working memory 3 - training Flashcards
Ericcson et al (1980) - training to increase memory span
230 hours of practice in a lab - someone was able to increase span from 7 to 79 digits
seemed like with a good mnemonic system there was no limit to memory performance
brain training ideas
define transfer
improvements in a practiced tasks lead to improvements in unpractised tasks
Ericcson et al (190) - training and transfer
switched from digits to letters after practicing for 3 months
showed no transfer from training
memory span reduced back down to 6 consonants
strategy-based training
introduction and acquisition of strategies
e.g. mnemonic to remember the order of planets
can be material or task specific
difficult to transfer to other contexts - e.g. a mnemonic only helps you remember that one specific thing
process-based training
repeated practice of specific tasks targeting cognitive processes
e.g. daily practice of complex span tasks
assumed to transfer to other contexts
functional overlap
transfer is expected if practiced and non-practiced tasks share underlying processes
WM as a core ability
variation in working memory is correlated with variation in many other abilities
WM:
- reasoning
- attention
- reading
- vocab learning
- storytelling
therefore by enhancing WM - could be able to improve wide range of related abilities
near and far transfer
marker of successful training
example:
practice = WM training task (n-back)
near transfer = untrained WM task (complex-span task)
far transfer = different but related cognitive ability (reasoning)
how to measure training effects
methodological rationale
compare pre-test (baseline) to post-test performance after training
pre-test –> training –> post-test
pre and post = all tasks (n-back, complex span, reasoning)
training = n-back only
use of control group in training and transfer studies ( 2 types + eval)
evaluate change relative to control group
passive group:
- no intervention
- good - test-retest effects ( performance improves with familiarity )
- bad - other factors that affected the period between test and retest
- bad - placebo effect
active group:
- alternative intervention
- good - other factors that affected the period between test and retest (e.g. motivation)
- good - reduces placebo effect
seminal training study - attention and ADHD - methodology
Klinberg et al (2002)
Can intensive working memory training help children with attention deficits such as in ADHD?
method:
- computerised training program with WM tasks
- test improvements relative to active control group in trained/untrained tasks
training task:
e.g. visuospatial WM task
transfer task:
e.g. raven’s progressive matrices (like non-verbal reasoning where you have to say which symbol comes next)
5 week training in conditions:
- intensive = 5x day WM training
- low-dose = 1x day WM training ( active control group )
post-test: look at change in training and transfer task
seminal training study - attention and ADHD - results
Klinberg et al (2002)
improved ability in intensive condition for both training and transfer task - some individuals by very large amounts
transfer task had a bigger spread of data - some improved by 12 points and some by 2
training task improved by 3-5 points - less spread of data
in low-dose - basically stayed the same, some got worse and some better (from -2 to 4 points)
seminal training study - attention and ADHD - evaluation
Klinberg et al (2002)
only sample of 7
first evidence for training and transfer effects
therefore unknown if its replicable
seminal training study - Klinberg (2005) - second study - method and results - adaptive vs non-adaptive training
multicentre, randomised controlled trial - 53 participants
5 weeks of training
- adaptive WM training group = change in difficulty of tasks as they improve
- non-adaptive WM training group = control = same difficulty level of training throughout
then 3 month gap before a follow up
results:
training tasks:
- span-board
- digit-span
larger benefits in adaptive group relative to nonadaptive in the practiced (training) tasks
transfer tasks:
- stroop task ( quicker = better )
- raven
larger benefits in adaptive relative to nonadaptive group in unpracticed inhibition and reasoning tasks
both groups improve but control not so much
seminal training study - Klinberg (2005) - second study - evaluation
the values in the results are corrected for differences in baseline score
uncorrected scores (raw data) shows control group stays same and experimental improves in both training tasks
uncorrected scores shows experimental group started at baseline with better scores in the raven transfer task
- they still improved but they were better to begin with
- both groups improved by basically the same amount
there is evidence for training and transfer effects
corrected differences are significant but uncorrected group differences are only small
seminal training study - Jaeggi et al (2008) - dual n-back training
8-19 training sessions
2 groups:
- WM training
- passive control ( no intervention )
measured change in training and transfer tasks at post-test
training:
dual n-back tasks
2 types of stimuli at the same time - visual and auditory targets (look at slides if confused)
transfer:
found improvement in Raven’s task (progressive matrices - non-verbal reasoning)
shows transfer effects
the hype of WM training - Redick et al (2013) method
pre-test
10 training sessions
mid-test
10 more training sessions
post-test
training:
- WM training
- active control (visual search)
- passive control ( no intervention)
results:
no significant near or far transfer effects in spatial or verbal working memory
hype of WM training - 2002-2015 WM training studies
hype and inconsistencies lead to this being a very active field of research in early 2010s
inconsistent evidence across large number of studies
need to find how these inconsistencies can be explained
reasons for inconsistencies with WM training studies - methodological (3)
- lack of active control - placebo
- task impurity problem - single tasks used for measuring cognitive abilities
- small samples - low statistical power and imprecise measurement
still not known why training works better in some studies than others
reasons for inconsistencies with WM training studies - theoretical (2)
many studies lack theoretical framework of training and transfer
without theory explaining mechanisms of transfer (i.e. why we expect effects), we can’t predict when we should observe effects
mechanisms of transfer - sources of variation frameworks (2)
von Bastian & Oberauer, 2014
multiple sources of variance framework:
- intervention-specific factors
- individual differences
these two impact training and transfer and therefore the observed effects
2 proposed mechanisms of transfer
enhanced capacity
enhanced efficiency
enhanced capacity (transfer)
training increases the number of information elements held in WM
i.e. larger broad focus of attention
prediction: training leads to broad transfer effects
enhanced efficiency
training supports a more efficient use of the existing capacity of WM
through strategies or faster processing
prediction: training leads to selective transfer effects
mechanisms of transfer study - methodology
De Simoni & von Bastian’s (2018)
pre-test –> 20 sessions training –> post-test
training:
- WM binding training
- WM updating training
- active control (visual search)
tested near and far transfer and trained abilities
binding task:
- show a symbol and number - bound together
- then recognise whether new presented pair was shown before (watch lecture if unsure here)
updating task:
- shown memoranda (2 numbers)
- then have 9 updating steps - either add or minus from original numbers
- then say what the answer is
visual task:
- e.g. find a 3 in an array of 8s
mechanisms of transfer study - results
large improvements in the trained tasks
no evidence for near or far transfer
training improved neither WM capacity nor efficiency
but: might be different for other WM tasks or measures of specific types of efficiency
maximising training benefits - individual differences
large individual differences are seen in training progress
often younger adults do better than older throughout sessions of training
3 hypotheses of who benefits most from training
magnification:
higher ability = gain more (rich get richer)
compensation:
lower ability = gain more
no difference
training - individual differences study
Guye et al.’s (2017)
how are initial training performance and slope in training progress related?
results:
- younger adults showed magnification of initial task performance
- little effect in older adults
intervention-specific factors - 2
study whether training and transfer effects depend on:
type of training task
dose of intervention
intervention-specific factors - Melby-Lervåg et al.’s (2016) meta-analysis
type of training task:
- cogmed = computerised training program
- compared cogmed, n-back, and complex span
- Cogmed (bran training) induces relatively larger verbal near transfer, but n-back yields relatively larger far transfer
training dose:
- dose has little effect
- only significant difference occurs for far transfer – but in the opposite direction (small dose shows larger improvement in far transfer)
possible indvidual differences in brain training (5)
age
gender
personality
motivation
beliefs
individual differences - Guye et al (2017) study
study of how demographics, personality, motivation, and beliefs are related to the slope in training progress
limited evidence for individual differences predicting slope in training progress
however, results may differ for samples with successful training and transfer
