Metacognition D&M Ch3 Notes

Chapter theme: metamemory research -- results, and methods beginning with Joseph Hart's 'feeling of knowing' work in the 1960s metamemory experiments were like flash cards -- do you know X, test if they're right, etc. tendency to judge study separately from retrieval

Nelson & Narens framework - shows areas of study of metamemory

Three stages of learning:

• acquisition: study
• retention: (...)
• retrieval: tests

Typical process:

1. study word pair
2. ask to make a "judgment of learning"
3. take a test
4. during test, ask for "retrospective confidence judgement"
5. during test, ask if there is "Tip of the Tongue" situation (soon able to recall, would recognize if they saw it)

	Name	Definition
Metacognitive Judgments
	Ease-of-learning judgments	Judgments of how easy or difficult it will be to learn any given item
	Judgments of learning	judgments of the likelihood of remembering recently studied items Immediate JOL - made immediately after studying an item Delayed JOL - made well after studying an item
	Feeling of knowing judgments	Judgments of the likelihood of recognizing currently unrecallable answers on an upcoming test
	Source-monitoring judgments	Judgments made during a criterion test pertaining to the source of a particular memory
	Confidence in retrieved answers	Judgments of the likelihood that a response on a test is correct (AKA 'retrospective confidence judgments')
Control Processes
	Selection of kind of processing	Selection of strategies to employ when attempting to commit an item to memory
	Item selection	Decision about whether to study an item for an upcoming test
	Termination of study	Decision to stop studying an item currently being studied
	Selection of search strategy	Selecting a particular strategy for producing a response during a test
	Termination of search	Decision to terminate the search for a response in memory

Example techniques:
feeling of knowing --> will you recognize?
source-monitoring judgments --> where did you learn? text, in class, etc.

Key Questions Defined and Explored In Research

• How do people monitor memory?
• How accurate is memory monitoring?
• Can monitoring accuracy be improved?
• How is monitoring used to control?

Metamemory: Collecting, Analyzing, and Interpreting Data

Typically people judge their memory and give them free rein on time allocation for memory task. Basic technique has been adapted and made more sophisticated over time. Example -- cognates vs non-cognates have different indexes, so people must use similarity clues from their memories when making metacognitive judgments of learning.

Basic approach: Compare your judgment with your performance. JoL compared to recall performance, FoK compared to recognition performance.

Two kinds of accuracy -- relative accuracy and calibration.
Relative accuracy or resolution, is the degree to which a person's metacognitive judgments predict the likelihood of correct performance on one item relative to another. something like, are you consistent across the sample set, or does your prediction proximity vary widely? If you were highly confident on ones you got wrong, and had low confidence on ones you got right, your resolution would be pretty poor. But if you were 75% confident on all your right answers, and 25% confident on all your wrong answers, that would be pretty good resolution. Consistency of your self-awareness. You can ask participants to judge this on any scale at all. Often computed using correlation -- range of -1 to 1. 0 means no accuracy, 1 is consistently correct, -1 is consistently incorrect. Must be computed for the individual.

Calibration is a ratio of your prediction to the actual answer So if you said 50% confidence for a bunch of answers that were all correct, you'd have a weaker calibration score. Reasonableness of your self-awareness. You have to use the same scale as the answer itself for this one -- say, 1 to 100 for both. The "signed difference score" -- called a measure of "bias" is often used; compute the magnitude of the judgment and the magnitude of the test performance, then subtract performance from magnitude. Negative values indicate under confidence and positive values indicate overconfidence. Kaylea question -- isn't a 'feeling of knowing' result highly susceptible to error? If all the answers but one are ridiculously wrong, doesn't it throw off the scale? And isn't "ridiculously wrong" rather relative, depending on how the thinker is feeling in their ToT? Example: State Capitals. If you list out cities that are capitals of other states, versus large cities in the same state, versus breakfast foods, versus misspelled versions of the correct answer, you may trigger different recognition levels in different people.

Another method: calibration curve
Plot mean level of test performance as a function of varying level of judgment magnitude. If performance exactly matches judgment, you'd have a 45-degree diagonal. Above the diagonal is under confidence (performance is above the judgment), below the diagonal is overconfidence -- performance is below your judgment).

The 'hard-easy effect'
People are often under confident when making lower judgments but overconfident when making higher judgments.

Frontal lobe damage -- relative FOK accuracy is impaired. Suggests that frontal lobe damage makes patients less able to do memory monitoring, and that frontal lobe is part of memory monitoring.

Study design is significant -- for example, since older adults have lower memory performance than younger, you must compensate for performance if you want to assess ability to do memory monitoring. Impact from number of alternatives, correct guessing.

How Is Monitoring Used To Control
Various experiments devised in order to test the overall model of memory. JoL during study --> determine whether to re-study an item. FoK after a test --> how long they persist at trying to remember an answer.