Rapid Serial Visual Presentation (RSVP)

Version: v1 (current)

A temporal attention paradigm presenting stimuli in rapid succession at a single location to study attentional processing over time.

Overview

Rapid Serial Visual Presentation (RSVP) is a fundamental method for studying temporal attention and visual processing. Items (letters, digits, words, images) appear one after another at the same screen location at rates typically ranging from 6-20 items per second. Participants must detect, identify, or remember target items embedded in this stream.

RSVP reveals temporal limits of attention: when two targets appear close together in time, the second target is often missed. This phenomenon, called the Attentional Blink (AB), demonstrates that attention operates like a "blink": after capturing one target, there is a brief refractory period (~200-500ms) during which subsequent targets are missed, even though they're clearly visible.

RSVP is used across cognitive psychology, neuroscience, and applied vision research to study attention, working memory, consciousness, and visual information processing under time pressure.

Scientific Background

Classic Findings:

Attentional Blink: Second target (T2) missed when presented 200-500ms after first target (T1)
Lag-1 Sparing: T2 actually detected better when immediately following T1 (lag 1) than at lag 2-3
Individual Differences: AB magnitude varies substantially across individuals; correlates with working memory capacity
No Blink for Same-Category: When T1 and T2 are same type (e.g., both letters), AB reduced
Temporal Resolution: Healthy adults can process ~10 items/second but attention "locks" on targets

Key Mechanisms:

Attentional Gating: Targets trigger attentional capture, temporarily blocking subsequent items
Working Memory Consolidation: T1 must be consolidated before resources free for T2
Interference: Distractors following targets interfere with encoding

Seminal Papers:

Raymond, Shapiro, & Arnell (1992): Temporary suppression of visual processing in an RSVP task: An attentional blink?
Chun & Potter (1995): A two-stage model for multiple target detection in RSVP
Dux & Marois (2009): The attentional blink: A review of data and theory

Why Researchers Use This Task

Temporal Attention: Study how attention operates over time; measure temporal resolution limits
Attentional Blink Research: Classic paradigm for AB; examine factors affecting magnitude
Individual Differences: AB magnitude predicts working memory capacity, fluid intelligence
Consciousness Studies: Explore boundary between conscious perception and processing
Clinical Assessment: Evaluate temporal attention deficits in ADHD, dyslexia, aging
Training Studies: Test whether attention training can reduce AB magnitude

Current Implementation Status

Fully Implemented:

✅ Per-trial presentation timing (stimulus duration and ISI defined per stream)
✅ Single highlighted target per stream, with forced-choice response options drawn from the stream
✅ Fully custom streams (any comma-separated text items: letters, digits, words)
✅ Practice trials with correct/incorrect feedback
✅ Per-trial accuracy and reaction-time summary, with optional per-condition breakdown
✅ Per-item stimulus_aoi bounds for gaze-to-AOI joining

Not Implemented:

❌ Dual-target (T1/T2) detection and lag manipulation
❌ Attentional-blink magnitude / T2|T1 metrics
❌ Adaptive rate adjustment based on performance
❌ Auditory RSVP variant; image stimuli in the stream

Configuration Parameters

The task is configured through a small set of display options plus two trial spreadsheets (main and practice). Presentation timing, stream contents, and the target are defined per trial row in the spreadsheet, not through global rate/lag parameters.

Display Options

Parameter	Type	Default	Range	Description
`rsvp_font_size_px`	number	48	8–400	Font size of the stream stimulus (px)
`rsvp_response_window_ms`	number	3000	100–10000	Default response window after the stream (ms), used when a trial row does not set its own
`rsvp_target_color`	string	`#FF0000`	—	Color used to highlight the target item
`rsvp_text_color`	string	`#000000`	—	Color of non-target stream items
`rsvp_background_color`	string	`#FFFFFF`	—	Stimulus background color
`rsvp_practice_enabled`	boolean	true	—	Whether the practice phase runs before the main trials

Instructions

Rich-text (HTML) instruction fields: main_instructions, hint_instructions, practice_instructions, and trials_instructions. The practice and trials instruction editors are only shown when practice is enabled.

Trial Spreadsheet

Each trial (main and practice tables share the same columns) is one row:

Column	Type	Description
Stream (comma-separated)	text	The full stream of items, comma-separated (e.g. `A,B,C,X,D,E,F`)
Target Position	numeric	1-based index of the target within the stream
Target Value	text	The target item; must equal `stream[Target Position − 1]`
Distractor Type	text	Free-text label for the distractor category (e.g. `letters`, `numbers`)
Block	text	Free-text block label
Condition	text	Free-text condition label (used for the summary breakdown)
Stimulus Duration (ms)	numeric	How long each item is shown (default 100, min 16, max 5000)
ISI Duration (ms)	numeric	Blank inter-stimulus interval between items (default 50, min 0, max 2000)
Response Window (ms)	numeric	Per-trial response window after the stream (default 3000, min 100); overrides the display-option default

Defining timing per row lets researchers vary presentation across trials, enabling designs such as mixed presentation rates within a session, speed-accuracy tradeoff manipulation (varying response windows), or gradual difficulty progression.

Note: The runtime checks that stream[Target Position − 1] equals Target Value; a mismatch is logged as a telemetry warning because it would cause the highlighted item to differ from the scored target.

Data Output

Note: This implementation presents a single highlighted target per stream; it does not implement the dual-target (T1/T2) or lag-based design described in the scientific background. Markers and response fields are per-trial and per-stream-item, not per-T1/T2.

Markers

Marker names for phase/trial boundaries are prefixed with the task slug (e.g. rsvp_trial_start, rsvp_practice_trial_start, rsvp_trial_end). Per-item and other event markers (stimulus_shown, target_onset, stimulus_aoi, feedback_shown, response_recorded) are emitted without the prefix.

Trial start (rsvp_trial_start):

{
  "type": "rsvp_trial_start",
  "ts": "2026-02-14T10:30:00.023Z",
  "hr": 1234.46,
  "data": {
    "trial_index": 1,
    "is_practice": false,
    "stimulus_id": "rsvp_0_1",
    "stream_length": 7,
    "target_position": 4,
    "target_value": "X",
    "condition": "standard",
    "block": "main",
    "distractor_type": "letters"
  }
}

Per-item stimulus (stimulus_shown):

{
  "type": "stimulus_shown",
  "ts": "2026-02-14T10:30:00.123Z",
  "hr": 1234.56,
  "data": {
    "trial_index": 1,
    "is_practice": false,
    "stimulus_id": "rsvp_0_1",
    "stream_position": 4,
    "stimulus_value": "X",
    "is_target": true,
    "condition": "standard",
    "block": "main"
  }
}

A dedicated target_onset marker is also emitted when the target item appears, and a stimulus_aoi marker records the stimulus element bounds once per trial (for gaze-to-AOI joining).

Response Data

Each response is recorded via the central recordResponse (trial grain), carrying these fields:

{
  "trial_index": 1,
  "is_practice": false,
  "stimulus_id": "rsvp_0_1",
  "source": "button",
  "target_value": "X",
  "response_value": "X",
  "response_correct": true,
  "responded": true,
  "response_options": ["X", "B", "D", "A"],
  "condition": "standard",
  "block": "main",
  "latency_ms": 850
}

source is one of button, timeout, or moderator_advance.
latency_ms is measured from target onset and is omitted when no target onset was captured.
response_correct is null when the trial did not count as a response.

Summary Artifact

On completion (non-preview), a JSON artifact named rsvp_summary_<taskIndex>.json is emitted:

{
  "task_kind": "rsvp",
  "task_index": 0,
  "total_trials": 10,
  "overall": {
    "total": 10,
    "valid_responses": 9,
    "correct": 7,
    "accuracy": 0.78,
    "mean_rt_ms": 850,
    "timeouts": 1
  },
  "breakdowns": {
    "condition": {
      "standard": { "total": 10, "valid_responses": 9, "correct": 7, "accuracy": 0.78, "mean_rt_ms": 850, "timeouts": 1 }
    }
  },
  "trials": [...],
  "practice": {
    "total_trials": 3,
    "overall": { "total": 3, "valid_responses": 3, "correct": 2, "accuracy": 0.67, "mean_rt_ms": 700, "timeouts": 0 },
    "breakdowns": {},
    "trials": [...]
  }
}

The practice block is only present when practice trials were run. Per-condition stats appear under breakdowns.condition only when trials carry a non-empty condition. There is no by_lag, attentional_blink, or T1/T2 field — the implementation does not compute attentional-blink metrics.

Example Research Configurations

Standard Attentional Blink Paradigm

Rate: 10 Hz (100ms per item)
Stream Length: 15 items
Targets: 2 letters (T1 and T2) among digit distractors
Lags: 2, 3, 7 (between T1 and T2)
Trials: 10 per lag (30 total)
Analysis: T2|T1 accuracy by lag; AB = accuracy at lag 7 - accuracy at lag 2

Single-Target Detection (Baseline)

Rate: 10 Hz
Targets: 1 letter per stream
Purpose: Establish baseline detection without dual-task demands
Comparison: T1 alone vs. T1 accuracy when T2 also present

Fast vs. Slow Presentation Rate

Conditions: 6 Hz (slow) vs. 15 Hz (fast)
Purpose: Test whether AB is due to temporal limits or fixed refractory period
Prediction: AB duration measured in items (lag), not time

Training Study (Pre-Post)

Pre: Standard AB task (30 trials)
Training: 10 sessions of RSVP practice with feedback
Post: Same AB task
Analysis: Reduction in AB magnitude after training

Participant Experience

Instructions: "Letters will flash one at a time. Spot the one letter shown in a different color. That is the target. Select it from the options afterward."
Practice Streams (if enabled):
- Watch the rapid sequence of items (each shown for its configured stimulus duration, default 100ms)
- One item appears in a different color. That is the target.
- Select the target from the response options (a checkmark or X is shown as feedback)
Main Task:
- For each trial:
  - Rapid sequence of items (very fast!)
  - Sequence ends
  - Prompt ("Which item was the target?"): select the target from the options
- No feedback during main trials
Completion: the task ends and emits its summary artifact; there is no on-screen score message.

Design Recommendations

General Guidelines

Presentation Rate: 10 Hz (100ms/item) is standard for AB research
Faster Rates: 15-20 Hz increase difficulty; used for training or high-performers
Slower Rates: 6-8 Hz reduce AB magnitude; easier for children or clinical groups
Stream Length: 12-18 items typical; longer streams increase fatigue
Target Distinctiveness: Letters among digits (or vice versa) ensures pop-out

Lag Selection for AB Studies

Lag 1: Often shows "sparing" (good T2 accuracy)
Lags 2-3: Peak AB deficit (lowest T2 accuracy)
Lags 7-8: AB recovery (baseline T2 accuracy)
Include Baseline: Lag 7+ for comparison to estimate AB magnitude

Distractor Design

Use category different from targets (digits if targets are letters)
Ensure distractors don't contain target-similar items
Randomize distractor order to prevent learning

Response Format

Free Report: Type the target (requires memory)
Forced Choice: Select from options (reduces memory load)
Order Matters: Report T1 first, then T2 (standard)

Population-Specific Adaptations

Children (8+ years):

Slower rate: 6-8 Hz
Shorter streams: 10-12 items
Larger font size
Practice until comfortable (may need 10+ practice trials)
Expect larger AB magnitude

Older Adults (65+):

Slightly slower rate: 8 Hz
High-contrast stimuli
Larger font
Allow extra time for responses
Generally similar AB to young adults (sometimes slightly larger)

ADHD:

Expect larger AB magnitude
Consider single-target baseline to rule out general detection deficit
May show reduced AB with medication

Dyslexia:

Some studies show larger AB
Consider non-letter stimuli to avoid reading-related factors

Common Issues and Solutions

Issue	Solution
Floor effects (T2 always missed)	Slow presentation rate to 8 Hz; reduce stream length; increase lag
Ceiling effects (no AB)	Increase rate to 12-15 Hz; reduce distinctiveness of targets
Participants miss T1 (>20% error)	Slow rate; practice more; ensure targets are perceptually distinct
No lag effect	Check lag timing calculation; ensure lags span AB window (1-5)
Participants lose fixation	Emphasize central fixation; make fixation point more salient
Anticipation errors	Randomize T1 position across trials

Data Analysis

Primary Measures:

T1 Accuracy: Overall detection of first target (should be high, ~85-95%)
T2 Accuracy: Overall detection of second target (lower due to AB)
T2|T1 Accuracy: T2 accuracy conditional on T1 correct (purest AB measure)
AB Magnitude: T2|T1 at baseline lag minus T2|T1 at peak deficit lag

Statistical Analysis:

Repeated-measures ANOVA with lag as within-subjects factor
Planned comparisons: lag 2 vs. lag 7; lag 3 vs. lag 7
Individual differences: Correlate AB magnitude with WM capacity, fluid intelligence

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.
Dux, P. E., & Marois, R. (2009). The attentional blink: A review of data and theory. Attention, Perception, & Psychophysics, 71(8), 1683-1700.
Martens, S., & Wyble, B. (2010). The attentional blink: Past, present, and future of a blind spot in perceptual awareness. Neuroscience & Biobehavioral Reviews, 34(6), 947-957.
MacLean, M. H., & Arnell, K. M. (2012). A conceptual and methodological framework for measuring and modulating the attentional blink. Attention, Perception, & Psychophysics, 74(6), 1080-1097.

Overview​

Scientific Background​

Why Researchers Use This Task​

Current Implementation Status​

Configuration Parameters​

Display Options​

Instructions​

Trial Spreadsheet​

Data Output​

Markers​

Response Data​

Summary Artifact​

Example Research Configurations​

Standard Attentional Blink Paradigm​

Single-Target Detection (Baseline)​

Fast vs. Slow Presentation Rate​

Training Study (Pre-Post)​

Participant Experience​

Design Recommendations​

General Guidelines​

Lag Selection for AB Studies​

Distractor Design​

Response Format​

Population-Specific Adaptations​

Common Issues and Solutions​

Data Analysis​

References​

See Also​