Cognitive Psychology Series Part 1: Memory Systems & Encoding

1. Core Memory Systems

Think of memory as a multi-story building. The ground floor (sensory memory) captures everything that enters your senses but holds it for mere fractions of a second. The middle floor (working memory) is your mental workspace where you actively manipulate information. The top floor (long-term memory) is your vast archive that can store information for a lifetime.

1.1 Sensory Memory

Sensory memory is the briefest form of memory — an automatic, unconscious buffer that holds raw sensory data for just long enough to be processed.

Type	Sense	Duration	Capacity	Key Researcher
Iconic Memory	Vision	~250-500 ms	~12 items	George Sperling (1960)
Echoic Memory	Hearing	~2-4 seconds	~5 items	Neisser (1967)
Haptic Memory	Touch	~2 seconds	Limited	Bliss et al. (1966)

Analogy: Imagine sensory memory as a camera's image sensor. It captures a complete, high-resolution snapshot of the visual field, but unless you "save" (attend to) part of it, the image fades almost immediately — like a message written on water.

1.2 Working Memory (Baddeley & Hitch Model)

Working memory is your brain's mental workspace — the system that holds and manipulates information you're actively thinking about. In 1974, Alan Baddeley and Graham Hitch proposed a multi-component model that replaced the simpler concept of "short-term memory."

The Baddeley & Hitch model includes four components:

Component	Function	Example
Central Executive	Directs attention, coordinates subsystems, switches between tasks	Deciding which task to focus on when multitasking
Phonological Loop	Stores and rehearses verbal/acoustic information	Repeating a phone number in your head
Visuospatial Sketchpad	Maintains visual and spatial information	Mentally rotating a 3D object or navigating a familiar route
Episodic Buffer	Integrates information from other components and long-term memory	Combining a person's face (visual) with their name (verbal) and the context where you met them

Analogy: Think of working memory as a project manager's desk. The central executive is the manager who decides what to work on. The phonological loop is like a voice recorder for verbal notes. The visuospatial sketchpad is a whiteboard for sketching diagrams. The episodic buffer is a folder that combines documents from different sources into a coherent brief.

# Simulating working memory capacity with a simple Python model
# Miller's Magic Number: 7 ± 2 items

import random

class WorkingMemory:
    """A simplified model of working memory with limited capacity."""

    def __init__(self, capacity=7):
        self.capacity = capacity
        self.phonological_loop = []    # Verbal information
        self.visuospatial_pad = []     # Visual/spatial information
        self.episodic_buffer = []      # Integrated representations

    def rehearse_verbal(self, items):
        """Add items to the phonological loop (verbal rehearsal)."""
        self.phonological_loop = items[:self.capacity]
        overflow = len(items) - self.capacity
        if overflow > 0:
            print(f"Capacity exceeded! Lost {overflow} items due to decay.")
        return self.phonological_loop

    def chunk(self, items, chunk_size=3):
        """Chunking: group items to effectively increase capacity."""
        chunks = []
        for i in range(0, len(items), chunk_size):
            chunk = ''.join(items[i:i + chunk_size])
            chunks.append(chunk)
        print(f"Chunked {len(items)} items into {len(chunks)} chunks")
        return self.rehearse_verbal(chunks)

    def demonstrate_serial_position(self, word_list):
        """Demonstrate the serial position effect."""
        recalled = []
        # Primacy effect: first items get more rehearsal
        recalled.extend(word_list[:3])
        # Middle items: less likely to be recalled
        middle = word_list[3:-3]
        recalled.extend(random.sample(middle, min(2, len(middle))))
        # Recency effect: last items still in working memory
        recalled.extend(word_list[-3:])
        return recalled

# Example: Digit span test
wm = WorkingMemory(capacity=7)
digits = ['4', '7', '2', '9', '1', '5', '8', '3', '6']
print("Attempting to remember:", ''.join(digits))
remembered = wm.rehearse_verbal(digits)
print("Remembered:", ''.join(remembered))

# Example: Chunking a phone number
phone_digits = list("5551234567")
print("\nPhone number digits:", ''.join(phone_digits))
chunked = wm.chunk(phone_digits, chunk_size=3)
print("Chunked representation:", '-'.join(chunked))

1.3 Short-Term vs Long-Term Memory

The distinction between short-term memory (STM) and long-term memory (LTM) is one of the most fundamental in cognitive psychology. While working memory is an updated concept of STM that emphasizes active processing, the core distinction remains:

Feature	Short-Term / Working Memory	Long-Term Memory
Duration	15-30 seconds (without rehearsal)	Minutes to lifetime
Capacity	7 ± 2 items (Miller, 1956)	Virtually unlimited
Encoding	Primarily acoustic (sound-based)	Primarily semantic (meaning-based)
Forgetting	Decay + displacement	Interference + retrieval failure
Brain Region	Prefrontal cortex	Hippocampus → distributed cortex

1.4 Explicit vs Implicit Memory

One of the most important distinctions in memory research is between explicit (declarative) and implicit (non-declarative) memory — the difference between memories you can consciously recall and those that influence your behavior without awareness.

Feature	Explicit (Declarative)	Implicit (Non-Declarative)
Awareness	Conscious recall	Unconscious influence
Types	Episodic (events), Semantic (facts)	Procedural (skills), Priming, Classical conditioning
Brain System	Hippocampus + medial temporal lobe	Basal ganglia, cerebellum, amygdala
Affected by Amnesia	Yes (impaired in H.M.)	Usually preserved
Example	"I remember my 10th birthday party"	Knowing how to ride a bicycle

2. Types of Long-Term Memory

Long-term memory is not a single warehouse but a collection of specialized storage systems, each with its own characteristics, neural substrates, and developmental trajectories.

2.1 Episodic Memory (Events)

Episodic memory, first described by Endel Tulving in 1972, stores personal experiences tagged with when and where they occurred. It's like a mental time-travel machine that lets you re-experience past events from a first-person perspective.

Characteristics:

• Autonoetic consciousness: The subjective feeling of "re-living" an event
• Contextual binding: Events are encoded with spatial, temporal, and emotional context
• Highly malleable: Susceptible to distortion, suggestion, and false memories
• First to decline: Among the earliest memory systems affected by aging and Alzheimer's

2.2 Semantic Memory (Facts)

Semantic memory stores general knowledge about the world — facts, concepts, word meanings, and categories — independent of personal experience. You know that Paris is the capital of France without remembering when or where you learned it.

Analogy: If episodic memory is a personal diary (dated entries about your life), semantic memory is an encyclopedia (organized knowledge without personal context).

Key features of semantic memory:

• Organized hierarchically: Concepts arranged in categories (animal → mammal → dog → golden retriever)
• Spreading activation: Activating one concept primes related concepts (Collins & Loftus, 1975)
• More resilient: Less affected by aging than episodic memory
• Schema-driven: New information integrated into existing knowledge frameworks

2.3 Procedural Memory (Skills)

Procedural memory is your memory for how to do things — motor skills, habits, and learned routines. It develops through repetition and practice, eventually becoming automatic.

Real-world examples:

• Typing on a keyboard (try describing exactly which fingers hit which keys!)
• Playing a musical instrument
• Driving a car in familiar conditions
• Athletic skills like a tennis serve or basketball free throw

2.4 Autobiographical Memory

Autobiographical memory is a blend of episodic and semantic memory that forms your personal life narrative. It includes both specific events (episodic: "my wedding day") and general personal knowledge (semantic: "I grew up in Chicago").

Martin Conway's model describes three levels:

1. Lifetime periods: Extended time spans ("when I lived in New York")
2. General events: Repeated or extended events ("my morning routine at that job")
3. Event-specific knowledge: Unique, vivid details of individual moments

3. Encoding Mechanisms

Encoding is the process of transforming sensory information into a form that can be stored in memory. Not all encoding is equal — the depth and richness of processing during encoding largely determines whether you'll remember something later.

3.1 Levels of Processing Theory

Craik and Lockhart (1972) proposed that memory depends not on which "store" information enters but on how deeply it is processed during encoding:

Processing Level	What It Involves	Example	Memory Outcome
Structural (Shallow)	Physical features of the stimulus	"Is the word written in capital letters?"	Poor retention
Phonemic (Intermediate)	Sound properties	"Does the word rhyme with 'train'?"	Moderate retention
Semantic (Deep)	Meaning and associations	"Does the word fit in this sentence: 'The ___ crossed the road'?"	Excellent retention

Elaboration & Association

Elaborative encoding involves connecting new information to existing knowledge through meaningful associations. The more connections you create, the more retrieval paths you build — making the memory easier to find later.

Techniques that leverage elaboration:

• Method of Loci (Memory Palace): Associate items with locations along a familiar route
• Self-Reference Effect: Information related to yourself is remembered better (Rogers et al., 1977)
• Generation Effect: Generating an answer yourself produces better memory than reading it (Slamecka & Graf, 1978)
• Testing Effect: Retrieval practice strengthens memory more than additional study (Roediger & Karpicke, 2006)

3.2 Dual Coding Theory

Allan Paivio's Dual Coding Theory (1971) proposes that information encoded in both verbal and visual formats creates two independent memory traces, effectively doubling your chances of retrieval.

Analogy: Imagine you're trying to find a book in a massive library. If you know only the title (one search path), your search is limited. But if you also know the color of the cover and the general shelf location (multiple search paths), you're far more likely to find it.

# Demonstrating Dual Coding Theory with a memory experiment simulation

import random

class DualCodingExperiment:
    """Simulates the dual coding advantage in memory recall."""

    def __init__(self):
        # Items with only verbal encoding (words alone)
        self.verbal_only = [
            "justice", "liberty", "theory", "concept",
            "truth", "hope", "value", "idea"
        ]
        # Items with dual coding (words + mental images)
        self.dual_coded = [
            ("elephant", "🐘"), ("mountain", "🏔️"),
            ("bicycle", "🚲"), ("lighthouse", "🗼"),
            ("guitar", "🎸"), ("telescope", "🔭"),
            ("butterfly", "🦋"), ("volcano", "🌋")
        ]

    def test_recall(self, trials=1000):
        """Simulate recall rates for verbal-only vs dual-coded items."""
        verbal_recall_rate = 0.45   # ~45% recall for abstract words
        dual_recall_rate = 0.72     # ~72% recall for concrete/imageable words

        verbal_hits = sum(1 for _ in range(trials)
                         if random.random() < verbal_recall_rate)
        dual_hits = sum(1 for _ in range(trials)
                       if random.random() < dual_recall_rate)

        print("=== Dual Coding Theory Simulation ===")
        print(f"Trials per condition: {trials}")
        print(f"\nVerbal-only (abstract words):")
        print(f"  Recalled: {verbal_hits}/{trials} ({verbal_hits/trials*100:.1f}%)")
        print(f"\nDual-coded (concrete/imageable words):")
        print(f"  Recalled: {dual_hits}/{trials} ({dual_hits/trials*100:.1f}%)")
        print(f"\nDual coding advantage: +{(dual_hits-verbal_hits)/trials*100:.1f}%")

        return verbal_hits / trials, dual_hits / trials

experiment = DualCodingExperiment()
experiment.test_recall()

3.3 Chunking

Chunking is the process of grouping individual items into larger, meaningful units — effectively bypassing the 7 ± 2 item limit of working memory. It's one of the most powerful encoding strategies available.

4. Memory Consolidation

Consolidation is the process by which fragile, newly formed memories are stabilized into durable long-term storage. Think of it as the difference between saving a document to RAM (temporary) versus writing it to your hard drive (permanent).

4.1 The Role of Sleep

Sleep is not merely rest — it's an active memory processing period. During sleep, the brain replays, reorganizes, and strengthens newly formed memories.

Sleep Stage	Memory Function	Evidence
Slow-Wave Sleep (SWS)	Declarative memory consolidation (facts, events)	Hippocampal-cortical dialogue; "sharp-wave ripples" replay daytime experiences
REM Sleep	Procedural & emotional memory consolidation	Motor skill improvement; emotional memory regulation
Sleep Spindles (Stage 2)	Memory integration and protection from interference	More spindles = better memory performance; correlated with IQ

4.2 Systems Consolidation vs Synaptic Consolidation

Consolidation operates at two levels:

Synaptic consolidation occurs within hours of learning at the level of individual synapses. Through a process called long-term potentiation (LTP), repeated activation of neural pathways strengthens the connections between neurons, making future firing easier. This requires protein synthesis — which is why drugs that block protein synthesis immediately after learning prevent memory formation.

Systems consolidation unfolds over weeks to years, gradually transferring memories from the hippocampus (temporary storage) to the neocortex (permanent storage). This explains why patients with hippocampal damage can recall childhood memories but not recent events — the old memories have already been "migrated" to the cortex.

# Modeling the forgetting curve and spaced repetition
import math

def ebbinghaus_forgetting_curve(time_hours, stability=1.0):
    """
    Ebbinghaus forgetting curve: R = e^(-t/S)
    R = retention (0 to 1)
    t = time since learning
    S = stability (strength of memory)
    """
    retention = math.exp(-time_hours / stability)
    return retention

def spaced_repetition_schedule(initial_stability=1.0, reviews=5):
    """
    Each successful review increases stability (spacing effect).
    Based on the SM-2 algorithm concept used in Anki.
    """
    stability = initial_stability
    schedule = []

    for review in range(reviews):
        # Review when retention drops to ~70%
        optimal_interval = -stability * math.log(0.70)
        schedule.append({
            'review': review + 1,
            'interval_hours': round(optimal_interval, 1),
            'interval_days': round(optimal_interval / 24, 1),
            'stability': round(stability, 2)
        })
        # Each review roughly doubles stability
        stability *= 2.2

    print("=== Spaced Repetition Schedule ===")
    print(f"{'Review':<10}{'Interval':<15}{'Days':<10}{'Stability':<12}")
    print("-" * 47)
    for item in schedule:
        print(f"{item['review']:<10}{item['interval_hours']:<15}{item['interval_days']:<10}{item['stability']:<12}")

    # Show forgetting curve before first review
    print("\n=== Forgetting Curve (Initial Learning) ===")
    for t in [0.5, 1, 2, 4, 8, 24, 48, 168]:
        r = ebbinghaus_forgetting_curve(t, initial_stability)
        bar = "█" * int(r * 40)
        label = f"{t}h" if t < 24 else f"{t//24}d"
        print(f"  {label:>4}: {r*100:5.1f}% {bar}")

spaced_repetition_schedule()

5. Retrieval & Forgetting

A memory that can't be retrieved is functionally equivalent to a memory that doesn't exist. Understanding retrieval mechanisms is crucial because most "forgetting" is actually a retrieval failure, not a storage failure — the information is still there, but you can't access it.

5.1 Retrieval Cues & Context-Dependent Memory

Tulving's Encoding Specificity Principle states that memory retrieval is most effective when the conditions at retrieval match the conditions at encoding.

State-dependent memory is a related phenomenon: your internal physiological state acts as a retrieval cue. Information learned while in a particular mood, under the influence of caffeine, or in a specific emotional state is better recalled when you're in that same state again.

5.2 Interference Theory

Interference theory proposes that forgetting occurs because other memories compete with or disrupt the target memory:

Type	Direction	Example
Proactive Interference	Old memories interfere with new learning	Your old phone number keeps coming to mind when you try to recall your new one
Retroactive Interference	New learning interferes with old memories	After learning French, your previously-learned Spanish vocabulary deteriorates

Analogy: Think of interference as a busy radio dial. Proactive interference is like an old station bleeding into the frequency of a new one you're trying to tune. Retroactive interference is like a powerful new station drowning out an older, weaker signal.

5.3 False Memories & The Misinformation Effect

Perhaps the most unsettling finding in memory research is that people can form vivid, detailed memories of events that never happened.

# DRM (Deese-Roediger-McDermott) False Memory Paradigm
# Demonstrates how associatively related words create false memories

class DRMExperiment:
    """
    The DRM paradigm: present lists of words all associated with
    a 'critical lure' that is NEVER presented. People reliably
    'remember' the critical lure with high confidence.
    """

    def __init__(self):
        self.word_lists = {
            'sleep': [
                'bed', 'rest', 'awake', 'tired', 'dream',
                'wake', 'snooze', 'blanket', 'doze', 'slumber',
                'snore', 'nap', 'peace', 'yawn', 'drowsy'
            ],
            'needle': [
                'thread', 'pin', 'eye', 'sewing', 'sharp',
                'point', 'prick', 'thimble', 'haystack', 'thorn',
                'hurt', 'injection', 'syringe', 'cloth', 'knitting'
            ],
            'sweet': [
                'sour', 'candy', 'sugar', 'bitter', 'good',
                'taste', 'tooth', 'nice', 'honey', 'soda',
                'chocolate', 'heart', 'cake', 'tart', 'pie'
            ]
        }

    def run_trial(self, list_name):
        """Simulate a DRM trial."""
        study_list = self.word_lists[list_name]
        critical_lure = list_name  # The word NEVER presented

        print(f"=== DRM Trial: Critical Lure = '{critical_lure}' ===")
        print(f"Study words: {', '.join(study_list)}")
        print(f"\nNote: '{critical_lure}' was NEVER in the list!")
        print(f"Yet ~55-80% of participants 'remember' seeing it.")
        print(f"Many rate their confidence as HIGH (4-5 on a 5-point scale).")

        # Typical false alarm rates from research
        print(f"\nTypical results:")
        print(f"  True recognition (studied words): ~72%")
        print(f"  False recognition (critical lure): ~62%")
        print(f"  False recognition (unrelated words): ~8%")

experiment = DRMExperiment()
experiment.run_trial('sleep')

6. Advanced Topics

6.1 Metamemory — Thinking About Memory

Metamemory refers to your knowledge and awareness of your own memory processes — knowing what you know, what you don't know, and how well your memory strategies work. It's a form of metacognition specific to memory.

Key metamemory phenomena:

• Feeling of Knowing (FOK): The subjective sense that you know something even though you can't currently recall it ("It's on the tip of my tongue!")
• Judgments of Learning (JOL): Predictions about how well you've learned something (often inaccurate — fluency ≠ learning)
• Tip-of-the-Tongue (TOT) state: You know you know a word, can often recall its first letter or number of syllables, but can't retrieve the full word
• Illusion of competence: Re-reading material feels easy, creating a false sense of mastery (Kornell & Bjork, 2008)

6.2 Neurobiology of Memory

Memory is not stored in a single location but distributed across brain networks. However, certain brain structures play specialized roles:

Brain Structure	Memory Function	Evidence
Hippocampus	Encoding new declarative memories; spatial navigation	Damage → anterograde amnesia (H.M.); London taxi drivers have enlarged hippocampi
Amygdala	Emotional memory; fear conditioning	Enhances consolidation of emotionally arousing events; damage impairs fear learning
Prefrontal Cortex	Working memory; strategic retrieval; source monitoring	Active during tasks requiring manipulation and monitoring of information
Cerebellum	Procedural memory; motor learning; classical conditioning	Essential for eyeblink conditioning; damage impairs motor skill acquisition
Basal Ganglia (Striatum)	Habit formation; reward-based learning	Damaged in Parkinson's disease, which impairs habit learning

6.3 Memory Disorders

Memory disorders provide a window into how normal memory works by revealing what happens when specific components break down:

Disorder	Description	Affected System	Preserved Abilities
Anterograde Amnesia	Cannot form new declarative memories	Hippocampus / medial temporal lobe	Procedural learning, pre-injury memories
Retrograde Amnesia	Cannot recall memories from before injury	Cortical storage areas	New memory formation (if hippocampus intact)
Alzheimer's Disease	Progressive loss of episodic → semantic → procedural memory	Hippocampus (early), then widespread cortical atrophy	Implicit memory preserved longest; emotional responses
Korsakoff's Syndrome	Severe amnesia + confabulation (filling gaps with fabricated memories)	Diencephalon (thiamine deficiency, often from alcoholism)	Intelligence, procedural memory
HSAM (Hyperthymesia)	Exceptional autobiographical memory — can recall nearly every day of their life	Enhanced hippocampal-cortical connectivity	All systems; but subject to false memories like everyone

Exercises & Self-Assessment

Conclusion & Next Steps

In this opening chapter of our Cognitive Psychology Series, we've explored the remarkable architecture of human memory — from the fleeting impressions of sensory memory to the vast, sometimes unreliable archives of long-term storage. Here are the key takeaways:

• Memory is not one system but a collection of interrelated systems (sensory, working, long-term) with different characteristics and neural substrates
• Working memory is a limited-capacity workspace with multiple components (Baddeley & Hitch model) that supports complex cognition
• Long-term memory divides into explicit (episodic, semantic) and implicit (procedural, priming) systems — as dramatically demonstrated by patients like H.M. and Clive Wearing
• Encoding depth matters: Semantic processing, elaboration, dual coding, and chunking create stronger, more retrievable memories
• Sleep is essential for memory consolidation, with different sleep stages supporting different memory types
• Forgetting is mostly retrieval failure, driven by interference and context mismatch — not storage decay
• Memory is reconstructive, not reproductive — making it vulnerable to false memories with serious real-world consequences