How beneficial or harmful is it to multitask?
- Janssen et al. (2015) multitasking and interruptions are ubiquitous --> in some environments e.g. offices, likely to affect productivity but unlikely to have direct dangerous consequences. Other settings e.g. aviation, healthcare and driving, they can impact safety
- Switching between tasks requires people to switch between physical and mental states --> operations required to make these changes take time and resources and thereby affect performance = will have some cost
- emotional costs associated with multitasking e.g. interruptions can increase feelings of stress and frustration by subjective (Mark et al., 2008) and physiological measures (Brumby et al., 2014)
- Janssen et al. (2011) in some settings multitasking and responding to interruptions can be considered adaptive and rational despite the costs incurred e.g. medic moving from one patient to the next incurs a variety of costs (e.g., moving wards, changing of gloves, reading of charts) but such moving is entirely rational if a patient requires emergency attention
- Might also switch activities if not making sufficient progress on the current activity, or because switching tasks might revel some new/useful info
- In other situations, people switch simply because they are bored (Jin and Dabbish, 2009) --> in monotonous tasks in particular, occasional multitasking can improve vigilance.
- potential to be used in the design of virtual, physical, and organizational interventions to help people manage tasks effectively e.g. preventing switching, providing info about the context in which an interruption or distraction takes place, providing training so that people better manage interruptions, designing systems to mediate the interaction between users and other tasks demanding their attention
Salvucci et al. (2009) claims multitasking exists on a continuum:
- looked at the time spent on one task before switching to another task
- some tasks we can easily do together e.g. talking whilst driving
- some tasks we cannot easily do together e.g. write a text message whilst driving
- thought to be because theres an increase in time before switching tasks: goes from concurrent multitasking to sequential multitasking
Wickens (1984) multiple resource model
- the human operator does not have one single information processing source that can be tapped, but several different pools of resources that can be tapped simultaneously
- depending on the nature of the task, these resources may have to process information sequentially if the different tasks require the same pool of resources, or can be processed in parallel if the task requires different resources
Threaded cognition (Salvucci and Taatgen, 2007)
- based on Wickens, proposed integrated theory of concurrent multitasking (performing 2 or more tasks at onee)
- idea that multitasking behaviour can be represented as the execution of multiple task threads, co-ordinated by a serial cognitive processor and distributed across multiple processing resources
- Multitasking and executive control behaves like a cook preparing food in a kitchen
- cook manages a number of resources in order to complete a stream of orders
- to prepare fish, pasta or cake a cook might execute certain steps, with each step generally involving the cook initiating some process, waiting for that process to complete and then moving on to the next step = each step is associated with the desired food order and the current state of the resources used
- if have to fill multiple orders simultaneously, some processes can be executed in parallel (e.g. fish bake whilst pasta boils). But several constraints limit amount of parallelism in filling order
- cook may experience resource conflict when 2 or more orders need same resource e.g. fish and cake need oven at same time with different temps
- 2 or more processes may require cook’s attention at the same time e.g. oven-preheating and water boiling processes end at same time = one order must be delayed.
- In threaded cognition, central procedural resource can be compared to the cook, and other resources (e.g. perceptual and motor) compared to the various cooking resources available to the cook.
- the procedural resource (cook) collects processing requests to the other resources
- goals (not orders) guide processing to directed action for desired task
- conflicts when multiple tasks require same peripheral resource (e.g. 2 require vision) or when multiple tasks require attention from the central procedural resource (e.g. auditory and visual tasks finish encoding simultaneously and need procedural resource to continue)
- threaded cognition initially relies on memorised instructions but improves skill through learning- just as a cook requires a cookbook in initial stages of learning
3 main assumptions:
- 1) Resource seriality assumption = all resources execute processing requests serially, one request at a time
- cognitive, perceptual and motor resources execute processing requests one at a time —> may be counter-intuitive as it is well known that there is parallelism in the brain’s neural circuitry
- but, still allows for parallelism at the level of multiple resources (e.g. parallel visual and aural processing) and within each resource (e.g. visual processing of scenes or parallel activation competition for declarative retrieval)
- BUT doesn’t account for role of memory —> forget first in series, and doesn’t account for priority e.g. fish more important than pasta due to cost, so prioritise correct cooking of fish
- 2) Resource usage assumption = threads acquire and release resources in a greedy, polite manner
- threads acquire resources in a greedy manner by requesting resources as soon as possible when needed (all resource requests occur in the actions of the rule currently firing, and the firing of this action coincides with the acquisition of the resource )
- after resource acquisition, a thread releases resources for other threads as soon as its processing is no longer required = in “polite manner”
- 3) Conflict resolution assumption = when multiple threads contend for the procedural resource, the least recently processed thread is allowed to proceed
- when a thread requires a resource and that resource is currently busy, will usually wait until current process is completed and then acquire that response —> but in certain cases 2 or more threads may be waiting to acquire a resource —> least recently processed thread is allowed to proceed
- this policy allows a high-frequency task to execute at high efficiency, but still allows the low-frequency task to acquire the procedural resource when attention is needed
Live in an era of ubiquitous computing (present, found everywhere)
- Weiser's vision: ubiquitous computers will come in different sizes all suited to a particular task
- Most profound technologies are those that disappear --> become indistinguishable from everyday life
- Pielot et al. (2014) p's received on average 63.5 mobile notifications per day.
- Engage in so much everyday multitasking e.g. talking on the phone whilst doing any other task
- Schuman (1984) = situated action theory --> users are likely to display intelligent adaptation when interacting with user interfaces, especially when they are novel/used in novel environments --> there's a value in looking at use of interfaces in natural settings
- +observe natural interaction with a system = insight into components that might not have been considered in the design process
- +allows to see how users adapt/learn to use a new interface --> can be used to design interfaces to enhance performance. Important considering a user experience cannot be created, but certain features can be put in to encourage/stimulate a user experience
- -degree of experimental manipulation is as minimal as possible --> increases the difficulty in obtaining accurate measurements
- -difficult to make predictions --> natural environments are noisy, have many alternative factors for consideration making it hard to pinpoint specific expectations and hypotheses
Cognitive modelling (e.g. Fitts' Law)
- +can be used to make predictions about users' interactions, especially when existing data has been put through the model and found to be robust in predicting observations
- +Field testing and observations don't have to be conducted too extensively = can focus on innovations and developments
- +can be applied to various fields of interest as it is generally a broad concept/area
- -models are limited in that they are only able to account for so many factors --> poses a challenge when models developed from lab-based studies are to be applied to in-situ findings e.g. Wobbrock et al. (2008) who extended the Fitts' Law model to include an error prediction component that it lacked
- -modelling assumes all users are going to act in the same way when they do not e.g. as seen with supertaskers
Watson and Strayer (2010) aim to identify individuals with extraordinary multitasking ability
- great number of people believe that the laws of attention do not apply to them --> e.g. they are the exception to being impaired at multitasking whilst driving
- used a driving simulator to pair the task of driving with an auditory version of the OSPAN task (artificial phone call) using 200 p's (better than Brumby, only used 8).
- Ps performed the driving and OSPAN tasks in combination and also performed each of the tasks separatel
- Predicted that most individuals would show substantial performance declines in driving and OSPAN when performed together compared with the single-task baseline measures --> pattern would highlight the role that executive attention plays in driving and would provide evidence that the dual-task interference was bidirectional (i.e., costs are observed for both driving and OSPAN tasks).
- By contrast, individuals with extraordinary multitasking ability, if they exist, might be able to perform these two tasks in combination without impairment --> Such a finding would challenge theories that postulate the existence of immutable bottlenecks in dual-task performance.
- some people were found to be ‘supertaskers’ --> had strikingly remarkable ability to successfully perform two attention-demanding tasks that over 97% of the population cannot perform without incurring substantial costs in performance
- But, decline in OSPAN scores for majority of p's implying that operating a motor vehicle also places demands on cognitive control networks associated with executive attention = demonstrated bidirectional interference using dual tasks
Why are we all not supertaskers? 2 possibilities suggested:
- 1) cost associated with being a supertasker - Grossberg (1987) organisms are faced with a stability/plasticity dilemma in which they must strike a balance between being overly rigid and overly flexible in their processing style. many clinical disorders are associated with an imbalance (APA, 1994).
= It may be that supertaskers excel at multitasking at the expense of other processing abilities.
- 2) there may be few costs (and possibly some benefits) associated with being a supertasker, but the environmental and technological demands that favour this ability are relatively new, and any selective advantage for being a supertasker has yet to propagate throughout the population.
- OSPAN task isn't the topic of normal phone conversations --> but argue that its naturalism wasn't the purpose of this task, used because can link to executive attention in the context of multitasking. Rules for task are objective and easily quantifiable compared to naturalistic conversations
- used a driving simulator- how much does this resemble real life? But, more complex than Brumby et al. --> had distractor vehicles that braked intermittently despite still being on a highway, p's controlled speed and unique driving scenarios for each participant
- Future research should focus more on individual differences --> individuals who perform significantly better (or worse) than the group average may help to shed light on the mechanisms of attention and cognitive control in a fashion similar to how failures of perception or memory (e.g., illusions) help us to understand the rules governing normal perception and memory.
- Attention has a limited capacity (Kahneman, 1973)
evident in situations in which people attempt to perform two or more attention-demanding tasks concurrently. In these situations, a reciprocal pattern emerges, wherein performance on one task prospers at the expense of the other (Wickens, 1980).
- Not uncommon for people to multitask in everyday situations. National Safety Council (2010) estimated that 28% of all accidents and fatalities on U.S. highways are caused by drivers using cell phones.
Theoretically possible that a small segment of the population may be capable of performing complex multitasking with little or no apparent costs. What allows them to exhibit behavior that seemingly violates cognitive scientists’ current understanding of attention and dual-task control?
- Ophir et al., 2009 individuals who report multitasking more frequently, multitask less well than those who are less frequent multitaskers
- Engle (2002) developed an operation span (OSPAN) task - sensitive to individual differences in executive attention.
- involves maintaining the task goal of memorizing items and recalling them in the correct serial order, while concurrently performing distracting math problems.
- Individual differences in OSPAN performance shown to predict behavior on a variety of classic cognitive tasks e.g. Stroop color naming and the antisaccade task (Engle, 2002), thought to require executive attention.