Topic 5: Memory

Memory Architecture & The Modal Model Framework

• Purpose and Function of Memory:

  • Core Purpose: Memory exists to store information so we can retrieve and use it later, while safely discarding details that are no longer useful.

  • Key Requirements: To work well, a memory system must have high accuracy (storing representations correctly) and high speed (retrieving information quickly in real time).

• The Information-Processing Approach:

  • Compares the human mind to a computer system that takes in, organizes, stores, and accesses data.

  • Encoding: Converting incoming sensory information into an internal mental format that can be stored in memory.

  • Decoding (Retrieval): Accessing stored information so it can be actively used for thinking, reasoning, calculating, or problem-solving.

• The Modal Model of Memory (Atkinson & Shiffrin, 1968; Waugh & Norman, 1965):

  • Proposes that memory moves through three separate structural stages ("stores") controlled by mental routines:

  • Sensory Memory: A very brief holding area for raw, unattended sensory inputs that fade almost instantly.

  • Short-Term Memory (STM / STS): Holds information we actively attend to for brief intervals ($20\text{--}30\text{ seconds}$ up to a few minutes); functions as an active mental workspace.

  • Long-Term Memory (LTM / LTS): A storage system with virtually unlimited capacity that keeps consolidated memories across days, years, or an entire lifetime.

• Separate Stores vs. Processing Continuum:

  • Multi-Store Perspective: Memory is split into separate mental compartments, each defined by its own storage capacity, time limit, and coding format.

  • Levels-of-Processing Perspective (Craik & Lockhart, 1972): Rejects the idea of separate storage boxes; argues instead that memory is a single continuum where how long something is remembered depends on how deeply and meaningfully it was analyzed during encoding.

Sensory Memory: Iconic & Echoic Stores

• General Properties of Sensory Buffers:

  • Modality-Specific: Every sense has its own temporary holding buffer (iconic for visual input, echoic for auditory input, plus touch, taste, and smell).

  • Raw and Unprocessed (Pre-Categorical): Preserves physical characteristics (such as color, brightness, layout, or pitch) without interpreting what the stimuli mean.

  • High Capacity, Extremely Short Duration: Takes in an entire visual or auditory scene, but the trace decays within milliseconds to seconds unless focused on.

• Iconic Memory (Visual Store):

  • A brief visual after-image that remains for a moment after a stimulus disappears (such as the glowing trail of a sparkler or a flash of lightning lingering after its physical duration of $< 333\,\text{ms}$).

  • Sperling's ($1960$) Letter-Grid Experiments:

  • Whole-Report Method: Participants saw a $3 \times 4$ grid of $12$ letters flashed for $50\,\text{ms}$. They could only report $4\text{ to }5$ letters ($33\text{--}40\%$), even if display time was increased to $500\,\text{ms}$, because the visual trace faded while they were speaking.

  • Partial-Report Technique: A high, medium, or low auditory tone played right after the display cued participants to read only the top, middle, or bottom row. Participants correctly recalled about $3$ out of $4$ letters from the cued row, proving that about $9\text{ out of }12$ letters ($75\%$) were briefly present in sensory memory.

  • Trace Decay: When the tone cue was delayed by $300\,\text{ms}$ to $1.0\,\text{second}$, performance dropped back down to the whole-report baseline ($4\text{ items}$), showing that iconic traces last $< 1\,\text{second}$.

  • Visual Masking (Averbach & Coriell, 1961): Flashing another visual shape (like a circle) in the exact same location right after the letters clears or overwrites the iconic trace.

  • Categorical Limitations: The partial-report advantage works well when cued by physical cues (like row position or color), but completely fails when cued by meaning or sound (such as "report only numbers" or "letters that rhyme with B").

• Echoic Memory (Auditory Store):

  • A temporary acoustic trace (for example, asking "What did you say?" and then immediately realizing what was spoken).

  • Moray, Bates, & Barnett ($1965$): Tested listeners using headphones with four separate streams of letters. Giving a light cue to report just one stream produced much higher proportional recall than asking for all four streams at once.

  • Duration & Capacity: Holds less spatial information at once compared to iconic memory, but lasts significantly longer—persisting up to $20\text{ seconds}$ (Watkins & Watkins, 1980; Crowder, 1976).

  • The Suffix Effect (Crowder, 1972): Adding an extra spoken word (the "suffix") to the end of a spoken list degrades recall of the final items. Irrelevant sounds like buzzes, tones, or visual cues do not cause this disruption; interference increases when the suffix sounds similar to the list items.

  • Everyday Example: A directory assistance operator saying "Have a nice day" right after giving a phone number acts as an auditory suffix, disrupting the caller's echoic buffer (Neath & Surprenant, 2003).

Short-Term Memory & Working Memory Dynamics

• Serial Position Effect & Free Recall (Murdock, 1962):

  • Recalling an unorganized list of items (e.g., $20$ words) produces a $U$-shaped recall curve depending on item position.

  • Primacy Effect: Better recall for items at the start of the list because there is enough time to rehearse them into Long-Term Memory; disappears if items are presented too quickly to rehearse.

  • Recency Effect: Better recall for items at the end of the list because they are still held in Short-Term or Sensory Memory; disappears if recall is delayed by a $30\text{-second}$ distractor task like counting backward (Postman & Phillips, 1965).

• Capacity Limits of STM:

  • The Magic Number ($7 \pm 2$) (George Miller, 1956): The standard number of unrelated items a person can hold in short-term memory (modern estimates suggest $5 \pm 2$, though $7 \pm 2$ remains the classic benchmark).

  • Chunking (Recoding): Combining separate items into larger, meaningful units using existing knowledge (for example, grouping the $12$ letters "NFLCBCFIBIMTV" into $4$ familiar acronyms: "NFL", "CBC", "FBI", "MTV").

• Coding Formats in STM:

  • Acoustic/Phonological Dominance: STM primarily relies on sound-based coding. Conrad ($1964$) showed that visually presented letters caused sound-alike mistakes (mixing up $P$ with $G$ or $C$) rather than visual mistakes (mixing up $P$ with $F$). Baddeley ($1966\text{a}$) showed that lists of words with similar sounds significantly degrade immediate recall.

  • Computational Workspace: STM acts as an active workbench where visual and spatial details are organized to solve problems (such as the visual differences between North American vertical alignment versus French horizontal layout during long division).

• Retention Duration & Forgetting Mechanisms:

  • Retention Duration: Without rehearsal, items in STM fade within $20\text{--}30\text{ seconds}$.

  • Brown-Peterson Task (Brown, 1958; Peterson & Peterson, 1959):

  • Participants hear a $3$-consonant trigram (like $BKG$), followed by a $3$-digit number (like $347$), and count backward by $3$s to prevent silent rehearsal.

  • After a $3\text{ seconds}$ delay, recall is roughly $80\%$; after an $18\text{ seconds}$ delay, recall drops to roughly $7\%$.

  • Decay vs. Interference:

  • Trace Decay Hypothesis: Memory traces naturally break down over time.

  • Proactive Interference (PI): Earlier learned material disrupts the ability to remember new information. Keppel & Underwood ($1962$) showed that forgetting does not occur on Trial $1$ of the Brown-Peterson task; it only appears across later trials as PI builds up.

  • Release from Proactive Interference (Wickens, Born, & Allen, 1963): Switching the category of items on Trial $4$ (such as switching from numbers to letters) restores recall accuracy back near $100\%$.

  • Adaptive Decay: Altmann & Gray ($2002$) argue that letting traces fade is necessary to clear out old details and avoid interference (such as updating speed limits while driving on different roads).

• Retrieval Dynamics in Short-Term Memory (Saul Sternberg, 1966, 1969):

  • Task: Participants memorize a short list of $1\text{--}6$ items (e.g., $BKFQ$) and decide as fast as possible whether a probe item (e.g., $K$) was in the list.

  • Potential Search Models:

  1. Parallel Search: All items are scanned at the exact same time; search time remains flat as list size grows.

  2. Serial Self-Terminating Search: Items are checked one by one, stopping when a match is found; "Yes" responses take half the time of "No" responses.

  3. Serial Exhaustive Search: Items are checked one by one through the entire list, even after a match is found; "Yes" and "No" responses show identical upward response slopes.

  • Finding: STM uses high-speed Serial Exhaustive Search ($\approx 25\text{--}40\,\text{ms}$ per item). The scanning process runs so quickly that completing the entire list before making a decision is more efficient than checking after each item.

• Baddeley's Working Memory Model (Baddeley & Hitch, 1974; Baddeley, 1986, 2000):

  • Dual-Task Evidence: Holding a $6$-digit load slowed down, but did not stop, sentence reasoning tasks, proving that short-term memory is an active system with multiple components rather than a single static box.

  • Central Executive: The attentional control center; coordinates processes, manages cognitive resources, links long-term knowledge to incoming stimuli, and relies on the prefrontal cortex.

  • Phonological Loop:

  • Phonological Buffer: Temporarily holds verbal and acoustic memory traces.

  • Subvocal Rehearsal Loop: The inner voice that refreshes items through silent rehearsal.

  • Working Memory Span (Daneman & Carpenter, 1980): Measured by reading sentences aloud while retaining their final words (e.g., autumn, delicatessen, necessary); predicts reading comprehension, language learning, and fluid intelligence.

  • Visuospatial Sketch Pad: The inner eye that holds and manipulates visual mental images and spatial layouts.

  • Episodic Buffer (Baddeley, 2000): A multimodal component that combines visual, spatial, and verbal details into ordered stories; explains why some amnesic patients can temporarily remember complete narratives.

Long-Term Memory Storage, Permastore & Forgetting

• Capacity Estimates:

  • Virtually unlimited; stores $50,000\text{--}100,000$ vocabulary words, motor skills, personal experiences, and concepts.

  • Landauer's ($1986$) Analysis: Compared brain synapses ($10^{13}$) and lifetime neural firings ($10^{20}$) with realistic human learning rates, estimating practical retrievable LTM capacity at roughly $\approx 1\text{ billion bits}$ ($10^9\,\text{bits}$) for a $35\text{-year-old}$ adult.

• Coding Format:

  • Primarily semantic and conceptual (meaning-based) rather than sensory.

  • Baddeley ($1966\text{a}$) showed that after a $20\text{-minute}$ delay, lists of words with similar meanings caused heavy confusion, while sound similarity had almost no effect.

• Retention Duration & Permastore:

  • Bahrick's ($1983, 1984$) Spanish Retention Study: Tested $733$ people across $1\text{ to }50\text{ years}$ after taking Spanish classes. Knowledge dropped over the first $3\text{ years}$, but then stabilized into a durable permastore state that remained flat for $30\text{ to }35\text{ years}$ before late-life declines.

• Ebbinghaus's ($1885/1913$) Forgetting Studies:

  • Memorized $85,000$ nonsense syllables (e.g., rur, hal) across $6,600$ lists during $830\,\text{hours}$ of self-study.

  • The Forgetting Curve: Forgetting is non-linear; loss is very rapid at first ($\approx 50\%$ lost in $1\,\text{hour}$, $\approx 70\%$ in $24\,\text{hours}$), then levels off with $\approx 20\text{--}25\%$ retained indefinitely.

• Interference Mechanisms in LTM:

  • Paired-Associates Learning ($A-B, A-C$):

  • Proactive Interference (PI): Prior list learning ($A-B$) impairs recall of the new list ($A-C$). Underwood ($1957$) showed recall drops steadily as the number of prior experimental trials increases.

  • Retroactive Interference (RI): Newly learned material ($A-C$) makes it harder to recall older material ($A-B$) (such as learning German vocabulary disrupting recall of previously learned French).

  • Cue Competition & Overload (Anderson & Neely, 1996): When multiple memories share the same retrieval cue (e.g., parking in various spots tied to "where did I park?"), the targets compete, making retrieval harder.

  • The Fan Effect (John Anderson, 1974): Learning more facts associated with a single concept increases the time needed to retrieve any specific fact about that concept.

  • Directed/Intentional Forgetting: Deliberately clearing out outdated information (like a cook forgetting a completed food order) protects upcoming tasks from proactive interference.

Retrieval Principles, Encoding Specificity & Context Effects

• Categorization & Semantic Clustering:

  • Bousfield ($1953$): Presenting $60$ scrambled words from $4$ distinct categories (animals, names, professions, vegetables) led participants to spontaneously group words by category during recall.

  • Subjective Organization (Mandler, 1967): Imposing your own organizational categories onto unrelated items significantly boosts recall.

  • Deficits: Older adults and individuals with frontal lobe damage often fail to group items spontaneously, contributing to memory difficulties (Stuss et al., 1994).

• Encoding Specificity Principle (Thomson & Tulving, 1970; Tulving & Thomson, 1973):

  • Memory retrieval works best when the cues available at retrieval match the exact cues and thoughts present during initial learning.

  • Weak cues present at encoding (e.g., pairing ground with COLD) are much more effective retrieval cues later than strong outside associates (e.g., hot) that were not present during study.

• Context Effects on Memory:

  • Environmental Context (Godden & Baddeley, 1975): Tested $16$ scuba divers memorizing and recalling $40$ words on land versus $20\text{ feet}$ underwater.

  • Finding: Words learned underwater were best recalled underwater; words learned on land were best recalled on land. Matching study and test environments aids free recall, but does not affect recognition memory (Godden & Baddeley, 1980).

  • Perspective Context (John Anderson): Participants touring a house under the mindset of either a homebuyer or a burglar recalled completely different features (buyers recalled roof condition and room layouts; burglars recalled jewelry, stereos, and exit paths).

• State-Dependent Learning (J. E. Eich, 1980):

  • Matching internal physical or chemical states (e.g., sober vs. intoxicated) during study and test improves free recall compared to mismatched states (though sober study and sober recall produces the highest absolute performance).

Long-Term Memory Subsystems & Semantic Networks

• Taxonomy of Long-Term Memory Systems:

  • Declarative (Explicit) Memory: Conscious, deliberate recall of information.

  • Episodic Memory (Tulving, 1972, 1983): Memory for personally experienced events tied to specific times and places ("Remembering when").

  • Semantic Memory: General world facts, concepts, and vocabulary without personal autobiographical context ("Knowing that").

  • Non-Declarative (Implicit) Memory: Unconscious memory shown through changes in task performance.

  • Procedural Memory: Automated physical and cognitive habits ("Knowing how", such as driving stick shift, playing an instrument, or laying brick).

  • Perceptual Representation System (PRS): Early perceptual priming for visual and auditory forms.

  • Classical Conditioning & Non-Associative Learning: Conditioned responses, habituation, and sensitization.

• Content-Addressable Architecture:

  • Memory works like a search engine: information is accessed directly by its content and features rather than by looking up an arbitrary catalog number (like library shelf codes).

  • Object-Orientation Limit: Semantic memory is indexed by concepts and objects rather than by abstract features. Finding all items that share an abstract feature (such as "fruits ending with the letter e" or "animals with claws") requires sequentially checking each concept one at a time.

• Hierarchical Semantic Network Model (Collins & Quillian, 1969):

  • Concepts are stored as nodes connected by directional links.

  • Principle of Cognitive Economy: General properties are stored at the highest possible category level to avoid redundancy (e.g., "can fly" is stored at Bird; "has skin" is stored at Animal).

  • Verification Latencies: Statements that require traversing fewer levels in the hierarchy are verified faster (Canary is a canary < Canary is a bird < Canary is an animal).

• Spreading Activation Model (Meyer & Schvaneveldt, 1971):

  • Activating one concept node causes excitation to spread outward along connected links, weakening with distance.

  • Semantic Priming: In a lexical decision task, people identify a target word faster when it is preceded by a related word (e.g., bread primes butter faster than chair).

  • Moses Illusion: People commonly answer "two" to "How many animals of each type did Moses take on the ark?" because Moses and Noah activate shared Biblical patriarch concepts, causing the Noah flood schema to fire automatically.

• Connectionist / Parallel Distributed Processing (PDP) Models:

  • Rejects single concept nodes; concepts are represented as distributed patterns of activity across many interconnected processing units.

  • Back Propagation: Connections begin with neutral weights ($0.5$); during training, output is compared to the desired target, and weights are adjusted iteratively to minimize error.

  • Graceful Degradation: Physical damage to units causes gradual, mild loss of function rather than total system collapse.

Implicit Memory, Priming & Process Dissociation

• Implicit vs. Explicit Memory:

  • Explicit Memory: Conscious, deliberate recall of past events.

  • Implicit Memory: Unconscious improvement in task performance resulting from previous experience without intentional awareness (Schacter, 1996).

• Repetition Priming:

  • Faster or more accurate processing of a stimulus because it was encountered recently (e.g., a $30\,\text{ms}$ flash of the word button makes a person more likely to complete _U_TO_ as button).

  • Morphological Sensitivity: Priming is strong for words sharing root meanings (sees primes seen), but weak for words that only sound or look similar (sees does not prime seed or seize). Nonwords produce little priming because they lack existing semantic nodes.

  • Persistence: Repetition priming can persist for over a week, even when participants have no conscious memory of seeing the words (Schacter, 1996).

• Dissociations in Amnesia:

  • Warrington & Weiskrantz ($1970$): Tested $4$ amnesic patients and $8$ control participants.

  • Result: Amnesic patients did very poorly on explicit tasks (free recall and recognition), but performed just as well as healthy controls on implicit tasks (word fragment identification and word stem completion).

  • Korsakoff's Syndrome (Korsakoff, 1889): An amnesic patient who received a mild electric shock forgot the experience explicitly, but refused to touch the shock generator case later, fearing he would be electrocuted.

• Process Dissociation Framework (Larry Jacoby, 1991):

  • Argues that memory tasks are rarely pure measures of one system; tasks involve a mix of Intentional (controlled) and Automatic (familiarity) processes.

  • The False Fame Paradigm (Jacoby, Woloshyn, & Kelley, 1989):

  • Participants studied non-famous names under Full Attention versus Divided Attention.

  • After $24\text{ hours}$, they rated the fame of names from a mixed list containing famous people, studied non-famous names, and new non-famous names.

  • Result: Participants in the divided-attention condition mistakenly rated previously seen non-famous names as famous because automatic familiarity remained, but they lacked the controlled recollection to remember where they saw the names.

Levels of Processing Framework & Encoding Principles

• Levels of Processing Theory (Craik & Lockhart, 1972; Craik & Tulving, 1975):

  • Memory retention depends entirely on how deeply and meaningfully information is analyzed when it is first encountered.

  • Incidental Learning Depths:

  1. Physical/Structural Processing (e.g., "Is the word in capital letters?") $\rightarrow$ Shallowest depth, poorest recall.

  2. Acoustic/Phonemic Processing (e.g., "Does the word rhyme with wait?") $\rightarrow$ Medium depth and recall.

  3. Semantic Processing (e.g., "Does the word fit in this sentence?") $\rightarrow$ Deepest depth, superior recall.

  • Elaboration Effect: Words placed in complex, descriptive sentences are remembered better than words in simple sentences because richer conceptual associations are formed.

  • Face Processing (Bower & Karlin, 1974): Rating faces for honesty (deep semantic judgment) produces better recognition memory than rating faces for gender (shallow physical judgment).

  • Critiques (Baddeley, 1978): Defining "depth" can be circular without independent measurement; theory fails to explain why sound-based matching can outperform semantic processing when the test relies on sound (Transfer-Appropriate Processing).

• Generation Effect:

  • Information that you actively generate yourself from cues or problem-solving is remembered much better than information you passively read or hear.

Reconstructive Memory, Autobiographical Recall & Eyewitness Testimony

• Reconstructive Memory & Schemata (Frederick Bartlett, 1932):

  • Rejects the view that memory is an exact recording; instead, retrieval is an active reconstruction that uses schemata (cultural mental frameworks) to fill in missing details.

  • Method of Serial Reproduction: Participants recalled the Native American folktale "The War of the Ghosts" across different time delays.

  • Distortions: Over time, retellings became shorter, omitted supernatural elements, and changed unfamiliar ideas into familiar Western concepts (e.g., "foggy and calm" became a "dark and stormy night").

• Autobiographical Memory Studies:

  • Marigold Linton's ($6\text{-Year}$) Self-Study: Recorded $2,003$ daily events on index cards. Proved that real-world life memories are remarkably durable, and that dating forgotten events relies on logical deduction (e.g., season, day of week, concurrent job).

  • William Brewer's ($1988$) Beeper Study: Students carried beepers going off randomly every $\approx 2\text{ hours}$ to log what they were doing. Retested at $21\text{ months}$ and $41\text{ months}$, participants remembered actions better than thoughts, and remembered rare, distinctive events (such as a first college Thanksgiving trip home) best.

• Flashbulb Memories (Brown & Kulik, 1977):

  • Highly vivid, detailed memories of learning about dramatic, shocking events (e.g., 9/11, Princess Diana's death, 1989 San Francisco earthquake).

  • Personal Account: A professor in a Stanford office during the $1989$ earthquake recalled building shaking, books falling, retrieving a dissertation floppy disk, bridge collapses ($30\text{ killed}$), and cold darkness.

  • Controversy: High emotion activates the amygdala to boost initial encoding, but frequent retellings (Neisser, 1982b) reshape the narrative over time. Objective tests show flashbulb memories lose accuracy at the same rate as ordinary memories, but personal confidence in their accuracy stays unusually high.

• Eyewitness Memory & The Misinformation Effect (Elizabeth Loftus):

  • Eyewitness accounts are persuasive to juries, even when inaccurate.

  • Stop/Yield Sign Study: Showing an auto accident and introducing misleading post-event questions ("Did another car pass the red Datsun while it was stopped at the stop sign?") dropped slide recognition accuracy from $75\%$ down to $41\%$.

  • Non-Existent Barn Study (Loftus, 1975): Asking "How fast was the sports car going when it passed the barn?" caused $17\%$ of participants to report seeing a barn a week later (compared to $< 3\%$ of control participants).

  • Sentence Integration (Bransford & Franks, 1971): Reading separate simple sentences caused people to later recognize complex integrated sentences with highest confidence, even though the integrated sentences were never presented.

• The Recovered / False Memory Debate:

  • Examines whether recovered trauma memories are genuine repressed memories or therapist-suggested false memories stemming from guided imagery and suggestive books like The Courage to Heal (Bass & Davis, 1988).

  • Lost in the Mall Study (Loftus & Pickrell, 1995): $24$ participants read booklets containing $3$ true childhood stories and $1$ false story (getting lost in a mall at age $5$). $29\%$ ($7/24$) endorsed the false event, and $25\%$ maintained false recollections across two follow-up interviews.

  • Hyman, Husband, & Billings ($1995$): Successfully led $\approx 25\%$ of participants to form false memories of childhood hospitalizations or clown birthday parties.

  • Brain Activity Differences (Cabeza et al., 2001): True word recognition activates both the posterior parahippocampal gyrus (perceptual details) and the hippocampus, whereas false semantic recognition activates the hippocampus alone.

  • Deese-Roediger-McDermott (DRM) Paradigm: Studying semantically related words (bed, rest, awake) triggers confident false recall of the missing theme word (sleep).

  • Guided Imagery Risks (Arbuthnott, 2005): Repeated guided imagery makes imagined events feel real, presenting legal and clinical risks.

Visual Imagery, Mental Rotation & Spatial Cognition

• Mnemonics & Imagery Mediators:

  • Visual images act as mental bridges connecting target items to retrieval cues.

  • Dual-Coding Hypothesis (Allan Paivio, 1969, 1971, 1983):

  • Long-term memory uses two independent formats: Verbal (language-based) and Imagery (mental pictures).

  • Concrete nouns evoke both codes; abstract nouns evoke only verbal codes.

  • Relational-Organizational Hypothesis (Gordon Bower, 1970b):

  • Imagery improves memory by multiplying interactive connections between items rather than just visual detail.

  • Bower ($1970\text{b}$): Interactive imagery ($53\%$ recall) vastly outperformed non-interactive imagery ($27\%$) and rote repetition ($30\%$).

• Empirical Investigations of Mental Imagery:

  • Lee Brooks ($1968$) Interference Task:

  • Visual Task: Mentally scanning corners of a block letter $F$ was $2.5\times$ slower when pointing to answers than when speaking.

  • Verbal Task: Scanning words in a sentence was slower when speaking than when pointing.

  • Conclusion: Visual imagery uses spatial processing systems separate from verbal processing systems.

  • Symbolic-Distance Effect (Moyer, 1973; Paivio, 1975): Deciding which imagined animal is larger takes less time when size differences are extreme (whale vs. cockroach) than when they are close (hog vs. cat), mirroring real vision.

  • Mental Rotation Experiments (Shepard & Metzler, 1971; Cooper & Shepard, 1973, 1975; Cooper, 1975, 1976):

  • 3D Objects (Shepard & Metzler, 1971): Decision time increased in direct linear proportion to rotation angle ($0^\circ\text{ to }180^\circ$), with identical speeds for picture-plane and depth rotations, proving continuous mental rotation.

  • Letter Rotation (Cooper & Shepard, 1973): Orientation cues presented $1,000\,\text{ms}$ early eliminated rotation time penalties.

  • Polygons (Cooper, 1975, 1976): Irregular shapes ($6\text{ to }24\text{ points}$) rotated at identical rates regardless of complexity, proving the entire object is rotated continuously.

  • Imaginal Scanning (Stephen Kosslyn, 1973, 1978):

  • Island Map: Participants memorized a map with $7$ landmarks and $21$ paths ($2\text{ to }19\,\text{cm}$). Scanning times between imagined locations showed a direct linear correlation with physical distance.

  • Distortions & Limits of Visual Imagery:

  • Map Heuristics (Barbara Tversky, 1981): Alignment rules distort mental maps (e.g., mistakenly thinking Reno is west of Los Angeles, or Cranbrook BC is east of Grande Prairie AB).

  • Ambiguous Figures (Chambers & Reisberg, 1992): People cannot mentally reverse an imagined duck/rabbit figure without drawing it, because mental images are tied to initial interpretations.

  • Reasoning Interference (Knauff & Johnson-Laird, 2002): Visual relations (cleaner-dirtier) slow down deductive reasoning compared to spatial relations (front-back) because generating mental images takes up working memory resources.

• Finke's Five Principles of Visual Imagery (Ronald Finke, 1989):

  1. Implicit Encoding: Unintentionally stored physical features can be extracted from mental images (e.g., counting kitchen cabinet doors from memory).

  2. Perceptual Equivalence: Visual imagery shares underlying neural systems with physical perception (Perky, 1910; Farah, 1985 letter detection).

  3. Spatial Equivalence: Mental layouts preserve physical distances and spatial arrangements (Kosslyn island map; Kerr, 1983 haptic map scanning in congenitally blind participants).

  4. Transformational Equivalence: Imagined transformations follow the continuous kinematic laws of physical motion (mental rotation).

  5. Structural Equivalence: Mental images are assembled piece-by-piece; complex images take longer to construct (Kosslyn et al., 1983).

• Theoretical Critiques of Imagery:

  • Tacit Knowledge & Demand Characteristics (Zenon Pylyshyn, 1981): Imagery tasks are "cognitively penetrable"; participants pause during image scanning because they expect scanning should take time.

  • Experimenter Expectancy Effects (Intons-Peterson, 1983): Unintentional experimenter cues bias participant speeds.

  • Propositional Theory (Anderson & Bower, 1973; Pylyshyn, 1973): Claims information is stored as language-like propositions rather than images. Disproved by Kosslyn ($1976$): using imagery leads participants to verify large physical features (cat has head) faster than small, high-association features (cat has claws).

• Neuropsychological Foundations of Imagery:

  • Cerebral Blood Flow (Roland & Friberg, 1985): Imagery selectively activates occipital and visual cortices.

  • PET & Image Size (Kosslyn et al., 1995): Small images activate posterior visual cortex; large images activate anterior visual cortex.

  • Auditory Imagery (Zatorre et al., 1996): Imagining song pitch changes activates secondary auditory cortex in temporal lobes.

  • fMRI Category Activation (O'Craven & Kanwisher, 2000): Imagining faces activates the Fusiform Face Area (FFA); imagining places activates the Parahippocampal Place Area (PPA).

• Spatial Cognition & Spatial Categories (Barbara Tversky, 2005):

  • Space of the Body: Knowledge of one's own bodily position and internal physical sensations.

  • Space Around the Body: Immediate surroundings mapped across three body axes (front-back, up-down, left-right).

  • Space of Navigation: Large-scale mental representations (cognitive maps) used to travel through environments.

Neuropsychology of Memory & Amnesic Syndromes

• Neuroanatomy of Memory:

  • Hippocampus & Medial Temporal Lobes (MTL): Consolidates new declarative memories from short-term into long-term storage.

  • Amygdala: Regulates emotional arousal and strengthens consolidation.

  • Prefrontal Cortex: Coordinates central executive operations, strategy, planning, and source monitoring.

  • Occipital & Temporal Cortices: Modality-specific visual and auditory perceptual processing and imagery.

• Cellular Foundations of Memory:

  • The Hebb Rule (Donald Hebb): Simultaneous activation of pre- and post-synaptic neurons strengthens their connection ("neurons that fire together, wire together").

  • Long-Term Potentiation (LTP): Repeated electrical stimulation in hippocampal pathways produces enduring increases in postsynaptic sensitivity.

• Clinical Amnesic Syndromes:

  • Anterograde Amnesia:

  • Inability to form new explicit/declarative long-term memories after injury.

  • Preserved Functions: Intact working memory, intact procedural skill learning, preserved repetition priming, and intact pre-injury semantic knowledge.

  • Retrograde Amnesia:

  • Loss of memories formed prior to brain injury.

  • Temporal Gradient (Ribot's Law): Memory loss is most severe for events right before the injury; older remote memories are better preserved because they have undergone more consolidation.

  • Standard Model vs. Multiple Memory Trace (MMT) Theory (Nadel & Moscovitch, 1997): The standard model claims hippocampal dependency fades over time; MMT theory claims the hippocampus is always needed for detailed episodic retrieval, and that spared remote memories are generalized "semanticized family lore." Ryan et al. ($2000$) fMRI confirmed that recalling $25\text{-year-old}$ memories activates the hippocampus just as much as recent memories.

• Classic Amnesic Case Studies:

  • H.M. (Henry Molaison) (Scoville & Milner, 1957):

  • Underwent bilateral medial temporal lobe resection in $1953$ to treat severe epilepsy. IQ rose to $112$ and short-term memory remained intact, but he experienced profound permanent anterograde amnesia and retrograde amnesia for several years prior to surgery, proving the MTL consolidates LTM but does not house STM.

  • Clive Wearing (Deborah Baddeley, 1990):

  • Encephalitis damaged temporal and frontal regions, leaving a memory span of only seconds to minutes; repeatedly wrote in his diary that he had just awakened ($3:10, 3:15, 3:20$), but retained procedural memory to play piano and conduct choirs.

  • Gene (Schacter, 1996):

  • A motorcycle accident damaged frontal/temporal lobes and the left hippocampus; suffered total loss of episodic memory (could not recall his brother's drowning or a local chemical spill), but retained intact semantic memory (could define technical terms and name former co-workers).

  • Frontotemporal Encephalitis Patient (Schacter, 1996):

  • The mirror image of Gene; lost semantic memory (forgot common word meanings, famous people, and animal colors), but retained rich episodic memories of her wedding, honeymoon, and father's death.