Quiz 6 - Transfer (and simulation)

Define near and far transfer.

Transfer is applying something we have learned into a different context. All skill / knowledge acquisition involves some degree of transfer (exact learning circumstances never quite match).

Near transfer: learning and performance context are very similar or nearly identical. This is easier than far transfer.

Far transfer: bigger difference between learning and performance context. Harder than near transfer.

Describe the Doctrine of Formal Discipline

Rival theories about transfer. Doctrine of Formal Discipline proposes the mind is made up of a set of general abilities. The exact content of what is learned is less important than the general strategies of learning and problem solving. (It’s more important that you’re learning how to learn). That allows you to do other things, even things you haven’t ben explicitly taught. The brain is seen as a muscle that can be developed through exercise. This is the idea behind traditional principles of education and brain training.

Critical thinking can be developed and transferred.

Describe the Theory of Identical Elements

Contrast to Formal Discipline.

The extent to which transfer occurs is down to the level of commonality between what you initially learn and how you perform it. The less commonality, the less transfer is likely to occur.

Thorndike emphasises “independent / specific capacities,” rather than generalisable skills in Formal Discipline

Thorndike says teach people what you want them to learn, not something else and expect them to figure out what you really want them to learn.

Why is the occurrence of far transfer important to the classical view of education?

Reword:
•The stakes in this debate are high: if far transfer is very rare then this raises a fundamental problem for education.

•Education could be viewed as an attempt to create experts from novices, using far transfer.

•The classical view of education is that by teaching people a range of topics, they will learn general skills they can apply to any domain.

•That is, it doesn’t matter if you teach people information they’ll never actually use (like Latin) because the skills they learn are transferable.

Give an example that illustrates the domain-specificity of expertise.

Reword: One memory champion, Chao Lu (who remembered pi to over 67,000 digits) only had an average digit span of random digits That is, maybe even what might be considered near transfer is less common than expected.

Why does our understanding of expertise raise problems for the likelihood of far transfer?

REWORD:
•Evidence from expertise indicates that it is extremely domain-specific, because it is based on having a large body of highly-organised domain-specific knowledge.

•That is, there appears to be little far transfer between different domains of expertise (Detterman, 1993).

Describe Carraher et al.’s (1985) experiment involving far transfer in Brazilian school children (methods, results, conclusions).

Reword:

Methods and results: Studied Brazilian school children who also worked as street vendors authors posed as customers and bought things from them, recording the number of correct calculations (e.g. 5 lemons at 35 cruzeiros each = 175). 98% of their calculations were correct.

They then invited the same children to the lab and gave them exactly the same sums as mathematical problems (e.g. 5 x 35 = ?). They only got 37% correct.

When the sums were written as word problems, performance improved to 74% (but still nowhere near how well they do under their real-world situation).

Conclusions: This was taken as evidence that far transfer failed in this context (and hence one goal of the education system had also failed). Even when maths problems are the same, their performance dips significantly through transfer. Implications for transferability of formal education.

Describe Burrage et al.’s (2008) study looking at the effects of formal education on working memory (participants, methods, results, conclusions).

Participants: kids who were two months either side of cut off for a particular school year. Nearly matched for age and other factors.

Methods: Tested them on cognitive abilities (executive function tests, such as auditory working memory).

Results: Children who went to school performed much better.

Conclusions: Potentially a causal benefit of attending school.

Describe Lehman et al.’s (1988) study looking at the effects of a university education on general reasoning abilities (participants, methods, results, conclusions).

Participants: Compared 1st and 3rd year uni students (cross-sectionally and longitudinally) across several degrees including psychology.

Methods: Tested them on general mental abilities (not content-specific) so general statistical / methodological reasoning, and conditional reasoning (two generalisable skills that go beyond content). Cross-sectionally (1st vs third years) and longitudinally (same students in 1st vs third year).

Results: GO TO 1:48pm. Mark has an answer here worded.

Conclusions: listen to presentation.

REWORD:

•They predicted medicine & psychology should improve general statistical & methodological reasoning (no effect for chemistry or law).

–E.g., spot the confounded variable; apply the law of large numbers.

•They predicted that law, medicine, & psychology should improve general conditional reasoning (no effect for chemistry).

E.g. Wason 4 card task

Results: only psychology students improved in general and conditional reasoning.

Conclusions: spontaneous far transfer is possible – but only if you do a psychology degree.

What did Barnett & Ceci (2002) propose was the reason behind differing outcomes amongst studies investigating far transfer?

REWORD:

Barnett & Ceci (2002) argued that the reason for the inconsistent views on far transfer is because not all transfer is equal.

Transfer can mean lots of different things, have lots of different dimensions. If we just treat it as a “it works or it doesn’t” absolute, we are unlikely to find evidence for it.

Describe six “take home” messages from the transfer literature (according to Horswill).

When designing training or educational intervention

1. Don’t assume transfer happens spontaneously

2. Make clear what kinds of transfer you want your training to achieve (not all transfer is the same). Are you trying to teach them something they need to apply in 5 years time in a different context. Consider your goals of what kind of transfer.

3. Design your program so it is geared towards achieving the kind of transfer you want to achieve, as per previous point.

4. Try to make far transfer (difficult) into near transfer (easy) wherever possible. E.g. design your assessment to match how you will use it in workforce.

5. Be explicit about how you want students to achieve transfer (if that is your goal). Tell students why they use this skill in future and how. May need to be persuasive. Don’t assume students will figure out what transfer a particular skill will apply to.

6. Treat transfer as its own skill that can be trained. Help students practice transferring, and to notice when transfer can apply.

What were three aspects of skill/knowledge transfer CONTENT that Barnett & Ceci (2002) proposed to understand the variety of outcomes in far transfer studies?

REWORD:

1.Consider specifically WHAT skill or knowledge you want people to transfer (e.g. some procedure or fact – or some principle or heuristic).

2.Consider HOW you want the trainee’s performance to change as a result of the transfer (e.g., improved accuracy, improved speed, improved strategic approach).

3.Consider whether trainees need to spontaneously apply the transferred skill to a new domain (no prompts = very hard) – or whether it will be obvious (e.g. they’ll get prompted) to use a transferrable skill.

Describe six aspects of skill/ knowledge transfer CONTEXT that Barnett & Ceci (2002) proposed to understand the variety of outcomes in far transfer studies

For marks: consider number of points. Don’t have entire table for 2 points, just can list them. Just need to get the gist.

1. Knowledge domain.

   1. Near = use SPSS in PSYC2010 and 3010.

   2. Far = apply principles of measurement theory from 3020 to astronomy.

2. Physical context

   1. Near = UQ lab to QUT lab.

   2. Far = Lab work to real world context (e.g. shopping centre).

3. Temporal context

   1. Near = same session, minutes later

   2. Far = Several years later

4. Functional context

   1. Near = PSYC4191 quizzes to PSYC4191 assignment

   2. Far = UQ course to real world job

5. Social context

   1. Near = training as individual then perform as individual

   2. Far = train as individual then perform in group team

6. Modality

   1. Near = MCQ for training and outcome

   2. Far = MCQ quiz for road rules for training, then real world driving for performance

Describe Gick & Holyoak’s (1980) experiment investigating problem solving by analogy (methods, results, conclusions).

Methods: had people solve two problems with the same underlying structure and solution, but framed with different context (how can you attack the fortress at the intersection of roads that have been mined without blowing everyone up AND how do you zap a tumour with radiation without harming surrounding healthy tissue). Gave people a go at one, let them solve it, then presented the second one and had them solve it.

• Solution to invading the castle problem is because you have all these different roads, break them up and send them down different road so if they hit a mine they don’t all hit the mines. Arrive at the castle at the same time.

• Radiation solution is have lots of small lasers at all different directions so each laser on its own won’t harm the healthy tissue, but when they all converge on the tumour it will be destroyed.

Results: participants didn’t notice that they were solving different descriptions of the same problem. However once prompted to notice the similarities (the deep structure rather than the superficial), they solved the second one much more quickly.

Conclusions: we should treat transfer as an acquired skill. It can happen when prompted / trained, but don’t assume it will happen spontaneously.

Describe three strategies for enabling students to transfer deep knowledge (Willingham, 2021).

1. Use lots of varied, familiar, concrete examples so people can notice what the examples have in common. They can identify the underlying principle by seeing what they have in common. They are then more likely to transfer to a future unprompted example.

2. Be very explicit that you are trying to get people to learn the deep structure, not the superficial elements

3. Give opportunities for people to practice the transfer of deep knowledge to novel contexts

Describe three strategies for improving motor skill transfer (Schmidt & Lee, 2014).

1. Point out similarities or differences among skills during cognitive stage of skill acquisition (involves declarative knowledge anyway, you are talking about things in an explicit way).

2. Use verbal cues to emphasize transfer (e.g. name the same skill used in different contexts with the same name)

3. Use variable practice. Practice the motor skill in as many contexts as possible (e.g. playing tennis on different court surfaces). Makes the schema more robust.

What were Simons et al.’s (2016) four skeptical conclusions regarding the brain training literature?

These paid apps are sold as staving off cognitive decline in ageing.

1. Extensive evidence these brain-training interventions improve performance on trained tasks

2. Less evidence they improve performance on closely related tasks

3. Little evidence that training enhances performance on distantly related tasks

4. Little evidence it improves everyday cognitive performance

What did Shah et al. (2017) conclude from their generally positive review of the brain training literature?

“…current evidence supports that at least some commercially available computerized brain training products can assist in promoting healthy brain aging”

Describe three methodological problems associated with the brain training literature.

REWORD

1. Problem with generating an appropriate placebo group

   1. “How similar does the placebo control intervention have to get before it is still not brain training but nonetheless allows us to dismiss motivational/expectation confounds?” People can tell when they’re not in a brain training condition.

2. Traditional neuropsychological tests look like brain training exercises. So, when brain training improves “working memory” – is this more “training to the test”?

3. Another issue is who is in your sample (most studies are on impaired or aged groups – but claims are made for all ages and health levels).

Describe one study that found far transfer effects for brain training.

•There is evidence for far transfer from brain training type interventions.

•Useful Field of View training reduces crash involvement in older drivers by 50% (Ball et al., 2010). However, see Simon et al. (2016) for a critique of this study.

•Other research points to improvements in everyday functioning for older adults (e.g., Rebok et al., 2014).

What are the arguments for and against engaging in brain training based on the current evidence?

•There is evidence for far transfer from brain training type interventions.

•Useful Field of View training reduces crash involvement in older drivers by 50% (Ball et al., 2010). However, see Simon et al. (2016) for a critique of this study.

•Other research points to improvements in everyday functioning for older adults (e.g., Rebok et al., 2014).•

However – these effects could still in principle be some sort of placebo/expectation effect.

•If brain training is only a placebo effect – but is nonetheless effective in, say, reducing dementia risk – then should people still do it?

•Are we being overcritical – where the weaknesses are in the available research methods – and this is a developing field (where it’d be amazing if it works)? If even part of their claims are true, this would be huge. Cheap, accessible.

•Or over-eager (maybe we can get the same effects with something more fulfilling and cheaper)?

What is part-task practice?

Breaking up a skill into subskills to be practiced independently

Describe two circumstances when we should consider part-task practice.

1. May be more effective to practice sub-skills independently for some tasks. Whole task is too hard to start with, need to start with sub-components.

2. It could be that typical performance of final skill involves too much time practicing skill components that have already been mastered, so just target the un-mastered sub-components.

Describe Schmidt & Lee’s (2014) three principles of part-task practice.

1. If your skill is slow, serial (do one bit after the other), and there is limited interaction between each component, then progressive part-task practice could be effective (1. practice parts, 2. combine bits into bigger bits and bigger again, 3. keep combining until you’re practicing the whole task)

2. If task is brief and involves some sort of programmed movement then part training could be counterproductive (e.g. cricket swing broken up into foot movement, hip turn, arm movement, head movement) doesn’t really make sense. Each part only makes sense when they’re together, because one influences the other.

3. The greater the level of interaction between parts of a skill, the less likely part-task practice will work.

Describe five situations in which we might consider the use of simulation for training.

1. When the real task is too expensive (e.g. flying a 747)

2. When the real ask is too dangerous or otherwise inappropriate for novices

3. When the real ask (or key components of it) are encountered too infrequently to allow for effective practice (e.g. war, or crash avoidance)

4. When practice factors, such as feedback, motivation, practice scheduling are not ideal for learning real life task / or are hard to achieve in real-life task (e.g. normal driving the feedback you’re getting is you’re doing okay, so no motivation to improve, even if you were driving dangerously).

5. When opportunities to practice real life task are constrained (e.g. cycling on a stationary bike when you don’t have access to a velodrome)

Describe how the Top Gun school achieved far transfer in aerial combat.

• Picked best pilots to be the trainers and used aspects of deliberate practice

• Trainers would engage in simulated dogfights with students (and trainers would win).

• They would then receive debriefing feedback, asking why they did certain things and what they could have done differently. They would repeated the above steps.

• Then the students would learn to ask the questions themselves, took over more and more of the training.

• The program is credited with having a massive impact on real-world battle success (far transfer)

What is the key question we need to consider when designing simulation-based training, according to Horswill?

To what extent does practice in the simulation transfer to the performance of the actual task? This should be evaluated through empirical research.

Name and define the two types of simulator fidelity we need to consider when designing a simulation for training.

1. Physical fidelity

   1. How realistic the simulator looks in terms of physical experience of being in the simulator. High physical fidelity may not be needed (or even desirable) depending on what the goal of the training is (e.g. part-task training or training while avoiding trainee trauma).

   2. (do you need a steering wheel to anticipate hazards in driving? No. But for drumming if you’re just doing it in the air you’ve lost the feedback from striking the drum).

2. Psychological fidelity

   1. how realistic the simulation “feels”. This is the extent to which people behave in the simulator as they would in the real task. For example, a driving simulator may have excellent physical fidelity (it is like being in a real car) but poor psychological fidelity (people know they won’t get hurt if they crash so they behave more riskily than they would in real driving).

   2. How close psychologically is sitting in the simulator to the real task?