For researchers · Run a study

研究者专区

Helibrain 提供经典认知任务的免费网页和手机版本，例如反应时间、记忆广度、斯特鲁普任务和气球风险任务。这些任务支持 113 种语言。每项任务都有方法页面，准确说明其运作方式：试次、时长、评分以及有效运行的规则。每项任务还附有可直接使用的引文，便于准确注明所用版本。

Create a study Explore the tasks

Free to use. Your survey handles information and consent; Helibrain runs the tasks.

From your survey to trial-level data

  1. Your survey

    Information, consent and a participant code

  2. Helibrain tasks

    A study link, with your tasks in order

  3. Back to your survey

    Completion redirect with the same code

Your dashboard → trial-level CSV / JSON

Here’s what you get

See the pattern, inspect quality flags, and take a readable report into your paper. Explore a full example with charts, Methods and Results text, PDF and Excel.

Explore the example report →

Example data · Stroop interference

Extra time for mismatched colours (ms)Mean: 107.3 msP011: -2.2 msP015: 22.5 msP013: 25 msP001: 40.5 msP006: 56 msP019: 62.2 msP018: 77.1 msP007: 106.6 msP014: 110 msP020: 112.2 msP017: 117.5 msP005: 122.5 msP004: 141.6 msP016: 146.6 msP002: 148.4 msP012: 148.6 msP008: 160.9 msP003: 164.5 msP010: 171.5 msP009: 213.1 ms-15.145.2105.5165.7226
Extra time for mismatched colours (ms)

The task setup is already done.

Spend your time on the research question. Inspect the methods before deciding whether a task fits your protocol.

Free, from setup to export

Create a study link and download the results without paying for access.

43 documented tasks and questionnaires

Classic paradigms alongside skill tasks and self-report measures. Read the trials, scoring, differences from the original and references on each methods page.

161 instruction languages

Word-based tasks use material made for the language where available, rather than simply translated English word lists. The dashboard shows the tasks available in your chosen language.

Every trial, ready to analyse

Export CSV or JSON with participant codes, responses, timing, task versions and scores. Go beyond the final score.

A focused participant experience

Study mode has no sharing buttons, public comparisons or ads. Participants follow your selected task order and return to your survey.

Consent stays in your workflow

Keep information, consent and any ethics approval with your own survey or study platform. Send participants to the tasks only after that step.

Separate study data

Study results stay separate from public player results and norms. Delete a study and its data in the dashboard, or request deletion. Optional contributions of anonymised aggregates are off by default.

Versions you can cite

Exports record the task versions used. Each methods page supplies a citation and original references. The methods document has a Zenodo DOI.

From survey to results in three steps.

  1. Create your study

    Sign in and confirm your email. Choose the language, add supported tasks in order, and set the completion URL. Create the study and copy its participant link.

  2. Add the link to your survey

    After consent, redirect participants to the study link. Pass an assigned code with ?pid=P001, replacing P001 for each person. After the last task, Helibrain returns them to your completion URL with the same pid; other URL parameters are preserved.

  3. Download your data

    Open your dashboard and download CSV or JSON. CSV has a row for each trial; JSON also includes study settings. Match the participant code to your survey responses.

Participant link format

https://helibrain.com/s/YOUR_STUDY?pid=P001

Keep the code-to-person key yourself. Use assigned codes, not names or email addresses. Codes must contain a letter and a digit; the dashboard link includes an example.

What the export looks like

These are actual column names and sample values from a local reaction-time study with simulated participants. This is a subset of the CSV, not a list of fields invented for this page. Responses and metrics vary by task.

Selected columns from the local study export
participant_codetask_codetask_versionlanguagesectiontrial
P005reaction_time1:231f48fef1dbceaaenreaction_times_ms{"index":0,"value":372.0}
P005reaction_time1:231f48fef1dbceaaenreaction_times_ms{"index":1,"value":518.0}
P005reaction_time1:231f48fef1dbceaaenreaction_times_ms{"index":2,"value":587.0}
Download the full sample CSV →

The full export also records task start and submission timestamps, configuration, device type, trial responses and task metrics. Times keep their original field names and units. Pilot the tasks and inspect the export before collecting your study.

Know the limits before you collect.

Browser timing includes the device

Screen refresh rate, browser scheduling and input hardware add delay and variation. A millisecond value in an export is not a guarantee of millisecond accuracy. Read the reaction-time timing notes and screen refresh-rate notes. Keep device and browser conditions consistent where your design needs this.

Online participation is unsupervised

Distractions, practice, comprehension and device differences can affect performance. Pilot your protocol, define exclusions in advance and check trial-level responses.

Validation is planned, not published

These implementations differ from established tests. Validation against established tests is planned and has not yet been published. The tasks are for research and demonstrations, not diagnosis, treatment or clinical decisions.

Questions before you start

What does it cost, and are there usage limits?

Study links and CSV/JSON exports are free. The Research Terms promise no fixed usage quota. The dashboard currently enforces study and participant caps, and service protections can slow or pause use. Check the dashboard and contact us before planning a larger study.

Who handles ethics approval and participant consent?

You do. Give participants information, obtain consent and arrange any required ethics approval in your own survey or study platform before sending them to Helibrain. You decide whether a task is suitable for your protocol. Read the Research Terms before collecting data.

Which survey tools can I use?

Any survey tool that can redirect to a URL with a participant-code parameter. Send the assigned code as ?pid=P001 on the study link, replacing P001 for each person. Set an HTTPS completion URL when creating the study; after the final task we append the same pid and preserve its other parameters. Participants do not need a Helibrain account.

Where is study data stored?

Study data is stored on servers in the United States. If your institution requires EU storage, contact us before collection; EU hosting can be arranged for larger studies. See the Research Terms for the processing agreement and transfer clauses.

Can I delete study data?

Yes. Delete the study and its participant data in your dashboard, or contact us with a deletion request. Export anything you need first. Backup copies are removed when the backups expire. Study results stay separate from public player results and norms; optional aggregate contributions are off by default.

Which languages are available?

Task instructions are available in 161 languages. Tasks that need words, passages or other language-specific material appear in the dashboard only where that material is available. Check each methods page for its language material and review status; do not assume every translation has been reviewed by a native speaker.

How should I cite Helibrain in a paper?

Use the citation on each task’s methods page, report the task version from your export, the language and your collection dates, and cite the original task papers listed there. For the methods document, use the Zenodo concept DOI 10.5281/zenodo.24052521.

How accurate is reaction-time measurement?

Timing uses the browser clock and includes display and input-device delay. Accuracy varies across devices and browsers. Read the reaction-time limitations and refresh-rate notes, pilot on your target devices and assess whether that precision fits your research question.

Are these validated tests or clinical assessments?

No validation of these implementations against established tests has been published; that work is planned. The methods pages explain differences from the classic tasks. Do not use the tasks for diagnosis, treatment or clinical decisions.

How can I discuss a study or ask for help?

Email hello@helibrain.com with your research question, the tasks and languages you need, and any hosting requirements. Contact us before collection if your institution needs EU storage or your study needs more capacity.

任务

下方每项任务都链接到其方法页面，其中列有版本号和参考文献。

43 documented tasks and questionnaires, grouped below. Each link opens the full method. The study dashboard is the current list of tasks available for a study link and language; not every documented task is supported in study mode.

Data quality flags

Study exports keep browser, operating system and available device model, screen and viewport sizes, display timing, and time away for each task run. Leaving or hiding the page, losing window focus, copying, cutting, pasting, text selections, context menus and delivered PrintScreen key events are counted. Clipboard contents and selected text are never recorded. Answer rows include time since the last return when it can be matched to the trial.

Flags point out time away, unusually fast or repeated answers, accuracy at or below a defined chance level, unreliable display timing, incomplete runs, copying, pasting into an answer field, a screenshot key and answers soon after returning. They are information for review: no flag removes a row or changes a score. JSON metadata gives the column dictionary and exact rules and thresholds; missing or inapplicable measurements are blank in CSV and null in JSON.

A browser cannot see screenshots taken on phones or Macs, a second device, or some in-browser side panels. These signals cannot establish whether someone used outside help.

引用方式

请引用所用任务及其版本，并附上方法页面链接：

Helibrain (2026). <Task name>, v<version>. helibrain.com/<slug>/methods

例如：Helibrain (2026). Attention Span Test, v1. helibrain.com/attention-span-test/methods

每当评分发生变化，版本号就会更新。不同版本的结果不应合并分析。也请引用任务的原始论文，该论文列在任务的方法页面上。

Helibrain (2026). Helibrain cognitive tests: methods and scoring. 10.5281/zenodo.24052521. This concept DOI always resolves to the latest version.

Use in class: free task demonstrations in 113 languages, with no sign-up.

Explore the classroom activities →

常模

我们将按语言和年龄段为每项任务发布汇总常模，作为带有 DOI 的开放数据集。只有当某个群体的参与者数量足以得出稳定数值时，我们才会发布该群体的数据集。每个数据集都会注明任务版本、样本情况以及运行数据的筛选方式。

任务和文本的制作方式

  • 任务。 每项任务都依据已发表的描述构建。所有数字（试次数、时长、阈值和评分）都集中保存在我们的服务器上，方法页面会展示实际使用的数值。
  • 翻译。 指令和项目文本采用机器辅助翻译，并进行自动化脚本与拼写检查，同时与母语文本语料库进行比对。
  • 语言专属材料。 对词语有依赖的任务（例如词表或阅读文章），只有在存在为该语言专门制作的材料时才会提供。我们不会直接翻译英文词表。
  • 审校状态。 每个方法页面都会按语言显示文本是否经过母语者审校，或仅经过自动检查。

联系我们

关于任务的问题、修正建议或想法：hello@helibrain.com

Ready to pilot your study?

Choose your tasks, try the participant link, and inspect the export before inviting your participants.

Create a study →Read the Research Terms