EEG vs. Pupillometry for Cognitive Load

Two Signals Your Body Can't Fake

Try something right now. Multiply 47 by 38 in your head. Don't grab a calculator. Just hold those numbers in your mind and work through it.

Did you get 1,786? (Don't worry if you didn't. The answer isn't the point.)

The point is what just happened inside your body while you tried. Two things changed, involuntarily, without your permission or awareness. First, your pupils dilated. Not a lot. Maybe a quarter of a millimeter. But measurably, reliably, within about 300 milliseconds of the moment the problem hit your working memory. Second, and simultaneously, the electrical patterns rippling across your cortex shifted. Theta-band activity (slow waves around 4 to 8 Hz) surged over your frontal midline. Alpha-band activity (faster waves around 8 to 13 Hz) dropped over your parietal and occipital cortex.

Two completely different physiological systems, one in your eyes and one in your brain, both responding to the same invisible thing: cognitive load.

Scientists have known about both of these signals for decades. Pupillometry, the measurement of pupil diameter changes, has been used as a cognitive workload index since the 1960s. EEG-based workload assessment has a similarly long history. Both are legitimate. Both are well-studied. Both give you real, reproducible data about how hard someone's brain is working.

But here's what most people don't realize: these two signals are not telling you the same thing. They're not even close. One is like a check engine light. The other is like a full diagnostic readout. And the difference matters enormously if you're trying to actually understand, track, or optimize cognitive performance.

The Pupil: A Tiny Window into a Massive System

Let's start with the signal that's easier to observe, because it's literally visible to the naked eye.

Your pupil is a hole. That's all it is. An opening in the iris that lets light reach the retina. Two muscles control its size: the sphincter pupillae, which constricts it, and the dilator pupillae, which expands it. These muscles receive input from the autonomic nervous system, and for most of human history, scientists thought pupil size was purely about light regulation.

Then, in 1964, a psychologist named Eckhard Hess published a paper that changed everything. Hess and his colleague James Polt showed people arithmetic problems of increasing difficulty while photographing their eyes. The harder the problem, the wider the pupils got. This wasn't a light reflex. The lighting in the room was constant. The pupils were responding to something happening inside the brain.

This discovery launched an entire field. Over the following decades, researchers confirmed that task-evoked pupil dilation is one of the most reliable physiological correlates of cognitive effort. It scales with working memory load, attentional demand, decision difficulty, and emotional arousal. It's been replicated thousands of times across hundreds of labs.

But what's actually causing it? Why would thinking hard make your eyes open wider?

The Locus Coeruleus Connection

The answer lives in a tiny, bilateral nucleus deep in the brainstem called the locus coeruleus (Latin for "blue spot," because it's pigmented with neuromelanin and actually looks blue in dissection). The locus coeruleus is tiny, roughly the size of a grain of rice on each side, but it's the brain's primary source of norepinephrine, a neurotransmitter that functions as the brain's arousal and alertness signal.

When cognitive demand increases, the locus coeruleus fires more intensely. It broadcasts norepinephrine widely across the cortex, essentially shouting "pay attention, something important is happening." This norepinephrine signal does many things simultaneously: it enhances sensory processing, sharpens attention, facilitates working memory, and, as a side effect, it activates the sympathetic pathway to the iris dilator muscle.

Your pupils dilate because the same brainstem nucleus that tells your cortex to work harder also happens to control your iris.

This is elegant. It's also limiting. Because what pupil dilation actually reflects is locus coeruleus activity, not cognitive load per se. The pupil is a proxy for a proxy. Task gets hard, locus coeruleus fires, norepinephrine goes up, pupils dilate. You're measuring the last domino in a chain, and by the time it falls, a lot of information has been lost.

What Pupillometry Is Good At

Credit where it's due. Pupillometry has real strengths, and there's a reason it shows up in so many research papers.

It's non-contact. Modern eye trackers can measure pupil diameter from a distance using infrared cameras. Nobody has to wear anything on their head. Nobody has to apply electrode gel. You just look at a screen while a camera watches your eyes. For certain research contexts, especially studies with children, clinical populations, or situations where headwear is impractical, this is a genuine advantage.

It's also simple. The signal is one-dimensional: pupil diameter goes up or down. You don't need to know signal processing or understand Fourier transforms. A larger pupil means more effort. A smaller pupil means less effort. The learning curve for interpretation is almost flat.

And within its lane, it's reliable. The correlation between task difficulty and pupil dilation has been replicated so many times across so many paradigms that it's one of the most established findings in cognitive psychophysiology. If you need a quick binary answer to "is this task harder than that task?", pupillometry will give you one.

But here's where things get complicated.

What Pupillometry Can't Tell You

The locus coeruleus doesn't just respond to cognitive load. It responds to emotional arousal, surprise, novelty, pain, pharmacological stimulation, and even ambient lighting changes that sneak past your experimental controls. Norepinephrine is a general-purpose arousal signal, and the pupil reflects that generality faithfully.

This creates the same problem that plagues skin conductance in stress measurement. If someone's pupils dilate, you know their locus coeruleus fired more intensely. You don't know why. Was it working memory overload? Emotional content in the stimulus? A confusing instruction? An interesting image? Boredom so extreme that the brain started seeking stimulation? All of these produce pupil dilation.

A 2011 study by Laeng, Sirois, and Gredeback in the journal Perspectives on Psychological Science cataloged the range of psychological states that produce task-evoked pupil dilation. The list is startlingly long: cognitive effort, emotional arousal, surprise, decision-making, memory retrieval, mental imagery, pain, sexual arousal, deception, and even musical pleasure. The pupil dilates for all of them.

And there's another limitation that's less obvious but equally important. Pupillometry gives you one number. Pupil diameter. That's it. There's no spatial information (because you're measuring one hole in one eye, not a distributed brain network). There's no frequency information (because pupil changes are slow and don't oscillate in the way brainwaves do). There's no way to decompose the signal into components that tell you what kind of cognitive process is driving the load.

This is like measuring the total electricity consumption of a building and trying to figure out what's happening inside. You can tell the building is "working harder" when power consumption spikes. But you can't tell whether the spike is from the air conditioning, the data center, the elevators, or someone on the 15th floor running a space heater. You just know more power is being drawn.

EEG, by contrast, is like having sensors on every floor.

The Cortex Speaks: How EEG Reads Cognitive Load

Now let's talk about the signal that comes from the source.

EEG, electroencephalography, measures the electrical activity produced by billions of neurons firing in your cerebral cortex. When large populations of cortical neurons synchronize their activity, the combined electrical field is strong enough to detect through the skull using electrodes on the scalp. This has been possible since Hans Berger recorded the first human EEG in 1929, but the relationship between EEG patterns and cognitive workload didn't become clear until the 1990s, when cheap computing made real-time spectral analysis practical.

Here's what happens in your brain when cognitive load increases, broken down by frequency band.

But the theta/alpha ratio is just the headline. EEG contains far more information about cognitive load than a single ratio can capture.

The Spatial Dimension: Where Is the Load?

Because EEG uses multiple electrodes distributed across the scalp, it can tell you which brain regions are working hardest. This is something pupillometry fundamentally cannot do.

A person performing a mental arithmetic task shows different topographic patterns than a person navigating a 3D environment, even if both tasks produce the same overall workload level. The arithmetic task lights up frontal theta and left-lateralized parietal activation. The navigation task engages right parietal and occipital regions more heavily. Same difficulty, same subjective effort rating, but very different brain signatures.

This spatial resolution means EEG can distinguish between types of cognitive load. Working memory load looks different from perceptual load. Attentional load looks different from decision load. Verbal processing loads different regions than spatial processing. A device with sensors across multiple brain regions can tell these apart. A camera pointed at someone's pupil cannot.

The Temporal Dimension: When Does the Load Happen?

EEG operates at millisecond resolution. A device sampling at 256Hz takes 256 measurements per second, each one a snapshot of the brain's electrical state at that precise moment. This means EEG can track cognitive load fluctuations that happen on the timescale of individual thoughts.

Consider reading a sentence. Some words are easy to process ("the," "and," "is"). Some are harder ("mitochondria," "defenestration," "electroencephalography"). EEG can detect the spike in processing demand at each difficult word, in real time, as it happens. Pupillometry, with its 500-to-1500-millisecond response latency, would smear all those individual spikes into one blurry hump.

This temporal precision matters for real-world applications. If you're designing an interface and want to know which specific moment confused the user, EEG can pinpoint it. If you're tracking your own cognitive state while working and want to know when during the hour your brain started flagging, EEG can show you the slope of decline minute by minute.

The Head-to-Head Comparison

Let's put everything on the table.

One pattern stands out from this comparison. Pupillometry's advantage is practical convenience: no head contact, fast setup, simple signal. EEG's advantage is informational richness: more dimensions, more specificity, more temporal precision, more spatial coverage. If your question is simply "is this person thinking hard right now, yes or no?", either method works. If your question is "what kind of thinking, in which brain regions, and how is it changing moment to moment?", only EEG can answer.

Brainwave data, captured at 256Hz across 8 channels, processed on-device. The Crown's open SDKs let developers build brain-responsive applications.

Explore the Crown

The "I Had No Idea" Part: Why These Two Signals Sometimes Disagree

Here's something that blew my mind when I first encountered it in the literature, and it reveals something fundamental about how the brain works.

EEG and pupillometry don't always agree. There are situations where cognitive load increases, frontal theta surges, alpha suppresses, and yet pupil dilation barely changes. And there are situations where pupils dilate dramatically but the EEG workload signatures remain stable.

How is that possible if they're both measuring cognitive load?

They're not. Not exactly. They're measuring different components of the cognitive load response, and those components can dissociate.

A 2018 study by Beatty and colleagues (building on Beatty's original 1982 work linking pupil dilation to mental effort) found that pupil dilation tracked most closely with the perceived difficulty of a task, which correlates with locus coeruleus firing and general arousal. EEG theta, on the other hand, tracked most closely with the actual working memory demands, regardless of whether the task felt difficult.

Think about what that means. You can have a task that feels easy but demands significant working memory resources (like an experienced programmer reading complex code that they're skilled enough to find non-threatening). The EEG would show elevated frontal theta. The pupil might barely budge.

Conversely, you can have a task that feels very difficult but doesn't actually tax working memory much (like an anxiety-inducing social situation where you're not doing any complex cognition, just feeling stressed). The pupils would dilate from the emotional arousal. The EEG might show no frontal theta increase at all.

This dissociation is not a failure of either method. It's telling you something real about the architecture of cognition. The brain has multiple resource pools, multiple systems that can be loaded independently. The locus coeruleus-pupil system reflects one global summary signal. The cortical oscillatory dynamics measured by EEG reflect the actual, distributed, multi-dimensional pattern of neural resource allocation.

If you only had the pupil data, you'd miss the programmer's hidden cognitive engagement. If you only had the pupil data, you'd mistake the anxious person's emotional arousal for intellectual effort. The EEG tells you what's really going on under the hood.

Where Pupillometry Still Wins (And Why It Might Not For Long)

Intellectual honesty demands acknowledging what pupillometry does better.

In UX research labs, where you need to evaluate whether a website or app interface is confusing users, pupillometry has a huge practical advantage. The person just sits in front of a screen and uses the product normally while a camera tracks their eyes. No headset. No electrodes. No setup. No concern about whether the measurement device itself is affecting the user's experience. For rapid, low-friction assessments of whether something is cognitively demanding, this is hard to beat.

Pupillometry also works well with populations where EEG is difficult. Infants and young children, who won't tolerate headgear. Patients with certain neurological conditions that affect scalp sensitivity. People with very short or very thick hair that makes electrode contact challenging. In these cases, a non-contact optical measurement is genuinely the better option.

And in multi-modal research setups, pupillometry complements EEG beautifully. The pupil signal, being subcortical in origin, provides information about brainstem arousal systems that EEG (which primarily captures cortical activity) might miss. Some of the most sophisticated cognitive workload studies combine both methods precisely because they capture different levels of the neuroaxis.

But the practical gap is narrowing. Consumer EEG devices have gotten dramatically easier to use. No gel. No lengthy setup. No technician required. You put the device on your head, and within a couple of minutes, you're getting real-time brainwave data. The Neurosity Crown, for instance, uses dry electrodes across 8 channels covering frontal, central, and parietal-occipital regions, and it provides spectral power data at 256Hz. The setup time advantage that pupillometry once held over EEG has shrunk from "hours vs. minutes" to "one minute vs. three minutes."

And the informational advantage has always favored EEG. It's just that the practical barriers used to be so high that many researchers and developers defaulted to the easier measurement, even though it told them less. Those barriers are falling.

Cognitive Load in the Real World: What You Actually Want to Know

Here's where all of this becomes personal.

If you're interested in cognitive load measurement, you probably fall into one of two camps. You're either a researcher studying cognition, or you're a person who wants to understand and optimize your own mental performance. The calculus is different for each.

If you're a researcher, you likely want both signals. Publish with EEG for the richness and specificity. Add pupillometry for the convergent evidence and the subcortical window. The field is moving toward multi-modal approaches because they answer more questions.

But if you're a knowledge worker, a developer, a student, or anyone who wants to understand when and why your brain is working hardest during your actual day, EEG is the signal you want. Here's why.

Cognitive load isn't a binary. You don't just have "easy" and "hard." Throughout a workday, your brain cycles through dozens of different cognitive states: deep focus, scattered attention, creative ideation, rote execution, confusion, flow, fatigue. Pupillometry could tell you that some of those states involve more effort than others. EEG can tell you which type of effort, in which brain regions, on what timescale, and how it's shifting.

Consider a concrete example. You're writing code for two hours. During that time, your frontal theta shows two distinct peaks: one at the 20-minute mark and one at the 75-minute mark. Your parietal alpha is suppressed throughout, but it shows brief recovery periods every 15 to 20 minutes. Your overall theta/alpha ratio climbs gradually over the two hours, suggesting accumulating cognitive fatigue.

That's a detailed performance profile. You can see when you were in your deepest focus. You can see when your brain took micro-breaks (the alpha recovery periods). You can see the slope of fatigue. You can use that information to structure your work sessions differently tomorrow.

A pupil measurement during those same two hours would show... that you were working. Which you already knew.

The Question That Matters More Than the Method

Let's zoom out.

Both EEG and pupillometry exist because scientists have been trying to answer a deceptively simple question for the last 60 years: how hard is this brain working right now?

But the more interesting question, the one that matters for the future, is: what can you do with the answer?

If all you can do is confirm that something is effortful, you have an interesting scientific tool but not a very useful personal one. Confirming effort is like confirming that water is wet. You could already feel the effort. The measurement just added a number to it.

But if you can decompose that effort into its components, if you can see the theta surge that signals working memory engagement, the alpha suppression that reveals cortical recruitment, the shifting topography that distinguishes verbal from spatial processing, then you have something genuinely new. You have a window into the microstructure of your own thinking. And with that window, you can start doing things that were never possible before.

You can identify which types of tasks drain your cognitive resources fastest. You can find the time of day when your frontal theta response is strongest (that's your peak working memory window). You can detect the onset of cognitive fatigue before it shows up in your performance. You can train your brain, through neurofeedback, to sustain the neural patterns associated with focused, efficient cognition.

None of that is possible from a pupil measurement. Not because pupillometry is bad science. It's excellent science. But it's the wrong level of analysis for these questions. It's like trying to understand a symphony by measuring how loud it is. Volume is real information. But it's not the music.

Your brain is constantly broadcasting the music of its own cognition. The theta rhythms of working memory, the alpha rhythms of cortical activation and idling, the complex interplay between frequency bands that reflects the full computational state of your cortex. For the first time in history, you don't need a research lab to hear it. You just need the right antenna.

The question isn't whether cognitive load is measurable. Hess proved that with a camera and some math problems in 1964. The question is whether you're willing to listen to the full signal, or just the echo.