Read this topic

Basic Neurosciences · Paper A

Salience network

The idea

Your brain runs (at least) two large-scale networks that cannot both be in charge at once, and a third network's whole job — on the leading model — is deciding which one gets the microphone. When you are staring out of a window, mind wandering over yesterday's conversation, your default mode network is running the show; the moment your phone buzzes with something that actually matters, control has to hand over to your central executive network so you can act on it. The salience network — built around the anterior insula and dorsal anterior cingulate cortex — is proposed to be the switch that makes that handover happen, by continuously scanning for what is behaviourally relevant, inside you and outside you, and flipping control accordingly. Nothing about this switching role has been directly proven rather than inferred from patterns of brain activity moving together, so hold it as the field's best current model, not as an established fact you would state without qualification.

How it works

The salience network is defined less by what it does on its own and more by the traffic it is thought to direct between two other networks. The default mode network — its core nodes are the medial prefrontal cortex, the posterior cingulate cortex/precuneus, and the angular gyrus — becomes most active when you are not engaged with the outside world: mind-wandering, recalling memories, thinking about yourself. The central executive network — its core nodes are the dorsolateral prefrontal cortex and posterior parietal cortex — becomes most active during effortful, externally directed tasks such as holding several numbers in mind while you work through them. These two networks are anticorrelated: when one is strongly active, the other typically quietens, which is part of why you cannot deeply mind-wander and hold a demanding working-memory task at full capacity at the same moment. The salience network's proposed job is to detect which of the two states the situation actually calls for and trigger the switch — the anterior insula flags a stimulus as behaviourally important, and the dorsal anterior cingulate cortex helps recruit the attentional and control resources that follow. Why the insula specifically: it receives converging input from bodily-state signals (interoception) and from external sensory streams, which is why it is well placed to compare "does this matter to me right now" across both internal and external information at once. This whole account rests on correlational functional-imaging evidence — you should treat "the salience network switches you between networks" as the field's working hypothesis, not as a mechanism watched happening in real time inside a single neuron.

How to spot it

A stem pairing the anterior insula and dorsal anterior cingulate cortex as the nodes detecting behaviourally relevant stimuli is describing the salience network.
A stem pairing the medial prefrontal cortex, posterior cingulate cortex/precuneus and angular gyrus as the nodes active during mind-wandering or self-referential thought is describing the default mode network.
A stem pairing the dorsolateral prefrontal cortex and posterior parietal cortex as the nodes active during an effortful, externally directed task is describing the central executive network.
A stem describing control shifting from internally directed thought to an externally directed task in response to something behaviourally important is describing the proposed switching role of the salience network between the other two networks.

Tell it apart

Salience network vs default mode network — the salience network's core nodes are the anterior insula and dorsal anterior cingulate cortex and it is proposed to trigger switching; the default mode network's core nodes are the medial prefrontal cortex, posterior cingulate cortex/precuneus and angular gyrus and it dominates internally directed, task-negative states — task-negative meaning active when you are not doing an external task.
Salience network vs central executive network — the salience network detects what matters and initiates a switch; the central executive network (dorsolateral prefrontal cortex, posterior parietal cortex) is the network that then actually executes effortful, externally directed cognitive work once switched in.
The salience network's insula/ACC nodes vs the mesolimbic dopamine pathway — the salience network is a large-scale functional-connectivity construct built from resting-state imaging, whereas the mesolimbic pathway is a specific anatomical neurotransmitter circuit; the two are proposed to interact (aberrant dopaminergic salience is a leading model in psychosis) but are not the same thing and should not be named interchangeably.
The triple-network switching model vs a settled diagnostic finding — the switching role is inferred from correlated activity patterns across studies, not confirmed by any single test that shows the switch happening, so it should be described as a model, not as an established mechanism.
Default mode network activity vs simple "doing nothing" — the default mode network is not the brain switched off; it is actively running a specific, internally directed mode of processing, which happens to be what dominates when you are not engaged in a demanding external task.

Watch for

⚠️ Stating the switching role of the salience network as a settled, directly observed mechanism rather than the field's leading model built from correlational connectivity data.
Swapping the node lists — assigning the anterior insula and dorsal anterior cingulate cortex to the default mode network, or the medial prefrontal cortex and posterior cingulate cortex to the salience network, is one of the exam's favourite pairing traps.
Assuming the default mode network is simply "the brain at rest" with no active function, rather than an actively engaged, internally directed processing mode with its own defined nodes.
Treating "salience network" and "mesolimbic dopamine pathway" as two names for the same thing because both involve the word "salience" — they are different levels of description (large-scale network vs specific neurotransmitter circuit).
Assuming salience network dysfunction is itself a diagnosis — it is a proposed contributing mechanism discussed across several conditions, not a clinical entity with its own diagnostic criteria.

If you catch yourself stating any of these five as settled fact rather than as a model, stop and re-check which network you actually mean.

Use it again

First identify whether the stem is naming node anatomy (insula/ACC vs medial PFC/PCC vs DLPFC/parietal) or describing a function (detecting relevance vs mind-wandering vs effortful task performance) — then match nodes to nodes and functions to functions separately, rather than mixing the two.
If the stem describes something being flagged as important and attention then shifting, default to the salience network's proposed switching role rather than to either of the two networks it switches between.
If the stem describes an internally focused, task-negative state, default to the default mode network; if it describes an effortful, externally focused task, default to the central executive network.
Never let the word "salience" alone pull you toward the mesolimbic dopamine pathway — check whether the stem is describing a large-scale imaging network (salience network) or a specific neurotransmitter circuit (mesolimbic pathway) before answering.
Hold the switching mechanism itself as a model when a stem asks how confidently it is established — do not answer as though it were a directly demonstrated mechanism.

In detail

Hook — a director for two networks that cannot both talk at once

Right now, part of your brain is either quietly running an internal narrative — replaying a conversation, planning what to have for dinner — or it has locked onto something outside you that actually demands effort, like working through a calculation. Those two modes do not comfortably run at full strength together, and something has to decide, moment to moment, which one you are actually in. Cognitive neuroscience calls that decision-maker the salience network, and this page builds, from nothing, what that means and how confidently you should believe it.

Plain definition — three networks, and what "network" means here

A "network" here does not mean a single brain region — it means a set of brain areas whose activity rises and falls together, discovered by watching functional MRI scans while people rest or perform tasks. Three such networks matter for this topic. The default mode network (medial , posterior cingulate cortex/precuneus, angular gyrus) is most active when you are not engaged with the outside world. The central executive network (dorsolateral prefrontal cortex, posterior parietal cortex) is most active during effortful, externally directed tasks. The salience network (anterior insula, dorsal anterior cingulate cortex) is proposed to detect what matters, inside you or around you, and to help hand control between the other two.

The critical caveat — a model, not a mechanism you can watch

Here is the trap almost every candidate falls into: treating the salience network's switching role as an established fact, the way you would treat "the heart pumps blood". It is not at that level of certainty. It is inferred from the observation that these three networks' activity patterns correlate and anticorrelate across many scans and many people, not from watching a single switch flip inside one brain in real time. Hold this as the field's leading working model — genuinely useful, widely cited, but a model — and you will not be caught out by a question testing exactly this distinction.

First principles — why an insula-based network is well placed to do this job

Start with what the anterior insula actually receives. It is one of the few cortical areas that gets dense input from your body's internal state — heart rate, gut sensation, breathing effort (this is called interoception) — and, separately, dense input from external sensory streams. Sitting at that junction, it is structurally well placed to compare "how much does this matter to me right now", combining both internal and external information into one relevance signal.

The dorsal anterior cingulate cortex, working alongside it, is heavily connected to the brain's attention and control systems, so once the insula flags something as relevant, the ACC helps recruit the resources needed to actually respond. Put those two together and you have a plausible reason why THIS pair of structures, rather than some other pair, ended up carrying the proposed switching role — not because it has been proven to switch anything, but because its wiring makes it a sensible candidate.

Structured walk-through — the anticorrelation, step by step

Here is the finding that makes the whole model hang together. Across resting-state scans, when the default mode network's activity is high, the central executive network's activity is reliably low, and vice versa — the two move in opposite directions closely enough that researchers describe them as anticorrelated. Picture it as a simple seesaw: if the default mode side goes up by a given amount, the central executive side tends to go down by a comparable amount, over and over, scan after scan. The salience network's activity, meanwhile, tends to rise just before a switch from one side of the seesaw to the other — which is the observational basis for calling it the network that triggers the switch, rather than simply one more network that happens to also be active.

Look-alikes & differentials — four things a stem could be describing

Before you commit to an answer, run the stem past this table and see which row it actually matches.

Looks like the salience network	How to tell it apart
Default mode network	Different nodes entirely (medial , posterior cingulate cortex/precuneus, angular gyrus) and dominant during internally directed, task-negative states (task-negative meaning active while you are not doing an external task)
Central executive network	Different nodes (dorsolateral prefrontal cortex, posterior parietal cortex) and dominant during effortful, externally directed tasks, not during detection or switching itself
Mesolimbic dopamine pathway	A specific circuit ( to ), not a large-scale cortical connectivity network, even though both are linked to the idea of "salience"
Simple attention or arousal	The salience network is a specific, anatomically defined network with two named core nodes, not a general term for being alert

Diagnosis — a group-average research finding, not a patient's result

You cannot order a test that confirms "this patient's salience network switched correctly" in routine care; the concept exists at the level of group-averaged research findings from functional MRI, not individual clinical assessment. Where clinicians reach for the model at all, they reach for it indirectly — for example, the aberrant-salience account of psychosis links a different, better-established finding ( dysregulation in the ) to the idea that ordinary stimuli become misattributed with importance, and the salience network is one candidate cortical system proposed to be involved in that broader story, not a stand-alone diagnostic finding you would look for on its own.

Management — treating the dopamine finding, one step below the model

Because the salience network is a research construct rather than a diagnosable condition, there is no treatment aimed at it and no guideline that names it. What follows from the broader model, in current practice, is indirect: treatments that reduce mesolimbic overactivity (antipsychotics) are the actual intervention used when aberrant salience is thought to be contributing to psychotic symptoms, and that decision rests on established psychosis guidance (NICE CG178), not on any test of the salience network itself. That is one step removed: the intervention treats the dopamine finding, not the network model built on top of it.

Exam pearl — pair the nodes, then grade the certainty

The question almost always tests one of two things: whether you can correctly pair anatomical nodes with the right network, or whether you can correctly identify a scenario as "detecting relevance and switching" rather than either of the two states it switches between. As a check on your own understanding, work out why daydreaming and doing mental arithmetic at full effort do not comfortably happen at the same moment — and which network you would blame if someone genuinely couldn't tell the two apart.

In one line — the proposed switch, and how sure we are

Remember it this way: your salience network (anterior insula, dorsal anterior cingulate cortex) is proposed — as the field's leading model, not a proven mechanism — to detect what is behaviourally relevant and trigger switching between your internally directed default mode network and your externally directed central executive network.