The nervous system doesn't register everything equally. It registers what matters — and what matters is determined by a layer of meaning that runs below conscious thought. Understanding salience is understanding why some things get in and others don't.
Not everything you encounter registers. Most of what arrives at your senses is processed, filtered, and discarded without ever reaching awareness. What rises to the level of being noticed — what gets treated as signal rather than noise — is salience. And salience is not neutral. It is not determined by intensity alone, or by objective importance. It is determined by a meaning system that assigns relevance based on learned predictions about what matters.
Understanding how salience works — and how it can go wrong — is essential to understanding why some things are impossible to ignore while others, equally present, seem to barely exist.
In the predictive processing framework, salience is essentially the precision weighting assigned to a signal. When the system predicts something is relevant — worth updating predictions about, worth attending to — it increases the gain on that signal. The signal is amplified. It rises to awareness. It recruits attention, action tendency, and interpretive processing.
This precision weighting is not determined by the signal itself. It is determined by the system's model of what is important. That model is built from prior experience, from current goals, from chronic threat patterns, and from learned meanings about what kinds of things typically require a response.
You notice what you've learned to notice. Your attention is not a neutral spotlight — it is a prediction about what deserves to be seen. The system amplifies the signals it has learned to treat as important and dampens everything it has learned to treat as background.
Meaning, in the technical sense being used here, is what the nervous system assigns to an incoming signal: what this signal predicts, what category it belongs to, what typically follows it, what it implies about the current situation. Meaning is not added consciously. It is generated automatically, at the conceptual layer, as part of the process of making incoming information interpretable.
This means that salience is downstream of meaning. Something becomes salient because the system has assigned it a meaning that predicts it matters. A particular facial expression registers not because of its physical properties — the arrangement of muscles — but because the system has categorized it as belonging to the concept "disapproval," which predicts negative social consequences, which triggers elevated attention and arousal. The meaning generates the salience. The salience generates the felt significance.
The salience system is calibrated to an organism's history. This is appropriate and adaptive — things that have been meaningful in the past are likely to be meaningful in the future, and it makes sense to be especially attentive to them. The problem is that historical calibration does not automatically update when circumstances change.
A person whose early environment was characterized by chronic unpredictability and threat will have a salience system calibrated for detecting subtle signs of danger: shifts in tone, changes in expression, deviations from routine. In that original environment, this calibration was adaptive. In a current environment that is genuinely safe, the same calibration produces persistent hypervigilance — the system is still amplifying signals that the current context does not require elevated attention to.
Threat oversalience: Neutral or ambiguous stimuli are assigned threat-consistent meaning and therefore elevated salience. Everything potentially dangerous gets noticed. Genuine safety cannot register because the filter is set to detect threat.
Reward undersalience: In depression, the system's model of what is worth attending to collapses. Things that previously generated salience — social connection, pleasure, future goals — lose their precision weighting. They are there, but the system no longer treats them as worth noticing. This is anhedonia, mechanistically understood.
Hyperspecific salience: In trauma, very specific cues — particular sounds, smells, interpersonal dynamics — have been assigned extreme precision weighting because of their historical significance. The system amplifies these signals dramatically, producing intrusive recall or acute reactivity, even when the current context shares only superficial similarity to the original.
Here is a counterintuitive consequence of understanding salience as meaning-dependent: analytical deconstruction of meaning can reduce salience — and not always in a useful way. When the meaning assigned to something is dissolved through analysis ("it's just a pattern," "this is just neurons firing," "there's no objective reason this should matter"), the salience can collapse with it. The thing stops getting in.
This is sometimes exactly what is needed — when a meaning is distorted, when salience is miscalibrated to historical threat, dissolving the meaning can reduce unnecessary activation. But when the meaning that is being dissolved is accurate and needed — when the salience is pointing at something real — this becomes a problem. It is possible to analyze your way out of genuine experience, to intellectualize salience into silence, and to lose contact with real signals in the process of trying to understand them.
This is one reason why purely analytical approaches to emotional life have limits: the analysis tool can flatten salience, and flat salience means flat experience. The goal is not to eliminate meaning but to calibrate it — to bring the salience system's precision weighting into closer alignment with current reality rather than historical prediction.