction, activation region, and observed endpoints.

WRONG

"One click predicts whole-task speed."

Search, decision, reaction, mode switching, dragging, and confirmation may dominate.

WRONG

"A high R-squared validates the interface."

A linear fit can coexist with high errors, poor labels, biased sampling, or an irrelevant task.

WRONG

"Throughput is a universal device constant."

Procedure, task range, calculation choices, participants, and strategy affect the estimate.

WRONG

"Mouse results transfer to touch or gaze."

Input mappings, biomechanics, noise, occlusion, and selection rules differ.

FITTS MODELSRapid aimed movement under defined spatial tolerance.FITTS DOES NOT MODELMeaning, choice, visual search, learning, or total workflow value.
10 / REFERENCES

Sources and further reading.

Original experiments, major HCI syntheses, the current ISO evaluation specification, and specialist measurement guidance are prioritized.

  1. Paul M. Fitts (1954) - The Information Capacity of the Human Motor SystemThe original reciprocal-tapping experiments relating movement amplitude, tolerance, speed, and information.Journal of Experimental Psychology
  2. Fitts & Peterson (1964) - Information Capacity of Discrete Motor ResponsesDiscrete target-reaching experiments examining amplitude, width, reaction time, and movement time.Journal of Experimental Psychology / PubMed
  3. I. Scott MacKenzie (1992) - Fitts' Law as a Research and Design Tool in HCIA major historical, theoretical, methodological, and application-focused review.Human-Computer Interaction
  4. Soukoreff & MacKenzie (2004) - Towards a Standard for Pointing Device EvaluationA detailed synthesis of recommended Fitts-law procedures for evaluating computer input devices.International Journal of Human-Computer Studies
  5. ISO/TS 9241-411:2012 - Evaluation Methods for Physical Input DevicesThe confirmed ISO technical specification for laboratory evaluation of keyboards and pointing devices.International Organization for Standardization
  6. MacKenzie (1989) - A Note on the Information-Theoretic Basis for Fitts' LawMotivation for the Shannon formulation and discussion of the model's information-theoretic interpretation.Journal of Motor Behavior / York University
  7. I. Scott MacKenzie - Throughput and Effective WidthTechnical documentation for effective amplitude, effective width, index of difficulty, and throughput calculations.York University
  8. Accot & Zhai (1997) - Beyond Fitts' LawIntroduces a quantitative model for trajectory-based interface tasks now known as the Steering Law.ACM CHI
  9. MacKenzie & Buxton (1992) - Extending Fitts' Law to Two-Dimensional TasksExamines target-width formulations for targets with two-dimensional geometry.ACM CHI
  10. Expanding Targets in Virtual Reality Environments (2023)An ISO-style study of dynamic target expansion during pointing in immersive VR.arXiv
CONTINUE EXPLORING

Related laws, with the relationship made explicit.

These are editorial connections, not claims that the laws are mathematically equivalent.

CONTINUE READING

Place this law inside the collection.

LAW 010 / 100 PUBLISHED