Read this topic

Behavioural Science & Sociocultural Psychiatry · Paper A

Types of long-term memory

The idea

Long-term memory splits into what you can consciously tell someone (declarative memory) and what your brain quietly gets better at without you having to think about it (procedural/implicit memory) — and the cleanest proof of this split is a patient whose conscious memory is wiped out by brain damage but who can still learn a brand-new skill perfectly normally, with no idea they have ever practised it.

How it works

Declarative (explicit) memory is anything that can be consciously brought to mind and put into words — it splits into episodic memory (personally experienced events tied to a specific time and place: 'what happened, when, where') and semantic memory (general facts and knowledge with no memory of the occasion they were learned on: 'what is known').
Non-declarative (implicit) memory is anything expressed only through performance, never through conscious recall — it includes procedural memory (a skill that improves with repeated practice, e.g. riding a bicycle), priming (a single earlier exposure making later identification of the same or a related stimulus faster, without conscious memory of that exposure), and simple associative/non-associative learning (classical conditioning, habituation).
The classic evidence for this split is a lesion dissociation: dense, bilateral damage to the medial temporal lobes (the hippocampus and surrounding structures) abolishes the ability to form NEW declarative memories (anterograde amnesia) while leaving procedural learning and priming intact, because these systems depend on different brain structures — declarative memory on the medial temporal lobes, procedural memory classically on the basal ganglia and cerebellum.
This dissociation recurs in Korsakoff's syndrome — amnesia from thiamine (vitamin B1) deficiency, classically seen with chronic heavy alcohol use — where dense anterograde amnesia and confabulation (unintentional fabrication of memories to fill gaps, with no intent to deceive) coexist with entirely preserved procedural learning.
Working memory (holding and manipulating information online for seconds) and sensory memory (a fraction-of-a-second raw sensory buffer) are not long-term memory at all, and sit outside this whole taxonomy — a stem describing information lost within moments is describing one of these, not a long-term store.
A few named phenomena sit at the edges of the taxonomy rather than inside one clean box: flashbulb memory is an unusually vivid (but not more accurate) subtype of episodic memory for a surprising or emotional event, and prospective memory is memory for a future INTENTION rather than a past fact or event.

How to spot it

A vignette naming both remembered FACTS and remembered PERSONAL EVENTS together, or giving no subtype detail at all, is describing declarative memory in general.
A vignette with rich time-and-place detail about something that happened to the person is episodic memory; a vignette with a fact stated confidently but no memory of learning it is semantic memory.
A vignette where a SKILL improves with repeated practice, and the person cannot describe what they have learned to do differently, is procedural memory — this is the pattern behind the classic amnesia case, where a patient with dense declarative amnesia still improves, session after session, at a motor task he has no memory of ever attempting.
A vignette where a single earlier exposure to a stimulus makes IDENTIFYING it (or something related) faster later, with no conscious memory of that exposure, is priming.
A vignette describing new memory formation being impaired AFTER a brain injury, with old memories intact, is anterograde amnesia; the mirror pattern (old memories lost, new ones forming normally) is retrograde amnesia.
A vignette with an amnesic patient confidently describing events that turn out never to have happened, without any intent to deceive, is confabulation.
A vignette describing remembering to DO something at a future point (a time or a cue) is prospective memory, not an ordinary declarative memory for the past.

Tell it apart

Declarative memory (consciously recalled, can be stated in words) vs procedural memory (expressed only through performance, never consciously recalled as an event) — the central split this whole concept tests.
Episodic memory (a specific personal event, tied to time and place) vs semantic memory (a general fact, with no memory of when or where it was learned) — the two declarative subtypes.
Episodic memory vs flashbulb memory — flashbulb memory is simply an unusually vivid, high-confidence episodic memory for a surprising event; the vividness does not make it more accurate.
Procedural memory (a SKILL improving over many repetitions) vs priming (a single earlier exposure making later IDENTIFICATION faster) — both are implicit and non-consciously-recalled, but one is gradual skill learning and the other is a one-off facilitation effect.
Classical conditioning (an association is deliberately formed between two stimuli) vs habituation (a response simply fades with repeated exposure to ONE stimulus, no association formed) — the two non-associative-versus-associative implicit learning processes.
Working memory / sensory memory (genuinely short-lived, seconds or less, not long-term at all) vs any long-term store (declarative or procedural, persisting over days to years once consolidated).
Anterograde amnesia (new memories AFTER the injury are impaired) vs retrograde amnesia (memories from BEFORE the injury are lost) — both are deficit PATTERNS, not memory types in their own right, and a question asking which memory SYSTEM is spared or impaired wants declarative/procedural, not the amnesia label itself.
Confabulation (unintentionally fabricated memory, no intent to deceive) vs ordinary memory loss — confabulation is a symptom that accompanies severe amnesia (classically Korsakoff's syndrome), not a store of memory being tested.

Watch for

⚠️ Treating 'declarative memory' and 'episodic memory' as interchangeable — episodic memory is only one of declarative memory's two subtypes; a stem naming both facts AND personal events, or giving no subtype detail, wants the broader term.
Naming the amnesia SYNDROME (anterograde amnesia, retrograde amnesia) when the question is actually asking which memory SYSTEM (declarative vs procedural) is spared or impaired — the syndrome name describes the deficit's timing, not the system.
Confusing procedural memory with priming — both are implicit, but procedural memory is a skill built up over repeated PRACTICE, while priming is facilitated identification from a single earlier EXPOSURE.
Assuming a vivid, emotionally charged personal memory must be 'flashbulb memory' by name, even when the stem gives no surprising/emotional detail — plain episodic memory is still the better answer without that specific quality.
Mistaking working memory or sensory memory for a long-term store because both can be described as 'holding information' — check the timescale: seconds or less means it is not long-term memory at all.
Forgetting that confabulation is a SYMPTOM that accompanies amnesia (fabricated memories, no intent to deceive), not itself a memory type being tested.

Use it again

First askis this being consciously recalled and stated (declarative), or only shown through performance without conscious recall (non-declarative/implicit)?
If declarativeis it a specific personal event tied to time and place (episodic), or a decontextualised fact (semantic)?
If implicitdid a SKILL improve over repeated practice (procedural), or did a single earlier exposure make something easier to identify afterwards (priming), or was an association deliberately formed between two stimuli (classical conditioning) versus a response simply fading with repetition (habituation)?
If the stem names an amnesia pattern or a fabrication, ask what the question is actually testing: the TIMING of the deficit (anterograde vs retrograde), the SYSTEM affected (declarative vs procedural), or a SYMPTOM riding alongside it (confabulation) — these are three different axes, not competing answers to the same question.
Do not select an option purely because it 'sounds like the memory term you remember from the stem' — check the exact timescale (seconds = not long-term at all) and mechanism (practice vs single exposure vs deliberate pairing) before committing.

In detail

Hook — the man who could not remember practising, yet kept getting better

Imagine a patient who wakes up each morning unable to recall anything that happened to him since his surgery years earlier — not what he had for breakfast, not the name of the doctor he met yesterday, nothing. And yet, hand him a new hand-eye coordination puzzle and ask him to practise it daily, and he gets steadily faster and more accurate, exactly like anyone else would. Ask him afterwards whether he has ever seen the puzzle before, and he will say no, completely sincerely. How can someone get better at something they have no memory of ever having done? The answer is that 'memory' is not one thing — it is at least two separate systems in the brain, and this case shows they can be damaged one without the other.

Plain definition

Long-term memory splits into declarative (explicit) memory — anything that can be consciously brought to mind and described in words — and non-declarative (implicit) memory — anything shown only through performance, never through conscious recall. Declarative memory itself has two parts: episodic memory (a personally experienced event tied to a specific time and place, like your first day at work) and semantic memory (a general fact stripped of when or where you learned it, like knowing Paris is the capital of France). Non-declarative memory covers procedural memory (skills that get better with practice, like riding a bicycle), priming (a single earlier exposure making something easier to recognise later), and simple learned associations (classical conditioning, habituation).

The critical caveat — the syndrome's name is not the memory system

Here is the trap almost every exam candidate falls into: hearing a description of amnesia and reaching straight for the amnesia label — 'anterograde amnesia' — when the question is actually asking something more specific: WHICH memory system is affected, or which one is spared. Anterograde amnesia and retrograde amnesia describe the TIMING of a memory problem (new memories after an injury, versus old memories from before it) — they say nothing on their own about declarative versus procedural memory. A patient can have dense anterograde amnesia for declarative memory while his procedural memory stays completely normal; naming the timing pattern is not the same as naming the system.

First principles — why the brain would build two separate systems at all

Think of the difference between knowing a recipe by heart (you could recite the ingredients to a friend) and your hands simply knowing how to knead dough after years of baking (you could not put the exact motion into words if you tried). The first is a fact you consciously hold and could state out loud; the second is a skill your body performs without needing conscious access to it at all.

Evolutionarily, it makes sense for these to be handled by different machinery. Conscious, flexible knowledge that can be combined and reasoned about (declarative memory) needs a system built for binding facts and events together — the and the surrounding medial structures.

Automatic, well-rehearsed skill performance needs a completely different kind of circuit, one that gets faster and more efficient with repetition without needing to consult conscious memory each time — classically the and . Because these are physically separate brain structures, damage to one can leave the other untouched.

Structured walk-through — the dissociation, worked through step by step

Suppose a patient with bilateral hippocampal damage is given a mirror-tracing task: tracing a star shape while only able to see his hand in a mirror, which reverses left and right. On day one he makes, say, 20 tracing errors and takes a long time to finish. Each day for a week he is brought back and asked to try the same task again — and each day, remarkably, his error count falls: 20, then 14, then 9, then roughly 4 by day five, following the same kind of practice curve a healthy volunteer would show.

But every single day, when asked whether he has ever seen this task before, he says no — and if you ask him to pick the mirror-tracing apparatus out from a set of photographs, he does no better than chance. His procedural memory for the SKILL is intact and improving exactly on schedule; his declarative memory for the EVENT of practising is completely absent.

Run the same logic on priming. Show him a word list once, with no request to memorise it, and later he will complete a word fragment from that list faster than an unfamiliar one — again with zero conscious memory of having seen the list. Two entirely different implicit effects (a skill curve, and a one-off facilitation) both survive in a brain that cannot lay down a single new declarative memory.

Look-alikes & differentials

Looks similar	How to tell it apart
Anterograde amnesia	Names the TIMING of a deficit (new memories after injury), not the system — ask separately whether declarative, procedural, or both are affected
Retrograde amnesia	The mirror timing pattern — memories from BEFORE the injury are lost, while new learning may still be possible
	A SYMPTOM riding alongside severe amnesia (unintentionally fabricated memories, no intent to deceive), not a memory type being tested
Flashbulb memory	Just an unusually vivid, high-confidence EPISODIC memory for a surprising event — the vividness does not make it more accurate
Working memory / sensory memory	Genuinely short-lived (seconds or less) — not long-term memory at all, however information-holding they sound

Diagnosis — how this is actually tested for

There is no single blood test or scan for 'which memory system is affected' — it is established through targeted neuropsychological testing that deliberately separates the systems rather than giving one overall 'memory score'. Declarative memory is tested with tasks a patient can fail completely while still learning normally elsewhere: word-list recall and recognition, story recall after a delay, and paired-associate learning where the patient has to consciously state what they remember. Procedural and implicit memory are tested with tasks scored by PERFORMANCE rather than by what the patient can report — a mirror-tracing or pursuit-rotor motor task scored by accuracy or speed across repeated sessions, or a word-fragment-completion task scored by response time, specifically so an examiner can demonstrate a (normal performance-based learning despite failed conscious recall) rather than just describing it from the history.

Management — what is actually done, and whose thinking it rests on

There is no drug that restores declarative memory once dense amnesia is established; management is about rehabilitation and support, using the systems that ARE working. Cognitive rehabilitation for amnesic patients, as practised in clinical neuropsychology, deliberately leans on preserved procedural and implicit learning: teaching a new daily routine or the use of a memory aid through repeated, guided, errorless practice (so the patient never rehearses a mistake), rather than relying on declarative strategies the patient cannot use.

Where the cause is reversible or treatable — most importantly thiamine deficiency in suspected Korsakoff's syndrome — urgent thiamine replacement is given to prevent the amnesia becoming permanent, since the underlying nutritional cause, unlike the memory systems themselves, can sometimes still be corrected in time. Longer-term support typically involves external memory aids (diaries, alarms, structured routines) built around what the patient's intact systems can actually do, rather than a plan that assumes declarative memory will simply recover.

Exam pearl

Whenever a vignette describes amnesia plus a skill or task that keeps improving despite no memory of practising it, stop and ask which of two different things is being tested: the TIMING of the amnesia (anterograde vs retrograde) or the SYSTEM doing the improving (procedural memory, spared alongside impaired declarative memory). Before you move on: could you explain to a friend, in one sentence, why someone can get better at a task they insist they have never seen before?

In one line

Declarative memory (consciously recalled facts and events) and non-declarative memory (skills and other performance-based learning) are separate brain systems, and the clearest proof is a patient whose declarative memory is destroyed while his procedural learning keeps improving right on schedule.