Disclaimer: English Kinda Thing

The sole purpose of the "English Kinda Thing" is to document my attempts to correct my own mistakes in standard English usage and to share the resources I find. In no way do I attempt to teach nobody English through these blurbs--just as I intend not to teach nobody to be a neurotic and psychotic handicap in Ratology Reloaded or Down with Meds! :-)

Friday, May 24, 2013

Nelson & Narens (1990). Metamemory: A theoretical framework and some new findings.

Nelson, T.O. & Narens, L. (1990). Metamemory: A theoretical framework and some new findings. In G.H. Bower (Ed). The Psychology of Learning and Motivation, 26, 125-173. New York: Academic Press

A: The abstract principles of metacognition

(The following three principles are the exact quotes)
  1. Principle 1: The cognitive processes are split into two or more specifically interrelated levels: the meta-level and the object-level.
  2. Principle 2: The meta-level contains a dynamic model of the object-level.
  3. Principle 3: There are two dominance relations, called "control" and "monitoring," which are defined in terms of the direction of the flow of information between the meta-level and the object-level. This distinction in the direction of flow of information is analogous to that in a telephone handset.
This diagram in Nelson and Narens (1990) (p. 126) shows the theoretical mechanism consisting of the meta-level and the object-level and their relations.  
This diagram above shows a theoretical mechanism consisting of the meta-level (e.g., metacognition) and the object-level (e.g., cognition).

The authors used the telephone handset analogy to explain the information flow between the meta and the object levels.  The way I interpret it is... its' sort of like the relations between Gordon Ramsey and the chefs in Hell's Kitchen (if not simply the chefs and their dishes)... with what goes on in your head the equivalent of "Hells Kitechen." (At least mine).  (Could cook something up with my psychosis though might as well go directly with cooking...)




The role of Gordon Ramsey is equivalent to your metacognition and the role of the chefs, cognition.  What the chefs cook up are like the cognitive outcome and how they do the cooking, the cognitive processes.  As the chefs are busy cooking down there, Ramsey (meta-level) gets information about what is going on with the chefs (object-level)--this is the monitoring process.  After Ramsey (meta-level) processes the information he gathers, he (metal-level) goes on yelling at these chefs (object-level) and tells them (object-level) what to do, and the chefs go back on to cook following the directions of Ramsey... this is what the control relation is about.

It might be an even better analogies if there are multiple people with a role similar to that of Gordon Ramsey... since some theories believe that how metacognition works in the brain is more or less like the notion of cloud computing... different regions associated with different metacognitive roles.

Also, one thing I kept on hearing Ramsey yelled out loud was... "Out! Out! Out!" Well, maybe your metacognition doesn't do it... mine surely does it on a daily basis... 

1. Control

The notion behind control is that the meta-level modifies the object-level... its state or process... and contribute to actions at the object-level: 1. to initiate and action, 2. to continue an action, or 3. to terminate an action.

2. Monitoring

The basic notion behind monitoring is that the meta-level is informed by the object-level... and result in the change in the model the meta-level has on the situation, including "no change" (i.e., Gorgon Ramsey's view of how the kitchen is running).  

Interestingly, as per Nelson and Naren (1990), "the opposite does not occur, i.e., the object-level has no model of the meta-level."  Don't think I really understand it... and thus... don't have the audacity to agree or disagree... Yet, in the process of working on the psychotic model of how my psychotic mind works... including at the metacognitive level, aim I not building up a cognitive model (object-level) of my metacognition (meta-level)? 8-O

Donno, maybe I haven't quite gotta the notion the authors were trying to convey or my example might be buggy and I am too blind to see it.

3.  Role of subjective reports about introspection for inferences about monitoring

Following are some notions the authors mentioned about the methodological issues they encountered.
  1. The measurement issue on introspective reports--the assumption behind using telescopes to observe and that when studying the use of telescopes as an instrument.  Were my head the equivalent of a telescope, I have no doubt doubting the validity of its outputs and distortions.  lol
  2. Views on introspective reports: descriptions of internal processes (as per the Structuralist's view) vs. data to be explained.  
  3. "A system that monitors itself (even imperfectly) may use its own introspections as input to alter the system's behavior."  I am in total agreement with this notion... the basic assumption of monitoring our psychotic mind with our disordered thinking.
  4. Perception could be conceptualized as sensation plus inference... the coexistence of accuracy and distortion--true value and error.  Apparently, even the normal have to deal with the error term...

B. The monitoring and control of human memory

This diagram in Nelson and Narens (1990) (p. 129) shows examples of monitoring and control tasks at different stages of learning.
The way I see it... correct or not... what's between monitoring and control is the processes occurring at the cognitive or object level... for the perspective of information processing theory... the middle section looks almost like the working memory (WM).  When discussing the theoretical framework represented by the above model, the authors recommended the context of student studying for an exam to the readers. 

Among all, one thing that differentiates between the model of Flavell (1979) and this one (Nelson & Narens, 1990) is that Nelson and Narens broke it down further... the processes at the meta-level and object-level (e.g., WM) and subdivided monitoring into ease of learning, judgments of knowing and feeling of knowing.

Most of the things the authors discussed are captured in the above diagram.  In the following section, I will only mention the notions of interest to me.  

1. Acquisition stage: In advance of learning

a. Determining one's goal: the person's norm of study

  1. A person's theory of retention is used to modulate how well the content has to be mastered in order for it to be remembered and recalled in the retention tasks.
  2. The norm of study is the product of an individual's theory of retention... the mastery the individual believes he or she should obtained during the acquisition phase.

b. Formulating a plan to attain the norm of study

Retrospective monitory vs. Protrospective monitoring in the context of seeing airplanes in the sky:
  • Retrospective monitoring (judgments of past performance): How well did I do the last time I saw an airplane flying in the sky?
  • Protrospective monitoring (judgments about subsequent performance): How well might I fare the next time I see an airplane flying in the sky?
The authors further subdivided protrospective monitoring into three categories:
  1. 1. Ease-of-learning (EOL): occurring in advance of acquisition.
  2. 2. Judgments of learning (JOL): occurring during or after acquisition to predict future performance on currently recallable items
  3. 3. Feeling of knowing (FOK): occuring during or after acquisition about whether a currently nonrecallable item is known and/or will be remembered in a subsequent test
Research found that EOL, JOL, and FOK are not highly correlated... meaning they measure different things... or these judgments might measure different aspects of memory with the structure underlying these judgments multidimensional.  

c. Ease of learning judgments.

The outcome of EOL can affect the amount of time allocated to study.

d. A Prior Choice of Processing judgments

After EOL, the person decides the kinds of processing to use on the study item.

e. Initial plan for the allocation of study time

Monitoring: this is a harder item to learn=>control: allocate more time
This diagram in Nelson and Narens (1990) (p. 132)
This is the diagram of my dream... since there's nothing new under the sun... I know I can found someone who has already gotten a model like this out... the separation of cognition and metacognition.  In this diagram, the upper portion (theoretical) represents metacognition while the lower portion, the processing in the working memory or cognitive level.

2. Acquisition stage: the ongoing learner


Following is the flowchart showing how learning works... at the metacognitive level and it's relation with the cognitive level.  One thing I love most that table titled metacognitive library, which, according to Flavell (1979) is rightfully encoded in LTM.  Why is this notion important? What is the likelihood that a certain strategy will be effective in intervening  a certain type of psychotic scenario?

This diagram in Nelson and Narens (1990) (p. 133)

Some processing involved in the above cycle...

a. Feeling of knowing for currently nonrecallable items.
FOK is linked to time allocation
b. Judgments of learning for currently recallable items:
Are you making judgments of future performance based on monitoring of information in the WM or long term memory?  Delayed recall might resolve this issue.
c. Updating the allocation of study time during a particular study trial of an item.
People study till the JOL for the item reach the norm of study though their judgments might be really off from the reality. Motivation might be a confounding factor.
d. Termination of study.
When JOL reaches the norm of study, study is terminated.

3. Retention stage

The maintenance of previously acquired knowledge might involve similar process as in the acquisition stage with the following additions:
  1. 1. discrepancy between the desired degree of mastery and the assessed degree of mastery
  2. 2. FOK-directed maintenance work

4. Retrieval stage: Termination

The distinction between self-directed and non-self-directed retrieval... Nelson and Narens focused on self-directed retrieval: the search itself and setting up the cues to initiate the search.  The following is the flowchart for the retrieval stage.

This diagram in Nelson and Narens (1990) (p. 136)

a. Quick initiation/termination of retrieval.

Possibly FOK-based with a latency far shorter than that for an actual recall. 
There is something very interesting about the notion of quick initiation and FOK... like... I know that feeling... can't give you the exact reason why but I think it's delusional... The FOK arrives before the identification.

b. Placement of retrieval termination:

The termination happens when someone is no longer willing to continue of the FOK no longer exceed the threshold of claiming to know the non-retrieved items. 

The good thing about my lack of creativity is that I seemed to have oversaturated the kinds of delusions my head is capable of creating... thus, not too many records for the head to search through when trying to answer the question of whether the observed event fit the profile of past delusions or not.  Given the limited number of search my head has to perform, and, I suspect, more sophisticated search algorithm it operates on as a result of being a more experienced psychotic (than in my younger years) plus my false-positive-allowed policy, there's gotta be an extremely high threshold for nonretrieved items.  Also, in the context of psychosis, premature termination might be something really risky... untreated delusions etc can become the weakest link my my head's security.

5. Retrieval Stage: Output of response.

People's threshold for outputting the retrieved answer might be affected by factors such as "cost vs. rewards.

6 Retrieval stage: Confidence judgments after recall

Commission error vs. omission error


After the description of the model, the authors went on discussion some issues concerning the methodology and relevant findings for their research.  Since my main interest is in the model itself, I will only mention the points of relevance to me.

1. Amount of information deposited in long-term memory is important for metacognitive monitoring. (That's why I went back to the psychiatric ward during the second full-blown episode.

a. Our early experiment
The FOK results
  1. Reliability: People are consistent in their FOK judgments
  2. Validity: The consistent judgments of people are not all that useful in predicting the outcome of future performance
b. Effect of degree of learning on retention interval on FOK accuracy
In order to perform monitoring, you need to have enough beef registered in the LTM.
With the retention rate, it looks like you have to lit it sit for some time to get the correct accessment.

c. Our first published experiment on metamemory
The degree of learning is important to FOK accuracy.  (Nelson, Leonesio, Shimamura, Landwehr & Narens, 1982)

d. Mechanisms for the overlearning effects on metacognitive accuracy:
  1. Metacognitive judgments attempt to discriminate between items.  Higher accuracy in discriminating items when the differences between the items increase
  2. Overlearning might enhance the stability of retention and increase the differences between items.
e. Empirical support for the item-discrimination mechanism
People are reasonably fine as measuring devices but their have limits in their ability to differentiate between objects.  The more different the item, the more likely for the metacognitive discrimination to be valid. Whether items could be validly discriminated is dependent on the person's ability to monitor and the degree of differences between items (difficulty level).


2. FOK may be perfectly valid at tapping a larger number of aspects of LTM but the accuracy of FOK for predicting criterion performance may nevertheless be imperfect.

This is an interesting one... in the context of psychosis, there is this FOK... the sense that the observation is delusional.  In reality, I treat all observations with uncertain intents as delusional--false positive allowed. For practical purposes, it doesn't have to be perfect... as long as no false negative.

FOK dependent on what's in LTM rather than in WM or unconscious memory.

3. Privileged access


a. Judge/observer experiments
People's prediction might benefit from the idiosyncratic information at their disposal during retrieval.
b. Normative predictions vs. the individual's own FOK prediction.
People's FOK predictions on their own subsequent performance on currently nonrecallable items  is better than the normative FOK prediction derived from the average prediction.  But individual's FOK predictions are worse than predictions derived from the normative probability of correct recall.  It has been found that the accuracy of people's FOK prediction is higher when given the normative probability of recall...Unfortunately, people seems to be poor at intuiting the normative probability of recall on items that can not recall.

4. Some factors underlying people's FOK judgments (vs. accuracy)

a. Overlearning affects not only FOK accuracy but als the magnitude of FOK.
b. Overlearning study trials vs. overlearning test trials.
Both the study trials and test trials affect the magnitude of subsequent FOK.
c. Actual overlearning vs. claimed overlearning.
Most effects of overlearning on FOK seemed to be mediated by people's belief about whether the items have been overlearned... as opposed to an "automatic" effect of overlearning.
The no-Magic Hypothesis for how FOK should be conceptualized: the person 1). considers particular recallable properties of the to-be-retrieved items) in conjunction with 2). rules about how the properties are related to the subsequent criterion performance that the person is trying to predict.

According to non-magic hypothesis, "FOK does not directly monitor a given unrecalled item in memory, but rather the FOK monitors recallable aspects related to that item, such as the item's acquisition history or partial/related recalled components."

5. Learning-to-learn effect for FOK judgment?

Can FOK accuracy be improved by sheer practice on FOK judgments? It doesn't matter whether you provide feedback or not on their FOK judgments... the practice effect is small shall it ever occur.

6. Relation between metacognitive monitoring and metacognitive control during acquisition: The allocation of self-paced study time.

The allocation of study time is also mediated by people's EOL and FOK in addition to a response to item difficulty.
If people believe that they have learned things well, it's useless to give them more time to learn.


7. Relation between metacognitive monitoring and metacognitive control during retrieval: termination of memory searching.


a. Role of "preliminary FOK judgments" prior to searching for an answer.
  1. Single-counter FOK hypothesis: one FOK component that only taps the presence of information in memory
  2. Dual-counter FOK hypothesis: tapping both the presence and absence of information in memory.
This diagram is in Nelson and Narens (1990) (p. 164)
Results indicated in the above diagram confirms the dual-counter hypothesis.

b. Relation between ongoing FOK and the latency of recall errors.
The substantial correlation between FOK and the latency of omission errors indicates that people's FOK affects whether the overall retrieval stage with be continued or terminated. However, the correlation between FOK and the latency of commission errors is nil for general-information items.  The difference could be considered as the manifestation of the differential mechanism involved as shown in figure 5 above.

c. Relation between metacognitive confidence judgments and the latency of recall.
While the latency of recall affects the confidence of judgment, distance between the the curves showed that something else is affecting people's confidence.  

This diagram is in Nelson and Narens (1990) (p. 167)

Love how the authors put it... "We do not yet know what that something is, but future research should attempt to isolate and identify it."

One thing I found most interesting is the notion of feeling of knowing (FOK)... 

A lot of times, this is how delusions manifest themselves... no hallucination and anything else attached... simply the FOK dropping by and landing on the working memory... like... I knew the the apocalypse was coming again, I know they are checking out everything I do including my internal blog, I know they are talking about this blurb I am working on now, or I know some other things a like out of no where that I have never known and maybe in contexts I have never been in.  How do I know it?  I could feel it... the gut feeling... I just know it, believe it. (Of course, this FOK I have spoken of up to this point might not be the FOK Nelson & Narens (1990 ) were talking about... unless... it's about one of the multiple manifestations of the same principle?  Donno.)  

Then, I would have the FOK about the above FOK... something fishy there... feels like delusions... Even before I have to explicate to myself which part of the scenario seems delusional, I intervene by means such as what Ramsey said to some chefs... "Out! Out! Out!"

So I came across the notion of FOK and the associated research findings... as well as the no-magic hypothesis... as summarized above... Then, I thought... a good way of conceptualizing the fishy kind of FOK... it surely could be true that, after being delusional for over a decade, I must have enough delusions encoded in my LTM to boost up the accuracy rate of my FOK.  My head might have, somehow, figured out the attributes of delusional dasein, which, theoretically, is the basis of how FOK works... in spite of my inability to have it articulated.  If this is really how it works, it would be really beneficial for my head since the last thing I need is to constantly access my memories of psychosis... not to mention that the constant access might strengthen the path towards them and make them easily activated.

At the same time, since I treat all observations with uncertain intents as symptomatic... allowing for false positive, I don't need the FOK to be accurate in all perspectives... as long as no false negative.

2 comments: