Introduction to Memory - AP Psychology

SECTION 1

Historical Context & Motivation

The scientific study of memory began not in a laboratory equipped with brain-imaging technology, but with a lone German philosopher methodically memorizing nonsense syllables in the 1880s. Hermann Ebbinghaus pioneered the experimental investigation of memory by using himself as the subject, painstakingly learning and relearning lists of meaningless consonant-vowel-consonant trigrams such as "DAX" and "BUP." His work demonstrated that memory could be measured quantitatively, revealing predictable patterns of forgetting over time — a finding that challenged the philosophical tradition of treating memory as an entirely subjective phenomenon. Ebbinghaus's research set the stage for over a century of increasingly sophisticated inquiry into how the brain encodes, retains, and recovers information.

1885 • Ebbinghaus's Forgetting Curve

Hermann Ebbinghaus published Über das Gedächtnis, demonstrating that memory decays rapidly at first and then levels off — the classic forgetting curve.

1890 • James's Primary vs. Secondary Memory

William James distinguished between primary memory (current conscious awareness) and secondary memory (the vast repository of past experiences), foreshadowing the short-term/long-term memory distinction.

1953 • Patient H.M. and the Hippocampus

Henry Molaison's bilateral hippocampal removal to treat epilepsy revealed that the hippocampus is essential for forming new explicit memories, revolutionizing the neuroscience of memory.

1968 • Atkinson-Shiffrin Model

Richard Atkinson and Richard Shiffrin proposed the three-stage model — sensory memory, short-term memory, and long-term memory — which became the dominant framework in cognitive psychology.

1974 • Baddeley's Working Memory Model

Alan Baddeley and Graham Hitch replaced the passive short-term store with a dynamic working memory system featuring a central executive, phonological loop, and visuospatial sketchpad.

These milestones reveal a recurring question that still drives memory research: how does a fleeting sensory experience become a lasting mental record, and why does that process sometimes fail? Understanding the mechanisms of memory is not merely an academic exercise — it informs everything from educational strategies to the reliability of eyewitness testimony to the treatment of Alzheimer's disease. In the sections that follow, we will build a comprehensive understanding of memory's architecture, its processes, and its vulnerabilities.

SECTION 2

Core Principles & Definitions

At its most fundamental level, memory refers to the persistence of learning over time through the processes of encoding, storage, and retrieval. These three stages form the backbone of virtually every memory model in cognitive psychology, and understanding each one is essential for grasping how information moves from the external world into a lasting mental representation — and how it can be recovered when needed.

Encoding

The process of converting sensory input into a form the brain can process and store. Encoding can be automatic (unconscious, effortless) or effortful (requiring deliberate attention and rehearsal).

Storage

The retention of encoded information over time. Storage varies in duration and capacity across the three memory stages: sensory memory (milliseconds to seconds), short-term/working memory (seconds to about a minute), and long-term memory (potentially a lifetime).

Retrieval

The process of accessing stored information when it is needed. Retrieval can occur through recall (generating an answer from memory), recognition (identifying a previously encountered item), or relearning (mastering material more quickly the second time).

Forgetting

The inability to retrieve previously stored information. Forgetting can result from encoding failure, storage decay, or retrieval failure — each representing a different point of breakdown in the memory system.

✦

KEY TAKEAWAY

Think of memory like a filing system in a large research library. Encoding is the librarian cataloging a new book — choosing which details go into the record and how finely the subject headings are classified. Storage is the shelf space itself, ranging from a temporary desk (sensory/short-term) to the permanent archive stacks (long-term). Retrieval is the search query you run to locate that book later. A failure at any stage — poor cataloging, damaged shelves, or a broken search index — means the information is effectively lost, even if it technically still exists somewhere in the system.

SECTION 3

The Atkinson-Shiffrin Model: A Visual Guide

The Atkinson-Shiffrin model (also called the multi-store model or modal model) remains the most widely taught framework for understanding memory's architecture. It proposes that information flows sequentially through three distinct stores — sensory memory, short-term memory, and long-term memory — with attention and rehearsal serving as the gating mechanisms that determine whether information advances to the next stage or is lost. The diagram below illustrates this flow, including the critical decision points where information may be transferred forward or discarded.

The Atkinson-Shiffrin model shows information flowing left to right through three stores. Sensory memory holds raw input briefly; attention selects what enters short-term memory. The rehearsal loop above STM keeps items active, while encoding transfers information to long-term memory. The dashed red arrows indicate points where forgetting can occur.

Notice that the model treats each store as having distinct capacity and duration characteristics. Sensory memory is vast in capacity but extremely brief — George Sperling's classic partial-report experiments demonstrated that iconic memory (visual sensory memory) lasts only about one-third of a second, while echoic memory (auditory sensory memory) persists for roughly three to four seconds. Short-term memory is far more limited: George Miller's famous "magical number" paper established a capacity of approximately 7 ± 2 chunks of information, and without active rehearsal, items decay within about 20 seconds — as demonstrated by the Peterson and Peterson (1959) trigram experiment. Long-term memory, by contrast, appears to have virtually unlimited capacity and can retain information for a lifetime, though retrieval is never guaranteed.

SECTION 4

Encoding Mechanisms & Levels of Processing

While the Atkinson-Shiffrin model emphasizes the structural stages of memory, the levels of processing framework proposed by Fergus Craik and Robert Lockhart in 1972 shifts the focus to how deeply information is processed during encoding. Their central insight is that the durability of a memory depends less on which "store" it occupies and more on the depth and elaboration of processing it receives. Shallow processing involves attending to surface features — the font a word is printed in, or the sound of a name — while deep (semantic) processing involves extracting meaning, making associations, and connecting new information to existing knowledge. Decades of research have consistently shown that semantic encoding produces the most durable and retrievable memories.

Types of Encoding

Encoding Type	Description	Depth	Example
Structural	Processing the physical appearance of a stimulus	Shallow	Is the word written in uppercase letters?
Phonemic	Processing the sound of a stimulus	Intermediate	Does the word rhyme with "train"?
Semantic	Processing the meaning and relating it to existing knowledge	Deep	Does this word fit the sentence: "The ____ crossed the road"?

Craik & Tulving (1975) demonstrated that deeper semantic processing leads to superior recall.

Encoding Strategies That Enhance Memory

✓ Elaborative rehearsal — Connecting new information to existing knowledge, as opposed to simple maintenance rehearsal (rote repetition). Elaborative rehearsal leads to deeper encoding and better long-term retention.
✓ Self-referencing effect — Information processed in relation to oneself ("How does this relate to my life?") is encoded more deeply than information processed in relation to others.
✓ Chunking — Organizing individual items into larger meaningful units (e.g., remembering a phone number as 555-867-5309 rather than ten separate digits), thereby increasing the effective capacity of short-term memory.
✓ Mnemonic devices — Techniques such as the method of loci, peg-word system, and acronyms that impose an organized structure on otherwise arbitrary material, facilitating both encoding and retrieval.
✓ Spacing effect — Distributing study sessions over time (spaced practice) produces stronger long-term memories than cramming all study into a single session (massed practice), a principle directly supported by Ebbinghaus's early research.

💡

AP EXAM TIP

The AP Psychology exam frequently tests the distinction between maintenance rehearsal (repeating information to keep it in short-term memory) and elaborative rehearsal (connecting information to meaning for long-term storage). Know that maintenance rehearsal is less effective for transferring information to long-term memory.

SECTION 5

Classification of Long-Term Memory

Long-term memory is not a single monolithic system. Research — particularly from studies of brain-damaged patients like H.M. — has revealed that long-term memory comprises at least two major subsystems that rely on different neural structures and serve different functions. The primary division is between explicit (declarative) memory, which involves conscious recollection, and implicit (nondeclarative) memory, which operates without conscious awareness. Each of these categories further subdivides, creating a taxonomy that is essential for the AP exam.

This taxonomy shows the two major divisions of long-term memory. Explicit memory splits into episodic (personal events) and semantic (factual knowledge), both dependent on the hippocampus. Implicit memory includes procedural memory and conditioned associations, relying on subcortical structures like the cerebellum and amygdala.

The case of patient H.M. (Henry Molaison) dramatically illustrates this division. After surgical removal of his hippocampus, H.M. lost the ability to form new explicit memories — he could not remember people he met minutes earlier or events from his recent past. However, he could still learn new motor skills, demonstrating intact implicit procedural memory. This double dissociation — one system impaired while the other remains intact — provides powerful evidence that explicit and implicit memory are neurologically distinct systems, not just conceptual categories.

SECTION 6

Worked Example: Identifying Memory Processes

AP Psychology free-response questions frequently present a scenario and ask students to identify which memory concepts are at work. Let us walk through a multi-part scenario step by step, as you would on the actual exam.

ℹ

SCENARIO

Maria is studying for her AP Psychology exam. She reads her textbook chapter on memory three times in a row the night before the test (Session A). Her friend Carlos, who studied the same material in shorter sessions over four days (Session B), scores significantly higher on the exam. During the exam, Maria recognizes a term from a multiple-choice option but cannot recall its definition for the free-response question. Months later, when Maria takes a college psychology course, she relearns the memory chapter faster than entirely new material.

Identifying Memory Concepts in the Scenario

1
Step 1 — Identify Maria's Study Strategy

Maria reads the chapter three times in one sitting. This is massed practice (also called cramming). Re-reading is a form of maintenance rehearsal — she is repeating the same information without actively elaborating on its meaning, which produces relatively shallow encoding.

Massed practice + maintenance rehearsal → shallow encoding
2
Step 2 — Explain Carlos's Advantage

Carlos studied across four separate days. This exemplifies the spacing effect (or distributed practice), which research consistently shows produces superior long-term retention compared to massed practice. Each study session allows time for consolidation, and each return to the material requires effortful retrieval, strengthening the memory trace.

Spacing effect → stronger encoding and consolidation
3
Step 3 — Explain the Recognition vs. Recall Discrepancy

Maria recognizes a term on the multiple-choice section but cannot recall its definition on the FRQ. Recognition requires only identifying a previously encountered stimulus — a less demanding retrieval task. Recall requires generating the information without cues, which demands a stronger memory trace. This is why multiple-choice tests are generally easier than free-response tests — they rely on recognition rather than recall.

Recognition (MC) is easier than recall (FRQ) — different retrieval demands
4
Step 4 — Identify Relearning as a Memory Measure

Months later, Maria relearns the memory chapter faster than entirely new material. This demonstrates the savings method (also called relearning), originally developed by Ebbinghaus. Even when information appears to be forgotten, residual memory traces reduce the time or trials needed to relearn the material — evidence that the memory was not entirely erased but had become inaccessible to conscious retrieval.

Relearning (savings) → memory trace persists even when recall fails

SECTION 7

Forgetting: Theories & Phenomena

Memory researchers have long debated whether forgetting occurs because memory traces physically decay over time or because other memories interfere with retrieval. In reality, both mechanisms likely contribute, and the AP exam expects you to distinguish among several explanations of forgetting. The table below compares the major theories, each of which addresses a different point of failure in the encoding-storage-retrieval chain.

Theory of Forgetting	Mechanism	Example
Encoding Failure	Information never entered long-term memory because it was not adequately attended to or processed.	You cannot recall what is on the back of a penny because you never encoded those details.
Storage Decay	Memory traces gradually fade over time if not accessed. Follows the Ebbinghaus forgetting curve.	Forgetting the details of a lecture attended months ago without any review.
Proactive Interference	Old memories interfere with the retrieval of new information (old disrupts new).	Calling your new teacher by your old teacher's name.
Retroactive Interference	New memories interfere with the retrieval of old information (new disrupts old).	After learning your new locker combination, you can no longer remember last year's combination.
Retrieval Failure	The memory exists in storage but cannot be accessed due to inadequate retrieval cues (tip-of-the-tongue phenomenon).	You know the actor's name but cannot produce it until a friend provides the first letter.
Motivated Forgetting	Unconscious suppression (repression) or conscious suppression of distressing memories.	Difficulty recalling traumatic childhood events (though repression as a concept remains controversial).

Major theories of forgetting tested on the AP Psychology exam

✦

KEY TAKEAWAY

A useful mnemonic for remembering the direction of interference: Proactive interference is like a prologue — what came before disrupts what comes after. Retroactive interference works in retrospect — the new information reaches backward to disrupt retrieval of old information. Think of it like a software update that overwrites your settings: the new version (retroactive) blocks access to the old configuration.

SECTION 8

Memory Distortion & Constructive Memory

One of the most important insights of modern memory research is that memory is not a faithful recording of events. Instead, memory is constructive — each time we retrieve a memory, we reconstruct it from stored fragments, filling in gaps with schemas, expectations, and post-event information. This reconstructive nature makes memory vulnerable to systematic errors and distortions that have profound implications for eyewitness testimony, therapy, and everyday life.

Concept	Definition	Key Research
Misinformation Effect	Exposure to misleading information after an event alters one's memory of the original event.	Loftus & Palmer (1974): Changing the verb in a question ("smashed" vs. "hit") altered speed estimates and false memories of broken glass.
Source Monitoring Error	Attributing a memory to the wrong source — confusing where, when, or how information was acquired.	Believing you heard a rumor on the news when a friend actually told you, or confusing a dream with a real event.
False Memories	Recalling events that never actually occurred, often created through suggestion or imagination inflation.	Loftus's "lost in the mall" study showed that vivid, detailed false childhood memories could be implanted through repeated suggestion.
Serial Position Effect	Items at the beginning (primacy) and end (recency) of a list are remembered better than middle items.	Primacy reflects transfer to LTM through rehearsal; recency reflects items still active in STM.

Memory distortion phenomena relevant to the AP Psychology exam

These findings connect directly to advanced topics you will encounter later in the course, particularly in the units on social psychology (how group pressure shapes memory) and clinical psychology (repressed memory debates in therapy). Elizabeth Loftus's research on the misinformation effect has had enormous practical impact on the legal system, leading to revised guidelines for police lineups and eyewitness interview protocols. The broader lesson is that human memory trades perfect fidelity for flexibility and efficiency — a tradeoff that is generally adaptive but occasionally leads to dramatic errors.

✦

KEY TAKEAWAY

Memory works less like a video camera and more like a Wikipedia article — it is continuously editable, influenced by contributions from multiple sources, and subject to revision each time it is accessed. The constructive nature of memory does not mean memories are unreliable; it means that confidence in a memory is not the same as accuracy, a distinction that is critical for evaluating eyewitness testimony.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

A student studies French vocabulary by simply reading the word list repeatedly before the test. According to the levels of processing framework, which type of encoding is the student primarily using, and why is it less effective for long-term retention?

A. The student is using deep semantic encoding, which is less effective because it creates too many associations that interfere with one another.
B. The student is using shallow encoding through maintenance rehearsal, which is less effective because it does not process meaning and creates weaker long-term memory traces.
C. The student is using elaborative rehearsal, which is less effective because it requires too much cognitive effort for short-term tasks.
D. The student is using procedural encoding, which is less effective because it is designed for motor skills rather than verbal material.

Show answer (B)

The student is using shallow (structural or phonemic) encoding combined with maintenance rehearsal. According to Craik and Lockhart's levels of processing framework, simply re-reading words processes them at a surface level — attending to their visual form or sound without engaging with their meaning. Deep (semantic) processing, such as creating sentences using the vocabulary words or connecting them to personal experiences, produces more durable memory traces because it creates richer, more interconnected neural representations. Maintenance rehearsal can keep information active in short-term memory but does not reliably transfer it to long-term memory.

Question 2: BASIC CALCULATION

After H.M.'s hippocampal surgery, he could no longer form new episodic or semantic memories, yet he could learn new motor tasks such as mirror tracing. Which of the following best explains this pattern?

A. The hippocampus is needed only for implicit memory, so removing it spared his explicit memory while impairing procedural learning.
B. Explicit memory formation depends on the hippocampus, whereas implicit procedural memory relies on the cerebellum and basal ganglia, which were intact.
C. The hippocampus controls sensory memory, and its removal eliminated the first stage of the Atkinson-Shiffrin model while leaving later stages functional.
D. The surgery reduced H.M.'s working memory capacity to zero, preventing all new learning except automatic procedural responses.

Show answer (B)

The correct answer is B. H.M.'s case demonstrates a double dissociation between explicit and implicit memory systems. The hippocampus is critical for forming new explicit (declarative) memories — both episodic and semantic — but procedural (implicit) memories depend on the cerebellum and basal ganglia, which were intact in H.M. Choice A is incorrect because it reverses the relationship: explicit memory was impaired, not spared, and the hippocampus is needed for explicit memory, not implicit memory. Choice C is incorrect because sensory memory is a separate short-lived store unrelated to the hippocampus's role in consolidation. Choice D is incorrect because working memory capacity, while possibly affected, does not explain the selective preservation of motor skill learning.

Question 3: INTERMEDIATE

Jenna learned Spanish in high school and is now studying Italian in college. She keeps accidentally using Spanish vocabulary words when trying to speak Italian. Later, when she tries to recall her old Spanish vocabulary for a conversation with a friend, she finds that Italian words keep intruding. Which types of interference is Jenna experiencing, respectively?

A. Proactive interference when Spanish disrupts Italian; retroactive interference when Italian disrupts Spanish.
B. Retroactive interference when Spanish disrupts Italian; proactive interference when Italian disrupts Spanish.
C. Retroactive interference in both cases, because the newer language always disrupts the older one.
D. Encoding failure in both cases, because Jenna did not deeply process either language.

Show answer (A)

The correct answer is A. When old Spanish vocabulary intrudes on new Italian learning, this is proactive interference — previously learned information interferes forward in time with the acquisition of new material. When newly learned Italian words intrude on attempts to recall old Spanish vocabulary, this is retroactive interference — new learning reaches backward to disrupt retrieval of older information. Choice B reverses the two types. Choice C incorrectly labels both instances as retroactive. Choice D incorrectly identifies both as encoding failure; Jenna clearly encoded both languages, as she can produce words from each — the problem is that similar memories are competing during retrieval.

Question 4: APPLIED

4 pts

A cognitive psychologist designs an experiment to test the misinformation effect. In the study, participants watch a video of a car accident. Afterward, half the participants are asked, "How fast were the cars going when they smashed into each other?" and the other half are asked, "How fast were the cars going when they hit each other?" One week later, all participants are asked whether they saw broken glass in the video (there was none). Using your knowledge of memory research: (a) Identify the independent variable and the dependent variable. (b) Predict which group will be more likely to report seeing broken glass and explain why, referencing the misinformation effect. (c) Explain how source monitoring errors contribute to the false memory of broken glass. (d) Describe one real-world application of this research and explain its significance.

Show answer

Rubric (4 points): Point 1 (Part a): The independent variable is the wording of the question ("smashed" vs. "hit"). The dependent variable is whether participants falsely report seeing broken glass one week later (or the estimated speed, though the broken glass report is the primary DV in the follow-up). Point 2 (Part b): Participants in the "smashed" condition will be more likely to falsely report seeing broken glass. The verb "smashed" implies a more severe accident, which introduces misleading post-event information that becomes integrated into the participant's memory of the original video. This is the misinformation effect: exposure to misleading information after an event distorts the original memory. Point 3 (Part c): Source monitoring errors occur when individuals cannot distinguish between information from the original event and information introduced afterward. Participants in the "smashed" group may form a mental image of broken glass based on the connotation of the word, and when later asked about glass, they misattribute this internally generated image to the original video — confusing what was suggested with what was actually seen. Point 4 (Part d): This research has significant implications for the legal system, particularly regarding eyewitness testimony. Police and attorneys can unintentionally alter witnesses' memories through leading questions during interviews. As a result, the Cognitive Interview technique was developed, which uses open-ended questions and avoids suggestive language to minimize the misinformation effect and produce more accurate eyewitness accounts.

Question 5: CRITICAL THINKING

4 pts

The Atkinson-Shiffrin model and Baddeley's working memory model both attempt to explain how information is temporarily maintained and processed in the mind. Construct an argument evaluating which model provides a more complete account of short-term information processing. Your response should: (a) Describe the Atkinson-Shiffrin model's conceptualization of short-term memory, including its key features. (b) Describe Baddeley's working memory model and identify at least two of its components. (c) Present evidence from research or clinical cases that supports the working memory model over the simpler short-term memory concept. (d) Identify one limitation of Baddeley's model or one strength of the Atkinson-Shiffrin model that the working memory model does not fully address.

Show answer

Rubric (4 points): Point 1 (Part a): The Atkinson-Shiffrin model conceptualizes short-term memory as a single, unitary store that holds approximately 7 ± 2 items for about 20 seconds without rehearsal. Information enters STM from sensory memory through attention and can be maintained through rehearsal or transferred to long-term memory through encoding. STM is treated as a passive holding area with limited capacity and duration. Point 2 (Part b): Baddeley's working memory model replaces the unitary short-term store with a multi-component system. The central executive directs attention and coordinates information from subsystems. The phonological loop processes and rehearses auditory/verbal information (consisting of a phonological store and an articulatory rehearsal process). The visuospatial sketchpad processes visual and spatial information. The episodic buffer (added in 2000) integrates information from the other components and long-term memory into coherent episodes. Point 3 (Part c): Research on dual-task performance supports the working memory model. If short-term memory were a single store, performing two tasks simultaneously should always produce interference. However, studies show that people can perform a verbal task and a spatial task simultaneously with little interference (because they use separate subsystems), but two verbal tasks interfere heavily with each other (because they compete for the phonological loop). Additionally, patients with damage to specific brain regions may show impaired verbal short-term memory but intact spatial processing, a dissociation that the unitary STM concept cannot explain but that maps naturally onto Baddeley's separate components. Point 4 (Part d): One limitation of Baddeley's model is that the role of the central executive remains somewhat vaguely defined — critics argue it functions as a "homunculus" (a little person in the head) rather than a well-specified mechanism, making it difficult to test empirically. Alternatively, one could argue that a strength of the Atkinson-Shiffrin model is its elegant simplicity and its clear depiction of the sequential flow of information from sensory input to long-term storage, which provides an accessible and empirically supported framework for understanding the broad architecture of memory, even if it oversimplifies the short-term stage.

SUMMARY

Summary: Introduction to Memory

Summary

Memory is the cognitive system by which the brain encodes, stores, and retrieves information. The Atkinson-Shiffrin model describes three sequential stages — sensory memory (brief, high-capacity), short-term/working memory (limited to about 7 ± 2 items for ~20 seconds), and long-term memory (potentially unlimited and permanent). Baddeley's working memory model refines the short-term stage into a multi-component system with a central executive, phonological loop, visuospatial sketchpad, and episodic buffer.

The levels of processing framework shows that deep semantic encoding produces more durable memories than shallow processing. Long-term memory divides into explicit (episodic and semantic, hippocampus-dependent) and implicit (procedural and conditioning, cerebellum/basal ganglia-dependent) subsystems. Forgetting occurs through encoding failure, storage decay, proactive and retroactive interference, and retrieval failure. Memory is fundamentally constructive, making it susceptible to the misinformation effect, source monitoring errors, and false memories — phenomena with critical real-world implications for eyewitness testimony and the justice system.