Stroop Task

Version: v1 (current)

Overview

The Stroop Task is one of the most widely used paradigms in cognitive psychology, measuring selective attention and cognitive control. Participants see color words (like "RED" or "BLUE") displayed in colored ink and must name the ink color while ignoring the word itself. When the word and ink color mismatch (e.g., the word "RED" in blue ink), response times slow and errors increase. This interference is known as the Stroop effect.

This task measures executive functions including response inhibition, selective attention, and processing speed. The interference from automatic word reading provides a window into how the brain resolves conflicting information streams.

Scientific Background

The Stroop effect, first documented by John Ridley Stroop in 1935, demonstrates that reading words is such an automatic process that it interferes with the more deliberate task of color naming. Key findings:

• Stroop, J.R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18(6), 643-662.
• MacLeod, C.M. (1991). Half a century of research on the Stroop effect: An integrative review. Psychological Bulletin, 109(2), 163-203.

The task typically includes three trial types:

• Congruent: Word and color match (e.g., "RED" in red ink). This condition produces the fastest responses.
• Incongruent: Word and color mismatch (e.g., "RED" in blue ink). This condition produces the slowest responses and the most errors.
• Neutral: Non-color word or symbol in color (e.g., "XXX" in red ink). This condition serves as the baseline.

The Stroop effect = mean RT(incongruent) − mean RT(congruent)

Why Researchers Use This Task

• Executive function assessment: Measures cognitive control and inhibition
• Clinical applications: Sensitive to ADHD, depression, anxiety, aging, and neurological conditions
• Individual differences: Stroop interference varies across populations and correlates with other executive measures
• Attention research: Probes automatic vs. controlled processing
• Bilingualism studies: Tests language control in multilingual individuals
• Neuroimaging: Robust activation in prefrontal and anterior cingulate cortex

Task Parameters

Response Mode

The Stroop task supports three response modes, which you select in the Options section of the task configuration panel:

Mode	Description	Mic acquired	Transcript recorded	Best for
Buttons	Participant clicks a button to advance each trial. No microphone used.	No	No	Lab studies, keyboard-only protocols
Verbal	Microphone level is displayed as a meter during each stimulus. No live speech-to-text. Responses are captured by the session recording and scored post-hoc.	Yes	No (null)	Studies with post-hoc Whisper transcription
Verbal with voice detection for auto-advance	Voice-onset detection automatically advances the trial as soon as the participant begins speaking. Live speech-to-text provides an in-session transcript. Requires Audio Calibration to be enabled on the study.	Yes	Yes	Studies that need in-session scoring

Verbal mode advance trigger (shown only when Verbal mode is selected):

Option	Behavior
Timeout only	Trial advances automatically after the stimulus duration
Button press only	Participant clicks a button to advance
Either	Trial advances on whichever comes first

Verbal with voice detection for auto-advance speech tolerance (shown only when Verbal with voice detection for auto-advance mode is selected):

• Minimum confidence score (0–1) for a speech recognition result to be accepted as a matched response.
• Higher values require more confident transcriptions. Default is 0.7.

Visual Configuration

Parameter	Type	Default	Description
Font size (px)	number	96	Font size for color words (8–400)

Speech Recognition

Parameter	Type	Default	Description
Speech recognition language	string	en-US	Language code for live STT (used in Verbal with voice detection for auto-advance mode only)

Practice Trials

Parameter	Type	Default	Description
Enable practice trials	boolean	true	Show a practice phase with visual feedback before main trials

Trial Configuration

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

Column	Description	Example
text	The word displayed	Red, Blue, Green, Yellow
color_hex	Ink color as hex code	#FF0000 (red), #0000FF (blue)
condition	Trial type	congruent, incongruent, neutral
expected_answer	Correct spoken response	Red, Blue (used for STT matching and accuracy scoring)
fixation_ms	Fixation cross duration	500
stimulus_ms	Word display duration	2000
block	Optional grouping label	Practice, Main, Block 1

Example Trials

| text   | color_hex | condition    | expected_answer | fixation_ms | stimulus_ms | block  |
|--------|-----------|--------------|-----------------|-------------|-------------|--------|
| RED    | #FF0000   | congruent    | Red             | 500         | 2000        | main   |
| RED    | #0000FF   | incongruent  | Blue            | 500         | 2000        | main   |
| BLUE   | #0000FF   | congruent    | Blue            | 500         | 2000        | main   |
| BLUE   | #FF0000   | incongruent  | Red             | 500         | 2000        | main   |
| XXX    | #00FF00   | neutral      | Green           | 500         | 2000        | main   |

Recommended Color Set

Use high-contrast, easily distinguishable colors:

• Red: #FF0000 (expected_answer: "Red")
• Blue: #0000FF (expected_answer: "Blue")
• Green: #00FF00 (expected_answer: "Green")
• Yellow: #FFFF00 (expected_answer: "Yellow")

Note: Yellow on white backgrounds can be hard to see. Consider #FFD700 (gold) as an alternative.

Participant Experience

Trial Sequence

1. Main Instructions: Participant reads task instructions
2. (Optional) Practice Instructions: If practice enabled, shown before practice trials
3. (Optional) Practice Trials: Each trial shows visual feedback (green check / red X) after response
4. (Optional) Trials Instructions: Shown after practice, before main trials
5. Main Trials: Each trial follows this sequence:
   • Fixation cross (+) appears (if fixation_ms > 0)
   • Color word appears in colored ink
   • Participant responds (speaks aloud, presses a button, or both depending on mode)
   • Trial advances when the response criterion is met

Microphone Behavior by Mode

In Verbal and Verbal with voice detection for auto-advance modes, the task displays an audio-level meter (five bars) while the stimulus is visible. Participants should:

• Speak clearly at normal volume
• Use the color names exactly as specified in the expected_answer column
• Allow microphone access when prompted

In Buttons mode, no microphone is used.

Data Output

Markers

Marker	Data Fields	Description
stroop_trial_start	trial_index, stimulus_id, word, color, condition, expected_answer, block, language, fixation_ms, stimulus_ms, is_practice	Logged when each trial begins (marker becomes stroop_practice_trial_start during practice)
stimulus_shown	trial_index, stimulus_id, word, color, condition, expected_answer, block, is_practice	Logged when the color word appears
speech_start	trial_index, stimulus_id	Logged when speech recognition starts listening (Verbal with voice detection for auto-advance mode only)
speech_error	error	Logged when a speech recognition error occurs
vad_trial_advance	trial_index, threshold_db	Logged when voice-onset detection fires (Verbal with voice detection for auto-advance mode only)
response_recorded	trial_index	Logged when the trial response is finalized

Response Data

One response object is recorded per trial, containing all fields needed for analysis:

Field	Type	Description
trial_index	number	Trial number (1-based)
stimulus_id	string	Unique stimulus identifier (e.g., stroop_0_1)
source	string	How the trial ended: 'button', 'verbal_no_advance', 'vad_stt', 'timeout', or 'moderator_advance'
word	string	The word displayed
color	string	Hex color code of the text
condition	string	'congruent', 'incongruent', or 'neutral'
expected_answer	string	The correct response (color name)
responded	boolean	Whether the participant produced any response on this trial (true for any concrete response; false for timeout with no response)
is_practice	boolean	Whether this was a practice trial
block	string	Trial block identifier
latency_ms	number	Time from stimulus onset to response (ms)
transcript	string or null	Speech recognition transcript (Verbal with voice detection for auto-advance mode); null in Buttons and Verbal modes
confidence	number or null	Speech recognition confidence score (Verbal with voice detection for auto-advance mode); null otherwise
transcribed_answer	string or null	Matched color name extracted from transcript; null if no match or non-STT source

No response_correct field: The Stroop task does not auto-score correctness. The task records what the participant said (response_value in vad_stt mode) and whether they responded at all (responded). Researchers determine accuracy in post-processing by comparing the transcript or video to the expected answer.

source values by mode:

• buttons mode: typically 'button' or 'timeout'
• verbal mode: typically 'verbal_no_advance', 'timeout', or 'button' (if advance trigger includes press)
• vad_stt mode: typically 'vad_stt' or 'timeout'
• Any mode: 'moderator_advance' when a researcher advances the trial in strictly moderated sessions

Inferred fields in Verbal with voice detection for auto-advance mode

When a trial in Verbal with voice detection for auto-advance mode produces no live transcript, the export may add best-effort *_inferred fields (such as transcript_inferred and answer_inferred) by aligning the trial window to the session's Whisper transcription. These fields are clearly labeled in the export as unverified inferences, not the participant's recorded response. Review them, or exclude them, when transcript accuracy matters for your analysis.

Summary Artifact

A JSON summary file with aggregated RT statistics by condition and the Stroop effect. Because Stroop does not auto-score correctness, there are no accuracy fields:

{
  "task_kind": "stroop",
  "task_index": 0,
  "overall": {
    "total_trials": 40,
    "responded_trials": 38,
    "mean_rt_ms": 758
  },
  "breakdowns": {
    "condition": {
      "congruent":   { "n": 20, "responded": 19, "mean_rt_ms": 652 },
      "incongruent": { "n": 20, "responded": 19, "mean_rt_ms": 823 },
      "neutral":     { "n": 20, "responded": 19, "mean_rt_ms": 698 }
    }
  },
  "effects": {
    "stroop_effect_ms": 171
  },
  "extras": {
    "auto_match_hint_count": 5
  },
  "trials": [ "/* per-trial data */" ]
}

Key metrics:

• stroop_effect_ms: Difference between incongruent and congruent mean responded RTs (larger = more Stroop interference). Computed over all responded trials, not restricted to "correct" trials.
• auto_match_hint_count: Number of trials where the in-session voice recognition fired a match hint. This is a non-authoritative indicator only; it does not mean the answer was correct.
• No accuracy or mean_correct_rt_ms fields: researchers score correctness from video, audio, or post-hoc transcripts in their own analysis pipeline.

Design Recommendations

Choosing a Response Mode

Study goal	Recommended mode
In-session accuracy scoring and RT measurement	Verbal with voice detection for auto-advance
Studies relying on post-hoc Whisper/human scoring	Verbal
Non-verbal or multilingual protocols, or when STT is unreliable	Buttons

Set Audio Calibration to Custom on the study and add at least one sentence when using Verbal with voice detection for auto-advance mode. The calibration step measures each participant's voice level and sets an appropriate voice-onset detection threshold. This is enforced automatically: selecting that Stroop mode locks the study-level Audio calibration setting on Custom.

Trial Design

Balanced design (recommended for most studies):

• 20 congruent trials
• 20 incongruent trials
• 10 neutral trials (optional, for baseline)
• Randomize trial order
• Use at least 4 different color words

Minimum for reliable Stroop effect:

• 12 congruent trials
• 12 incongruent trials

Timing Guidelines

Parameter	Recommended range	Notes
Fixation (ms)	300–800 ms	500 ms is standard
Stimulus (ms)	1500–3000 ms	Older adults may need longer

Practice Trials

Include 6–12 practice trials:

• Mix of congruent and incongruent
• Feedback helps participants understand the task
• Use all color words that appear in main trials

Color Selection

For standard Stroop:

• Use primary colors: Red, Blue, Green, Yellow
• Ensure high contrast against white background
• Avoid colors that are linguistically ambiguous

For specialized populations:

• Consider color-blind friendly palettes (avoid red-green only)
• Test color visibility on the actual display devices

Common Issues and Solutions

Speech Recognition Not Working

Problem: Microphone icon does not appear or speech is not recognized.

Solutions:

• Check that the browser supports speech recognition (use Chrome or Edge)
• Ensure microphone permissions are granted
• Consider switching to Verbal or Buttons mode

High Error Rates

Problem: Accuracy below 80%.

Possible causes:

• Speech recognition language mismatch
• Ambiguous expected_answer values
• Participants misunderstanding instructions
• Stimulus duration too short

Solutions:

• Verify stroop_language matches the participant's language
• Check expected_answer matches how participants naturally say colors
• Extend practice trials with clearer feedback
• Increase stimulus_ms for slower populations

No Stroop Effect

Problem: Incongruent RT not longer than congruent RT.

Possible causes:

• Too few trials (high variability)
• Stimulus timing is too brief or too long

Solutions:

• Increase trials per condition (minimum 12 each)
• Emphasize speed in instructions
• Use standard timing (500 ms fixation, 2000 ms stimulus)

Example Study Configurations

Standard Adult Stroop (Verbal with voice detection for auto-advance)

• 40 trials total (20 congruent, 20 incongruent)
• 4 colors (Red, Blue, Green, Yellow)
• 500 ms fixation, 2000 ms stimulus
• Verbal with voice detection for auto-advance mode (study-level audio calibration set to Custom with at least one sentence)
• Practice: 8 trials with feedback

Post-Hoc Scoring Version (Verbal)

• 40 trials total
• Verbal mode: live transcript is skipped; session recording used for post-hoc Whisper scoring
• 500 ms fixation, 2000 ms stimulus, timeout advance trigger

Clinical or Child Version (Buttons)

• Large font size (120 px)
• Longer stimulus duration (3000 ms)
• Buttons mode: no microphone required
• Extensive practice with feedback
• Fewer trials (24 total) to reduce fatigue

References

• Stroop, J.R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18(6), 643-662.
• MacLeod, C.M. (1991). Half a century of research on the Stroop effect: An integrative review. Psychological Bulletin, 109(2), 163-203.
• MacLeod, C.M., & MacDonald, P.A. (2000). Interdimensional interference in the Stroop effect: Uncovering the cognitive and neural anatomy of attention. Trends in Cognitive Sciences, 4(10), 383-391.