Classification & Assessment in Psychiatry · Paper A
Mental state examination: technique
The mental state examination is a snapshot, not a story. The history tells you what has happened to a patient over weeks or years; the MSE tells you what is true about them right now, in this room, in this ten minutes — and it does it by walking the same eight stops in the same order every single time, so nothing gets missed and nothing gets double-counted.
A vignette is testing MSE technique — not diagnosis — whenever it hands you a single observed or reported finding in isolation and asks which DOMAIN it belongs to, rather than asking what condition the patient has. The tell is the question stem itself: "which part of the mental state examination…", "which domain…", or a description that deliberately gives you only one layer (only what was observed, or only what the patient said) so you have to know which box that layer goes in.
Naming the surface finding instead of the domain it belongs to — e.g. calling pressured speech "thought disorder" because it sounds disorganised, when speech FORM (rate, volume, quantity) and thought FORM (how ideas connect) are scored as two separate domains even when they appear together.
Defaulting to "mood" for any emotional finding, even when the vignette explicitly describes something the examiner OBSERVED rather than something the patient reported. The observed layer is always affect, never mood, no matter how strong the emotion looks.
Filing passivity phenomenaThe delusional experience of one's own thoughts, feelings, or actions being controlled by an outside force.Read more → under perception because the patient's language sounds sensory ("they're putting thoughts in my head"). The loss of ownership over one's own thoughts or actions is a thought-content finding, a first-rank symptom, not a perceptual one.
Treating insightA person's awareness and understanding that they are unwell and need help.Read more → as a yes/no box. Insight sits on a spectrum (full, partial, absent) and needs to be rated as such, and it is never interchangeable with judgement or with treatment compliance.
Reaching for "orientation" as if it were its own MSE domain, rather than recognising it as one part of cognition alongside attention and memory.
Whenever a question isolates one finding and asks which domain it belongs to, run two checks before answering: first, was this REPORTED by the patient or OBSERVED by the examiner (splits mood from affect); second, is this about the STRUCTURE of thinking or the SUBSTANCE of what's believed (splits thought form from thought content). Everything else in the MSE family sorts itself once those two questions are answered, because appearance/behaviour, speech, perception, cognition and insightA person's awareness and understanding that they are unwell and need help.Read more → each have only one home and rarely get confused with anything except each other's easy near-neighbours (orientation-inside-cognition, judgement-beside-insight).
Every psychiatric interview eventually asks the same question in disguise: what, exactly, is going on in this person's mind right now? The history can tell you what happened last month or last year, but it can't answer that question on its own — you need a second, separate tool that captures the present moment with enough structure that two different examiners, seeing the same patient an hour apart, would write down basically the same thing. That tool is the mental state examination, usually shortened to the MSE, and it is arguably the single most repeated skill in the whole of psychiatric training, because you use it in every single assessment, from a five-minute ward review to a two-hour new-patient clerking.
The MSE is a fixed checklist of eight things the examiner assesses, in the same order, every time: appearance and behaviour, speech, mood, affect, thought (its form and then its content), perception, cognition, and . Think of it as a cross-sectional photograph rather than a video — it describes the patient AT THIS MOMENT, whereas the history is the video of everything that led here. Each of the eight domains has its own narrow job, and the examiner's skill is not remembering the list (that part is easy) but correctly sorting a given observation into the right box.
The one thing almost every learner gets wrong at first is treating "mood" and "affect" as two words for the same thing, and treating "thought" as one domain instead of two. They are not interchangeable. Mood is what the patient TELLS you they feel, sustained over days or weeks — you elicit it by asking directly, and the gold standard is recording it in the patient's own words. Affect is what you, the examiner, OBSERVE them displaying during the ten minutes you're actually in the room with them — their range of expression, whether it shifts appropriately with the conversation, whether it matches what they're describing.
A patient can report feeling fine (mood) while looking flat and unreactive throughout the interview (affect), and MSE technique requires you to record both, separately, without forcing them to agree with each other. The same split applies to thought: form is the SCAFFOLDING connecting one idea to the next, and content is the ACTUAL SUBSTANCE of what's believed or experienced. Disorganised, loosely-connected speech with no single false belief in it is a thought-form finding; a fixed false belief held despite good evidence against it is a thought-content finding, even if the patient expresses it in a single, perfectly grammatical sentence.
Why split emotion and thought each into two layers instead of one? Because reported and observed information behave completely differently as evidence, and because structure and substance fail independently of each other. A patient's self-report of their mood can be distorted by exactly the illness you're trying to assess — someone with severe depression may report their mood as "okay" out of habit or hopelessness, while their observed affect (a flat voice, no eye contact, tearfulness they don't acknowledge) tells a more reliable story. If mood and affect were merged into one box, you'd lose that cross-check entirely.
The same logic applies to thought: a person's speech can be perfectly logical in its GRAMMAR while the CONTENT is (a coherent sentence describing an impossible belief), or the grammar itself can be broken while every individual claim, on its own, is mundane and true. Separating structure from substance means a fault in one doesn't get incorrectly blamed on the other.
Walk it in the standard order. Appearance and behaviour comes first because you can assess it before the patient says a word — note dress and self-care, eye contact, rapport, and any abnormal movement, including the catatonic signs (, where a limb stays wherever it's placed, , repetitive mannerisms). Speech is judged purely on its form: is it fast or slow, loud or quiet, spontaneous or sparse, easy or impossible to interrupt?
Mood is elicited by directly asking how the patient has been feeling in themselves, ideally over a defined recent period, and recorded in their own words. Affect is what you watch happen in real time — its range (full, restricted, blunted, or flat, from most to least expressive), its reactivity to what's being discussed, and its congruence with the content of speech. For example, imagine a patient who says her mood has been fine this month (that's the reported mood) while sitting throughout the interview with a flat voice and no change in expression even when discussing something upsetting (that's the observed affect) — both get written down, side by side, without forcing them to agree.
Thought form is read off the connections between the patient's sentences (are they linear, or do they drift, block, or loop). Thought content is what the patient actually believes and experiences, including (experiences where a thought, feeling, impulse or action feels like it has been inserted, withdrawn, broadcast, or made by an outside force) — passivity phenomena belong here even though they can sound like a perception, because the defining feature is losing ownership of a mental event, not a sensory experience with no cause.
Perception itself covers (a sensory experience with NO real external stimulus at all), illusions (a genuine external stimulus that is misperceived — mistaking a coat on a hook for a person), and (an internal experience the patient themselves recognises as not real, arising inside their own mind).
Cognition covers orientation to time, place and person, attention and concentration (classically tested with serial sevens or reciting the months of the year backwards), and memory — usually screened with a brief instrument such as the ACE-III, where a total score below roughly 88 out of 100 is a commonly used flag for further cognitive assessment. closes the sequence: does the patient accept they are unwell, accept the proposed explanation, and engage with treatment — rated across a spectrum from full to partial to absent, never as a single yes-or-no tick.
The two classic confusions are mood-versus-affect and thought-form-versus-content, covered above, but two more are worth naming. versus ordinary : both can be described by a patient in vivid, sensory-sounding language, but a passivity phenomenon is defined by loss of OWNERSHIP over a thought, feeling or action ("my hand moved on its own, someone else was controlling it"), while a is a sensory experience with no matching ownership question at all. versus judgement: insight is about understanding illness and accepting treatment; judgement is the separate, practical skill of predicting the likely outcome of a decision and acting sensibly on that prediction — a patient can retain everyday judgement (managing money, avoiding obvious danger) while having very poor insight into a psychiatric diagnosis, and the exam expects you to keep the two apart.
The MSE itself isn't diagnosed — it's the tool used to gather findings that FEED a diagnosis — but establishing each domain's finding reliably has its own discipline. Cognitive findings are corroborated with a validated bedside instrument (MMSE, MoCA, or ACE-III) rather than left to unstructured impression, because unstructured impressions of cognition are known to miss mild deficits. and other first-rank symptoms are elicited with specific, direct questioning rather than inferred from vague complaints, because a vague complaint ("I feel controlled") can mean several different things until it's pinned down to a specific, defined experience. is checked by asking, in sequence, whether the patient thinks they are unwell, what they think is wrong, and whether they will accept the offered treatment — three separate questions, because a single global impression of "insight" is exactly the kind of shortcut that produces disagreement between examiners.
Management here means what the MSE findings are actually used FOR, since it isn't itself a condition to treat. A finding of impaired combined with a treatment-refusing decision feeds directly into a formal mental assessment, governed in England and Wales by the framework described in NICE guideline NG108 on decision-making and mental capacity — insight and capacity are related but are assessed and recorded separately, and neither one automatically decides the other. A finding of or other first-rank symptoms, combined with the rest of the MSE, feeds into risk formulation and treatment planning for psychosis along the lines set out in NICE guideline CG178 on psychosis and in adults, which is one of the reasons an accurate MSE is treated as a patient-safety document, not just an academic exercise — errors here change real decisions about admission, capacity and treatment.
Whenever a question gives you a single finding and asks which MSE domain it belongs to, ask two questions before answering: was this REPORTED by the patient or OBSERVED by the examiner (splits mood from affect), and is this about the STRUCTURE of thinking or the SUBSTANCE of a belief or experience (splits thought form from thought content). Those two questions resolve almost every genuinely hard MSE-domain item, because the remaining domains — appearance and behaviour, speech, perception, cognition, — each have only one home and are rarely confused with anything except their own easy near-neighbours. Before moving on, cover this page and try to say all eight domains back in the correct order from memory — that short retrieval effort is what actually makes the sequence stick, far more than re-reading it a second time.
The MSE is eight fixed stops taken in the same order every time — appearance/behaviour, speech, mood, affect, thought form, thought content, perception, cognition, — and the exam almost always tests whether you know which single stop a given finding belongs to, not whether you can recite the list.