Cognitive Load and Planning'

1
Cognitive Loadand Planning.
2
Introduction.
3
Planning.

• To plan is to model a sequence of actions in
  preparation for carrying out a particular task
  (Shallice, 1982).
• Before embarking on an action sequence which is
  novel or complex, we usually spend some time
  thinking what we are about to do, how best to
  achieve the goal, in what order to perform the
  individual actions, and how much time and effort
  will need to be allocated to the task. Memory is
  involved in formulating such plans, holding the
  elements and sequence in mind while the plan is
  being assembled, evaluated, revised and
  implemented (Cohen, 1989, p. 50).

4
Reducing Cognitive Load.

• Therefore, planning places a heavy load on
  cognitive resources.
• How can this cognitive load be reduced and
  planning facilitated?
• Performance is dependent on
• i) an individuals WM capacity.
• ii) the cognitive load made by the task.

5
External Representation.

• Common view that all representations in cognition
  occur solely in the mind and that external
  objects are, at best, peripheral tools (Zhang
  Norman, 1994).
• But external representations (ERs)
• Can give access to knowledge and skills
  unavailable from internal representations.
• ER formalisms include tables, lists, and set
  diagrams (e.g., Venn diagrams).
• Graphical ERs influence cognition by reducing
  search and working memory load by organising
  information by location (Cox, 1999).
• In choosing to represent cognition externally,
  benefits must outweigh costs (Zhang, 1997).

6
External Representation contd.

• Scaife and Rogers (1996) identify three aspects
  of external cognition
• computational offloading- how differential
  external representations reduce the amount of
  cognitive effort needed to solve problems that
  are informationally equivalent.
• Compare solving geometry diagrams to sentences
  containing the same information.
• Re-representation- how different external
  representations, but with the same abstract
  structure, make problem-solving simpler or more
  difficult.
• Compare the ease of multiplying LXVIII x X to 68
  x 10!
• Graphical constraining- graphical elements in a
  graphical representation can constrain the kinds
  of inferences that can be made about the
  underlying represented world.

7
The Experiment.

• Will examine how external representation supports
  planning.
• Will control for memory span.

8
The Debrief.
9
External Representation and Problem-Solving.

• External representation
• reduces the load on internal working memory.
  Storing less information in internal working
  memory means there is less chance of forgetting
  information. This reduces errors.
• may make the task seem to be less cognitively
  complex, because of the reduced load on working
  memory. Hence, participants feel more confident
  about solving it.
• allows the user to become more focused on solving
  the problem as opposed to remembering the rules.
• (Noyes Garland, 2003 p. 580).

10
Working with Plans.

• Five specific types of cognitive load in
  planning
• The memory load required to store the plan.
• Identification of options for action at each step
  of the plan.
• Decision about which of these possible actions
  should actually be chosen as the next step in the
  plan.
• Keeping track of dependencies between planned
  actions.
• Keeping track of the effect of the plan as a
  whole with respect to overall goals. Analysing
  the combined effect of multiple actions.

11
Working Memory.

• Limited capacity memory system involved in the
  temporary storage and processing of information.
• Maintains, integrates, and manipulates
  information from different sources.
• Model of Baddeley (1986).

12
The Multi-component Model of Working Memory.
Central Executive
Phonological Loop
Visuospatial Sketchpad
Long-term Knowledge Structures Visual
Language Semantics
13
Measuring the Working Memory System.

• Simple span tasks require only the storage
  operation of the PL or VSSP.
• Working memory span tasks require storage in the
  PL or VSSP and simultaneous processing which
  draws on the CE.
• Key
• CE Central executive.
• PL Phonological loop.
• VSSP Visuospatial sketchpad.

14
The Importance ofWorking Memory.

• Working memory involved in, for example
• language acquisition (Gathercole and Baddeley,
  1989).
• reading comprehension (Daneman Carpenter,
  1983).
• mental arithmetic (Logie, Gilhooly, Wynn,
  1994).
• problem-solving (e.g., Johnson-Laird, 1983) and
  hypothesis generation (e.g., Gilhooly et al.,
  1993).

15
Planning and the Tower Tasks.

• Shallices (1982) Tower of London planning task.
• Objective move from start state to goal state.
• Constraints
• Only one disc can be moved at a time.
• All discs not being moved must be placed on a
  peg.
• A larger disc must not be placed on a smaller
  disc.

16
External Representation and Tower Tasks.

• Role for working memory in Tower tasks
• Involvement of executive processes (planning) and
  spatial working memory (e.g., Goel Grafman,
  1995 Phillips et al., 1999).
• External representation aids performance
• Zhang and Norman (1994).
• Noyes and Garland (2003)
• Participants most successful and fastest with
  computerised display on Tower of Hanoi. But more
  trial-and-error approach.
• Mental solutions led to more strategies and fewer
  moves to solution.

17
External Representations and Planning contd.

• But this is one, very abstract, planning task.
• Designed to have very little specific domain
  knowledge.
• Is there a similar influence of ERs and working
  memory on other types of planning task? And in
  more realistic domains?

18
The Experiment.
19
Overview of the Experiment.

• Designed to examine
• the relative contribution of verbal and
  visuospatial working memory to planning.
• the impact of external representations on
  performance.
• which aspects of the interface may help planning.

20
Tasks.

• Working memory measures
• Simple span tasks-
• Verbal digit span.
• Visuospatial Corsi block span.
• Working memory span tasks-
• Verbal verification-digit span.
• Visuospatial spatial working memory span.
• Planning task.

21
Order of Presentation.

• The tasks were presented in two orders
• Digit span, verification-digit span, Corsi Block
  span, spatial working memory span, planning.
• Corsi Block span, spatial working memory span,
  digit span, verification-digit span, planning.
• i.e. presentation was blocked by modality
  (verbal, spatial) and then simple span task
  followed by working memory span task.

22
Verbal Working Memory Measures.
23
Digit Span Task.

• Recall a series of digits in the correct serial
  order by entering them on calculator-style
  keypad.
• Number of digits to recall increases gradually
  over the course of the task, with 2 trials at
  each level.
• Span level defined as the last level at which 1/2
  trials are answered correctly.

24
Verification-Digit Span Task.

• Adapted from Shah and Miyake (1996).
• Processing component indicate with a
  button-press whether simple sentences are true or
  false.
• e.g., Trees are living things. (True)
• Butter is a snack. (False)
• Storage component remember simultaneously
  presented digit for later recall in serial order,
  using keypad.
• Span level increases in same way as digit span.

25
Visuospatial Working Memory Measures.
26
Corsi Block Span Task.

• Corsi Block span task (Corsi, 1973).
• Blocks highlighted in serial order in an array,
  with two trials at each level.
• Remember spatial sequences.
• Reproduce them by clicking on blocks in an empty
  array.
• Span measure (progress to next level if at least
  1/2 correct).

27
Corsi Blocks Span Task.
28
Spatial Working Memory Span Task.

• Spatial working memory span task.
• Processing component indicate whether more cells
  highlighted in top or bottom half of screen.
• Storage component recall positions of cells
  marked with checked pattern, by clicking on
  blocks in an empty array.
• Span level calculated in same way as for Corsi
  block span.

29
Spatial Working Memory Span Task.
30
The Planning Task.
31
The Medical Planning Task.

• Initial familiarisation with interface.
• Graded problems of increasing difficulty, with
  increasing constraints (pregnancy, surgery) and
  number of illnesses to reduce risk of
  contracting.
• Task instructions presented on paper.
• Different levels of planning support offered by
  the interface timeline, timeline graph,
  timeline graph constraint warnings.

32
Design.

• For both tasks, the IV is interface support (3
  conditions timeline, timeline graph, timeline
  graph constraint warnings).
• Span tasks
• Dependent variable span recall accuracy.
• Planning task
• Dependent variables
• Time -Time taken to solve the planning problem.
• GraphAarea Final area under Graph for Line A
  (i.e. overall success of the plan- lower score
  better).
• Insertions Number of events inserted into plan.
• Deletions Number of events deleted from plan.

33
Statistics Note.

• Note Due to a design flaw, many participants
  failed to contribute a spatial working memory
  span score. Therefore, instead of analysing span,
  use the recall accuracy measures for each task
  instead. You could also look at the verification
  (processing) component for both of the working
  memory span scores too, if youre feeling keen.

34
Null Hypotheses.

• Null hypotheses (H0)
• No relationships between span and planning tasks.
• No difference in planning scores between the
  three interface support conditions.

35
Alternative Hypotheses.

• Alternative hypotheses (H1)
• Relationship expected between span tasks and
  planning.
• Planning task to have stronger correlations with
  working memory span tasks than simple span tasks.
• Strongest correlation between spatial working
  memory span and planning, due to visuospatial
  nature of interface.
• Significant difference in planning scores between
  conditions. More interface support should lead to
  more successful performance.

36
Data Analysis.
37
Variable Definitions.

• The data are already formatted as an SPSS data
  file.
• All variables are labelled in the Variable View
  window in SPSS.

38
Computing Overall Scores.

• Use SPSSs Compute function to calculate
• mean verbal and visuospatial spans and an overall
  mean memory span.
• an overall mean for each of the planning
  dependent variables (i.e., collapsed across the
  three problems).
• To do this
• Pull down the Transform menu,
• select Compute,
• then Name your Target Variable (e.g. PlanTime),
• select MEAN(numexpr, numexpr),
• then enter (between the brackets) the DVs you
  want to convert into a mean value. For example,
  MEAN(PlanTime1, PlanTime2, PlanTime3).
• A new variable should appear with the mean value
  in it- check the calculation is correct and then
  repeat.

39
Planning Task Analyses.

• Run one-way ANOVAs on each of the converted
  planning DVs to determine whether the level of
  external representation has an influence on the
  planning task and on which dependent variables it
  has most impact.
• Use post hoc Bonferroni comparisons to determine
  which of the conditions differs significantly
  from the others.
• Plot group performance for each planning DV.

40
Working Memory Analyses.

• Run an ANOVA to determine whether there are any
  differences in overall recall performance between
  the three conditions.
• Use post hoc Bonferroni comparisons to determine
  which, if any, of the conditions differs
  significantly from the others.
• Report the group accuracy means for each span
  task.

41
Planning Working Memory Correlations.

• Run bivariate correlations on the accuracy scores
  for the four working memory tasks and the
  converted planning scores.
• To determine whether there is any relationship
  between planning and working memory performance.
• Which of the four working memory measures
  correlates most highly with planning ability?

42
Planning Working Memory Linear Regression.

• Run linear regression analyses on any of the
  planning DVs that correlate significantly with
  span tasks (using recall accuracy on the span
  tasks as the IVs).
• How much variance (r²) in planning ability can be
  explained by the span task?

43
Discussion.
44
Discussion.

• To include a discussion of
• how strongly do the various span tasks correlate?
  Is there a dissociation between verbal and
  visuospatial working memory?
• how did participants do on the working memory
  tasks? Were they better on the simple span tasks
  or the working memory span ones?
• how did external representation influence
  planning performance?
• what contribution did recall accuracy make to
  planning performance? Is it a good predictor of
  planning ability?

45
Discussion contd.

• Consider any flaws in the design and how they
  might have influenced the results. How could the
  design be improved upon in the future?

46
Applications of the Work.

• Greater understanding of
• role of working memory in planning.
• support offered by external representations to
  planning.
• Medical planning
• REACT (Risk, Events, Actions, and their
  Consequences over Time).
• Supports communication of risks and implications
  of planned courses of actions.
• Interface split into 3 main sections planning,
  outcome measures, and argumentation.
• For an overview, visit www.acl.icnet.uk/lab/react
  .html

47
References.

• Baddeley, A.D. (1986). Working Memory. Oxford
  Clarendon Press.
• Cohen, G. (1989). Memory in the real world. Hove
  Lawrence Erlbaum Associates.
• Cox, R. (1999). Representation construction,
  externalised cognition, and individual
  differences. Learning and Instruction, 9,
  343-363.
• Corsi, P. (1973). Human memory and the medial
  temporal region of the brain. Dissertation
  Abstracts International, 34(2-B), 891.
• Daneman, M., Carpenter, P.A. (1983). Individual
  differences in integrating information between
  and within sentences. Journal of Experimental
  Psychology Learning, Memory, and Cognition,
  9(4), 561-584.
• Gathercole, S.E., Baddeley, A.D. (1989).
  Evaluation of the role of phonological STM in the
  development of vocabulary in children A
  longitudinal study. Journal of Memory and
  Language, 28, 200-213.
• Gilhooly, K.J., Logie, R.H., Wetherick, N.E.,
  Wynn, V. (1993). Working memory and strategies in
  syllogistic-reasoning tasks. Memory Cognition,
  21(1), 115-124.
• Goel, V., Grafman, J. (1995). Are frontal lobes
  implicated in planning functions? Interpreting
  data from the Tower of Hanoi. Neuropsychologia,
  33(5), 623-642.
• Johnson-Laird, P.N. (1983). Mental models.
  Cambridge Cambridge University Press.

48
References contd.

• Logie, R.H., Gilhooly, K.J., Wynn, V. (1994).
  Counting on working memory in mental arithmetic.
  Memory and Cognition, 22, 395-410.
• Noyes, J.N., Garland, K.J. (2003). Solving the
  Tower of Hanoi does mode of presentation matter?
  Computers in Human Behavior, 19, 579-592.
• Phillips, L.H., Wynn, V., Gilhooly, K.J., Della
  Sala, S., Logie, R.H. (1999). The role of
  memory in the Tower of London task. Memory, 7,
  209-231.
• Scaife, M., Rogers, Y. (1996). External
  cognition how do graphical representations work?
  International Journal of Human-Computer Studies,
  45, 185-213.
• Shah, P., Miyake, A. (1996). The separability
  of working memory resources for spatial thinking
  and language processing An individual
  differences approach. Journal of Experimental
  Psychology General, 125(1), 4-27.
• Shallice, T. (1982). Specific impairments of
  planning. Philosophical Transactions of the Royal
  Society of London, 298, 199-209.
• Zhang, J. (1997). The nature of external
  representations in problem solving. Cognitive
  Science, 21(2), 179-217.
• Zhang, J., Norman, D.A. (1994). Representations
  in distributed cognitive tasks. Cognitive
  Science, 18, 87-122.