Process Models of Memory and Forgetting Study Guide

Evaluation of Memory Store Models vs. Process Models

• Transitioning from the Modal Model: The "Modal Model" of memory, proposed by Atkinson & Shiffrin (1968), characterizes memory as a linear progression through structurally distinct stores: Sensory Store, Short-Term Store (STS), and Long-term Store (LTM).
• The Four Central Questions regarding Process Models:
  1. Is the "Sensory Store" actually a memory store or a persistence of visual processing?
  2. Should the "Short-Term Store" be replaced by a functional "Working Memory" model?
  3. Is all rehearsal equally effective for long-term retention?
  4. Do humans ever truly forget information, or is forgetting merely a failure of retrieval?

Sensory Store: Decay from Memory vs. Visible Persistence

• Classic Perspective (Sperling, 1960):
  • Methodology: Using a tachistoscope, Sperling presented a 12-letter matrix for a very brief duration (typically $<100\,ms$).
  • Findings: Participants in the "Full Report" condition could only recall 3 or 4 letters, describing a sensation of items fading before they could be written down (the "reporting bottleneck"). In the "Partial Report Technique," participants were cued for a specific row after the stimulus disappeared and could report nearly 100% of the row (3 or 4 out of 4 letters).
  • Interpreted Conclusion: Humans have a sensory store with high capacity but very brief duration ($<1\,sec$).
• Dot-Pattern Fusion (Eriksen & Collins, 1967):
  • Procedure: Sequential presentation of two random dot patterns ($S1$ and $S2$). If fused, they form a syllable (e.g., "VOH").
  • The Inter-Stimulus Interval (ISI) Manipulation: Authors increased the gap between $S1$ and $S2$. Performance degraded over a few hundred milliseconds, supporting the notion of an "iconic memory" decaying over roughly $300\,ms$.
• The Challenge by Di Lollo (1970s, 1980s):
  1. Visible Persistence Theory: Visual processing begins at stimulus onset and persists for a fixed duration ($150-300\,ms$), regardless of stimulus duration.
  2. Critical Manipulation: S1 duration.
  3. Contrasting Predictions:
     • Sensory Memory Account (Eriksen & Collins): Decay starts at stimulus offset. Increasing S1 duration with a constant ISI should NOT affect fusion performance.
     • Visible Persistence Account (Di Lollo): Processing starts at onset. Increasing S1 duration (thereby increasing the Stimulus Onset Asynchrony or SOA) should decrease effective overlap and fusion performance, even if there is no ISI.
  • Result: Hogben & Di Lollo (1974) and Di Lollo (1977, 1980) found that fusion performance gets worse as S1 duration increases. This provided evidence for the visible persistence explanation rather than decay from a sensory memory store.
• Refining the Concept (Coltheart, 1980): Suggested three distinct phenomena:
  1. Neural Persistence: Extremely brief overlap in neural activity.
  2. Visible Persistence: Overlap in visual processing (Di Lollo's findings), duration $<200\,ms$.
  3. Informational Persistence: The "icon" that decays (Sperling's findings), duration $~150-300\,ms$.
• Critical Evaluation (Haber, 1983): Criticized iconic memory research for ecological invalidity, famously stating: "The notion of an icon… cannot possibly be useful in any typical visual information-processing task except reading in a lightning storm."

From Short-Term Memory to Working Memory

• Functions of the Short-Term Store (STS):
  • Gating conscious report.
  • Facilitating rehearsal and transfer into LTM.
  • Recognized as an active "executive" system rather than a passive storage bin.
• The Working Memory Model (Baddeley & Hitch, 1974):
  • Dual-Task Paradigm: If STS has limited capacity, performing two tasks simultaneously should cause interference. Participants rehearsed 0-8 digits (secondary task) while performing reasoning tasks like "A precedes B: AB (True)" (primary task).
  • Results: Error rates remained constant. Reasoning time increased significantly but only by $35\%$. This suggested STS can "divide to conquer," leading to the multi-component Working Memory (WM) model.
• Components of the Working Memory Model:
  1. Central Executive: An attentional system that coordinates the two "slave" systems, manages retrieval strategies, and selective attention. It is a workspace where storage and processing demands compete.
  2. Phonological Loop: Consists of the Phonological Store (inner ear; holds traces for 1-2 seconds) and the Articulatory Control Process (inner voice; subvocal rehearsal).
  3. Visuospatial Sketch-pad: A workspace for storing and manipulating images to guide behavior.

Detailed Evidence for the Phonological Loop

• Phonological Confusability Effect: Lists that sound similar (e.g., G, C, P, T) are harder to recall than those that sound distinct.
• Word-Length Effect (Baddeley, 1975): Memory span is greater for short words than long words. This correlates with articulation rate rather than the number of syllables.
  • Language Statistics:
    • Chinese: Articulation rate $265\,ms/digit$, Digit span $9.9$.
    • English: Articulation rate $321\,ms/digit$, Digit span $6.6$.
    • Welsh: Articulation rate $385\,ms/digit$, Digit span $5.8$.
• Articulatory Suppression: Preventing rehearsal by repeating irrelevant items (e.g., saying "the, the, the…") eliminates the word-length effect for written words because the Articulatory Control Process is occupied. Spoken words bypass this and enter the Phonological Store directly.
• Utility of the Phonological Loop:
  • Learning to Read: Reduced span correlates with lower reading ability.
  • Vocabulary Acquisition: Correlation between non-word repetition ability and vocabulary size (Gathercole & Baddeley, 1989).
  • Language Comprehension: Patients with STS impairments struggle with complex sentence syntax.

The Visuospatial Sketch-pad

• Anecdotal Evidence: Baddeley (1997) noted that visualizing a radio broadcast of an American football game caused him to weave his car across the freeway because the spatial imagery interfered with driving.
• Brooks Matrix Task (1967):
  • Spatial condition: Remembering sentences like "In the next square to the right put a 2."
  • Non-spatial condition: "In the next square to the quick put a 2."
  • Findings: Spatial performance was better with auditory instructions (spatial-spatial interference with written text), while non-spatial was better with written instructions.
  • Dual-Task Distraction: A secondary pursuit rotor task (spatial distractor) disrupted the spatial condition but not the nonsense (non-spatial) condition (Baddeley et al., 1975).

Challenges to the Working Memory Model

• Articulatory Suppression Paradox: It does not fully prevent the registration of visually presented words.
• Neuropsychological Data: STM patients often show similar deficits in both visual and verbal spans (e.g., digit span of 2, visual span of 4), contrary to independence.
• Rehearsal Questions: It is unclear how pre-verbal children or how adults rehearse non-verbal visual information.
• The Binding Problem: How does consciousness "bind" modality-specific info? This led to the addition of the Episodic Buffer (Baddeley, 2000) in later models.

Levels of Processing (LoP)

• Core Concepts (Craik & Lockhart, 1972):
  • Memory is a byproduct of perception and comprehension.
  • Rejection of structural stores; retention depends on the "depth" of processing.
• Hierarchy of Processing Depth:
  • Orthographic (Shallow): Physical structure (e.g., "Is the word in upper case?"). Poor retention.
  • Phonological (Medium): Sounds (e.g., "Does the word rhyme with mat?").
  • Semantic (Deep): Meaning (e.g., "Does the word fit in this sentence?"). Best retention.
• Types of Rehearsal:
  • Type I (Maintenance): Rote repetition; does not improve long-term retention.
  • Type II (Elaborative): Linking info to existing knowledge; increases retention.
• Experimental Support:
  • Craik & Watkins (1973): In a task to remember the last word starting with "g," the amount of time spent rehearsing (maintenance) did not improve subsequent recall.
  • Craik & Tulving (1975): High elaboration (complex sentences) led to better retention than low elaboration.
• Criticisms of LoP:
  • Circularity: Defining depth by the fact it leads to better memory, then explaining better memory by depth.
  • Transfer-Appropriate Processing (Morris, Bransford & Franks, 1977): Memory depends on the match between encoding and test processes. If the test is a rhyme test, "shallow" rhyme encoding actually outperforms "deep" semantic encoding.

Theories of Forgetting

• The Forgetting Curve (Ebbinghaus, 1885): Using nonsense syllables, Ebbinghaus identified a logarithmic loss of information over time. The formula for the curve is $y = \frac{1.84}{(\log(t))^{1.25} + 1.84}$.
• Theories for Forgetting Mechanisms:
  1. Decay: Memory fades with time; however, decay is argued to be an active process.
  2. Interference: Memory traces are disrupted by other learning.
     • Cockroach Study (Minami & Dallenbach, 1946): Moving cockroaches forgot $70\%$ whereas immobilized cockroaches only forgot $25\%$.
     • Proactive Interference (PI): Prior learning disrupts new learning. Performance declines over successive trials but recovers when the category changes ("Release from PI," Wickens, 1970).
     • Retroactive Interference (RI): New learning disrupts old learning (Slamecka, 1980).
  3. Retrieval Failure (Cue-Dependent Forgetting):
     • Encoding-Specificity Principle (Tulving, 1979): Success depends on informational overlap between encoding and retrieval contexts.
     • Extrinsic Context: Godden & Baddeley (1975) showed divers who learned lists underwater recalled them better underwater than on land (and vice versa). This effect occurred for free recall but not recognition.
     • Intrinsic Context: Barclay et al. (1974) showed cues like "something melodious" worked for "The man tuned the PIANO," but poorly for "The man lifted the PIANO."
     • Mood Congruency: Memory is better if the mood at study and test matches (Eich & Metcalfe, 1989).
  4. Motivated Forgetting/Repression:
     • Freud (1901): Unconscious repression of trauma.
     • Anderson & Green (2001): "Think/No-Think" paradigm; suppressing a memory reduces its later recall.
     • Emily Holmes (2009): Playing Tetris after a traumatic event reduces intrusive flashbacks.
  5. Consolidation Failure:
     • Molecular Consolidation: Protein synthesis and synaptic strengthening.
     • Systems Consolidation: The Hippocampus mediates strengthening of cortical connections; eventually, the Hippocampus is no longer needed (Dudai, 2004).

Questions & Discussion

• Q: Fusion of two consecutive, brief visual displays is dependent on what?
• A: The SOA (Stimulus Onset Asynchrony), which supports the visible processing view.
• Q: Does word-length effect depend on spoken duration for written words?
• A: Yes, suggesting written words are articulated in working memory.
• Q: What is the negative effect of learning previous lists on the ability to learn a current list?
• A: Proactive interference.
• Case Study (The woman who never forgets): Mentioned as a counter-example to standard forgetting (AJ).
• Neurostimulation (Penfield, 1958): Evoked "memories" by stimulating temporal lobes in $40$ out of $520$ patients; most resembled dreams rather than absolute historical records.