Skip to main content

Navon Task

Version: v1 (current)

A hierarchical stimuli paradigm measuring global versus local processing biases in visual attention.

Overview​

The Navon task uses hierarchical letter stimuli, where large "global" letters are composed of smaller "local" letters, to investigate how attention is allocated between different levels of visual structure. For example, a large letter "H" might be made up of small letter "S"s. Participants are instructed to identify either the global letter (H) or the local letters (S) by pressing the corresponding key.

This elegant paradigm reveals fundamental principles of visual perception: typically, global features are processed faster than local features ("global precedence"), and incongruent local features interfere with global identification more than vice versa. These effects vary with cognitive style, clinical conditions, and cultural background, making the Navon task a valuable tool for investigating individual and group differences in attentional scope.

The task is widely used in cognitive psychology, neuroscience, and clinical research on autism, schizophrenia, and aging.

Scientific Background​

Classic Findings:

  • Global Precedence: Global letters are identified faster than local letters
  • Asymmetric Interference: Incongruent local letters slow global identification, but incongruent global letters have less effect on local identification
  • Congruency Effects: Performance is better when global and local levels are congruent (e.g., large H made of small H's) versus incongruent (large H made of small S's)
  • Individual Differences: Cognitive style (holistic vs. analytic) modulates global-local balance

Key Mechanisms:

  • Low Spatial Frequency Advantage: Global structure conveyed by low spatial frequencies, processed first
  • Attentional Scope: Broad attention facilitates global processing; narrow focus facilitates local processing

Seminal Paper:

  • Navon, D. (1977). Forest before trees: The precedence of global features in visual perception. Cognitive Psychology, 9(3), 353-383.

Why Researchers Use This Task​

  1. Attention Research: Study the scope and flexibility of visual attention
  2. Clinical Assessment: Evaluate global-local processing in autism (local bias) and schizophrenia
  3. Cognitive Neuroscience: Investigate visual processing pathways and hierarchical perception
  4. Individual Differences: Measure analytic vs. holistic cognitive styles
  5. Cross-Cultural Studies: Explore cultural differences in attention to global vs. local features

Configuration Options​

Response Mode​

ParameterTypeDefaultDescription
Time-based trialsbooleanTrueIf enabled, trials auto-advance after stimulus duration; if disabled, participant responds via button

Visual Settings​

ParameterTypeDefaultDescription
Global font sizenumber120Font size for the large letter shape (8-400 pixels)
Local font sizenumber16Font size for the small letters forming the shape (8-400 pixels)
Letter spacingtext0.5emCSS letter-spacing value for local letters
Line heighttext1.2CSS line-height value for local letter rows

Level Cue​

A per-trial text label can appear above the stimulus telling the participant which level to attend to (global vs. local). The label text is configurable and is read directly from the stored parameters (not translated at runtime).

ParameterTypeDefaultDescription
Show level cuebooleanTrueShow a text label above the stimulus on each trial indicating the target level
Global level labeltextGlobal (large letter)Participant-facing cue shown on global-target trials
Local level labeltextLocal (small letters)Participant-facing cue shown on local-target trials

Practice Trials​

ParameterTypeDefaultDescription
Enable practicebooleanTrueShow practice trials with visual feedback before main trials
Practice trialsarray(default set)Array of practice trial configurations

Keyboard Shortcuts​

Researchers can customize the keyboard bindings used during the task:

ParameterTypeDefaultDescription
Show keyboard hintbooleanTrueDisplay an on-screen hint showing the configured key
Response keykeyAny letter (A-Z)Key accepted for letter identification response
Response action labeltext"to match the target letter"Label shown in the keyboard hint

The default "Any letter" setting accepts any letter key A-Z, which is appropriate since the task requires identifying specific letters.

Trial Configuration​

Each trial is defined in the Trials spreadsheet with the following columns:

ColumnDescriptionExample Values
global_letterThe large letter shape formed by the patternH, S, E, L, T
local_letterThe small letter used to form the global shapeH, S, E, L, T
target_levelWhich level the participant should respond toGlobal, Local
conditionCongruency label for the trialCongruent, Incongruent, Neutral
expected_responseThe correct keyboard responseH, S, E, etc.
blockOptional grouping labelGlobal Block, Local Block
fixation_msFixation cross duration before stimulus500
stimulus_msStimulus display duration2000

Example Trials​

Global Block (respond to large letter):

| global_letter | local_letter | target_level | condition    | expected_response | fixation_ms | stimulus_ms |
|---------------|--------------|--------------|--------------|-------------------|-------------|-------------|
| H             | H            | global       | congruent    | H                 | 500         | 2000        |
| H             | S            | global       | incongruent  | H                 | 500         | 2000        |
| S             | S            | global       | congruent    | S                 | 500         | 2000        |
| S             | H            | global       | incongruent  | S                 | 500         | 2000        |

Local Block (respond to small letters):

| global_letter | local_letter | target_level | condition    | expected_response | fixation_ms | stimulus_ms |
|---------------|--------------|--------------|--------------|-------------------|-------------|-------------|
| H             | H            | local        | congruent    | H                 | 500         | 2000        |
| H             | S            | local        | incongruent  | S                 | 500         | 2000        |
| S             | S            | local        | congruent    | S                 | 500         | 2000        |
| S             | H            | local        | incongruent  | H                 | 500         | 2000        |

Optional Neutral Condition​

The neutral condition is an optional third value. It is researcher-labeled: the runtime does not infer it from the letter pair. The standard operationalization is to place a control letter outside the study's response set at the non-target level. The expected_response is set in the same way as for any other trial; correctness scoring is unchanged.

Examples:

Global-target neutral (participant attends to the global letter; local level carries a control letter outside the response set):

| global_letter | local_letter | target_level | condition | expected_response | fixation_ms | stimulus_ms |
|---------------|--------------|--------------|-----------|-------------------|-------------|-------------|
| H             | F            | global       | neutral   | H                 | 500         | 2000        |

Local-target neutral (participant attends to the local letter; global level carries a control letter outside the response set):

| global_letter | local_letter | target_level | condition | expected_response | fixation_ms | stimulus_ms |
|---------------|--------------|--------------|-----------|-------------------|-------------|-------------|
| F             | H            | local        | neutral   | H                 | 500         | 2000        |

Note: neutral trials can look visually similar to incongruent trials. The condition column and the choice of control letter are the researcher's design responsibility. Neutral is entirely optional; a study that uses only congruent and incongruent trials is fully valid.

Practice Trials​

Practice is controlled by a single Enable practice trials toggle (default: on):

  • Disabled: The task begins directly with the main trials.
  • Enabled: When practice is on and at least one practice trial is configured, the participant first sees the practice-instructions page, then completes all practice trials, then sees the trials-instructions page before the main (scored) trials begin.

During practice, participants receive visual feedback after each response (green checkmark for correct, red X for incorrect). Practice responses are recorded separately and do not count toward the main scored statistics.

Participant Experience​

Trial Sequence​

  1. Main Instructions: Overview of the task structure
  2. (Optional) Practice Instructions: If practice enabled
  3. (Optional) Practice Trials: With visual feedback
  4. (Optional) Trials Instructions: Shown before main trials
  5. Main Trials: Each trial follows this sequence:
    • Fixation cross (+) appears (if fixation_ms > 0)
    • Hierarchical stimulus appears (large letter made of small letters)
    • Participant presses the appropriate letter key
    • Trial advances automatically (time-based) or after response (button-based)

Response Methods​

Keyboard (recommended):

  • Press the letter key corresponding to the target level (default accepts any A-Z key -- configurable by researcher)
  • For global trials: respond to the large letter shape
  • For local trials: respond to the small letters

Buttons (if time-based mode disabled):

  • Click the appropriate response button

All keyboard bindings are configurable by the researcher in the study configuration. The keys listed above are the defaults.

Data Output​

Markers and Responses​

Each trial also emits a trial-onset marker named navon_trial_start (or navon_practice_trial_start during practice), along with generic fixation_onset / fixation_offset, stimulus_aoi, and response_recorded markers. The per-trial index field is trial_index in the main phase and practice_trial_index during practice (with is_practice on every record).

Stimulus marker (stimulus_shown):

{
  "type": "stimulus_shown",
  "ts": "2024-01-01T00:00:01.000Z",
  "hr": 1234.56,
  "data": {
    "trial_index": 1,
    "stimulus_id": "navon_0_1",
    "global_letter": "H",
    "local_letter": "S",
    "target_level": "global",
    "condition": "incongruent",
    "expected_response": "H",
    "block": "global_block"
  }
}

Response Data:

{
  "trial_index": 1,
  "stimulus_id": "navon_0_1",
  "source": "keyboard",
  "raw_key": "h",
  "global_letter": "H",
  "local_letter": "S",
  "target_level": "global",
  "condition": "incongruent",
  "expected_response": "H",
  "response_value": "H",
  "response_correct": true,
  "latency_ms": 623,
  "ts": "2024-01-01T00:00:01.623Z",
  "hr": 1857.56
}

Summary Artifact​

A JSON file (navon_summary_<taskIndex>.json) with comprehensive statistics:

{
  "task_kind": "navon",
  "task_index": 0,
  "total_trials": 60,
  "overall": {
    "total": 60,
    "valid_responses": 58,
    "correct": 53,
    "accuracy": 0.91,
    "mean_rt_ms": 512,
    "mean_correct_rt_ms": 498,
    "timeouts": 2
  },
  "breakdowns": {
    "target_level": {
      "global": { "total": 30, "valid_responses": 29, "correct": 27, "accuracy": 0.93, "mean_rt_ms": 480, "mean_correct_rt_ms": 478, "timeouts": 1 },
      "local": { "total": 30, "valid_responses": 29, "correct": 26, "accuracy": 0.90, "mean_rt_ms": 544, "mean_correct_rt_ms": 542, "timeouts": 1 }
    },
    "condition": {
      "congruent": { "total": 24, "valid_responses": 24, "correct": 23, "accuracy": 0.96, "mean_rt_ms": 466, "mean_correct_rt_ms": 465, "timeouts": 0 },
      "incongruent": { "total": 24, "valid_responses": 23, "correct": 20, "accuracy": 0.87, "mean_rt_ms": 546, "mean_correct_rt_ms": 545, "timeouts": 1 },
      "neutral": { "total": 12, "valid_responses": 11, "correct": 10, "accuracy": 0.91, "mean_rt_ms": 511, "mean_correct_rt_ms": 510, "timeouts": 1 }
    },
    "target_and_condition": {
      "global_congruent": { "total": 12, "valid_responses": 12, "correct": 12, "accuracy": 1.0, "mean_rt_ms": 450, "mean_correct_rt_ms": 450, "timeouts": 0 },
      "global_incongruent": { "total": 12, "valid_responses": 12, "correct": 11, "accuracy": 0.92, "mean_rt_ms": 506, "mean_correct_rt_ms": 506, "timeouts": 0 },
      "global_neutral": { "total": 6, "valid_responses": 5, "correct": 5, "accuracy": 1.0, "mean_rt_ms": 489, "mean_correct_rt_ms": 489, "timeouts": 1 },
      "local_congruent": { "total": 12, "valid_responses": 12, "correct": 11, "accuracy": 0.92, "mean_rt_ms": 480, "mean_correct_rt_ms": 480, "timeouts": 0 },
      "local_incongruent": { "total": 12, "valid_responses": 11, "correct": 9, "accuracy": 0.82, "mean_rt_ms": 584, "mean_correct_rt_ms": 584, "timeouts": 1 },
      "local_neutral": { "total": 6, "valid_responses": 6, "correct": 5, "accuracy": 0.83, "mean_rt_ms": 531, "mean_correct_rt_ms": 531, "timeouts": 0 }
    }
  },
  "effects": {
    "global_precedence_ms": -64,
    "global_interference_ms": 54,
    "local_interference_ms": 66
  },
  "trials": [ ]
}

Each per-cell stats object contains total, valid_responses, correct, accuracy, mean_rt_ms, mean_correct_rt_ms, and timeouts. RT means are null when there are no qualifying responses. When practice is enabled and practice trials ran, an additional top-level practice block is appended with the same overall / breakdowns shape (target level and condition only) plus its own trials array.

Key metrics (in effects):

  • global_precedence_ms: Global mean correct RT minus local mean correct RT (negative value means global is faster)
  • global_interference_ms: Incongruent minus congruent mean correct RT for global-target trials
  • local_interference_ms: Incongruent minus congruent mean correct RT for local-target trials

The effects block uses only congruent and incongruent trials. Neutral trials appear in breakdowns.condition.neutral and the breakdowns.target_and_condition cross-tabs. The neutral buckets are always present in the JSON even when no neutral trials were run (they then contain zero counts and null RT values).

Instructions​

The task uses a four-tier instruction system:

  1. Main Instructions: Shown on a dedicated page before the task begins
  2. Practice Instructions: Shown before practice trials (if practice enabled)
  3. Trials Instructions: Shown before main trials after practice (if practice enabled)
  4. Hint Instructions: Quick-reference help available via "?" button during task

All instruction text can be customized in rich-text format during study configuration.

Design Recommendations​

Trial Design​

Balanced design (recommended):

  • Equal numbers of global and local trials
  • Equal numbers of congruent and incongruent trials within each target level
  • Randomize order to prevent anticipation
  • Counterbalance target level order (global-first vs. local-first) across participants

Minimum for reliable effects:

  • 20 trials per target level (40 total)
  • 10 congruent + 10 incongruent per target level

Timing Guidelines​

ParameterStandardBriefExtended
Fixation (ms)500ms300ms800ms
Stimulus (ms)1500-2000ms150-250ms3000ms+

Note: Brief presentations (150-250ms) enhance global precedence effects. Longer presentations allow more time but may reduce processing biases.

Practice Trials​

Include 8-16 practice trials:

  • Mix global and local target levels
  • Mix congruent and incongruent conditions
  • Provide feedback so participants learn the task
  • Consider starting with congruent trials, then introducing incongruent

Letter Selection​

Recommended letter set:

  • Use letters with distinct shapes: H, S, E, L, T
  • Avoid confusable letters (e.g., O and Q)
  • Letters should form clear global shapes when repeated

Enhancing Global Precedence​

  • Brief stimulus duration (150-250ms)
  • Use monospace font with adequate letter spacing
  • Larger size ratio (global letter 10x larger than local)

Enhancing Local Processing​

  • Long stimulus duration (500ms+)
  • Reduce global-local size ratio
  • Use clear, high-contrast local letters

Common Issues and Solutions​

No Global Precedence Observed​

Problem: Local and global RTs are similar

Possible causes:

  • Stimulus duration too long
  • Size ratio not large enough
  • Participants treating task as unspeeded

Solutions:

  • Reduce stimulus_ms to 200-300ms
  • Increase global font size or decrease local font size
  • Emphasize speed in instructions ("Respond as quickly as possible")

High Error Rate (>15%)​

Problem: Accuracy below 85%

Possible causes:

  • Stimulus duration too short
  • Participants confusing target level instructions
  • Local letters not clearly visible

Solutions:

  • Increase stimulus_ms to 2500-3000ms
  • Provide clearer instructions with visual examples
  • Increase local font size
  • Extend practice with feedback

Keyboard Responses Not Working​

Problem: Letter keys don't register

Solutions:

  • Ensure task window has focus (click on task area)
  • Check browser compatibility (works best in Chrome/Edge/Firefox)
  • Use button-based mode as alternative

Population-Specific Adaptations​

Children (8+ years)​

  • Larger stimuli (global: 150px, local: 20px)
  • Longer exposure (2500-3000ms)
  • Simpler letter set (3-4 letters total)
  • Extended practice until child demonstrates understanding
  • Button-based mode may be easier than keyboard

Older Adults (65+)​

  • Larger stimuli and longer exposure
  • High contrast (black on white)
  • Generous response window (3000ms+)
  • Clear block instructions with examples
  • Self-paced mode (button-based)

Clinical Populations​

  • Autism: May show reduced or reversed global advantage (local bias)
  • Schizophrenia: May show disrupted global processing
  • Adapt timing and size to individual capabilities
  • Focus on accuracy over speed for some populations

References​

  • Navon, D. (1977). Forest before trees: The precedence of global features in visual perception. Cognitive Psychology, 9(3), 353-383.
  • Lamb, M. R., Robertson, L. C., & Knight, R. T. (1989). Attention and interference in the processing of global and local information: Effects of unilateral temporal-parietal junction lesions. Neuropsychologia, 27(4), 471-483.
  • Dale, G., & Arnell, K. M. (2013). Investigating the stability of and relationships among global/local processing measures. Attention, Perception, & Psychophysics, 75(3), 394-406.
  • Happé, F., & Frith, U. (2006). The weak coherence account: Detail-focused cognitive style in autism spectrum disorders. Journal of Autism and Developmental Disorders, 36(1), 5-25.

See Also​