Skip to main content

Trail Making Test

Version: v1 (current)

A neuropsychological assessment of visual attention, processing speed, task switching, and executive function through number and letter sequencing.

Overview​

The Trail Making Test (TMT) is one of the most widely used neuropsychological instruments worldwide. It consists of two parts: Part A requires connecting numbered circles in ascending order (1-2-3...), measuring processing speed and visual scanning. Part B requires alternating between numbers and letters in sequence (1-A-2-B-3-C...), adding a cognitive switching component that taps executive function.

The difference between Part B and Part A completion times (B-A) provides a relatively pure measure of executive function (task switching and mental flexibility) while controlling for processing speed. The TMT is sensitive to brain injury, dementia, ADHD, and normal aging, making it valuable for clinical assessment and cognitive screening.

Originally developed for military personnel assessment, the TMT has been validated across countless studies and is included in virtually all comprehensive neuropsychological batteries.

Scientific Background​

Classic Findings:

  • Part A (Numbers Only): Mean completion time ~25-30 seconds in healthy adults
  • Part B (Number-Letter Switching): Mean completion time ~60-75 seconds in healthy adults
  • B-A Difference: Mean ~40 seconds; represents executive function "cost"
  • Error Rate: Healthy adults make 0-1 errors; increased errors suggest executive dysfunction
  • Age Effects: Strong age gradient; older adults significantly slower, especially on Part B
  • Brain Sensitivity: Sensitive to frontal lobe damage, diffuse injury, white matter disease

Key Mechanisms:

  • Processing Speed: Visual scanning and motor speed (Part A)
  • Task Switching: Alternating between number and letter sequences (Part B)
  • Working Memory: Maintaining two sequences simultaneously
  • Cognitive Flexibility: Shifting between mental sets
  • Visual Search: Locating next target among distractors

Seminal Papers:

  • Reitan (1958): Validity of the Trail Making Test
  • Lezak, Howieson, Bigler, & Tranel (2012): Neuropsychological Assessment (comprehensive review)
  • Sánchez-Cubillo et al. (2009): Construct validity of Trail Making Test: Role of task-switching

Why Researchers Use This Task​

  1. Neuropsychological Assessment: Screen for cognitive impairment across disorders
  2. Dementia Screening: Sensitive to early Alzheimer's disease and MCI
  3. Executive Function: Measure task-switching and mental flexibility
  4. Traumatic Brain Injury: Assess recovery and persistent deficits
  5. ADHD Assessment: Evaluate attention and executive control
  6. Aging Research: Track cognitive decline across lifespan
  7. Clinical Trials: Outcome measure for cognitive interventions

Current Implementation Status​

Fully Implemented:

  • ✅ Part A (numbers 1-25)
  • ✅ Part B (alternating numbers 1-13 and letters A-L)
  • ✅ Completion time measurement
  • ✅ Error detection and self-correction tracking
  • ✅ Touch/mouse drawing of paths
  • ✅ Practice trials

Fixed by Design (not configurable):

  • Standard Reitan layout, authored in a fixed 1030×1030 px coordinate system (for normative comparability). The canvas is displayed at a uniform scale of that layout, fit to the available viewport width and height — the ratio is never stretched, and the canvas can be enlarged as well as shrunk depending on screen size
  • Immediate error feedback (wrong segment flashes red)
  • Error correction required (the participant must reach the correct next circle to advance)

Not Yet Implemented:

  • ❌ Pressure/stylus sensitivity for motor analysis
  • ❌ Multiple equivalent forms for repeat testing (layout is fixed)

Configuration Parameters​

The Trail Making Test exposes a deliberately small configuration surface. The node layouts, canvas size, circle sequences, error feedback, and correction behavior are fixed in code to preserve comparability with the standard Reitan administration; they are not configurable.

Options​

ParameterTypeDefaultDescription
test_partsstring[]['A', 'B']Which parts to administer; built from the Part A / Part B checkboxes (defaults to both if neither is selected)
practice_enabledbooleantrueInclude guided practice trials before each part

Instructions​

ParameterTypeDefaultDescription
main_instructionsstring (HTML)''Main instructions shown before the task
hint_instructionsstring (HTML)''Optional hint text
part_a_instructionsstring (HTML)''Instructions shown during Part A practice
part_b_instructionsstring (HTML)''Instructions shown during Part B practice
during_part_astring (HTML)''Hint available (via the ? button) during Part A
during_part_bstring (HTML)''Hint available (via the ? button) during Part B

Data Output​

Markers​

The shared task runner emits the task_start marker centrally at the Start click, so this task does not emit its own. The task records the following per-part markers:

Part start (part_start) — one per part (and per practice run):

{
  "type": "part_start",
  "ts": "2026-02-14T10:30:00.123Z",
  "hr": 1234.56,
  "data": {
    "part": "A",
    "is_practice": false,
    "sequence_length": 25,
    "circle_positions": [
      {"label": "1", "x": 739, "y": 589},
      {"label": "2", "x": 489, "y": 754}
    ]
  }
}

Part complete (part_complete) — carries { part, is_practice, resize_metadata }. resize_metadata records the canvas's display/scale state at that moment: authored_width/authored_height (always 1030), rendered_width/rendered_height (the on-screen size), scale_x/scale_y (always equal — the canvas is scaled uniformly, never stretched), viewport_width/viewport_height, and device_pixel_ratio.

Task complete (task_complete) — emitted with no extra data when all selected parts finish.

Response Data​

Responses are recorded for individual events during each part. Each response includes an event field indicating its type.

Circle clicked (circle_clicked):

{
  "event": "circle_clicked",
  "part": "A",
  "is_practice": false,
  "circle_index": 5,
  "circle_label": "6",
  "expected_label": "6",
  "correct": true,
  "sequence_position": 6,
  "segment_duration_ms": 1245
}

Error (error):

{
  "event": "error",
  "part": "B",
  "is_practice": false,
  "error_type": "alternation_error",
  "from_label": "7",
  "to_label": "F",
  "expected_label": "D"
}

error_type is one of sequence_error, skip_error, or alternation_error (alternation errors apply to Part B only).

Part complete (part_complete):

{
  "event": "part_complete",
  "part": "A",
  "is_practice": false,
  "completion_time_ms": 28400,
  "completed": true,
  "total_errors": 1,
  "error_types": {"sequence_error": 1},
  "sequence_breaks": 1,
  "resize_metadata": {
    "authored_width": 1030,
    "authored_height": 1030,
    "rendered_width": 824,
    "rendered_height": 824,
    "scale_x": 0.8,
    "scale_y": 0.8,
    "viewport_width": 1024,
    "viewport_height": 900,
    "device_pixel_ratio": 2
  }
}

Task complete (task_complete): recorded as { "event": "task_complete" } only.

Summary Artifact​

A JSON file (tmt_summary_<index>.json) with headline statistics for the administered (non-practice) parts:

{
  "schema_version": 1,
  "task": { "kind": "trail_making", "version": "v1", "index": 0 },
  "measures": {
    "part_a_completion_time_ms": 28400,
    "part_b_completion_time_ms": 72100,
    "tmt_b_minus_a_ms": 43700,
    "part_a_errors": 1,
    "part_b_errors": 2,
    "total_errors": 3,
    "error_ratio_b_to_a": 2
  },
  "meta": {
    "parts_administered": 2,
    "parts_completed": 2,
    "part_a_completed": true,
    "part_b_completed": true,
    "summary_valid": true,
    "resize_metadata": {
      "authored_width": 1030,
      "authored_height": 1030,
      "rendered_width": 824,
      "rendered_height": 824,
      "scale_x": 0.8,
      "scale_y": 0.8,
      "viewport_width": 1024,
      "viewport_height": 900,
      "device_pixel_ratio": 2
    }
  }
}

Key metrics:

  • part_a_completion_time_ms / part_b_completion_time_ms: Total time to complete each part (milliseconds); null unless that part was administered and finished (not force-advanced/abandoned)
  • tmt_b_minus_a_ms: Part B − Part A difference (executive-function index); null unless both parts finished
  • part_a_errors / part_b_errors: Error count per part; null unless that part finished
  • total_errors: Combined error count across all administered parts; null unless every administered part finished (a partial sum over an abandoned part would be misleading)
  • error_ratio_b_to_a: Ratio of Part B to Part A errors; null unless both parts finished and Part A has at least one error

Run metadata (meta):

  • parts_administered: Number of non-practice parts that were run, regardless of outcome
  • parts_completed: Number of those parts that were genuinely finished (not force-advanced)
  • part_a_completed / part_b_completed: Per-part completion flag (null if that part was not administered)
  • summary_valid: true only when at least one part was administered and every administered part finished — check this before trusting the headline measures above; if false, the trail was abandoned partway and some measures will be null
  • resize_metadata: canvas display/scale state at the most recent part completion (see Markers above); useful for confirming the trail was displayed close to its authored size, or for correcting spatial analyses if it was scaled significantly

Note: the implementation does not compute percentiles, normative interpretations, or age/education-adjusted norms — those are analysis steps left to the researcher. Per-node error and click events (with their own total_errors per part_complete record) always ride the response stream regardless of whether the part finished; only the summary's headline aggregates are withheld for an incomplete part.

Example Research Configurations​

Standard Clinical Administration (Reitan Version)​

Part A: Connect 1-25 in order
Part B: Alternate 1-A-2-B-3-C through 13-L
Layout: Fixed standard Reitan positions (identical for every participant)
Timing: Millisecond completion time per part
Errors: Immediate feedback, correction required to advance
Scoring: Time in milliseconds, number of errors

Dementia Screening Battery​

Part A: Standard 1-25
Part B: Standard alternating
Administration: Part A first, brief rest, then Part B
Analysis: Compare to age-adjusted norms; B-A > 75s suggests impairment
Include: Error rate (>2 errors raises concern)

Longitudinal Study (Repeat Testing)​

Baseline: Standard TMT
Follow-ups: Use equivalent forms with different layouts
Interval: Minimum 6 months between administrations
Analysis: Track change over time; increases >10 seconds = meaningful decline

Research Protocol (Computerized)​

Version: Digital with touch/mouse input
Advantage: Precise millisecond timing, automated error detection
Caution: May not align with pen-and-paper norms
Analysis: Use computerized norms or compare within-subjects

Participant Experience​

  1. Instructions - Part A: "You will see numbered circles on the screen. Using your mouse/finger, draw a line connecting them in order: 1 to 2, 2 to 3, and so on. Work as quickly and accurately as you can."

  2. Practice - Part A:

    • See circles numbered 1-8 scattered on screen
    • Click/touch 1, drag to 2, drag to 3... up to 8
    • If error: "Oops! That's not the next number. Go back to [X]."
    • Complete practice successfully

  3. Main Task - Part A:

    • See circles numbered 1-25
    • Timer starts automatically
    • Connect all numbers in sequence as fast as possible
    • Timer stops when reaching 25

  4. Instructions - Part B: "Now you'll see both numbers and letters. Connect them in alternating order: 1 to A, A to 2, 2 to B, B to 3, and so on."

  5. Practice - Part B:

    • See circles with 1, A, 2, B, 3, C, 4, D
    • Connect in alternating order: 1-A-2-B-3-C-4-D
    • Practice until correct

  6. Main Task - Part B:

    • See circles with 1-13 (numbers) and A-L (letters)
    • Connect in alternating sequence: 1-A-2-B-3-C... 12-L-13
    • Timer records completion time

When the final selected part finishes (or the participant clicks Skip during the main trials of a part), the task ends and the session moves on. Completion time and error counts are saved with the participation; no on-screen summary is shown to the participant.

Design Recommendations​

General Guidelines​

  • Order: Always administer Part A before Part B (standardization)
  • Practice: Mandatory for valid results; ensure understanding
  • Environment: Minimize distractions; quiet testing space
  • Equipment: Touch screen or mouse with precision
  • Instructions: Clear, consistent across participants

Scoring and Interpretation​

Normative Data (Approximate, Healthy Adults 20-50 years):

  • Part A: Mean ~25-30 sec, SD ~8-10 sec
  • Part B: Mean ~60-75 sec, SD ~20-25 sec
  • B-A Difference: Mean ~40 sec, SD ~15 sec
  • Errors: 0-1 typical; 2+ suggests concern

Age-Adjusted Norms (critical for interpretation):

  • Ages 20-39: A ~22 sec, B ~52 sec
  • Ages 40-59: A ~28 sec, B ~72 sec
  • Ages 60-79: A ~40 sec, B ~110 sec
  • Ages 80+: A ~55 sec, B ~165 sec

Impairment Cutoffs (vary by source):

  • Mild Impairment: A >40 sec or B >110 sec (for ages 20-59)
  • Moderate Impairment: A >60 sec or B >180 sec
  • Severe Impairment: A >90 sec or B >300 sec (timeout)
  • Executive Dysfunction: B-A >75 sec or errors >2

Population-Specific Adaptations​

Children (Ages 8-15):

  • Part A: 1-15 (fewer nodes)
  • Part B: 1-A through 8-H
  • Slower expected times; use pediatric norms
  • More practice trials tolerated

Older Adults (75+):

  • Standard version appropriate
  • Allow generous time limit (no pressure)
  • Very sensitive to white matter disease
  • Expect 2-3x longer than young adults

Low Education/Literacy:

  • Part A less affected by education
  • Part B may be more challenging
  • Ensure familiarity with alphabet sequencing
  • Use education-adjusted norms

Clinical Populations:

  • Alzheimer's: Dramatically slowed, especially Part B; many errors
  • Frontal Lobe Damage: Severe Part B deficit; B-A often >100 sec
  • ADHD: Moderately slowed; more errors on B
  • Depression: Slowed but fewer errors; improves with treatment
  • Parkinson's: Slowed motor speed; B-A ratio often normal

Common Issues and Solutions​

IssueSolution
Participant lifts pen/fingerInstruct to maintain continuous path; doesn't invalidate if rare
Many errors on Part BRepeat instructions; ensure understanding of alternation
Very slow but accurateNormal strategy; record time; don't pressure for speed
Fast but errors ignoredEmphasize accuracy as well as speed in instructions
Digital vs. paper normsUse digital-specific norms or compare within-subjects
Learning effects (retest)Use alternate forms; wait 6+ months between administrations

Data Analysis​

Primary Outcomes:

  1. Time to Completion: Part A, Part B (in seconds)
  2. Error Count: Total errors on each part
  3. B-A Difference: Executive function index
  4. B/A Ratio: Alternative executive measure (less age-sensitive)

Secondary Measures:

  • Self-Corrections: Errors caught and fixed by participant
  • Path Efficiency: Total path length (if tracking X-Y coordinates)
  • Segment Times: Time between each node pair
  • Velocity Changes: Speed-up or slow-down patterns

Statistical Considerations:

  • Age Adjustment: Critical; use age-stratified norms
  • Education Adjustment: Some norms adjust for education level
  • Outliers: Ceiling at 300 seconds (timeout) for statistical analysis
  • Skewness: Times are positively skewed; consider log transformation

References​

  • Reitan, R. M. (1958). Validity of the Trail Making Test as an indicator of organic brain damage. Perceptual and Motor Skills, 8(3), 271-276.

  • Lezak, M. D., Howieson, D. B., Bigler, E. D., & Tranel, D. (2012). Neuropsychological Assessment (5th ed.). Oxford University Press.

  • Tombaugh, T. N. (2004). Trail Making Test A and B: Normative data stratified by age and education. Archives of Clinical Neuropsychology, 19(2), 203-214.

  • Sánchez-Cubillo, I., Periáñez, J. A., Adrover-Roig, D., Rodríguez-Sánchez, J. M., Ríos-Lago, M., Tirapu, J., & Barceló, F. (2009). Construct validity of the Trail Making Test: Role of task-switching, working memory, inhibition/interference control, and visuomotor abilities. Journal of the International Neuropsychological Society, 15(3), 438-450.

  • Arbuthnott, K., & Frank, J. (2000). Trail Making Test, Part B as a measure of executive control: Validation using a set-switching paradigm. Journal of Clinical and Experimental Neuropsychology, 22(4), 518-528.

See Also​