Five Layers, Not One Face
"Deep" and "superficial" are the two most-used words in injectable training and among the least examined. They are relational terms, and a relation needs something to be relative to. A practitioner who can say which layer they are in, and what lies immediately above and below it, is working from anatomy. One who can only say "deep" is working from a habit learned on a course.
This is the structural half of the anatomy the work requires. Its companion argues that vessels sit in a distribution rather than at an address, and ends without answering the obvious question: if the artery could be anywhere, what do you do with the needle. Part of the answer is here.
The layered model, and the part that gets dropped
The face is conventionally described in five layers: skin; subcutaneous fat with its fibrous retinacula cutis; the superficial musculo-aponeurotic system; deep fat; and periosteum or deep fascia [1].
That much is taught. The sentence immediately after it usually is not — that the arrangement varies between facial regions, particularly once the line of ligaments is brought into the model [1]. The scheme is a framework for reading a region, not a constant to carry from the temple to the lip.
Two of the five deserve more care than the shorthand gives them.
Layer three is a continuum, not one sheet. It changes name by region — galea in the scalp, temporoparietal fascia in the temple, SMAS in the cheek, orbicularis oculi around the eye, orbicularis oris in the lip, platysma in the neck. "Deep to the muscle" therefore describes different depths in different places.
Layer four is a glide plane, and its contents are the point. Cadaveric and histological study of the lower face describes a space that is rhomboidal, lined by membrane and reinforced by retaining ligaments, with a floor of thin dense connective tissue — and, critically, with the branches of the facial nerve passing beneath that floor rather than through the space [3]. That is why this space is described as a bloodless and safe plane to dissect. It is also the most useful thing the layered model gives an injector, and it is routinely lost when layer four is summarised as "the deep fat".
Fat is not a sheet
Subcutaneous facial fat is partitioned into compartments separated by septal boundaries rather than spread as a continuous blanket. The paper that established this in the aesthetic literature drew the conclusion that matters clinically: the face does not age as a confluent or composite mass, and shearing between adjacent compartments may itself contribute to soft-tissue malposition [2].
Boundaries and ligaments are related, though how is disputed. The compartment work proposes that some structures called "retaining ligaments" are simply fusion points where abutting septal barriers meet [2]. The ligament review disagrees that the two vocabularies describe one thing, treating the compartment literature's "septal boundary" as a different use of the same word [5]. The disagreement is worth knowing about, because it is the seam where two bodies of teaching are usually spliced without comment.
They are not, however, all the same kind of structure. The term "ligament" is generally reserved for a full interconnection between bone and skin, while "fascia" is used where the attachment relates to the parotid or masseteric fascia [5]. The zygomatic ligaments anchor cheek skin to the inferior border of the zygoma and the mandibular ligaments tether skin to the anterior mandible [4]; the zygomatic ligament runs from bone to dermis [5]. The masseteric ligament is less settled than most teaching admits: long described as arising from the masseteric fascia, its upper part has more recently been reported to originate from the maxillary buttress, and published accounts still differ on where along the muscle it arises [5].
The useful reading is narrower than "ligaments are septal fusions" and more useful than a list of names: the face is built with internal partitions, they differ in kind and in strength, and some of them are anchored to bone.
What the living imaging actually shows
Here the article has to part company with how facial ageing is usually taught, because the cadaveric account and the living-subject account do not agree.
A retrospective computed tomographic study compared repeat scans in 262 patients, mean age 46, median 56 months apart. Upper and middle compartments decreased in volume with age — and the inferior compartment, along with the total across the three compartments measured, increased [8]. A magnetic resonance study of living white women, stratified by age, imaged 100 faces and reported superficial temporal fat in 85 of them — around a third larger in old age than in youth [9].
So the robust finding is redistribution — upper and middle compartments shrinking while lower ones grow — and not the uniform deflation that "volume loss" implies. Total volume is genuinely contested, and by largely the same investigators: a smaller longitudinal magnetic resonance series found volume and thickness decreased in all three compartments [13], while the larger computed tomographic series from an overlapping author group reports that total rising [8]. Both agree on the direction of travel between compartments — width falling above and rising below.
That distinction matters at the chairside, because "the face empties, so refill it" is a different treatment logic from "the face redistributes, so decide which compartment you are addressing and why".
Ultrasound of 24 living volunteers, 48 midfaces between them, adds the mechanical half: during smiling the superficial midfacial compartment moved measurably in a cranial direction while the deep compartments essentially did not [10]. Superficial and deep are not a naming convention. They behave differently in a moving face.
The platform moves too
Soft tissue is not the whole account. Three-dimensional computed tomography of the bony orbit in 60 living subjects found orbital aperture width and area increasing with age, the superior rim receding medially in both sexes, and the inferior rim receding laterally in women and across its whole extent in men [6]. A compartment that has neither emptied nor moved can still sit lower and further back because the bone beneath it has withdrawn.
One caveat belongs with that finding rather than in a footnote. Those subjects were all white, and a three-dimensional computed tomographic study of 107 Koreans found no significant change in orbital aperture area with age in either sex — a result its authors explicitly contrast with the earlier white cohorts [7]. The skeletal ageing pattern most European courses teach as universal has not replicated in the one East Asian series that looked for it. For a practice treating a mixed population, that is not a small footnote.
Three regions where five layers is the wrong model
The clearest way to see that the scheme is a framework rather than a constant is to take it somewhere it does not apply.
The temple. Published layer counts here vary between authors, which is itself the point. What is not in doubt is that the region contains structures the five-layer scheme has no slot for: cadaveric study of Korean specimens found the middle temporal vein running between the superficial and deep layers of the deep temporal fascia, with a splitting and reuniting pattern in more than a quarter of cases [11]. An instruction to go "deep" in the temple does not name a plane; it names a choice between several, one of which contains a large vein.
The lip. There is no SMAS and no equivalent of layers four and five. The arrangement is skin, orbicularis oris, submucosa, mucosa. A practitioner asking "which layer" at the lip is asking a question the region does not answer.
The nose. Here the model does not merely fail — it inverts the habit built on it. Anatomical and clinical study of nasal vasculature found that, other than the lateral nasal veins, the major arteries, veins and lymphatics run superficial to the musculoaponeurotic layer, and concluded that dissection in the areolar plane below that layer preserves the vascular supply [12]. An injector carrying "superficial is safer" from the cheek to the nose has carried it into the region where it is most wrong.
What the evidence is, and is not
The compartment and layer literature rests substantially on cadaveric dissection, and the living imaging that exists is narrow. The bony orbit study is white-only [6], and the one series in another population found a different result [7]; the upper-face magnetic resonance series is all female and all white [9]; the ultrasound mobility study is 24 volunteers of one ancestry [10]. Sex, ancestry and age range constrain every one of them, and none was designed to test what happens when product is placed.
Nothing cited here establishes that filling a named compartment reproduces the appearance of a younger face. That claim is made constantly in marketing and is not supported by this literature.
One more limitation belongs to the reader rather than the studies. Every distribution described here was measured in faces that had not been treated. A patient booked for a fifth round of filler has scarred glide planes, spaces that no longer open cleanly, and boundaries that have been crossed repeatedly under pressure. The layered model describes the face they were born with.
What this changes in the consultation
Name the plane by region, not by adjective. "Deep" in the midface means supraperiosteal, beneath the plane the nerve branches occupy, against a hard floor — the plane the mobility study concluded its findings justify for the deep compartments [10]. "Deep" in the temple means a choice among fascial planes, one containing a vein. "Deep" at the lip means very little. A protocol that uses the same word in two regions is describing a technique rather than an anatomy, and it will transfer only as far as the anatomy happens to agree.
Ask which change you are looking at before reaching for volume. The living imaging says compartments redistribute — upper and middle down, lower up [8][9]. Migration and volume loss are not the same problem and do not have the same answer, and the literature does not establish that either is corrected by filling.
Work in the planes the anatomy already provides. This is where the arterial companion's open question gets an answer. The defence against a vessel that could be anywhere is not steering around a remembered line but choosing a plane with a floor and few structures crossing it — which is what the premasseter space was shown to be [3]. The caveat travels with it: that was open dissection under vision, and no study cited here tests whether the same plane protects a needle.
Treat boundaries as resistance, not as walls. Septa and ligaments are the structural reason a deposit stops where it stops, but what was shown is that boundaries confine diffusing dye in cadaveric specimens [2] — not how product behaves in living tissue under injection pressure. When product does not go where you expected, a boundary you did not account for is still the first hypothesis.
Ask whether the study population looks like your patient. The orbital finding that did not replicate in Korean subjects [7] is the clearest single reason to hold the standard ageing account loosely, and it applies to more of this literature than anyone teaching it usually says.