ERP Research Lab

Event-related potential research lives or dies on timing. The components you're measuring are tens of milliseconds wide, so a few milliseconds of trigger jitter smears the average and buries the effect. The difference between a clean ERP and noise is often nothing more than how precisely the stimulus was timestamped.

This use case shows how a cognitive neuroscience lab runs ERP studies on Qusp with sub-millisecond trigger alignment, reproducible preprocessing, and BIDS-ready exports — so effects survive averaging and results survive replication.

The setup

The lab streams EEG and stimulus markers onto one shared clock, so every trigger lands exactly where it happened. Preprocessing runs as a pinned, scripted pipeline rather than a series of manual clicks, and every dataset exports in a standard structure ready for analysis or sharing.

The deployment has four parts:

• Precise triggers — Stimulus markers ride the same time base as the EEG over Lab Streaming Layer, holding trigger jitter under a millisecond.
• Continuous capture — The full session is recorded continuously, so epoch windows and baselines are chosen at analysis time, not locked at capture.
• Pinned preprocessing — Filtering, epoching, and artifact rejection run as a version-pinned script that produces the same result every time.
• BIDS-ready export — Data and metadata export together in a standard layout, ready for group analysis or open data sharing.

How a typical study runs

The paradigm sends stimulus markers into Qusp over LSL, where they're timestamped against the same clock as the EEG. There's no separate trigger box to reconcile and no drift to correct after the fact — alignment is correct at capture.

The session records continuously. Because nothing is epoched at capture time, the researcher can re-window, re-baseline, and re-reject as the analysis evolves, all from the same recording.

Preprocessing runs as a script: the same filters, the same epoch definition, the same rejection thresholds, pinned to specific versions. Re-running it on the same data produces an identical result.

Finished datasets export to a BIDS-compatible structure, so a collaborator — or the lab six months later — can pick them up and reproduce the analysis without guesswork.

What every session captures

Each ERP session records to the same specification:

• 64-channel montage — A 64-channel cap at 1000 Hz, giving the spatial and temporal resolution ERP work depends on.
• Marker stream — Stimulus and response markers carried on the shared LSL clock, timestamped with the EEG.
• Provenance metadata — Every output records the exact preprocessing steps and parameters that produced it.

Compare that to the common setup where triggers arrive over a separate channel with unknown latency and preprocessing is done by hand in a GUI. Jitter creeps into the average and no two analysts get quite the same numbers. Locking timing and pinning the pipeline removes both problems.

The outcomes

A lab running this pattern typically sees:

1. Cleaner averages — Sub-millisecond trigger alignment keeps ERP components sharp instead of smeared.
2. Reproducible results — A pinned pipeline produces the same numbers on every run, by every analyst.
3. Reusable data — Continuous recording lets the same session support new epoch windows and reanalyses for years.
4. Shareable datasets — BIDS exports make data ready for collaborators and open repositories without reformatting.
5. Faster onboarding — New lab members inherit the pipeline and structure instead of reinventing preprocessing.

Where it doesn't fit

This setup fits timing-critical ERP and evoked-response research. Three situations are weaker fits.

First, paradigms with no precise event structure — resting-state or free-viewing work doesn't need sub-millisecond triggers, though it still benefits from the pipeline discipline.

Second, stimulus hardware that can't emit markers. The timing guarantee depends on the paradigm publishing markers to LSL; hardware that can't has to be characterized for latency separately.

Third, one-off exploratory pilots. If you're changing the paradigm daily, formalizing a pinned pipeline can wait until the design stabilizes.

Standing it up

Most labs get this running in a couple of weeks. The first step is wiring the paradigm's markers into LSL and verifying timing against a photodiode or sync pulse. The second is scripting the preprocessing pipeline and pinning versions. The third is validating the BIDS export on a complete dataset.

The researchers who design the paradigms should own the pipeline. They know which components matter and which artifacts to reject, and that knowledge is what makes the automated preprocessing trustworthy.

Bring this to your lab

If you run ERP or evoked-response studies, Qusp can give you sub-millisecond trigger alignment, a reproducible preprocessing pipeline, and BIDS-ready exports. Talk to our team about wiring your paradigm, pinning your pipeline, and standardizing your lab's data.