What Brain Imaging Can and Cannot Tell Us About Awareness

Article ID: DBK-068
Article Type: Deep Dive
Role: Signature
Evidence Label: Supported
Subject World: Consciousness & Reality
Subject Area: Nature of Consciousness
Collection: The Problem of Consciousness

A brain image is not a photograph of consciousness

Modern neuroscience can observe the living brain with extraordinary precision.

Functional MRI can track changes related to blood oxygenation. EEG can record electrical activity from the scalp with high temporal resolution. MEG measures magnetic fields generated by neural activity. Intracranial EEG records directly from electrodes placed inside the skull in selected clinical contexts.

These tools have transformed consciousness research.

But the phrase "brain imaging shows consciousness" can be misleading.

None of these methods produces a photograph of subjective experience.

They measure physical signals that correlate with neural activity, and researchers infer relationships between those signals, behavior, task performance, and states of awareness.

That distinction is central.

What fMRI measures

Functional magnetic resonance imaging, or fMRI, often relies on the blood-oxygen-level-dependent signal.

Active neural tissue changes its metabolic demands. Local blood flow and oxygenation change in response. The scanner detects those hemodynamic differences.

The result is powerful spatial information about patterns associated with a task or state.

But it is indirect.

The scanner does not detect "redness," "painfulness," or "awareness" as such.

It detects a physiological signal that researchers model in relation to experimental conditions.

This matters because interpretation always depends on the design of the study.

What EEG and MEG add

EEG and MEG provide much finer information about timing.

Neural processing unfolds over milliseconds, while the hemodynamic signal measured with fMRI is slower.

EEG captures voltage differences at the scalp.

MEG records magnetic fields associated with electrical currents in the brain.

Both can reveal rapid patterns that help researchers ask when information becomes globally available, when sensory processing diverges between conscious and unconscious perception, and how brain states change across sleep, anesthesia, or injury.

Intracranial EEG can provide even more direct and spatially precise electrical recordings, but only in limited clinical populations where electrodes are medically justified.

Each method therefore sees a different part of the picture.

Consciousness research needs converging measures

No single signal is a universal consciousness meter.

Researchers often combine:

• behavioral response;
• verbal report;
• task performance;
• fMRI;
• EEG;
• MEG;
• intracranial recordings;
• physiological measures;
• carefully controlled stimulus conditions.

The goal is convergence.

If several independent measurements change systematically with conscious perception, confidence rises that the underlying process is relevant.

But relevance is not identity.

A neural correlate of consciousness is not automatically the complete explanation of consciousness.

Recent theory testing shows both progress and limits

In 2025, a large international adversarial collaboration directly tested predictions from two influential neuroscientific theories: Integrated Information Theory and Global Neuronal Workspace Theory.

The study involved 256 participants and used fMRI, MEG, and intracranial EEG.

Researchers preregistered predictions before the final analyses and used theory-neutral teams to reduce confirmation bias.

The results were valuable precisely because they were not a simple victory for either camp.

Some findings aligned with predictions from both theories, while other findings substantially challenged important tenets of each.

That is what strong science often looks like.

A theory does not have to collapse completely for evidence to force revision.

And a brain-imaging experiment does not need to "solve consciousness" to make meaningful progress.

Brain data can test predictions, not metaphysics by itself

Suppose one theory predicts that conscious content should be sustained primarily in posterior cortical regions.

Another predicts characteristic global broadcasting involving prefrontal systems.

Those claims can be compared with data.

If the predicted pattern does not appear, the theory faces pressure.

This is empirically powerful.

But the same experiment does not automatically answer:

• Why does neural activity feel like anything?
• Is consciousness fundamental or emergent?
• Is physicalism true?
• Is panpsychism true?
• Is phenomenal consciousness illusory?

Those are broader philosophical questions.

A neural theory may inform them. It does not simply replace them.

The clinical case makes the distinction even more important

Disorders of consciousness provide one of the clearest examples of both the power and the limits of neural measurements.

Some patients with severe brain injury do not visibly respond to verbal commands.

Yet a person may still follow instructions internally in ways that can sometimes be detected through fMRI or EEG.

A large 2024 New England Journal of Medicine study examined 353 adults with disorders of consciousness across six international centers.

Among 241 participants who showed no observable response to commands, 60 showed evidence of performing a cognitive task on fMRI or EEG.

That phenomenon is called cognitive motor dissociation.

The finding has major clinical and ethical importance because behavior alone can underestimate preserved cognitive capacity in some patients.

But the result must be stated carefully.

Detection is meaningful; non-detection is not the mirror image

If a patient reliably follows an instruction through a task-based neural signal, that is strong evidence that more cognitive processing is present than bedside behavior alone revealed.

But failure to detect such a signal does not prove absence of consciousness.

Why not?

Because a patient may fail the task for many reasons:

• impaired hearing or language comprehension;
• fluctuating arousal;
• memory limitations;
• inability to sustain attention;
• technical noise;
• movement;
• damage affecting the required neural response;
• fatigue;
• limitations of the chosen paradigm.

A test can therefore have meaningful positive evidence without having perfect sensitivity.

This is a general lesson for consciousness science:

absence of a detected marker is not automatically evidence of absent experience.

State markers are not content readers

Research can also distinguish broad states.

A 2024 Communications Biology study examined dynamic relationships between brain structure and functional connectivity across wakefulness, anesthesia, and slow-wave sleep. The authors identified brain-dynamic patterns that generalized across states and could serve as signatures related to level of consciousness.

This is useful.

But a marker that helps distinguish waking from unconscious states does not tell us exactly what a person is experiencing.

Level of consciousness and content of consciousness are different problems.

A system might be awake without seeing a particular image. A patient might have internal awareness without being able to report it. A neural state can be compatible with multiple subjective contents.

Decoding is not mind reading

Neuroimaging studies sometimes reconstruct or classify information about what a participant is seeing, hearing, imagining, or intending.

Headlines may call this "mind reading."

That phrase overstates the result.

Decoding systems depend on:

• training data;
• predefined task categories;
• statistical regularities;
• known experimental conditions;
• individual or group calibration;
• limited output spaces.

A model can classify which of several known stimuli best matches a neural pattern without gaining unrestricted access to a person's private inner life.

As decoding improves, ethical questions will become more serious.

But technical capability should still be described accurately.

Correlation, mechanism, and explanation

There are at least three levels of claim.

Correlation

A neural pattern changes when awareness changes.

This is an empirical relationship.

Mechanism

A particular process is causally necessary or sufficient for some aspect of conscious access or experience.

This requires stronger experimental evidence.

Ultimate explanation

The mechanism explains why subjective experience exists at all.

This is the strongest claim, and it is not automatically established by either correlation or causal intervention.

Keeping these levels separate prevents brain science from being both undervalued and oversold.

What brain imaging can tell us

With appropriate experimental design, brain measurements can help researchers:

• identify neural correlates of conscious perception;
• compare waking, sleep, anesthesia, and injury states;
• test competing predictions from theories;
• detect some forms of covert command following;
• investigate temporal and spatial dynamics;
• examine how attention, report, memory, and perception interact;
• constrain theories that make neural predictions.

These are substantial achievements.

What it cannot currently tell us

Brain imaging cannot currently provide:

• direct access to raw subjective experience;
• a universal yes/no consciousness detector for every patient;
• proof that non-detection equals unconsciousness;
• a complete translation from neural signal to private mental content;
• a final answer to the hard problem;
• empirical proof of a metaphysical worldview.

Those limits do not reduce the value of neuroscience.

They define the boundary between what the measurement supports and what we would merely like it to support.

DarkBrain assessment

Brain imaging is one of the strongest tools available for studying the neural conditions associated with awareness.

Recent multimodal research shows that competing theories can be tested rather than protected indefinitely from evidence. Clinical work shows that neural measures can reveal cognitive responses that behavior alone may miss.

At the same time, every signal is mediated by a method, a task, a model, and an interpretation.

The disciplined conclusion is neither "the scanner reads consciousness" nor "brain imaging tells us nothing about experience."

It is:

brain measurements reveal increasingly precise relationships between neural activity and awareness, while subjective experience itself remains an inference rather than a directly imaged object.