Lecture 2 Notes: Structure of Memory

Structure of Memory

Influential Models of Memory

Sensory Memory
Short-Term Memory
Long-Term Memory

Learning Objectives

  • Demonstrate knowledge of influential models of memory and their possible short comings.

  • Demonstrate knowledge of experimental work that may support / contest these models.

  • Demonstrate knowledge of classic memory phenomena.

  • Demonstrate knowledge of recent debates in memory research.

  • Demonstrate an appreciation of various experimental methodologies that have been used in memory research.

Short Term Memory versus Working Memory

Short-term memory is defined as “a simple storage buffer, the capacity of which is determined by practiced skills and strategies, such as rehearsal and chunking”.

Working memory is more complex. It consists of a storage component as well as an attention component.

The function of working memory is to maintain memory representations in the face of concurrent processing, distraction, and/or attention shifts.

The extent to which a task demands Working Memory Capacity (WMC) is determined by the extent to which it requires the maintenance of activation to memory representations that could otherwise be lost from the focus of attention due to interference or decay.

Short-Term Memory Tasks

Four versions of a word span task are used to assess short-term memory.

Working Memory Tasks

Reading span, operation span, and counting span tasks, along with two tests of intelligence, are used to assess working memory.

Terms

The terms "short-term memory" and "working memory" have somewhat different connotations and detailed meanings, but these are inconsistent among investigators.

Recall

Free Recall

For example, given the sequence 3, 6, 2, 8, 5, the participant recalls the numbers in any order.

Span Task

For example, given the sequence 3, 6, 2, 8, 5, the participant recalls the numbers in the order presented.

Cued Recall

For example, the paired associates test:
dog-file
tree-car

Given the cue "file," the participant recalls "dog."

Recognition

Exposure Phase

Participants are exposed to a set of items (e.g., new, mouse, library, old).

Test Phase

Participants are presented with a mix of old and new items (e.g., road, mouse, snow, library) and must identify which items they recognize from the exposure phase.

Outcomes

Recognition can be related to old items, with "lures" used to assess false alarms.

  • Hit: Correctly identifying an old item.

  • Miss: Failing to identify an old item.

  • False Alarm: Incorrectly identifying a new item as old.

  • Correct Rejection: Correctly identifying a new item as new.

Memory Strength

Memory strength is represented as a distribution, with separate distributions for old and new items.

d′d' (d prime) is the distance between old and new distributions, reflecting item familiarity and memory performance.

Decision Making in Recognition

Response bias plays a role in recognition, influencing whether participants are more likely to guess or be cautious in their responses.

Signal Detection Theory

Signal detection theory is applied to recognition memory, with new items representing noise and old items representing a signal.

  • Hits: Correctly identifying old items (signal).

  • Correct Rejections: Correctly identifying new items (noise).

  • Misses: Failing to identify old items (signal).

  • False Alarms (FA): Incorrectly identifying new items as old (noise).

Response Bias

  • Cautious: Participants are less likely to say "old" unless they are very sure.

  • Liberal: Participants are more likely to say "old," even if they are not very sure.

Recognition reflects the placement of the criterion on the x-axis and biases associated with the response.

Measuring Sensitivity

d′=Z<em>hits−Z</em>falsealarmsd' = Z<em>{hits} - Z</em>{false alarms}
d' represents the sensitivity or the ability to discriminate between old and new items.

  • d′=0d' = 0 indicates no discrimination ability.

  • 0<d′<10 < d' < 1 indicates some discrimination ability.

  • d′=1d' = 1 indicates good discrimination ability.

Cowan’s K Value

Cowan's K value is used in the change detection paradigm to measure working memory capacity.

K=SetSize×d′K = Set Size × d'

Development of Working Memory Capacity

Studies show how capacity develops over time.

Influence of Long-Term Memory

Long-term memory influences current memory performance.

Factors Influencing Memory Performance

Chunking: Working memory capacity is limited not by the number of items, but by the number of “chunks” of information.

Slots versus Resources

Slots Model
Resource Model

In the resource model, attention and resources are allocated dynamically.

The precision of visual working memory is set by allocation of a shared resource. Mixture model Bays PM, Catalao RFG & Husain M. The precision of visual working memory is set by allocation of a shared resource. Journal of Vision9(10): 7, 1-11 (2009)

The allocation of the “working memory resource” can be biased by selective attention and towards targets of upcoming eye movements.

Summary

  • Recognition versus recall

  • Signal detection theory

  • Short-term/working memory capacity

  • Chunking

  • Slots model versus Resources model

Retrieval Cues

Spreading activation: Retrieval cues activate related memories.

Dual Coding Hypothesis

Paivio (1969, 1971) proposed the dual-coding hypothesis:

For imageable words (e.g., orangutan, wheelbarrow), there are two routes to retrieval: verbal and visual.

Abstract words (e.g., hope, wealth, beauty) have only a verbal route.

Encoding Specificity

Cortical reinstatement: Retrieval involves the same pattern of brain activity that was present when the memory was encoded.

Retrieval cues will be more effective the more similar they are to the conditions present at encoding (Tulving & Thomson, 1973).

Declarative versus Non-Declarative Memory

Declarative / Explicit Memory

Memories that can be consciously recalled (or