MCAT topics → Cognition, Learning, Memory, and Language → Memory Systems (Sensory, Short-Term, Long-Term, Working)

Memory Systems (Sensory, Short-Term, Long-Term, Working)

MCAT trap: Applies STM capacity limits to long-term memory. STM has a limited capacity (~7 ± 2 chunks) and short duration (~30s), while LTM has essentially unlimited capacity and indefinite duration.

Memory systems is one of the most frequently tested cognition topics on the MCAT, and it rewards students who can do more than recite definitions — you need to trace how information moves through systems and predict what breaks when specific structures are damaged. The core architecture runs from sensory memory (iconic for visual, echoic for auditory — lasting milliseconds to a few seconds) through short-term memory (roughly 7 ± 2 chunks, ~30 seconds without rehearsal) into long-term memory (essentially unlimited capacity and duration). Long-term memory then splits into declarative (explicit: semantic facts + episodic events) and non-declarative (implicit: procedural skills, conditioning). Know this taxonomy cold.

The MCAT tests this in three main ways: straight definition questions about capacities and durations, mechanism questions about Baddeley's working memory model (phonological loop, visuospatial sketchpad, central executive, episodic buffer), and passage-based case studies where a patient's deficit pattern tells you exactly which system is compromised. The H.M. case — bilateral hippocampal removal, intact procedural learning, devastated declarative consolidation — is essentially a blueprint the exam returns to repeatedly. Expect passages that describe a novel patient and ask you to map their deficits onto systems and brain regions.

What makes this topic tricky is that students conflate systems that sound similar. Working memory gets collapsed into STM, STM capacity limits get incorrectly applied to LTM, and the hippocampus/cerebellum distinction for declarative vs. procedural memory gets reversed under pressure. These aren't random errors — they reflect genuinely confusable concepts. The fix is building a clean functional model: each system does a specific job, has a specific capacity/duration profile, and maps to specific neural hardware.

Common misconceptions

Common mistake
Wrong: Long-term memory has a limited capacity similar to short-term memory.
Right: STM has a limited capacity (~7 ± 2 chunks) and short duration (~30s), while LTM has essentially unlimited capacity and indefinite duration.
STM's ~7 ± 2 chunk limit and ~30-second duration are defining features of that specific store — they do not scale up to LTM. LTM is considered functionally unlimited in both capacity and duration; forgetting from LTM is typically a retrieval failure or interference issue, not a storage overflow. If you find yourself thinking 'LTM must eventually fill up,' that's STM logic bleeding into the wrong system.
Common mistake
Wrong: H.M.'s hippocampal lesion impaired all forms of memory, including procedural learning.
Right: H.M. had severe anterograde amnesia for declarative memory but retained the ability to acquire new procedural skills (e.g., mirror drawing), demonstrating hippocampal independence of procedural memory.
H.M.'s case is famous precisely because his lesion was selective: bilateral hippocampal removal wiped out his ability to form new declarative memories (anterograde amnesia) but left procedural learning intact — he could learn to do mirror drawing better over sessions even though he had no conscious memory of practicing. This dissociation proves that procedural memory does not depend on the hippocampus, and it's the single most important data point for the hippocampus/procedural independence distinction on the MCAT.
Common mistake
Wrong: Working memory is simply another name for short-term memory.
Right: Working memory (Baddeley's model) is an active system that manipulates information via the phonological loop, visuospatial sketchpad, and central executive — it is more than passive short-term storage.
Short-term memory is a passive storage buffer; working memory is an active manipulation system. Baddeley's model adds dedicated subsystems — the phonological loop actively rehearses verbal information, the visuospatial sketchpad processes spatial layouts, and the central executive allocates attention across tasks. You use working memory when you do mental arithmetic or follow a multi-step argument, not just when you're passively holding a phone number. They overlap but are not the same thing.
Common mistake
Wrong: The cerebellum is involved in consolidating declarative memories.
Right: The cerebellum (along with basal ganglia) supports procedural/motor memory; the hippocampus consolidates declarative memories.
The cerebellum's role in memory is motor and procedural — it handles timing, coordination, and classically conditioned motor responses (like eyeblink conditioning). Declarative memory consolidation is a hippocampal job. A quick way to keep this straight: damage the hippocampus and the patient can still ride a bike (procedural, cerebellum/basal ganglia intact) but can't remember learning to. Damage the cerebellum and the coordination suffers, but factual and episodic memory are fine.
Free Deck audit

See if your Anki deck covers this topic.

Upload your deck →
Guided session

Stuck on this? An AI tutor that probes your understanding.

Start a session →

What the exam tests

  1. Know the properties of each store: sensory memory (iconic/echoic, milliseconds to ~2s), STM (~7 ± 2 items, ~30s without rehearsal), and LTM (unlimited capacity, indefinite duration) — including the declarative/procedural and semantic/episodic subdivisions of LTM.
  2. Understand Baddeley's working memory model as an active processing system with four components: the phonological loop (verbal/auditory rehearsal), visuospatial sketchpad (spatial and visual information), central executive (attentional control and coordination), and episodic buffer (integrating information across systems).
  3. Apply memory system knowledge to patient case studies — given a deficit pattern (e.g., can't form new declarative memories but learns motor skills normally, or loses semantic knowledge while episodic memory is intact), identify which memory system and which brain region are affected.
  4. Connect memory systems to their neural substrates — hippocampus for declarative memory consolidation, basal ganglia and cerebellum for procedural/motor memory — and predict which functions survive vs. fail after specific lesions.

Can you avoid these mistakes?

A patient sustains damage to the hippocampus bilaterally. Which of the following would you expect to remain intact: (A) forming new episodic memories, (B) recalling semantic facts learned decades ago, (C) learning a new motor sequence through practice, or (D) consolidating new factual information? Explain your reasoning for each.
What is the functional difference between the phonological loop and the central executive in Baddeley's working memory model? Give a concrete example of a task that specifically taxes the central executive rather than the phonological loop.
A researcher claims that long-term memory must have a capacity limit because people forget things over time. Identify the flaw in this argument using what you know about the distinction between storage and retrieval in LTM.
You read a passage describing a patient who cannot recognize faces or recall the names of famous people (semantic memory deficit) but can still describe what she did last Tuesday (episodic memory intact) and learns new physical therapy exercises normally. Map each preserved and impaired function to the correct memory system and, where applicable, to the relevant brain structure.

Related topics

Information Processing Model of Cognition Piaget's Stages of Cognitive Development Problem Solving and Decision Making Heuristics and Cognitive Biases

See how your Anki deck covers this topic.

Upload your deck for a free audit →