Why Your Brain Decides What You're Seeing Before You See It

Perception feels like collecting information: light reaches your eyes, sound reaches your ears, your brain puts the pieces together and you experience the result. The actual order is closer to the reverse, with your brain generating a prediction of what's out there and using the incoming signal mainly to check and correct it.

The framework is usually called predictive processing, and it has become one of the more influential ideas in neuroscience because it explains so much of ordinary experience once you know where to look.

A brain that waited for the data would always be late

Raw sensory data is noisy, incomplete and delayed, so assembling everything from scratch would leave you operating slightly in the past, which is a poor arrangement for catching a ball or crossing a road.

Running a model avoids that problem, because your brain keeps an ongoing prediction of what's likely and attends mostly to the difference between what it expected and what arrived, a gap known as prediction error, and only that gap gets passed up for further processing. Most of what surrounds you at any moment is unsurprising, and analysing it in full would waste effort.

You can catch it happening

• You read straight past typos because your brain supplies the word it was expecting
• A song lyric you misheard once keeps sounding wrong even after you've learned the real one
• Visual illusions mostly work by exploiting assumptions your visual system makes automatically
• You hear your own name across a noisy room, where the signal was weak but the prediction was strong
• An unfamiliar accent is hard to follow at first and much easier after a while, as your model updates

The subtitle effect is the demonstration most people have lived through. Switch subtitles on for a muffled line of dialogue and you suddenly hear it clearly, even though the audio hasn't changed and only your prediction has.

Expectation reaches well beyond what you see

The same mechanism appears to sit behind placebo effects, where expecting relief produces measurable physical changes, including release of the body's own opioids, which drugs that block opioid receptors can cancel out. The expectation is doing real physiological work and isn't just changing what someone reports.

The effect also runs in the other direction, since being told a treatment may cause side effects raises the chance of experiencing them, and it's well enough established to have its own name, the nocebo effect.

For anxiety, the model is the problem

If your brain runs largely on predictions, a model that expects threat will produce experiences consistent with threat. Prediction explains several things about anxiety that otherwise look puzzling: ambiguous situations get read as hostile because the model resolves uncertainty toward what it expected, neutral bodily sensations get taken as danger signals in a way that sits at the centre of how panic escalates, and a blank face across the table reads as disapproval.

None of that is deliberate misreading, and all of it is the model filling gaps the way it always does.

The same logic explains why reassurance achieves so little. A single "it's fine" can't shift a model built from years of accumulated experience, and models only update through repeated encounters with evidence that contradicts them, which is what exposure-based treatment is designed to provide. You gather enough prediction error that the model has to revise itself.

Pain turns out to be a prediction as well

Chronic pain is one of the most significant applications. Pain was long treated as a direct readout of tissue damage, with the amount of pain reflecting the extent of the injury, and that model never fitted the evidence well, since people can have severe pain with no detectable damage and significant damage with very little pain.

Predictive processing offers a better account in which the brain produces pain as an assessment of threat to the body, drawing on sensory input along with expectation, context and prior experience. The pain is entirely real, and it's an output of the system rather than a simple measurement of damage.

The shift has changed treatment for persistent pain, since approaches that explain the mechanism and gradually rebuild confidence in movement work partly by revising the predictions driving the pain. Anxiety and low mood reliably make pain worse for the same reason, because both push the model toward expecting threat.

Hold your first impressions loosely

Your experience of any situation is partly constructed from what you brought to it, which is worth holding lightly rather than treating as a reason to distrust everything. Strong first impressions deserve a little scepticism, and two people can leave the same conversation having genuinely experienced it differently.

If your predictions have settled somewhere unhelpful and situations reliably feel more threatening than they turn out to be, that pattern responds well to treatment. Our psychologists work with anxiety and panic at our St Kilda clinic and via telehealth, or you can read more about how we approach therapy.