EEG explained

A brand appears on screen, someone hears an unexpected sound or enters a new space. If you want to investigate how responses develop around such moments, EEG may be a suitable method. Its value emerges when you connect the recording to a concrete question, a well-executed task and a justified analysis.

EEG stands for electroencephalography: recording electrical voltage differences associated with brain activity. This article concerns recordings made with electrodes on the scalp. The recording itself is called an electroencephalogram. Buzsáki et al., 2012

You can use EEG to investigate patterns during a task or experience. Moving from a pattern to a statement about attention, experience or choice requires an appropriate study design. Start with the question you want to answer.

An important historical source is Hans Berger’s 1929 Über das Elektrenkephalogramm des Menschen, a publication on human EEG. Its terminology is recognisable in the modern word electroencephalogram. Berger, 1929

The signal combines contributions from neurons, particularly synaptic currents. Intervening tissue affects the recording; an electrode therefore does not measure an isolated cell or sharply bounded brain region. Buzsáki et al., 2012

Document amplification, digitisation, reference, electrode positions and recording settings to support interpretation. Keil et al., 2014

When assessing a proposal, look beyond the number of sensors. Ask which measurements your comparison needs and how the team will assess whether they are usable.

An event-related potential (ERP) is a pattern associated with an event, often estimated by averaging repeated presentations. Time-frequency analysis describes how signal features at different frequencies change over time. Both approaches are discussed in the original EEGLAB publication. Delorme and Makeig, 2004

For your brief, ask whether you want to compare a repeated, brief event or investigate changes during a longer task. That choice helps guide the analysis discussion. Selecting a software package is not yet an analysis plan.

Names such as delta, theta, alpha, beta and gamma denote frequency ranges. The clinical glossary by Kane and colleagues defines the alpha band as 8–13 Hz, for example. Hertz means cycles per second here. Band boundaries should be explicitly stated in the study protocol. Kane et al., 2017

Do not treat a band name as a standalone label for someone’s feelings. If a report links a change to relaxation, attention or mental effort, ask which task, comparison and evidence support that interpretation.

Use these six steps when discussing a study design. They provide a practical briefing structure, not a fixed recording protocol.

1. Specify the decision. Do you want to compare two presentations of information, for example? Describe what you want to choose or change afterwards.
2. Define the task and participants. Record who takes part, what they do and which conditions should remain comparable.
3. Prepare the recording. Discuss electrode placement, signal checks and how events will be marked.
4. Run the task. Record which version each person receives, what happens and any deviations during the session.
5. Assess and analyse the data. Use predefined criteria for usable data and show which portions are excluded from analysis.
6. Connect findings to the decision. Explain what the comparison supports and which questions remain open.

Ask who is responsible for each step and what will be documented. That allows you to trace the eventual recommendation back to the study’s execution.

Eye movements, muscle activity and technical interference can affect recordings. Report how the data were processed. Keil et al., 2014

An artefact here is a contribution to the recording other than the brain signal you want to investigate. Ask how the team identifies these contributions and justifies exclusion or correction. A smooth-looking graph is not independent evidence of good data.

For a comparison around a particular moment, also ask how that moment was recorded. A note saying a logo appeared somewhere halfway through a video provides a different basis from a checked event marker at its actual presentation.

Mobile EEG enables recording during more natural behaviour. Gramann and colleagues describe the importance of recording movement, environment and events together, and separating brain signals from other sources. Gramann et al., 2014

For research in a shop, museum or another environment, prepare an execution plan: which moments will you compare, what differs between participants and which records will help reconstruct the context? Test the setup in the intended situation beforehand.

A mobile device does not automatically make a study equivalent to everyday experience. The task, supervision and equipment may change the situation; include that consideration when explaining the findings.

These examples help translate a business question into a research discussion. They are proposed designs, not demonstrated results.

Ask what additional information EEG provides. If a usability test or another direct measurement can sufficiently support your decision, include that option in the assessment.

This example is fictional and contains no research findings from Neurofactor.

An organisation wants to choose between two short instructional videos. Version A presents all the information at once; version B separates it into steps. The practical objective is correct execution of the procedure.

The team can first decide which performance measure should guide the choice. It then considers a useful additional EEG question, such as whether predefined measurements differ during the instructional steps. Presentation order and materials need to be designed to make the comparison useful.

If an EEG outcome differs, that alone does not establish which version explains the procedure better. Assess it alongside task execution and other data. A follow-up test may be needed to evaluate the proposed change.

Translating a brain response directly into one mental state requires evidence. Poldrack discusses this as reverse inference: reasoning backwards from measured activity to the process proposed to explain it. Poldrack, 2006

For EEG, we apply this general interpretive principle as follows. Look for three steps in a report: what was measured, what meaning is assigned to it and which decision is proposed? Show the data or assumptions needed for each step. A custom index called ‘engagement’ or ‘purchase intention’ needs an explanation of its calculation and validation.

Also consider whether a statement about an average fits the decision you want to make. A group comparison is not an automatic prediction for every individual customer.

EEG also has clinical applications, including investigations of epilepsy. The NHS describes assessment of these recordings by specialist healthcare professionals. NHS: EEG

Consumer or experience research addresses a different question. Do not treat its report as a medical result. Discuss the session’s purpose, what participants will do, how data will be used and where questions about participation can be raised.

EEG can make a valuable contribution when you specify which question it answers. Document the comparison, execution and analysis, and connect interpretation to the data you actually have. That allows you to assess what the study contributes to your decision.