Chapter 11. Attention, Time-Sharing and Workload

1
Chapter 11. Attention, Time-Sharing and Workload

• MECHANISMS OF TIME-SHARING
• Automaticity and Resources
• automatic process
• a consistent mapping between stimulus and some
  further categorization
• rapid, accurate, and relatively resource free
• motor program (schema) highly overlearned
  manual skills as a consequence of learning not
  depend on guidance from visual feedback low
  attention demand, single-response selection,
  consistency of outcome
• readily and efficiently timeshared with other
  more demanding tasks
• the factors of effort and difficulty suggest that
  automaticity is not really an all or none
  phenomenon, but rather can be thought of as a
  continuum which can be related, inversely, to the
  mental resources demanded by a task
• performance resource function (PRF) data-linked
  (limited), resource-limited
• Time-Sharing and the PRF
• Automaticity and Difficulty
• practice and difficulty primary task vs.
  secondary task in dual-task situation
• Performance Strategies and the PRF
• heuristics (mental shortcuts) vs. optimal
  compensatory strategy
• utility in relation to effort and performance ?
  trade-off between performance and effort
• investing resources improve performance if the
  task is resource-limited (single vs. dual)

2

• Resource Allocation and Switching Strategic
  Control
• Allocation Skill
• optimal selective attention allocated visual
  attention to a channel, characterized by its
  information content and value (chapter 3)
• selective attention in cue seeking (chapter 8),
  and the choice of allocating or conserving effort
  between different decision strategies
• without training, nonoptimal fashion (allocating
  more resource to automatic task) at the expense
  of the resource-limited task
• allocating and switching attention improved
  time-sharing performance
• Optimal Allocation and Switching
• prescribe an optimal allocation schedule to tasks
  as a function of their importance
• relationship between optimum and actual
  allocation of resources to tasks
• inertia effect cost for switching between tasks
  ? tendency to continue performing a
  lower-priority task longer than optimal if the
  need to perform a task of higher priority
  suddenly arises
• the concept of switching attention suggests a
  movement metaphor scanning distance, not
  necessarily apply for cognitive distance,
  similarity causes interference
• saliency more rapid switching
• simplification for task management because it is
  a source of high cognitive workload or resource
  demand and, hence, would be self-defeating at the
  very time they might be most necessary
• tend to be more proactive in modest workload,
  more reactive in high workload
• well-designed automation support for task
  management skills

3

• Structural Factors in Time-Sharing Efficiency
• Bottleneck Theory
• the primary bottleneck for perception and
  response at the stage of response selection --
  for two RT tasks, two independent responses
  cannot be selected at the same time
• resource available for response selection must be
  allocated in an all or non fashion to one task or
  the other
• concurrent perceptual activity for two tasks can
  also compete ? reading and listening cannot be
  perfectly time shared
• separate resources for the two processes,
  perception and response should avail more
  efficient concurrent processing, compared to the
  use of a single source for two activities
• Multiple Resources
• three important structural dichotomies within the
  human brain in time-sharing efficiency
• vertical modality dichotomy between auditory and
  visual resources -- perception
• code distinction between verbal and spatial
  processes all stages of processing
• the stage of processing dimension -- perceptual
  and cognitive (working memory) different from
  action selection and execution
• Stages
• relative independence of perceptual demand and
  response selection process in concurrent RT tasks
  physiological evidence with P300 of ERP ? stage
  dichotomy related to brain structure
• speech and motor activity tend to be controlled
  by frontal regions in the brain (forward of the
  central sulcus), while perceptual and language
  comprehension activity tends to be posterior of
  the central sulcus

4

• Perceptual Modalities
• cross-modal time-sharing is better than
  intramodal ? not because of the result of
  separate perceptual resources within the brain
  but because of the result of the peripheral
  factors that place the two intramodal conditions
  as a disadvantage
• visual scanning is a factor that dual-task
  interference can be reduced by off-loading some
  info channels from the visual to the auditory
  modality
• Focal-Ambient Vision
• focal vision foveal, fine detail and pattern
  recognition
• ambient vision heavily involves peripheral
  vision, sensing orientation and ego motion
• supporting efficient time-sharing
• being characterized by qualitatively different
  brain structures
• being associated with qualitatively different
  types of information processing
• Processing Codes
• spatial and verbal processes, or codes, depend on
  separate resources and that this separation can
  often be associated with the two cerebral
  hemispheres ? the separation accounts for the
  high degree of efficiency with which manual and
  vocal outputs can be time-shared, assuming that
  manual responses spatial in nature and vocal ones
  are verbal
• the near perfect time-sharing efficiency with
  which auditory shadowing and visual-manual
  transcription tasks by skilled operators
  require separate resources
• manual control may disrupt performance in a task
  environment imposing heavy demands on WM, whereas
  voice control may disrupt performance of tasks
  with heavy verbal demands
• resource conflict, rather than compatibility, is
  the more dominant of the two forces in dual-task
  situations

5

• Confusion and Similarity
• proximity benefits (cooperation) and costs
  (confusion)
• Cooperation
• the improvement of time-sharing efficiency by
  increasing similarity from a common display
  property, mental set, processing routine, or time
  mechanism
• similarity in information-processing routines
  leads to cooperation and facilitation of task
  performance, whereas differences leads to
  interference, confusion and conflict
• Confusion
• the increasing similarity of processing material
  may reduce rather than increase time-sharing
  efficiency may be closely related to
  interference effects in memory
• multiple-resources theory (greater similarity
  producing greater interference) not
  appropriate, but more likely based on confusion,
  or outcome conflict ? Stroop task
• confusion is not always present nor always an
  important source of task interference
• PRACTICAL IMPLICATIONS
• three categories of applications of attention
  theory
• to system and task design most concern
  predicting multiple-resources model imposed by
  different task environments or system design
  features
• to operator training development of
  automaticity and training of time-sharing skills
• to operator selection which tests to predict
  success in time-sharing

6

• Predicting Multiple-Task Performance
• relative predictions of task interference
• allow the designer to know ahead of time which of
  two or more configurations will provide better
  multiple-task performance
• the role of confusion in dual-task performance is
  also relevant to performance
• presence of music as background or entertainment
  for operators engaged in various tasks relative
  independence
• relative predictions may be based on any
  quantifiable variables that are predicted to
  increase mental workload count variables ?
  higher levels along these count scales often
  produce lower time-sharing performance than lower
  levels
• absolute predictions
• at what level would be considered an overload
  condition
• time-line analysis absolute workload and
  performance prediction enable to profile the
  workload that operators encounter during a
  typical mission
• workload is proportional to the ratio of the time
  occupied performing tasks to total time available
  peak workload greater than 100 as potential
  bottlenecks
• time-line analysis needs to be extended to
  accurately predict performance
• operators may not necessarily choose to
  time-share two tasks if they have an opportunity
  to reschedule (i.e., postpone) one or the other
• must incorporate covert activities such as
  planning or rehearsal cognitive task analysis
• must be sensitive to the differences in resource
  demands of different tasks
• should incorporate the multiple-resources
  concept two tasks overlapping ? either very
  efficient or very disruptive, depending on their
  degree of resource conflict

7

• Assessing Mental workload
• Importance of Workload
• workload prediction, the assessment of workload
  imposed by equipment (optimize the system), and
  the assessment of workload experienced by the
  human operator (selection and training)
• workload is defined by resource supply- task
  demand (supply-demand function)
• underload region workload inversely related to
  reserve capacity
• overload region workload inversely related to
  the level of task performance
• Equipment Assessment
• to measure the workload of a system already
  existing at some stage of production to identify
  bottlenecks where resource demands momentarily
  exceed supply and performance breakdown
• if the absolute level of workload by a system is
  above or below a given absolute criterion level
• Assessing Operator Differences
• the level of skill or automaticity by operators
  with same primary-task performance
• operators may be monitored on-line in real task
  performance adaptive automation
• Criteria for Workload Indexes
• sensitivity diagnosticity selectivity
  obtrusiveness bandwidth and reliability
• Primary-Task Measures
• primary task the target of evaluation
• four reasons that primary-task performance may be
  insufficient to reveal clearly the merits of the
  primary task
• two primary tasks may lie in the underground
  region of the supply demand space cannot
  discriminate between them

8

• two primary tasks to be compared may differ in
  how they are measured or what those measures mean
• sometimes it is impossible to obtain good
  measures of primary-task performance
• two primary tasks may differ in their
  performance, not by the resources demanded to
  achieve that performance, but by differences in
  data limits
• The Secondary-Task Technique
• measure of residual resources or capacity not
  utilized in the primary task ? inversely
  proportional to the primary-task resource demands
  ? may reflect differences in task resource
  demand, automaticity, or practice that are not
  reflected in primary-task performance
• the investigator is interested in variation in
  the secondary-task decrement to infer differences
  in primary-task demand
• loading task technique ? different allocation
  instructions provided ? devote all necessary
  resources to the loading task first then compare
  differences between primary tasks
• Secondary Task Examples
• rhythmic tapping task, random number generation,
  probe reaction time task, time production and
  time estimation techniques
• Benefits and Costs of Secondary Tasks
• benefits
• a high degree of face validity
• can be applied to two very different primary
  tasks and give workload measure in the same units
• costs
• must account for the fact that there are
  different kinds of resources not always
  sensitive
• may interfere with and disrupt performance of the
  primary task obtrusiveness

9

• Physiological Measures
• Heart-Rate Variability
• variability or regularity of HR as a measure of
  mental load
• variability decreases as the load increases
  more sensitive than diagnostic
• Pupil Diameter
• correlates closely with the resource demands of a
  large number of diverse cognitive activities
• highly sensitive, although undiagnostic
• Visual Scanning
• dwell time an index of the resource required
  for information extraction from a single source
• scanning as a diagnostic index of the source of
  workload within a multi-element display
  environment
• Costs and Benefits
• relatively continuous record of data over time
• not obtrusive into primary-task performance
• physically obtrusive, must be used to infer
• Subjective Measures
• NASA Task Load Index (TLX) workload assess on
  each five 7-point scale
• subjective workload assessment (SWAT) workload
  on three 3-point scales
• Costs and Benefits
• do not disrupt primary-task performance,
  relatively easy to derive

10

• Relationship Between Workload Measures
• dissociation compared conditions have different
  effects on different workload measures
• dissociation between primary task and subjective
  measures effort and the number of tasks ? do
  not always influence performance in the same way
• in the underload region, the greater resources on
  the more difficult task will not yield better
  performance
• performance will differ but the resource invested
  (and therefore the subjective workload
  experienced) will not differ
• if two systems are compared, one of which induces
  a greater investment of effort, the latter will
  probably show higher subjective workload, even as
  its performance is improved
• a very strong influence on subjective workload is
  exerted by the number of tasks that must be
  performed at once
• Consequences of Workload
• increases on workload do not inherently have
  bad consequences it is the low levels of
  workload with boredom, fatigue, or sleep loss
  have negative influence on human performance
• 4 types of adaptation in excessive workload
• people may allow performance of tasks to degrade
• people may perform the tasks in a more efficient,
  less resource-consuming way
• people may shed tasks altogether, in an optimal
  fashion, eliminating performance of those of
  lower priority
• people may shed tasks in a nonoptimal fashion,
  abandoning those that should be performed