fNIRS explained

How can you investigate what happens when someone follows instructions, uses an interface or learns a task? fNIRS can add another layer of measurement. It uses light to track changes in blood-related signals. Turning those measurements into useful findings requires a clear match between the question, task and measurement setup.

fNIRS stands for functional near-infrared spectroscopy. It investigates brain activity indirectly through changes in oxygenated and deoxygenated haemoglobin. Measurements are primarily sensitive to superficial cortex beneath the measurement locations. A signal difference alone is not a score for preference, understanding or purchase intention. Scholkmann et al., 2014

In 1977, Frans Jöbsis published work on non-invasive infrared monitoring of oxygen-related processes, including in the brain. This provides a historical foundation for NIRS. It does not validate present-day marketing applications. Jöbsis, 1977

The addition of functional refers to investigating function. A research brief should therefore explain which task or comparison will be used and what conclusion the findings are intended to support.

Light sources and detectors placed on the head are called optodes. Oxygenated haemoglobin (HbO) and deoxygenated haemoglobin (HbR) absorb light differently. Common continuous-wave systems estimate relative concentration changes from intensity changes using the modified Beer–Lambert law. They do not measure electrical brain activity or directly provide percentage oxygen saturation. Scholkmann et al., 2014

When examining a graph, first ask what its lines represent. Are they HbO, HbR, differences between conditions or a derived index? What units and reference are used? Without this information, a colour or peak is difficult to assess.

Scalp blood flow and changes in systemic physiology can affect the recorded signal. An effect may consequently be attributed to brain activity incorrectly or go undetected. Tachtsidis and Scholkmann, 2016

Short-separation channels are predominantly sensitive to superficial tissue. Researchers can use them to estimate and reduce contributions from outside the brain. Contact quality, movement and excluded data also need reporting. Such correction does not guarantee a completely isolated brain signal. Yücel et al., 2021

Turn this into a practical question: how does the study establish that its preferred explanation is more plausible than a change in movement, physical effort or sensor contact?

Wearable setups are available. A systematic review of postural and walking research illustrates the potential of fNIRS for movement science, while identifying methodological differences that make studies difficult to compare. Herold et al., 2017

For an applied study, start with a pilot in which participants perform the actual task. Assess whether the setup suits those activities. A successful seated demonstration does not establish feasibility during bending, speaking and walking. Agree in advance when insufficient usable data should trigger a change in task or method.

These are briefing examples, not validated applications of a particular device.

Do not substitute a physiological outcome for the actual research objective. If the aim is fewer errors, errors also need to be measured.

1. Name the decision. For example, choosing between two instructions for the same action.
2. Define the comparison. Specify what stays constant and what deliberately changes.
3. Select the outcomes. Identify the performance, experience or choice that matters alongside fNIRS.
4. Request a justified setup. Ask which regions are sampled and which quality criteria apply.
5. Specify the analysis beforehand. Explain how exclusions, confounds and uncertainty will be handled.
6. Agree on the next step. Decide when the findings warrant a revision, another test or no change.

This is an editorial briefing aid. The actual design and sample require their own scientific justification.

Imagine an organisation comparing written instructions with instructions using illustrations. Both explain the same action at a workstation. This example is fictional; it contains no measurements or client findings.

Agree how conflicting outcomes will be discussed. Instructions could produce fewer errors while a physiological measure changes. This alone does not justify calling them “easier”. The conclusion should reflect all relevant evidence and its uncertainty.

• Treating a higher line as better. Ask which hypothesis supports that interpretation.
• Using a figure without a reference. Show the task, comparison, units and variability.
• Leaving exclusions invisible. Ask how much data was usable and whether exclusions could bias the comparison.
• Generalising from one task to all customers. State which participants and circumstances the conclusion covers.
• Inferring an associative target group from one signal. In this knowledge base, this means a bounded group making decisions on the basis of shared associations. That working definition requires research into associations and decisions; an fNIRS pattern alone does not establish the grouping.

Assess fNIRS through the question you need to answer. Ask what it adds, how alternative explanations will be examined and which decision the findings can support. Also agree on consent, access to data and the use of conclusions. This article explains a method; it is not a diagnostic protocol or a tool for evaluating individual employees.