Probabilistic Reward Task

Field	Value
Name	Probabilistic Reward Task
Version	html (v0.1.0-dev)
URL / Repository	https://github.com/TaskBeacon/H000039-probabilistic-reward-task
Short Description	Browser companion for the Chinese probabilistic reward task with short-vs-long mouth discrimination and deferred reward carryover
Created By	TaskBeacon psyflow-web py2js port
Modality	Behavior
Language	Chinese

1. Task Overview

This HTML companion mirrors the canonical local T000039-probabilistic-reward-task. Participants first read Chinese instructions, complete two practice trials, and then perform three experimental blocks of 100 trials each. On every trial they see a cartoon face, briefly flash a short or long mouth, and identify the mouth length with a counterbalanced key mapping.

The browser port preserves the local reward logic: one stimulus type is rich and receives three times as many reward opportunities as the lean stimulus, missed reward opportunities remain pending until the next correct response to the same stimulus type, and the cumulative score starts at 0.

2. Task Flow

Block-Level Flow

Step	Description
Load Config	Load config/config.yaml and build the shared stimulus bank.
Collect Subject Info	The shared runner collects a three-digit subject_id.
Counterbalance	subject_id mod 4 assigns the short/long key order and rich-stimulus mapping.
Show Instructions	Present the Chinese instruction text and optional instruction voice asset.
Practice	Run two practice trials with correctness-only feedback.
Experimental Blocks	Run three blocks of 100 trials with 30 rich and 10 lean reward opportunities per block.
Rest Breaks	Show a fixed 30-second break after each block.
Final Summary	Show the final cumulative score and export data through the shared runner.

Trial-Level Flow

Step	Description
Fixation	Show a central + for 500 ms.
Face Preview	Show the face outline and eyes for 500 ms.
Mouth Flash	Flash the short or long mouth for 100 ms.
Response Window	Keep the face visible while collecting a response for up to 2.5 s.
Outcome Feedback	Practice trials show 正确 / 错误; experimental trials show reward feedback only when a reward is delivered, otherwise a neutral fixation.
ITI	Show a short fixation for 500 ms before the next trial.

Controller Logic

Component	Description
Schedule generator	generate_prt_schedule(...) creates balanced short/long trial orders and controlled reward-opportunity lists.
Deferred reward rule	RewardTracker keeps pending rewards per stimulus type and delivers them on the next correct response of the same type.
Counterbalancing	resolve_counterbalance(...) maps subject_id mod 4 to key order and rich-stimulus assignment.
Signal-detection metrics	Block summaries compute reward totals, accuracy, and the log_b / log_d audit metrics.

3. Configuration Summary

All settings are defined in config/config.yaml.

a. Subject Info

Field	Meaning
subject_id	Three-digit participant identifier used for deterministic counterbalancing.

b. Stimuli

Stimulus ID	Purpose
instruction_text	Chinese task instructions.
instruction_text_voice.mp3	Canonical instruction voice asset preserved for parity.
fixation	Central fixation cross.
face_outline	Cartoon face outline.
eye_left / eye_right	Eyes shown during preview and response screens.
mouth_short / mouth_long	Line stimuli for the short/long mouth flash.
practice_feedback_correct / practice_feedback_incorrect	Practice-only correctness feedback.
reward_feedback	Reward feedback with the running score.
block_break	Fixed-duration rest screen between blocks.
good_bye	Final completion screen.

c. Timing

Phase	Duration
Practice fixation	0.5 s
Face preview	0.5 s
Mouth flash	0.1 s
Response window	2.5 s
Reward feedback	1.75 s
Practice feedback	1.75 s
ITI	0.5 s
Block break	30 s

d. Reward Rules

Parameter	Value
rich_reward_trials	30 per experimental block
lean_reward_trials	10 per experimental block
reward_win	5
reward_loss	0
counterbalance_policy	subject_id_mod_4
no_response_policy	random

e. Browser Notes

Parameter	Meaning
task.save_path	Browser output target handled by the shared runner.
task.voice_enabled	Kept false by default so the task remains text-only.
Runtime renderer	The shared psyflow-web runtime renders PsychoPy-style line stimuli and wrapped Chinese text.

4. Methods

Participants completed a browser-based probabilistic reward task that measures reward responsiveness and response bias. On each trial, a cartoon face is previewed and then flashed with either a short or long mouth. Participants identify the mouth length using a counterbalanced response-key mapping. Reward feedback is delivered according to a controlled schedule that favors one stimulus type three-to-one over the other, and missed reward opportunities are carried forward until the next correct response of the same stimulus type.

The browser companion keeps the local analysis logic intact while moving shell behavior into the shared psyflow-web runtime. Reduced output remains one logical trial per row, and the task-specific Chinese text stays in YAML so it can be reviewed without editing the trial logic.

Run

1. Open psyflow-web in the shared workspace.
2. Run npm install if dependencies are not installed yet.
3. Start the runner with npm run dev.
4. Open http://127.0.0.1:4173/?task=H000039-probabilistic-reward-task.

Notes

• The browser companion preserves the canonical 3-block, 300-trial structure.
• The local trigger codes stay in config for provenance, but the browser runtime does not emit serial triggers.
• Pushes to main can refresh the shared runner through .github/workflows/notify-psyflow-web.yml when TASKBEACON_ORG_DISPATCH_TOKEN is configured.