Neurotech
Non-invasive neurotech and the physics of the skull
Reading brain activity through bone is limited by physics rather than by engineering budget, which sets a ceiling on what consumer devices can do.
Consumer neurotechnology — headbands, earbuds and headsets claiming to read focus, stress or intention — rests on measuring electrical or optical signals through the skull.
What is achievable through that route is constrained by physics, and knowing the constraint makes most product claims easy to assess.
What EEG measures
Electroencephalography records voltage fluctuations at the scalp, produced by the summed activity of large populations of neurons.
Two things limit it fundamentally.
Volume conduction. The signal passes through cerebrospinal fluid, skull and scalp, each with different conductivity. The skull in particular is a poor conductor and smears the signal spatially.
The consequence is that a scalp electrode records a blurred sum from a large cortical area, not from a specific location. Spatial resolution is on the order of centimetres, and no amount of signal processing recovers what the physics has averaged away.
Amplitude. The signals are tiny — microvolts — and are competing with electrical noise from muscles, eye movements, heartbeat, mains hum and the electronics themselves.
Muscle artefact in particular is orders of magnitude larger than the brain signal. Clenching the jaw produces a dramatic trace that has nothing to do with cognition.
This matters for consumer devices, because a system that responds to jaw tension while claiming to measure focus will appear to work.
What EEG is genuinely good at
Excellent temporal resolution — millisecond scale — which is why it remains a core research and clinical tool.
Detecting oscillatory activity in defined frequency bands, which correlates with broad states such as drowsiness and relaxed wakefulness.
Sleep staging, where it is the clinical standard.
Seizure detection and epilepsy monitoring.
Event-related potentials — averaged responses to repeated stimuli — which underpin several practical BCI paradigms.
The workable BCI paradigms
Non-invasive BCIs that actually function use specific tricks rather than general mind reading.
P300 spellers flash rows and columns of letters; the brain produces a characteristic response roughly three hundred milliseconds after the attended letter flashes. Averaging over repetitions identifies it.
Slow, reliable, and requires the user to attend to a flashing grid.
Steady-state visual evoked potentials. Targets flicker at different frequencies; the visual cortex response contains that frequency. Choosing a target means looking at it.
Faster, and requires flickering stimuli that many users find unpleasant.
Motor imagery. Imagining movement of the left or right hand produces detectable changes over the corresponding motor cortex.
Requires training, works poorly for a meaningful proportion of users — a phenomenon researchers call BCI illiteracy — and yields a small number of commands.
All three deliver a few bits per second at best. That is the honest bandwidth of non-invasive control.
Functional near-infrared spectroscopy
Measures blood oxygenation changes optically through the skull.
Better spatial localisation than EEG for superficial cortex, far worse temporal resolution — because it tracks blood flow, which lags neural activity by seconds.
Portable and useful in research, and not a route to fast control.
Assessing a consumer claim
Some questions that resolve most of them.
How many electrodes, and where? A single frontal electrode cannot distinguish activity across the brain. Two electrodes on the forehead are heavily contaminated by eye and muscle activity.
What exactly is being measured? "Focus" is not a physical quantity. A band power ratio is. Ask which.
Is there peer-reviewed validation against an established measure, from a group not selling the device?
Could the effect be artefact? If the device responds when you clench your jaw or move your eyes, it is measuring muscle.
Is there a plausible placebo effect? Neurofeedback studies with proper sham controls have frequently found much smaller effects than uncontrolled ones.
Where the honest value is
Sleep tracking, where EEG-based consumer devices measurably outperform accelerometer-based ones and are the correct tool.
Relaxation and breathing feedback, where the device provides a plausible signal and the benefit is largely from the guided practice.
Research and education, where affordable hardware has genuinely widened access.
What is not available, at any price, from outside the skull: reading specific thoughts, decoding language, or high-bandwidth control. Those are blocked by the skull rather than by the engineering.
