Memory Systems and Acquisition Study Guide to Acquisition

Exam 1 Results and General Announcements

Performance Statistics

• Mean Score: $73.4\%$

• Median Score: $77\%$

• Standard Deviation: $16.8\%$

Administrative Notes

• Students are encouraged to review their individual exam results during Office Hours (OH).

• A curve estimate will be created after Exam 2. This estimate will provide a precise calculation of the current course grade inclusive of the curve.

Chapter 6: Acquisition of Memory and the Short-Term/Working Memory System

Lecture Outline

• General Structure of Memory

• The Modal Model

• Working Memory

• Entering Long-term Storage

• Elaborate Encoding

• Organizing and Memorizing

• Links Among Acquisition, Storage, and Retrieval

What Is Memory?

Formal Definition

Memory refers to the processes involved in retaining, retrieving, and using information about stimuli, images, events, ideas, and skills after the original information is no longer present.

The Three Stages of Memory

1. Encoding: The process where incoming information is acquired and processed. Example: When studying for an exam, this may involve listening to a lecture, reading a textbook, or observing an experiment.

2. Storage: The creation of a record of the encoded information which is then stored in the brain. Example: When learning a new language, storage occurs as vocabulary words and grammar rules are retained for future use.

3. Retrieval: The process of recovering information from memory stores to bring it to consciousness or use it to execute a behavior (e.g., a motor act).

• Forgetting: This occurs when memory fails and can happen at any of the three listed stages.

Human Memory Caveats

• Human memory stages are not as distinct or independent as those in a computer system.

• Dependency on Prior Knowledge: New learning is inextricably linked to previously learned information.

• Future Utility: "Successful" acquisition depends heavily on how that information is intended to be used later.

The Modal Model of Memory

Historical Context

• Proposed by Atkinson and Shiffrin (1968).

• This model uses the computer as a metaphor for human cognition.

• It posits that memory is an integrated system that processes information by acquiring, storing, and retrieving it. The components of this system do not act in isolation. Example:

System Constraints

Memory has a limited capacity characterized by:

• Limited space

• Limited resources

• Limited time

Model Architecture

1. Incoming Information enters Sensory Memory.

2. Information moves to Short-term Memory (STM).

3. Maintenance Rehearsal keeps information in STM; otherwise, it is Lost.

4. Information is transferred to Long-term Memory (LTM).

5. Retrieval brings information from LTM back into STM.

Control Processes

These are active processes that can be controlled by the individual:

• Rehearsal

• Encoding Strategies: Used to make a stimulus more memorable.

• Attentional Strategies

Everyday Example: Ordering a Pizza

• (a) Sensory Input: You see the number "555-5100" on a screen; it enters sensory memory.

• (b) Focus: You focus on the number, and it enters STM.

• (c) Rehearsal: You repeat the number to keep it in STM while making the call.

• (d) Storage: You memorize the number, storing it in LTM.

• (e) Retrieval: When you want to order again later, you retrieve the number from LTM, moving it back into STM for awareness.

Sensory Memory

Characteristics

• Registers most or all information that hits sensory receptors.

• Iconic Memory: Visual sensory memory.

• Echoic Memory: Auditory sensory memory.

• Information decays very quickly but is initially very accurate.

• Functions: Collects information for initial processing and "fills in the blanks" (e.g., persistence of vision in film frames or a sparkler’s trail).

Measuring Capacity and Duration (Sperling, 1960)

Sperling used an array of 12 letters flashed briefly to determine sensory memory limits.

• Whole Report Method: Participants asked to report as many letters as possible. Average result: $4.5$ out of $12$ letters ($37.5\%$).

• Partial Report Method: Participants heard a tone indicating which specific row to report. Average result: $3.3$ out of $4$ letters ($82.5\%\text{ per row}$). This suggests participants had access to all rows initially.

• Delayed Partial Report: The tone was delayed by a fraction of a second. Performance decreased rapidly as time increased, showing the rapid decay of sensory memory.

Short-Term Memory (STM) vs. Long-Term Memory (LTM)

Short-Term Memory

• Function: Stores small amounts of information for a brief duration.

• Content: Includes both new sensory information and info recalled from LTM.

• Example: ?

Long-Term Memory

• Capacity: Theoretically unlimited.

• Duration: Unlimited time period.

• Accessibility: Getting information in or out of STM is generally easier than LTM, though LTM benefits significantly from cues and hints.

Modern Perspectives on the Modal Model

• While the 1968 model is a textbook staple, Atkinson and Shiffrin originally advanced it as a general-purpose modeling framework.

• Working Memory Mention: They actually used the term "working memory" in 1968, describing it as having verbal and visual short-term stores.

• Patient HM: They justified the STM/LTM distinction using the amnesic patient HM, who had an intact verbal STM but could not form new LTMs.

• Depth of Processing: The model acknowledged that some "coding processes" are more effective for LTM transfer than others.

Working Memory (WM)

Definition and Distinction

• Working Memory: A limited capacity system for temporary storage and manipulation of information during complex tasks (comprehension, learning, reasoning).

• Key Difference: Unlike the original STM (often viewed as a single storage "box"), WM consists of multiple parts and is concerned with processing/manipulation during cognition.

Serial Position Curve: Evidence for the Modal Model

• Pracy Effect: Better memory for items at the start of a list. This occurs because rehearsal allows transfer from WM to LTM. Retrieval at test comes from LTM. Linked to the hippocampus.

• Recency Effect: Better memory for items at the end of a list. These items are still fresh in WM and are retrieved directly from there. Linked to the perirhinal cortex.

• Manipulations:     - A $30\,s$ filled delay (task-heavy) eliminates the recency effect.     - A $30\,s$ unfilled delay does not wipe out recency.     - Slow presentation of items aids the primacy/pre-recency items but leaves the recency effect unchanged.

Working Memory: Duration

• Measurement: Participants read three letters, a number, and then count backward by threes to prevent rehearsal.

• Results:     - After $3\,s$ of counting: $80\%$ recall.     - After $18\,s$ of counting: $10\%$ recall.

• Conclusion: Without rehearsal, WM lasts approximately $15$-$20\,s$.

• Proactive Interference: This occurs when previously learned information interferes with the learning of new information. It is often the cause of forgetting in WM rather than simple decay.

Working Memory: Capacity

• Digit Span: The number of digits a person can remember (typically $5$-$8$ items).

• Chunking: Combining small units into larger, meaningful units (chunks). Chunks are elements strongly associated with each other but weakly associated with other chunks.

• Case Study: S.F. (Ericsson et al., 1989): A student with an initial digit span of $7$ trained for $230$ hours and reached a span of $79$ digits by chunking them into meaningful units.

• Chess Study (Chase & Simon, 1973): Chess masters outperformed beginners when pieces were in legitimate game positions (due to chunking) but performed the same as beginners when pieces were randomly placed. (study this)

Operation Span

• A modern measurement of WM as it is "working."

• Reading Span: Reading sentences and remembering the last word of each.

• Arithmetic Task: Performing math while remembering words.

• Significance: Operation span correlates strongly with verbal SAT scores, reasoning ability, and intelligence tests, whereas digit span does not.

Individual Differences and Efficiency (Vogel et al., 2005)

• Used ERP (Event-Related Potentials) to measure brain activity during a memory task involving red items and distractors.

• Findings: High-capacity participants were more efficient at ignoring distractors, showing less unnecessary neural activation compared to low-capacity participants.

Coding in Working Memory

Types of Representation

1. Physiological: Representation via neuron firing.

2. Mental: Representation in the mind.

Preferred Coding Methods

• Auditory Coding: Conrad (1964) found that errors in letter recall usually involve letters that sound alike (e.g., mistaking "F" for "S") rather than look alike (e.g., "F" for "E"). This is the default form of WM coding.

• Visual Coding: Della Sala (1999) showed participants could recreate patterns of up to $9$ items for visual info that is hard to verbalize.

• Semantic Coding: Wickens et al. (1976) demonstrated "release from proactive interference." When participants switched to a new category of words on the 4th trial, memory increased, proving meaning is used in WM.

Baddeley & Hitch’s Working Memory Model

Components

1. Phonological Loop: Deals with verbal and auditory information.     - Phonological Similarity Effect: Confusion between similar-sounding letters/words.     - Word-length Effect: Memory is better for short words because they take less time to rehearse (subvocal speech) and produce.     - Articulatory Suppression: Repeating a redundant word (e.g., "the, the, the…") prevents rehearsal, reduces memory span, and eliminates the word-length effect.

2. Visuospatial Sketchpad: Used for visual imagery and spatial reasoning.

3. Central Executive: The attention controller. It divides/switches attention, suppresses irrelevant info/habits, and sets goals.

4. Episodic Buffer (Added 2000): A backup store that communicates with LTM and WM. It has a greater capacity and holds info longer than the loop or sketchpad.

Component Overload (Brooks, 1968)

• Task 1 (Sentence/Phonological): Deciding if words are nouns. Responding verbally (phonological) was hard; responding by pointing (visuospatial) was easy.

• Task 2 (Mental 'F'/Visuospatial): Scanning corners as interior/exterior. Responding by pointing (visuospatial) was hard; responding verbally (phonological) was easy.

• Principle: Performance suffers when the task and response draw on the same specific resource (component). WM can process different types of info simultaneously but struggles with similar types presented at once.

Working Memory and the Brain

• Prefrontal Cortex (PFC): Responsible for processing incoming visual and auditory info. Damage to the PFC results in failure on delayed-response tasks (monkeys perform at $50\%$).

• Neural Firing (Funahashi et al., 1989): Single-cell recordings show neurons in the PFC fire during the delay between stimulus presentation and response, keeping info available as long as they fire.

Entering Long-Term Storage

Rehearsal Types

• Maintenance Rehearsal: Simple recitation. Unlikely to transfer info to LTM (e.g., a phone verification code).

• Elaborative (Relational) Rehearsal: Linking info to other knowledge. Significantly more likely to transfer to LTM.

Levels of Processing Theory

• Shallow Processing: Superficial attention to physical traits; poor memory.

• Deep Processing: Close attention to meaning; good memory.

• Intent vs. Depth: Hyde & Jenkins found that intent to learn has little effect; performance is driven by the depth of processing during study. However, intent may indirectly help if it leads to using better strategies.

• Brain Correlates: Successful encoding shows greater activation in frontal areas, the hippocampus, and adjacent areas.

The Role of Meaning and Organization

• Indexing: Deep processing creates retrieval paths to information.

• Elaborate Sentences: Craik & Tulving (1975) showed that words in complex sentences (e.g., "The great bird swooped down…") are remembered better than in simple sentences ("She cooked the chicken") because they provide richer retrieval paths.

• Mnemonics:     - Peg-word systems: Hanging items on known pegs ("One is a bun").     - First-letter mnemonics: (e.g., ROY G. BIV).

• Understanding: Ambiguous material is remembered better if it is identified/understood first (Wiseman & Neisser, 1974).

Questions & Discussion

Learning Check #1

• Prompt: How many words did you remember from the first five, middle five, and last five?

• Concept: Demonstrates the serial position curve (primacy and recency effects).

Learning Check #2

• Question: Sound-based mistakes in working memory suggest that:     - a) Working memory is able to code visual information     - b) Working memory is influenced by semantic information     - c) Working memory’s preferred method of coding is auditory     - d) Working memory has limits to capacity and duration     - e) All of the above

• Answer: c) Working memory’s preferred method of coding is auditory.

Learning Check #3

• Which statement(s) about encoding is/are true? (Check all that apply)     - A: Counting vowels vs. fitting in a sentence (False; vowel counting is shallow, sentence is deep).     - B: Asking pleasantness vs. telling people "learn these" (True; pleasantness is deep; intent doesn't matter more than depth).     - C: Elaborative sentences overload WM (False; they aid retrieval paths).     - D: Neurological data shows differences in activation for remembered vs. forgotten items (True).     - E: Intent to learn produces strong traces because it correlates with depth of processing (True).

• Correct Selections: B, D, E.