Attentional Blink Task

Version: v1 (current)

A rapid serial visual presentation (RSVP) paradigm measuring the temporal limits of visual attention.

Overview

The Attentional Blink task measures a fundamental limitation in human attention: when two targets appear in quick succession within a stream of distractors, the second target is often missed if it appears 200-500ms after the first. This phenomenon reveals the temporal dynamics of conscious perception and attentional selection.

Participants view a rapid stream of letters or digits presented one at a time at the same location. Their task is to identify two targets (e.g., letters in a different color) among distractors (e.g., black letters). The "blink" refers to the period after detecting the first target when attention is unavailable for processing the second target.

This task is widely used in cognitive neuroscience to study attention, working memory, and consciousness, with clinical applications in ADHD, aging, and neurological disorders.

Scientific Background

Classic Findings:

• Temporal Window: The attentional blink is most pronounced when T2 appears 200-500ms after T1 (lag 2-5 at 100ms/item)
• Lag-1 Sparing: T2 is often detected well if it immediately follows T1 (no blink at lag 1)
• Individual Differences: Blink magnitude correlates with working memory capacity and age
• Recovery Time: Attention recovers by ~500-800ms after T1

Key Mechanisms:

• Reflects capacity limits in consolidating information into working memory
• Related to P3 ERP component timing and attentional resource allocation
• Demonstrates the bottleneck in conscious perception

Seminal Papers:

• Raymond, Shapiro, & Arnell (1992): Original demonstration of the attentional blink
• Chun & Potter (1995): Two-stage model of temporal attention
• Olivers & Meeter (2008): A boost and bounce theory of the attentional blink

Why Researchers Use This Task

1. Attention Research: Study temporal dynamics and capacity limits of visual attention
2. Working Memory: Investigate the relationship between attention and memory consolidation
3. Clinical Assessment: Measure attention deficits in ADHD, schizophrenia, and neurological conditions
4. Aging Studies: Track age-related changes in attentional capacity and processing speed
5. Consciousness Research: Explore the neural correlates of conscious perception and awareness

Configuration Options

Timing Parameters

Parameter	Type	Default	Description
Stream duration (ms)	number	200	Duration each item is displayed in the RSVP stream (50-1000ms).
ISI (ms)	number	0	Inter-stimulus interval (blank screen between items, 0-500ms).
Response timeout (ms)	number	5000	How long to wait for participant response (1000-30000ms).

Visual Parameters

Parameter	Type	Default	Description
Font size (px)	number	48	Font size for stream items (8-400px).
T1 color	string (hex)	#FF0000	Color for first target (T1).
T2 color	string (hex)	#00FF00	Color for second target (T2).
Distractor color	string (hex)	#000000	Color for distractor items.
Background color	string (hex)	#FFFFFF	Background color.

Practice Trials

Parameter	Type	Default	Description
Enable practice	boolean	False	Enable practice trials with visual feedback.
Practice trials	array	[]	Array of practice trial configurations (same structure as main trials).

Response Options

Response set size and dynamic option design are your responsibility as the researcher; the app validates that every trial can still be answered unambiguously and records exactly what was shown.

Parameter	Type	Default	Description
Option generation mode	dropdown	Stream-derived	Controls how the T1/T2 option sets are generated for trials that do not author their own options in the sheet (per-trial options always take priority). Stream-derived (default): every unique item in that trial's own stream, reproducing the original behavior. All target letters: a shared master alphabet across every trial. Target + matched foils: the correct answer plus a configurable number of other stream items. Fixed yes/no detection: no stream-derived choices for the T2 step (T1 still falls back to stream-derived, since it has no yes/no concept).
All target letters	text (comma-separated)	(empty)	Used only by the "All target letters" mode: the fixed alphabet of possible targets shown at every trial's T1/T2 response step.
Foil count	number	3	Used only by the "Target + matched foils" mode: how many other stream items accompany the target.
Option layout rows / columns	number	0 / 0	Leave both at 0 for the neutral automatic layout (recommended). Set both to a positive number only when a fixed on-screen grid shape has scientific or usability value; the exact rows/cols are recorded with every trial result.

Every trial in the spreadsheet may also author its own T1 Options / T2 Options columns (comma-separated), which override the generation mode for that trial only. The "T2 not present" choice is always shown on the T2 step and is never part of the authored/generated option set itself.

Instructions

Four instruction types are available:

Instruction Type	Description
Main instructions	Shown before task starts on a dedicated page with camera/screenshare preview.
Hint instructions	Quick reference shown via "?" button during task execution.
Practice instructions	Shown before practice trials begin (if practice enabled).
Trials instructions	Shown after practice, before main trials (if practice enabled).

Trial Configuration

Each trial in the trials spreadsheet is defined by:

Column	Type	Description
Stream Items (comma-separated)	string	Comma-separated list of items in stream (e.g., 'A,B,C,X,D,E,F,Y,G').
T1 Position	number	Position of first target in stream (1-indexed).
T2 Position	number	Position of second target in stream (1-indexed).
Lag	number	Lag value (T2 position - T1 position), stored alongside the trial.
Block	string	Optional block label for grouping trials in analysis.
Fixation (ms)	number	Fixation cross duration for this trial.
Stream Duration (ms)	number	Per-item display duration for this trial.
ISI (ms)	number	Inter-stimulus interval for this trial.
Response Timeout (ms)	number	Response window for this trial.
T1 Options	string (comma-separated, optional)	Explicit T1 response-option set for this trial; blank uses the response-option settings above.
T2 Options	string (comma-separated, optional)	Explicit T2 response-option set for this trial; blank uses the response-option settings above. The "T2 not present" choice is always added on screen and is never part of this column.

The runtime also reads an optional t2_present field per trial (default true) to determine whether T2 is actually shown, but it is not a column in the configuration spreadsheet.

Example Trial Sheet

Stream Items          | T1 Position | T2 Position | Lag | Block | Fixation (ms) | Stream Duration (ms) | ISI (ms) | Response Timeout (ms)
----------------------|-------------|-------------|-----|-------|---------------|----------------------|----------|----------------------
A,B,C,1,2,D,E,F,G,H   | 4           | 5           | 1   | main  | 500           | 200                  | 0        | 5000
X,Y,Z,3,W,4,V,U,T,S   | 3           | 5           | 2   | main  | 500           | 200                  | 0        | 5000
A,B,C,1,D,E,2,F,G,H   | 3           | 6           | 3   | main  | 500           | 200                  | 0        | 5000

Design Tips:

• Stream Length: 15-20 items typical; longer streams increase task difficulty
• Target Positions: Place T1 in middle third of stream to allow time for attention to engage
• Lag Range: Include lag-1 (sparing), lag-2/3 (peak blink), and lag-7+ (recovery)
• Item Duration: 80-120ms is standard; slower (150ms) for clinical/elderly populations

Lag Calculation

Lag = T2 position - T1 position

• Lag 1: T2 immediately follows T1 (typically shows "lag-1 sparing")
• Lag 2-4: Peak attentional blink period
• Lag 7+: Recovery period (T2 detection improves)

Participant Experience

Unmoderated/Moderated Mode

1. Main Instructions: Participant reads task instructions and clicks "Start"
2. Practice Phase (if enabled):
   • Practice instructions appear
   • Practice trials with feedback (✓ or ✗)
   • Feedback shows whether both targets were correctly identified
3. Trials Instructions (if practice was enabled): Brief reminder before main trials
4. Main Trials:
   • Fixation cross appears (if configured)
   • Rapid stream of items appears (e.g., letters/digits at 100ms each)
   • Two items in the stream are colored differently (targets)
   • After stream ends, participant reports both target identities
   • No feedback during main trials
5. Completion: Task ends after all trials

Strictly Moderated Mode

Same flow, but researcher controls advancement via moderator dashboard. Participant cannot advance trials independently.

Response Collection

After the RSVP stream ends, participants:

• See a response screen with prompts for the first and second target letters
• Type or select the two target items they saw (prompts configured via ab_t1_prompt and ab_t2_prompt parameters)
• Click "Submit" to proceed to next trial
• Have a configurable timeout period to respond

Data Output

Markers and Responses

The task records high-resolution data in two separate collections:

Markers (attentional_blink_trial_start):

{
  "type": "attentional_blink_trial_start",
  "ts": "2024-01-01T00:00:01.000Z",
  "hr": 1234.56,
  "data": {
    "trial_index": 1,
    "is_practice": false,
    "stimulus_id": "ab_0_1",
    "stream_length": 20,
    "t1_position": 7,
    "t1_value": "X",
    "t2_position": 10,
    "t2_value": "K",
    "t2_present": true,
    "lag": 3,
    "block": "main"
  }
}

During the practice phase this marker is emitted as attentional_blink_practice_trial_start and carries practice_trial_index instead of trial_index. The runtime also emits stimulus_shown, t1_onset, t2_onset, fixation_onset, fixation_offset, and stimulus_aoi markers.

Markers (stimulus_shown):

{
  "type": "stimulus_shown",
  "ts": "2024-01-01T00:00:01.700Z",
  "hr": 1934.56,
  "data": {
    "trial_index": 1,
    "is_practice": false,
    "stimulus_id": "ab_0_1",
    "stream_position": 7,
    "stimulus_value": "X",
    "is_t1": true,
    "is_t2": false,
    "block": "main"
  }
}

Response Data:

{
  "trial_index": 1,
  "is_practice": false,
  "stimulus_id": "ab_0_1",
  "source": "button",
  "condition": "3",
  "stream_length": 20,
  "t1_position": 7,
  "t2_position": 10,
  "t1_value": "X",
  "t1_response": "X",
  "t1_correct": true,
  "t2_value": "K",
  "t2_response": "Y",
  "t2_correct": false,
  "t2_correct_given_t1": false,
  "responded": true,
  "t2_present": true,
  "lag": 3,
  "t1_options": ["X", "M", "Q", "W"],
  "t2_options": ["K", "Y", "P", "not_present"],
  "option_layout": { "mode": "auto" },
  "latency_ms": 3200,
  "block": "main",
  "ts": "2024-01-01T00:00:04.200Z",
  "hr": 4434.56
}

source is one of button, timeout, or moderator_advance. condition is the lag encoded as a string. t1_options/t2_options record exactly what was shown on each response step (an authored per-trial list, or generated per the configured option-generation mode); option_layout is { "mode": "auto" } unless the researcher configured explicit grid rows/cols.

Response Accuracy Measures

Measure	Description
t1_correct	Whether T1 was correctly identified
t2_correct	Whether T2 was correctly identified (regardless of T1)
t2_correct_given_t1	T2 correct AND T1 correct (key measure for blink magnitude)

Summary Artifact

At task completion, a JSON summary is generated (ab_summary_<taskIndex>.json):

{
  "task_kind": "attentional_blink",
  "task_index": 0,
  "overall": {
    "total": 30,
    "valid_responses": 30,
    "t1_correct": 27,
    "t1_accuracy": 0.9,
    "t2_correct": 20,
    "t2_accuracy": 0.667,
    "t2_correct_given_t1": 18,
    "t2_accuracy_given_t1": 0.667,
    "mean_rt_ms": 3200,
    "timeouts": 0,
    "total_trials": 30
  },
  "breakdowns": {
    "lag": {
      "2": {
        "total": 10,
        "t2_accuracy_given_t1": 0.4,
        "t1_accuracy": 0.9
      },
      "3": {
        "total": 10,
        "t2_accuracy_given_t1": 0.45,
        "t1_accuracy": 0.9
      },
      "7": {
        "total": 10,
        "t2_accuracy_given_t1": 0.85,
        "t1_accuracy": 0.9
      }
    }
  },
  "effects": {
    "attentional_blink_magnitude": 0.45
  },
  "trials": [...]
}

Each lag entry under breakdowns.lag is the full calcStats object (same keys as overall); only a subset is shown above for brevity. When practice trials are run, an additional top-level practice block is emitted with the same overall / breakdowns / trials shape.

Key Metric: effects.attentional_blink_magnitude = T2|T1 accuracy at long lags (7+) - T2|T1 accuracy at short lags (2-4). Positive values indicate a blink effect.

Design Recommendations

General Guidelines

• Item Duration: 80-120ms is standard; slower (150ms) for clinical/elderly populations
• Stream Length: 15-20 items typical; longer streams increase task difficulty
• Target Positions: Place T1 in middle third of stream to allow time for attention to engage
• Lag Range: Include lag-1 (sparing), lag-2/3 (peak blink), and lag-7+ (recovery)

Recommended Trial Counts

Study Type	Trials per Lag	Total Trials	Lag Conditions
Standard Research	10-12	40-60	1, 2, 3, 7
Clinical Assessment	6-8	24-32	2, 4, 7
Detailed Lag Curve	6-8	42-56	1, 2, 3, 4, 5, 6, 7

Population-Specific Adaptations

Children (8-12 years):

• Slower presentation (120-150ms/item)
• Shorter streams (12-15 items)
• Fewer trials (20-24 total)
• Mandatory practice with extensive feedback

Older Adults (65+):

• Slower presentation (120-150ms/item)
• High-contrast colors
• Larger font size (60-72px)
• More practice trials

Clinical Populations:

• Slower timing and fewer trials
• Consider single-target control condition
• Include breaks between trial blocks
• Use digits instead of letters if reading impairment present

Example Configurations

Standard Attentional Blink:

Stream: 20 items, 100ms each
T1 Position: 7
T2 Lags: 1, 3, 5, 7
Targets: Colored letters among black letter distractors
Trials: 10 per lag condition (40 total)

Clinical Assessment:

Stream: 15 items, 120ms each (slower for clinical populations)
T1 Position: 5
T2 Lags: 2, 4, 6
Targets: Colored digits among black digit distractors
Practice: Mandatory with feedback
Trials: 8 per lag (24 total)

Lag-1 Sparing Study:

Stream: 20 items, 100ms each
T1 Position: 8
T2 Lags: 1, 2, 3, 4, 5, 6, 7
Focus on comparing lag-1 (sparing) with lag-2/3 (blink)
Trials: 6 per lag (42 total)

Common Issues and Solutions

Issue: T1 Accuracy Too Low

Symptoms: T1 accuracy below 70%

Solutions:

• Increase item duration to 150-200ms
• Reduce stream length to 12-15 items
• Use high-contrast colors for targets
• Increase font size

Issue: No Attentional Blink Observed

Symptoms: T2|T1 accuracy similar across all lags

Solutions:

• Verify lag timing is correct (should be 200-500ms between targets at 100ms/item)
• Ensure T1 position allows attention to engage (position 5-8 in stream)
• Check that item duration is fast enough (80-120ms)
• Increase distractor-target similarity

Issue: Ceiling/Floor Effects

Symptoms: T2|T1 accuracy >90% or <20% across all lags

Solutions for Ceiling:

• Reduce item duration (faster RSVP)
• Increase stream length
• Make distractors more similar to targets

Solutions for Floor:

• Increase item duration (slower RSVP)
• Reduce stream length
• Use more distinctive target colors
• Add more practice trials

Issue: Participants Missing Instructions

Symptoms: Confusion about reporting both targets

Solutions:

• Emphasize both targets must be reported in instructions
• Practice until participants understand the dual-target requirement
• Provide examples showing both target positions

Issue: Response Timeout Issues

Symptoms: High rate of incomplete responses

Solutions:

• Increase response timeout window to 8000-10000ms
• Simplify response interface
• Remind participants they need time to recall both targets

References

• Raymond, J. E., Shapiro, K. L., & Arnell, K. M. (1992). Temporary suppression of visual processing in an RSVP task: An attentional blink? Journal of Experimental Psychology: Human Perception and Performance, 18(3), 849-860.

• Chun, M. M., & Potter, M. C. (1995). A two-stage model for multiple target detection in rapid serial visual presentation. Journal of Experimental Psychology: Human Perception and Performance, 21(1), 109-127.

• Olivers, C. N., & Meeter, M. (2008). A boost and bounce theory of temporal attention. Psychological Review, 115(4), 836-863.

• Dux, P. E., & Marois, R. (2009). The attentional blink: A review of data and theory. Attention, Perception, & Psychophysics, 71(8), 1683-1700.

---

Related Tasks: Visual Short-Term Memory, Change Blindness, Backward Masking