Chi, Michelene T.H. (1987). Presenting knowledge and metaknowledge: implications for interpreting metamemory research

An interesting article making arguments on why some meta stuffs aren't really meta.  At the same time, isn't it true that the outcome of meta-processes can be encoded for later recall... the conversion of meta to non-meta?  Donno.  Just a thought.

Chi, Michelene T.H. (1987). Presenting knowledge and metaknowledge: implications for interpreting metamemory research. In F. E. Weinert & R. Kluwe (Eds.), Metacognition, motivation, and understanding (pp. 239-266). Hillsdale, N.J.: L. Erlbaum Associates.

A framework of representing knowledge

"The author borrows heavily from the work of Anderson (1976), Greeno (1978), and Norman and Rumelhart (1975); however, any deviations and modifications of their work are the results of intuition and preconceived biases, suitable for explaining developmental findings." (p. 245)

Domain knowledge

Domain knowledge can take two forms: declarative and procedural.

Declarative knowledge is factual in nature...  "Declarative knowledge can be represented in terms of a semantic propositional network, where a concept (such as a dog) may be represented as a node, and links specify the relationships among the nodes... The degree of complexity of the semantic network should correspond to the amount, elaborateness, and organization of a child's declarative knowledge." (P. 245-246)

"Procedural knowledge is knowledge about how to do things.... Furthermore, procedural knowledge can be represented as a set of production rules, which are condition-action pairs."  (p. 246-247)

Strategy

The notion of strategies seem to have a special status in developmental research because "they have he property of being general and global, rather than domain-specific."  P2 in the above figure is an example of a strategy--a general procedure applicable to different domains.

The four properties of a strategy: (p. 247)

1. General and domain-independent
2. Like all procedures, it has a goal
3. Like all procedures, it can have several components
4. The number of strategies in memory is small (there should be fewer number of P2 than P3-P9).

The implications of the above definition: (p. 248)

• Strategies might have the properties of procedural knowledge and they are represented in the same way as production rules.
• The separation between strategies and procedural knowledge is artificial but serve a purpose.
• Studies have been conducted on the training of strategies for the purpose of generalization.  

The observed failure to transfer might be due to the fact that what was taught was domain-specific procedures (e.g., P3, P4) rather than domain-independent strategies (e.g., P2).  Younger children need to be told explicitly to use the strategies learned and it could be due to a problem in the encoding of goals because when the goal is missing, the procedure might not fail to be activated.  The reason why older children learn to learn and adopt the strategies faster than younger children could be due to the fact that they already have different versions of the a production (e.g., P3, P4, P5) in memory and it makes it easier to acquire the general form, the strategy production (P2).

A good way of looking at my coping with psychosis during the onset, right after the onset, and now.  Can't say getting all that good though I guess and I hope I am getting better learning to cope.  Sort of like, you don't know it's delusion, you don't know that you have to intervene.  Whether the intervention is procedural knowledge or a strategy is not really an issue for me.  Though, shall strategies be considered as a more general form and content-independent of production rules, it surely would be harder to extract a strategy rule if you do not have enough procedural knowledge.

Metaknowledge

What is metaknowledge?

• The straightfoward way-->metaknowledge is cognitive knowledge and metamemory would be memory knowledge.
• The more complex way-->second-order knowledge.

Meta-delcarative knowledge (p. 249-250)

1. Meta as a reference to cognitive knowledge?  "Meta-declarative knowledge would be factual knowledge about cognition like factual knowledge about any other domain.
2. Meta refers to second-order knowledge, or a function? In this case, "meta-declarative knowledge would be the same thing as a procedure or a strategy... Hence, theoretically, meta-declarative knowledge would take the same form as either procedural knowledge or strategies." For instance, how much do I know about animals?  To answer the question, you have to search or activate relevant nodes in the semantic network and make assessment thereafter.  "The complexity of the processes needed to answer meta-declarative type of questions depends on whether or not the question addresses a prestored declarative knowledge, or if a procedure is needed to assess existing declarative knowledge about cognition."

Let's replace it with this question... "Is this delusional?"  And... remember... delusions morph and don't always appear in the same fashion... though contains certain "delusional" attributes, which should make the notion of delusion detection something mighty interesting...

Meta-strategies: (P. 250-251)

"Again, there are two ways to represent meta-strategic knowledge.  One way is direct prestored declarative knowledge"... For instance, "Do you remember better than your friends?" In this case, the term meta is used only to refer to knowledge about strategies or activities of remembering."

"Alternatively, one can also view meta-strategic knowledge as second-order operations.  In this case, meta-strategy would be a rule that evaluates another rule (See P1).  If it's the first time you encounter the question, "Do you remember better than your friends?"  There is no restored answer... and one has to engage in assessments to perform evaluations at two levels: the level of the output of each individual memory strategy and the level of the sum of all the available strategies for remembering.

"Is this delusional?" Chances are... though delusions are fairly similar in certain principles, it's highly unlikely for them to be the same. Thus, the lower degree of likelihood for you to find a stored answer.  Yet, the head must have some interesting way of conducting pattern matching when performing delusion detection... possibly... partial identification of attributes of delusions for the sake of efficiency.

"However, because meta-strategies can themselves be represented as production rules, they are not different in form of quality from other production rules, such as strategies and procedural knowledge.  Hence, there is agreement with Flavell's (1979) assumption that 'metacognitive knowledge and metacognitive experiences differ from other kinds only in their content and function, not in their form or quality'." (p. 251)

Meta-procedural knowledge: (p. 251)

Similar to meta-strategies, meta-procedural knowledge are rules that evaluate other domain-specific rules.  

Summary and discussion

Interesting notion...

"Beyond these rules, there should be a set of executive rules, controlling the entire system's on going activities (whether the system is a child, a program, or an adult).  Executive rules are analogous to an interpreter than determines the sequence of processes to be executed.  A variety of processes, such as monitoring one's own state of processing and allocating ongoing attention, has now been given a new status of meta-processes... Why these executive processes have been given a new name, other than the possibility that they control memory processes, is unclear." (P. 254) 

Metamemory research

This is a somewhat random selection [the research], simply because it is nearly impossible to be exhaustive for a paper of this length.  The selected examples serve to illustrate the complexity and fuzziness of some of the metamemory research.

Metaknowledge of person variable

Whether or not an age effect can be observed (p. 254)

1. total amount of related experience
2. whether greater amount of experience can produce a more accurate and robust assessment
3. the consistency of early encounters

"For example, if a minimal amount of experience is sufficient to produce a robust assessment, then a stable knowledge of person variable could probably be reached early in life; consequently no age effect will be observed." (p. 254)

Why the onset is inevitable?  Why the occurrences of reinstitutionalization and the institutionalizable grade of mental state?

Sensitivity

"a young child has in memory only production rules that approximate the final mature rules (such as P6), but not the mature rules themselves (such as P2).  Therefore, the overt request to memorize is not a sufficient cue to trigger the strategic rule P2 and other similar rules."

"Sensitivity may reflect only a deficit in existing cognitive rules.

Would say I agree with it.  Why did I get institutionalized the second time?  The delusions in the onset was about fame and wealth etc.  The delusions in the second major episode... the highlight... the Appocalypse.  The P2 actually was not mature enough since it ain't got no idea about delusions associated with Appocalypse etc.  You can't ientify what you don't know and thus failure to intervne (regardless the effectiveness of interventions.)

Monitoring and checking

Based on the study of expert and novice physicist, the author came up with the argument that "younger children's inability to accurately monitor their current state of knowledge (such as preparedness for recall), as well as their inadequate allocation of attention, is attributed to an inadequacy in part of their domain knowledge related to the stimulus items, rather than strictly underdeveloped monitoring process." (p. 260)

Check is a notion related to monitoring.  In the physicist study, by mistake, "the numerical value for one of the known variables in the problems was too large, so that the actual numerical solution was unusually small.  Guess what the outcome might be... of course... results indicated a difference between expert and novice.

"One possible interpretation for such results is that over years of experiences, the experts have acquired schemata of problem types, and one of the slots of a schema is potential values of the solution.  Hence, when a value does not correspond to the range dictated in the schema, it signals to him that he should doublecheck... checking is totally an outcome of the presence of the relevant domain knowledge in memory, and not a meta-strategy that some individuals have and others do not." (p. 261)

As I grow older in my psychotic age, guess I also started to acquire schemata of problem types with one of the associated schema as the potential solution.... therefore... defects and workarounds.

Summary and discussion

Sensitivity and executive processes such as monitoring and checking need not be called meta.  With the personal variable type of study, it's difficult to tease apart what is meta and what is not.

General discussion

"It was conjectured that the reason that children may not, for example, check their solution as readily as adults, reflects not so much deficits in their control or monitoring processes, but rather, the lack of a relevant schema in the declarative knowledge base to tell them that the answer was inappropriate." (p. 263)

The little relation between memory and metamemory performance? A definition problem. (p. 263-264)

• A variety of meta-type tasks are really not meta, therefore, one would not expect performance on thoses tasks to correlate with memory performance
• metaknowledge is not unlike knowledge of any other domain, and it is conceivable that knowledge of cognition and knowledge of noncognitive domains develop simultaneously and independently, so that one need not bear one the other
• the only kind of processes that must have some relation to cognition are those that evaluate other strategies or rules.