Exposure Hierarchy: What It Is and How to Build One

What is an exposure hierarchy?

An exposure hierarchy is an ordered list of concrete situations the patient is going to face in a predictable sequence, from the most accessible to the most demanding. That is as far as the textbook definition goes. What the definition hides is that each rung is not a situation but a combination of values: which stimulus, at what distance, for how long, with whom present and with how much advance information. The column the patient sees on the sheet is the visible face of a table of combinations sorted by one of its columns.

The purpose of that ordering has changed over the last two decades, and it is worth saying so plainly. In the classical framing, the list existed so that arousal would come down inside each rung before moving up to the next. In the inhibitory learning and retrieval framework described by Craske and colleagues, extinction does not erase the original fear association: it builds a competing association that then has to be made retrievable at the moment it is needed. What an exposure hierarchy orders, then, is not tolerable suffering but the sequence of predictions the patient will be able to disconfirm. Each item exists because there is a concrete expectancy attached to it.

Habituation does not drop out of the picture. It remains an observable phenomenon inside the session and a useful sign that the rung is genuinely being worked. What no longer carries the weight of the design is its predictive value: a fall in the rating inside a single session is a poor guide to the medium-term outcome. A column built to maximise that fall and a column built to maximise the discrepancy between what was predicted and what occurred look very similar on paper and differ at every rung.

The tool should also be told apart from its ancestor. The ordered list of situations was born with Wolpe's systematic desensitisation, paired with progressive relaxation on the logic of reciprocal inhibition. Contemporary exposure inherited the ordering and discarded the pairing, because relaxation stopped being regarded as the active ingredient. Anyone building an exposure hierarchy today is working with a skeleton from 1958 and with a theory of change different from the one that produced it.

How do you build an exposure hierarchy?

Building an exposure hierarchy breaks down into five movements. They are not phases of treatment but workbench operations, carried out on paper, with the patient present, and before the headset comes into play.

1. Take an inventory of the items. Collect the raw material: situations the patient avoids, tolerates with effort, or faces only under very specific conditions.
2. Choose the dimensions of difficulty. Decide which parameters are going to move along the column and which are held still.
3. Rate and order. Turn the inventory into a column governed by the difficulty the patient anticipates.
4. Fix the number of levels and the jump between levels. Decide how many rungs there are and how much distance sits between two consecutive ones.
5. Define the ceiling and the review schedule. Establish where the hierarchy ends and when it gets looked at again with a pencil in hand.

The first movement is the one most often skimped on and the one that costs most. The inventory does not come out of a manual or a list of typical situations: it comes from the avoidance record of this case, from what the patient has stopped doing and from the conditions under which he or she still does it. Those conditions are gold, because they flag in advance which modifiable parameter is holding the problem up.

Three requirements make an exposure item usable. It must be observable, in the sense that two clinicians would see the same thing; reproducible under the real conditions of the practice; and completable within one session. “Getting over my fear of dogs” is not an item. “Staying three minutes at two metres from a medium-sized dog on a lead, with the owner present” is one. The item is written as a scene, with a subject, a distance and a duration of contact.

It pays to over-inventory and cut back afterwards: twenty or thirty candidates leave room to choose, whereas a badly spread dozen forces improvisation in the sixth session. Where the material is going to be presented in virtual reality, the inventory is also crossed against what the environment can represent. In the dog phobia exposure scenario in the catalogue, for instance, the size of the animal, the lead, the distance and the barking are independent parameters, and that decides which items are buildable and which stay reserved for in vivo work.

Step 2: which dimensions of difficulty get modulated

An item becomes difficult for concrete reasons, and those reasons are the dimensions of difficulty of the case. Identifying them is the operation that separates a usable exposure hierarchy from a list put in rough order. Most presentations admit five or six dimensions, and almost none of them orders well on a single one.

• Distance to the stimulus. The most obvious axis and the first to run out, because there comes a point at which getting closer no longer changes the patient's expectancy.
• Duration of contact. How long the task is held. It modulates well at the intermediate rungs and badly at the first ones.
• Presence of a companion. A textbook safety signal, powerful and convenient to withdraw in stages: the clinician inside the environment, the clinician outside but audible, the patient alone.
• Magnitude or number of stimuli. One dog or four, one interlocutor or a full auditorium, one syringe or a whole tray.
• Predictability. Knowing how long the task will last and what will happen inside it lowers anticipated difficulty more than almost any other parameter.
• Escape route. Whether the patient keeps the option of stopping, and under what conditions it can be exercised.

When two parameters move independently, the column stops being a single file and becomes a mesh. One item can be designed with minimal distance and maximal predictability, and another with generous distance and no information at all about duration. A multidimensional exposure hierarchy is more laborious to build and considerably harder to exhaust halfway through treatment. The claustrophobia lift scenario is the clearest case to look at, with cabin size and the length of the stoppage working as two axes at the same time.

The question that governs the whole step is which of the parameters actually holds the problem up. It is answered by moving one and leaving the rest still, which is also the reason a jump between levels should never move two at once.

Step 3: rating the items and ordering the column

The rating is given by the patient and recorded by the clinician. It is asked for in anticipatory mode: how much anxiety the patient expects to feel on facing that item tomorrow, on whatever scale of subjective units of distress the practice is already using. The number matters less than its nature. An anticipated rating is a prediction, and a prediction is the only thing a task can disconfirm.

The procedure is deliberately austere. The item is read aloud exactly as it was worded, the figure is written down, and the date is written down. Nothing else. This is not the place to argue about the number or to correct a patient who rates an apparently comfortable item high, because that discrepancy is the most valuable finding of the whole step.

Inversions turn up in almost every case, and almost always for the same reason: the clinician was ordering by the wrong dimension. A patient afraid of dogs may rate a loose puppy above a mastiff on a lead, and then the parameter governing the case is not the magnitude of the stimulus but the predictability of its behaviour. When the column is ordered and an item sits out of place, the previous step is revisited before anything else moves forward.

It is worth keeping a notebook with the literal wording of each item, its anticipated rating and the date of each reading. Reconstructing from memory in session eight is the fast route to an exposure hierarchy that no longer describes the patient who signed it. The checklist for a virtual reality practice includes the recording section that holds that notebook up.

Step 4: number of levels and the jump between levels

These are two decisions with a single logic behind them. The number of levels and the jump between levels are the same variable seen from its two ends: how many rungs fit between the most accessible item and the ceiling. For a monothematic specific phobia, a column of six to ten rungs usually covers the ground without leaving gaps the patient cannot absorb. Where two themes coexist in the same case, the sensible move is not to lengthen the column but to raise two parallel hierarchies with ceilings of their own.

Fine graduation is not a universal requirement either, and that is worth holding in mind before manufacturing fifteen rungs. The single-session treatment Öst described for specific phobias works with prolonged exposure and therapist modelling in one long encounter, and it produces learning without an extensive column behind it. The reasonable reading is not that the hierarchy is redundant, but that the number of levels is a clinical decision rather than a constant of the procedure.

The size of the jump is calibrated by the prediction and not by the clock. Too short a jump generates no discrepancy at all: the patient predicts accurately what is going to happen, is proved right, and the task adds no new information to the competing association. Too long a jump does the symmetrical damage: the item is abandoned halfway and what the task teaches is that avoidance worked. As a starting point, a difference of ten to fifteen points of anticipated difficulty between consecutive rungs leaves reasonable room in both directions.

From there comes the operational rule that pays off most in supervision: one jump moves one modifiable parameter. If the next rung shortens the distance to the stimulus, the duration of contact and the presence of a companion all at once, whatever learning occurs will not be attributable to anything in particular. The acrophobia protocol with its six-level height hierarchy illustrates the minimal version of that discipline, with height and the perceived protection of the edge kept apart as distinct parameters.

Step 5: the ceiling of the exposure hierarchy and its review

The ceiling of the exposure hierarchy is the highest rung that is going to be worked, and choosing it is a clinical decision rather than a technical one. It is set by the patient's life goal, not by the maximum the equipment can represent. Flying to Lisbon with a small child at Easter is a ceiling. “Tolerating the most intense scenario in the catalogue” is not, because it gives the patient back no part of his or her life.

The two errors are symmetrical. A low ceiling produces premature discharges and relapses that arrive with the first real situation sitting above the column. An unreachable ceiling produces a rung that is never attempted and a hierarchy the patient ends up reading as a verdict. The check is simple: if the ceiling item does not appear on the list of things the patient wants back, it has been chosen badly.

Reviewing the exposure hierarchy closes the workbench and reopens it every few sessions. Items get cheaper as treatment advances, so what was rated eighty may be rated forty six weeks later, and a column that is never re-rated stops discriminating. New items appear as well, because dismantling one avoidance uncovers the ones underneath it. A cadence of every four or five sessions works well, and each reading deserves the same discipline applied to recording outcome assessment instruments in the practice.

One closing frame, so that the part is not mistaken for the whole. The NICE CG113 guideline places exposure as a component of a cognitive behavioural treatment and not as a self-sufficient intervention, and the exposure hierarchy inherits exactly that status. It is an instrument for programming inside a wider plan, and the plan rests on the judgment of the licensed clinician in charge of the case.

Design errors that send the exposure hierarchy back to the workbench

What follows is the inventory of failures that surface in supervision, ordered by frequency rather than by severity. None of them is exotic and all of them are repairable on paper, provided they are spotted before the fourth session.

• A borrowed column. The hierarchy is copied from a manual or a published protocol and applied without being crossed against the avoidance record of the case. It puts an average patient who does not exist into impeccable order.
• A single dimension of difficulty. The whole design hangs off distance to the stimulus. When the patient reaches the third rung and the rating does not move, there is no modifiable parameter left in reserve.
• Items worded as categories. “Large dogs” cannot be faced; one specific dog, at a specific distance, for a specific length of time, can. Vague items produce unrepeatable tasks and records that cannot be set against one another.
• The jump measured in sessions. Moving up a level because a fortnight has gone by is administering the calendar, not the learning. The criterion is a prediction put to the test and disconfirmed.
• Safety signals with no withdrawal plan. The companion, the phone held in the hand or the seat next to the door enter at the first items and stay until discharge. Every signal that is introduced needs the rung at which it leaves.
• A ceiling set by the equipment. The column ends where the catalogue runs out, not where the patient needed to get to. This is the error that produces discharges the patient does not recognise as such.

There is little to argue about regarding the soundness of the procedure all this sits inside. The meta-analyses by Powers and Emmelkamp and by Carl and colleagues converge on virtual reality exposure performing equivalently to in vivo exposure in the presentations where the two have been compared, with neither format prevailing over the other, and the work of Opriş and colleagues adds that the gains hold at follow-up. What matters for this article is what none of them calls into question: the exposure hierarchy travels intact between media. What changes is the way the rung is presented, not the logic that orders the column.

That logic is medium-agnostic, but its execution is not. The modifiable parameters have to exist as controls and be movable during the task without taking the patient out of the environment; where the scene has to be restarted in order to change the distance to the stimulus, the design of the exposure hierarchy ends up bending to the tool. Anyone who wants to see how a column of items translates into configurable parameters can book a demonstration of the system and run the exercise with a case of their own on the table.

This article is for informational purposes for psychology professionals. It is not clinical advice for any individual case and does not replace the judgment of the licensed psychologist in charge. VRET is professional clinical-support software, not a CE-marked medical device.