Multitasking Flashcards
COGNITIVE CAPACITY
- even in singular tasks, cognitive capacity = limited
- memory retrieval/decision making = time consuming
- input processes (ie. syntactic parsing) handling limits
- representational/storage capacity (ie. WM) = limited
- limits = obvious when resources shared via tasks ie. more than 1 in certain time/some = time critical; simultaneous/switch between
MULTI-TASKING
- ie. cooking/ironing/baby monitoring/phone calling
- ie. chef/pilot/taxi driver
- limitations = important theoretically (global computational architecture of brain) + practically (efficiency/risk/human error)
MUTLI-TASKING DEMANDS
- competition for shared resources in simultaneous tasks (dual-task interference)
- task-switching = set-shifting/task shifting costs/retrospective memory (what’s left to do?)/prospective memory (ie. trigger monitoring (is it time to do X?)/trigger meaning (ie. what is X?))
- OVERALL not single competence
EXECUTIVE CONTROL DEMANDS
- planning/scheduling/prioritising/coordinating the two task streams
- trouble shooting/problem solving when it goes wrong/unexpected conditions arise
- overall critical
USING PHONE WHILE DRIVING
- epidemiological studies = ^ accidents; relative risk similar to driving at legal alcohol limit
- observational studies = delayed braking at T-junction
- experimental studies = impaired braking/hazard detection especially in young drivers
RELATIVE RISK X USING PHONE WHILE DRIVING
STRAYER, DREWS & CROUCH (2006)
- pp in simulator; follows pacer car in motorway 15m; tries distance maintenance; pacer sometimes brakes
- baseline VS alcohol (80mg/100ml) VS casual hh/hf mobile talk (initiated before; terminated after measure)
- phone = slower reactions/recovery/^ tail-end collisions
- alcohol = ^ aggressive (closer following/harder braking)
- NO STATSIG between hh/hf!
DRIVING SIMULATOR STUDIES +
STRAYER (2015)
HF MOBILE
- < 50%: anticipatory glances to safety critical areas (ie. parked lorry blocking zebra)/later recognition memory of driving environment objects/P300 amplitude to brake light onset in followed car
- ^ unsafe lane change prob
- crash risk data suggests dif effect in direct talking; passengers = sensitive to driver’s load (ie. stop talking/wait for reply); help spot hazards/distractions
LAB DUAL-TASK INTERFERENCE MEASURING
- only 2 tasks designed measurement/manipulation
- typically measures performance on tasks A/B alone/A+B; questions performance deterioration in cs
DUAL-TASK INTERFERENCE SOURCES
- slower/inaccurate performance possibly via:
- specialised domain-specific resource competition (ie. body effectors/sense organs/brain modules/processes/representations)
- general purpose processing capacity competition (ie. central processor/GP processing resource pool)
- limited executive control mechanism capacity (sets up/manages system info flow)/subliminal control strategies
DOMAIN-SPECIFIC RESOURCE COMPETITION
- two continuous speech inputs cannot be understood/repeated simultaneously BUT both can be monitored for target word-meaning
- spatial tracking task interferes w/visual imagery to remember (both use visuo-spatial working memory)
- 2 dif tasks use same perceptual processes/response mechanisms/central translation/coordination processes = dual-task interference UNLESS info rate low enough to switch resource use
- question of concurrent performance by any task pair = some interference?
GENERAL-PURPOSE PROCESSOR COMPETITION
BROADBENT’S P-SYSTEM (1958)
- analogous to standard late 20th century digital computers w/single CPU
- assumed required for pattern recognition/memory access/decision making/action selection/awareness
- others (ie. POSNER (1978) identified consciousness w/central processor
GENERAL-PURPOSE RESOURCE POOL COMPETITON
KAHNEMAN (1975)
- first proposed shared concurrent task effort
- capacity varies on:
- over/within people (alertness)
- sustained/available attention (diminishes w/boredom/fatigue; ^ w/time of day (NOT post-lunch dip)/moderate stressors (noise/heat)/emotional arousal/conscious effort
CENTRAL PROCESSOR/RESOURCE POOL ASSUMPTION
- capacity shared by any 2 tasks
- cap demand sum + NO exceeding total = no interference; exceeding = interference/^ 1 task difficulty = cap reduction for other
- hard to know a priori how much cap task needs
- test theory via pair use in tasks; obvs each would require all/most central capacity
DEMANDING COMBINED TASKS W/O INTERFERENCE
ANTONIS & REYNOLDS (1972)
- Reading uni Y3 music students (competent pianists)
- A = G2(e)/G4(h) sight read; B = Austen(e)/Old Norse(h) novel shadow prose; little practice (10m shadowing for 2 no omission trials; 2m sight reading, 7m dual tasks)
- 2 sessions of dual tasks for easy/hard combos + 1m sight reading/shadowing alone (order balanced)
RESULTS
- shadowing/errors = no dif w/w/o concurrent sight; concurrent shadowing = no ^ sight errors
- ^ h shadow/sight errors (difficulty manipulations work) BUT neither affected by session 2
DEMANDING COMBINED TASKS W/O INTERFERENCE (MORE EXAMPLES)
SHAFFER (1975)
- tasks combined w/o apparent interference
- skilled visual-type copying combined w/prose shadowing = no interference
NORTH (1977)
- one task insensitive to difficulty of other
- continuous tracking/digit = key task of: key press/before present one/successive pair digit identification
- no difficulty effect of task on tracking delays
NO CENTRAL GENERAL-PURPOSE PROCESSOR/RESOURCE
ALLPORT (1980)
- complex input-output translation task pairs can be combined w/little/no inter if using non-overlapping modules (dif input codes/modalities/action systems/central representations/networks)
- SO general-purpose central processor = unnecessary
- BUT even when tasks use dif modules some inter arises via coordination/control demands (ie. specialised executive processes load consequence)
NO CGP X DRIVING HYPOTHESIS
- rebooted hypothesis:
a) driving/navigation (visuo/spatial input -> hand/foot response) + conversation (speech input meaning speech production) = v dif module use? - YES/NO! dif input/output modalities BUT both require mental model construction (ie. driving = route/goals/progress/road-signs/observed events/potential hazard prediction etc.)
- driving mental model construction interfered via convo asking for visuo-spatial arrangements (ie. “when we get to X, do you know how to find Y?”)/imagine movements
IMPORTANCE OF PRACTICE
SPELKE HIRST & NEISSER (1976)
- uncombinable w/o inter tasks = easier combined w/practice (ie. changing gear)
- 85h practice reading stories (comprehension) simultaneously w/dictation writing (6w) then reading concurrent w/writing spoken words (11w)
- some pp = little dual-task inter as:
- practicing 1 task automates it (reduces executive control of constituent processes need)
- practicing combo tasks develops optimal control strategies for combining the specific task pair
BROADBENT’S OBJECTION TO ALLPORT EXPS
BROADBENT (1982)
- continuous task pairs (ie. shadowing/sight reading) =
- some input predictability/anticipation
- substantial input/output lag (temp WM storage)
- SO could still be central processor task switching (time-sharing); while servicing one task, input/output in other stored in WM buffers
- processor-switching revealed in tests is concurrent tasks w/small input/output lags used + next stimuli = unpredictable (ie. RT tasks)
PRP (PSYCHOLOGICAL REFRACTORY PERIOD)
WELFORD (1952)
- 2 choice RT tasks
- stimulus onsets separated via variable/short interval (SOA (stimulus onset asynchrony))
SCHUMACHER et al (2001)
- 1 = audio-vocal (low/m/high tone = 1-3)
- 2 = visuo-manual (000 = index/m/ring finger)
- PRP effect (robust to practice) occurs even when stimuli/responses for 2 tasks = dif modalities
PASHLER’S PRP THEORY
PASHLER (1990)
- response selection = bottleneck; performed only p/task; if stimuli 2 arrives/identified, must wait til response selection mechanism = free
- capacity limitation must be in central translation processes (perceptual processes -> central trans -> response execution)
PRP CRITICISMS
- RT task pairs w/no PRP interference exist (ie. GREENWALD & SHULMAN (1973); repeat spoken letter word + move lever w/arrow); all w/natural/practiced input/output mapping
- observed PRP effect = not via structural bottleneck but cautious control strategy pps adopt to avoid responding to stimuli 2 (ie. MEYER & KIERAS (1997))
- if pps trained w/liberal strategy = PRP disappears
- dual task cost in PRP paradigm via bottleneck (response selection mechanism) VS soft capacity limits w/control strategies = still questioned
- response-selection mechanism = NOT GPP
SUMMARY
- much inter between simultaneous tasks = via specialised resources competition
- general purpose processing capacity for high cognition (consciousness?) = popular BUT few evidence
- dual task inter cases interpreted as via GPP limited capacity competition/generic response-selection mechanism = better explained as:
- domain-specific resource competition
- specialised executive control mechanism capacity
- inefficient control strategy use