Skip to content

Introduction to HED

Context: Open and Reusable Neuroscience Data

Modern neuroscience increasingly relies on open, well-structured, and reusable datasets. Projects such as the Austrian NeuroCloud aim to make research data easier to share, process, and analyze using standardized formats and reproducible workflows. Within this ecosystem, standards such as BIDS (Brain Imaging Data Structure) play an important role in organizing neuroimaging and electrophysiology data.

However, while BIDS standardizes how data files are organized, researchers still need a clear way to describe what actually happens during an experiment. This is where Hierarchical Event Descriptors (HED) becomes important.

HED provides a standardized and machine-readable way to describe experimental events so that datasets become easier to understand, search, reuse, and analyze.

What is HED?

Hierarchical Event Descriptors (HED) is a standardized framework for describing events that occur during an experiment. It is mainly used for time-series and experimental data such as:

  • EEG
  • MEG
  • fMRI
  • physiological recordings
  • behavioral experiments

In many datasets, events are stored using short labels or numeric codes such as:

1
2
stim_on
resp

While these labels may make sense to the original researcher, they are often ambiguous or undocumented for others.

HED solves this problem by providing a controlled vocabulary organized as a hierarchy. Researchers can use this vocabulary to describe events in a structured and consistent way. These descriptions are called HED tags.

A HED annotation can describe many aspects of an event, for example:

  • the stimulus that was presented
  • the participant action
  • the task context
  • experimental conditions
  • timing information

Because HED annotations follow a shared schema, they can be understood by both humans and software tools.

HED is also integrated into widely used data standards such as BIDS (Brain Imaging Data Structure) and NWB (Neurodata Without Borders), making it especially useful for modern neuroscience data infrastructures.

Why should researchers use HED?

Researchers often collect valuable experimental data, but the meaning of the events in those datasets is not always documented well enough for other people — or even for the original research team months later — to understand them easily.

HED addresses this problem by making event information explicit, structured, and machine-readable.

1. It makes event annotations understandable

Traditional event codes are often too short to explain what really happened. A label like 3 or target may only make sense to the person who created the dataset.

HED replaces unclear event labels with annotations that capture the actual meaning of the event in a human-readable way.

2. It supports machine‑actionable analysis

HED annotations are not just notes for humans. They are designed so that software tools can:

  • validate event descriptions
  • search datasets
  • extract specific types of events
  • automate analysis pipelines

This makes datasets significantly easier to work with in reproducible research workflows.

3. It improves sharing and reuse

When datasets are annotated consistently, they become easier to share with collaborators, repositories, and the broader scientific community.

Researchers who access the dataset later can understand the experiment without relying on undocumented lab-specific conventions.

4. It enables cross‑study comparison

Different studies often use different local terms for similar events. HED provides a shared vocabulary, which makes it easier to:

  • compare experiments
  • search across datasets
  • combine data for meta-analyses or large-scale studies

5. It creates analysis‑ready datasets

A well-annotated dataset is closer to being ready for downstream analysis. HED helps preserve the experimental meaning of events, conditions, and task structure directly with the data.

This reduces ambiguity and improves long-term usability of research data.

In simple terms

HED helps researchers answer one essential question:

What exactly happened during this recording or experiment?

It provides a structured way to store that answer so that:

  • people can read it
  • software tools can process it
  • datasets can be reused
  • scientific results become easier to reproduce and compare