Quiz 3 - PSYC4191 Expertise Part. 3, Automaticity, Knowledge Acquisition Pt. 1

What study did Ward et al. (2013) conduct into expertise in soccer (methods, results, conclusions)?

Participants: Expert and novice soccer players.

Methods: show them videos from real games, pause the video and ask them what happens next.

Results: Expert players showed more accuracy in predictors and had better memory for where players where on the field and where they were heading (better situation awareness).

Conclusions: Expert higher accuracy and awareness was due to better mental representations in long term memory, which allowed more sophisticated perception of the patterns of action (like expert chess players). Like the chess study where novices just see individual pieces. Experts see patterns.

What sort of stimuli have experts across a range of domains been found to remember better?

Experts have been found to remember structured stimuli that is representative of their domain better than novices (e.g. for chess players, a chess board).

List seven domains in which experts have been found to have better memory for domain-relevant material.

Examples include bridge, Othello, snooker, medicine, electronic circuit diagrams, dance, basketball, field hockey, volleyball, figure skating, and football (see Ericsson & Lehman, 1996, page 293 for a list of references).

To what extent does expertise in one domain tend to transfer to other similar domains?

It does not transfer. Faloon was not better at practicing letter strings. He was specifically better at number strings (his domain). His strategies did not translate).

What did Ericsson et al. (2004) find regarding the memory of pi-reciting champion, Rajan Mahadevan?

Mahadevan could recite pi to 31,811 digits. However his memory for lists of symbols (e.g. ?, @, #, *) was only 6 (same as college students) – until he started using strategies to recode the symbols as numbers.

Describe evidence that indicates expertise tends to be very domain specific.

1. Ericsson & Pool (2016) reports that Steve Faloon wasn’t better at remembering strings of letters – just numbers. That is, his purposeful practice hadn’t improved his memory in general.

2. The same has been found for other memory experts. Rajan Mahadevan could recite pi to 31,811 digits. However his memory for lists of symbols (e.g. ?, @, #, *) was only 6 (same as college students) – until he started using strategies to recode the symbols as numbers (Ericsson et al., 2004).

3. Similarly – chess experts’ memory is only better for game positions not random positions (e.g. Chase and Simon 1973).

What are the three stages of skill development, according to Fitts and Posner (1967)?

1: The Cognitive Stage

2: The Associative Stage

3: The Autonomous Stage

Describe the cognitive stage of skill development, according to Fitts and Posner (1967).

Develop declarative representations of (((FACTS RELEVANT TO SKILL))) (e.g. a set of explicit problem-solving operations to complete parts of the skill).

Learners typically rehearse these facts as they first perform the skill.

Slow and lots of errors ((DUE TO HEAVY COGNITIVE LOAD)).

It’s the cognitive stage because you’re using your cognition. Having to really think about it.

Describe the associative stage of skill development, according to Fitts and Posner (1967).

Eliminate errors in initial understanding

Associations between inputs and outputs are strengthened.

Connections between elements of the skill are strengthened (effectively start to chunk them, form associations)

REDUCTION IN COGNITIVE LOAD. TASK BECOMES LESS EFFORTFUL. FREES UP SPACE.

E.g. play a standard drum beat in cognitive stage goes from conscious thought about what each hand and foot is doing, in associative stage “play a standard drum beat” becomes its own single chunk.

(((((The learner gradually drops conscious step-by-step control with more practice.)))))

This stage comes after having done a good amount of practice in cognitive stage.

Describe the autonomous stage of skill development, according to Fitts and Posner (1967).

((((Reached after long time (possibly years) in associative stage)))

The skill becomes even faster and more free of errors.

Elements of the task become more automatic (they require fewer and fewer attentional resources to implement successfully).

In theory, this means it should be easier to do something else at the same time as performing the skill (the dual task experimental paradigm is a classic test for the level of automaticity of a skill). Basically, if you can do two tasks at once, you’ve probably automated one of them (because it requires so few cognitive resources)

What is declarative knowledge?

Declarative knowledge - knowledge about facts and things; ((explicit knowledge we can report and of which we are consciously aware)).

What is procedural knowledge?

Knowledge about how to do things. Usually (though not always) constructed from declarative knowledge. How to solve problems.

What is proceduralization?

Anderson (2015) describes skill development in terms of “proceduralization” – which is converting declarative knowledge into procedural knowledge. This all aligns with/builds upon Fitz and Posner.

If the solution to a problem is encoded as procedural knowledge, PEOPLE CAN JUST RECALL THE SOLUTION RATHER THAN HAVING TO WORK IT OUT.

For some skills, this may be about making an appropriate response to a given situation into a reflex.

For other skills, it might be about relevant cues “popping out” of an environment without an effortful search.

What are the parallels between Anderson’s concept of proceduralization and Ericsson’s concept of mental representations?

Proceduralization means you can solve the problem by just recognising the pattern: a move from computation to retrieval (see chess evidence).

((This speeds things up.

This reduces errors and frees up cognitive resources.))

Means that the skill becomes more automatic (or the problem can be solved with less thinking).

Maps onto Fitts & Posner’s model of skill acquisition and Ericsson’s mental representations.

Similarity with Ericsson: underlying idea is the same. It’s all about developing skill into long-term memory to reduce cognitive load while performing task.

Basically: when you’re acquiring a skill, you’re basically memorising stuff. You’re going from someone who’s working everything out, to someone who is just remembering it. Aligns with Ericsson.

Once something is automated, you have all the pieces arranged in long term memory (memory traces) so you can just recall them. That’s automaticity.

What verbal commentary evidence supports the idea of proceduralization?

Anderson’s (1982) verbal commentary studies on geometry appear to demonstrate the idea of proceduralization (and automaticity).

Participants were shown complex geometry problems, before proceduralisation they had to think aloud while they solved geometry proofs, and worked slowly. ((SUGGESTS HIGH COGNITIVE LOAD)).

After lots of practice, they were: Faster, 2. without verbal rehearsal (not thinking out loud), and 3. the application was one step (just solved it in one step)

What three things changed following proceduralization during Anderson’s (1982) geometry studies?

After lots of practice, participants solving geometry problems were: 1. Faster, 2. did so without verbal rehearsal (not thinking out loud), and 3. the application was one step (just solved it in one step)

How did Logan (1988) define skill acquisition in terms of problem solving?

Logan (1988) suggested that skill acquisition can be thought of as remembering solutions to problems that one has previously solved (i.e., effortful processing to automatic recall). ((He proposed AUTOMATICITY involves accumulating enough memory traces for a task)).

What did Logan and Klapp (1991) find in their experiment examining automaticity (methods, results, conclusions)?

Methods: direct participants to identify which letter is X number along from Y letter. Number of letters required to be counted was varied.

Results: Initially, participants took much longer for higher numbers (because they were counting in their heads: bigger numbers meant a longer counting time). After automatization (i.e. they had memorized the answers), they took the same (short) time to do a letter 5 spaces into the alphabet as 2.

Conclusion: Practice led to automisation. When you’re a novice, the number of spaces you have to count makes a difference. They’re working it out in their head.

Describe Jenkins et al.’s (1994) PET scanner study looking at automaticity (methods, results, conclusion).

Methods: Taught participants a novel finger tapping sequence

Results: With more and more practice, participants’ brain regions associated with conscious effortful processing (frontal) decreased in activity, while regions associated with implicit/procedural motor memory increased

Conclusions: Learning is accompanied by a change in how neural systems work during performance, in a manner consistent with the idea that skills become automated with more practice (via proceduralisation). Say: literally using a different part of your brain (going from hard thinking brain to automatic movement brain). Skill acquisition is accompanied by neural reorganisation.

What experiment did Rowe and McKenna conduct to investigate automaticity in expert tennis players (methods, results, conclusions)?

Participants: expert and novice players

Methods: showed participants videos of a tennis player winding up to hit the ball. Instructed them to press a button when you realise the ball is going across the court (rather than down the line). In one version of the experiment, participants did two versions of the test. One as per previous description, and the other while doing a secondary distractor task (generating a stream of random letters with no abbreviations or alphabetical order). This task is difficult and requires full attention.

Results: ((Main effect significant: experts better than novices at predicting, confirming their expertise)). Also: Interaction significant: groups are made worse from the dual task, but the effect is much bigger for novices. Simple effects were significant.

Conclusions: Indicates that the the experts have automated the tennis task. Their performance dipped much less while doing distractor task because their cognitive resources were not being preoccupied by tennis task, as it is automated.

What did Morgan & Pollack (1977) find out about the thought processes of expert marathon runners compared with novices?

Reword: Expert marathon runners continually monitor their physiological state and the effectiveness of their running whereas novice runners think about things unrelated to running to minimize their pain. Automaticity frees up your resources to think at a higher level about the task (e.g. running technique)

How can we reconcile the automaticity account of expertise with experts reporting thinking during skilled performance?

Experts remember more from task-relevant stimuli than novices (e.g. chess players remember more of the board layout when surprised with a memory test). This shows expert performance is not completely automated.

Expert thinking is at the highest strategic level (e.g. planning, reasoning, and anticipation). Experts can think at this high level BECAUSE they don’t have to think about lower level things (which have been automated). Helps explain their expertise. Automate the lower level stuff to think more about the higher level strategic thinking

What is a componential analysis?

Dissecting a skill into its key components. With practice, these components would be the things converted from declarative to procedural knowledge.

What is mastery learning?

Follows componential analysis. A kind of tiered or progression-based learning structure where each component from componential analysis must be mastered before the learner can progress to the next component.

How does a mastery learning approach differ from traditional education practices?

Traditional learning approaches allow for progression without mastery (e.g. a student progressing from 2nd year stats to 3rd year stats with only a 50% grade on 2nd year). Suggests they don’t understand some core concepts, but they still progress. This is not allowed in mastery learning.

What are Intelligent Tutoring Systems?

A form of ((AI)) programming that uses componential analysis and mastery learning for teaching. They put Anderson’s proceduralisation theory into practice by using componential anaylsis to turn it into a computer program.

Describe an example of an Intelligent Tutoring System that teaches computer coding.

Anderson’s Intelligent Tutoring System that taught LISP ((main AI programming language in 80s and 90s)). The system could write LISP coding functions and offer feedback and solutions to student errors. The system used complex IF/THEN statements to allow students to progress only when mastery of that stage was achieved.

Give an example of an Intelligent Tutoring System, based on Anderson’s approach, that is currently in use.

MATHia (distributed by Carnegie Learning) is currently in use for hundreds of thousands of American school students learning maths.

What advantages and disadvantages do Intelligent Tutoring Systems based on Anderson’s approach have compared with generative AI?

1.Tutoring systems don’t hallucinate, don’t make mistakes (because they run only on IF/THEN statements)

2.((Progress is transparent))

3.This is great for domains where everything is locked down and well-understood

4.Disadvantages: that’s not true for every domain, so future solutions might involve a hybrid of generative AI and Intelligent Tutoring Systems, AND takes a lot of workload to make an Intelligent Tutoring System.

Describe five of Willingham’s principles for learning in education together with what needs to be known about the students, and classroom implications.

1. Knowledge precedes skill

2. Memory is the residue of thought

3. While people are curious, they only willingly think under certain circumstances

4. We understand new things in the context of things we already know

5. Proficiency requires practice