Download
c hapter 9 selection of action n.
Skip this Video
Loading SlideShow in 5 Seconds..
C hapter 9. Selection of Action PowerPoint Presentation
Download Presentation
C hapter 9. Selection of Action

C hapter 9. Selection of Action

113 Vues Download Presentation
Télécharger la présentation

C hapter 9. Selection of Action

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. Chapter 9. Selection of Action • OVERVIEW • skill-based behavior – the most automated level; a rapid automatic responses with a minimum investment of resources; extensive training and experience • rule-based behavior – action by bringing into WM a hierarchy of rule – less automatic and timely • knowledge-based – entirely new problems; neither rule nor automatic mappings exist – diagnoses, decisions, troubleshooting • selection of skill-based actions – response time or reaction time (RT) – simple and choice RT • VARIABLES INFLUENCING BOTH SIMPLE AND CHOICE REACTION TIME • stimulus Modality • simple RT to auditory stimulus (130 msec) is faster than visual stimuli (170 msec) • Stimulus Intensity • simple RT decreased with increases in intensity of the stimulus to an asymptotic value (Fig9.1) • aggregation over time of evidence in the sensory channel until a criterion is exceeded • temporal Uncertainty • the degree of predictability of when the stimulus will occur • manipulated by warning interval (WI) between a warning signal and imperative stimulus to which the person must response • short and constant WI  imperative stimulus is highly predictable (temporal uncertainty is low)  short RT • long or variable WI  high temporal uncertainty  long RT

  2. Expectancy • RT increases as the average WI of a block of trials becomes longer (temporal uncertainty) • an opposite effect is observed within a block having randomly varied long and short WIs – the concept of expectancy • VARIABLES INFLUENCING ONLY CHOICE REACTION TIME • The Information Theory Model: The Hick-Hyman Law • choice RT was longer than simple RT  RT was a negatively accelerating function of the number of stimulus-response alternatives • Hick (1952) and Hyman (1953) applied information theory to quantify the uncertainty of stimulus events (Fig 9.2) • choice RT increased linearly with stimulus information (log2N) • RT = a + bHs (Hick-Hyman law) • human has a relatively constant rate of processing info, defined by inverse slope (1/b) (bits/sec) • The Speed-Accuracy Trade-off • they tend to make more errors as they try to respond more rapidly • bandwidth as Ht/RT (bits/sec) constant bandwidth model of human performance is not quite accurate • Howell and Kreidler (1963, 1964) – easy and complex choice RT tasks by different instructions – fast, accurate, fast and accurate, maximize Ht • instructions changed RT and error rate; speed instruction having the largest effect • easy choice RT – max. Ht obtained by maximize Ht instruction • complex choice RT – highest level of performance efficiency with speed set instruction

  3. The Speed-Accuracy Operating Characteristic (SAOC) • RT is on the x-axis and accuracy (error rate) on the y axis (Fig 9.3) • info. transmission (performance efficiency) is optimal at intermediate speed-accuracy sets • accuracy (=log[P(correct)/P(errors)])  linear SAOC (Fig 9.4) • one important aspect of the speed-accuracy trade-off is its usefulness in deciding what is “best” • The Speed-Accuracy Micro-Trade-off • compare the accuracy of fast and slow responses within a block of trials, using the same system (or experimental condition) – depend on the particular nature of the RT task • when the criterion is conservative  processing full info, taking longer time  high accuracy • when the criterion is risky  response initiated rapidly, based on little evidence  errors will be likely • in the extreme  “fast guess” – a random response initiated as soon as the stimulus is detected  error RTs are faster than correct RTs when RTs are short and stimulus quality is good • if poor stimulus, long processing or working memory load – opposite form of micro-trade-off  error responses tend to be slower than correct ones • DEPARTURE FROM INFORMATION THEORY • Stimulus Discriminability • RT is lengthened as a set of stimuli are made less discriminable from one another • similarity or difference – the ratio of shared features to total features within a stimulus • discriminability difficulty reduced by deleting shared and redundant features where possible

  4. The Repetition Effect • repetition effect, the advantage of repetitions over alternations, is enhanced by increasing N (the number of S-R alternatives), by decreasing S-R compatibility, and by shortening the interval between each response and the subsequent stimulus • No repetition effect • long intervals between stimuli and may be replaced by an alternation effect (faster RT with a stimulus change) – gambler’s fallacy  do not expect a continuous run of the same sort • rapid repetition of the same finger slower than alternations • Response Factors • RT is lengthened as the confusability between responses is increased • RT is lengthened by the complexityof the response • Practice • practice decreases the slope of the Hick-Hyman law function relating RT to info • compatibility and practice appear to trade off reciprocally in their effect on this slope • Executive Control • it takes time to load or activate these rules when they are first used or shift from one to another  the function of central executive control • Stimulus-Response Compatibility • Location Compatibility • provided by human’s intrinsic tendency to move or orient toward the source of stimulation

  5. colocation principle – controls next to the relevant display – not always possible to achieve (Fig 9.5) • congruence – congruence between spatial controls and displays (Fig 9.6); often defined in terms of an ordered array rule • increase from left to right, aft to forward, clockwise, bottom to top • far-right to top when left-right array mapped to a vertical display • top-down ordering is not strong  vertical display (or control) arrays that are not congruent with control (display) arrays should be used with caution • put a slight cant, or angling, of one array in a direction that is congruent with the other(Fig 9.7) • Movement Compatibility • the set of expectancies that an operator has about how the display will respond to the control activity: Cognitive-Response-Stimulus (C-R-S) compatibility • congruence principle of location compatibility applied to the compatibility of movement • when congruence violated, a common mapping of increase; also governed by a principle of movement proximity in Fig 9.8 (Warrick principle) -- not related to congruence • Compatibility Ambiguities • mental model • movement proximity principle was far less pronounced for psychology students than ME students, ME having the strong mental model of the mechanical linkage • a design for the vertical speed of an aircraft (Fig 9.9) • frame or reference -- exocentric viewpoint (compatible S-R movement), egocentric viewpoint • distinction between status and command displays

  6. Transformations and Population Stereotypes • any S-R mapping that requires some transformation will be reduced in its compatibility • population stereotypes define mappings that are more directly related to experience • Consistency and Training • be wary of possible violation of consistency to optimize the compatibility of each • training can also be used to formulate correct mental model and enhance the agreement between the mental model and the correct dynamics • Knowledge in the World • should provide an invitation to the appropriate actions (affordance) or forcing function, as well as a “lockout” of the inappropriate actions (Fig 9.5 and Fig 9.10) • STAGES IN REACTION TIME • The Subtractive Method • delete a mental operation entirely form the RT task – the decrease in RT is assumed to reflect the time required to perform the absent operation • Additive Factors Technique • confirming evidence for the existence and identity of processing stages (Fig 9.11) • to define the existence and distinctiveness of different stages by manipulating variables that are known to lengthen reaction time • interactive (influence a common stage of processing); additive (influence different stages) • Experimental Techniques • make inferences about what manipulations influence what stages of processing • patterns of additivity and interactions (Fig 9.12)

  7. the response selection is a major bottleneck in speeded information processing • stimulus probability appears to affect two stages • improbable stimuli require longer to be recognized • their associated responses take longer to be selected • Applications of Additive Factors Methodology • how information processing speed is influenced by aging, poisoning, mental workload • Problems With Additive Factors • the assumption that stages proceed strictly in series  convincing evidence that information processing does not strictly proceed in a serial fashion • underadditive relationship – delay by increasing the difficulty at one stage of processing is actually smaller at the more difficult level of the other stage • The Event-Related Brain Potential as an Index of Mental Chronometry • event-related brain potential (ERP) • a direct estimate of the timing of processes up to the intermediate stage of stimulus categorization – a series of electric voltage from the surface of the scalp • The Value of Stages • the separation of processing stages should not taken too literally • some overlap in time between processing in successive stages – parallel processing • the stage concept more than compensates for any limitations in its complete accuracy • SERIAL RESPONSES • The Psychological Refractory Period • PRP – a situation in which two RT tasks are presented close together in time

  8. ISI (interstimulus interval) – the separation in time between the two stimuli – SOA • the second stimulus response is delayed by the processing of the first under short ISI • human being as a single-channel processor of information • the processing of S1 temporarily captures the single-channel bottleneck  S2 must wait until S1 is finished  anything that prolongs the processing of S1 will increase the PRP delay of RT2 in Fig 9.14 (simple reaction vs. choice reaction) • perceptual analysis of S2 can proceed even as the processor is fully occupied • the delay in RT2 will increase linearly with a decrease in ISI and with an increase in the complexity of RT1 (Fig. 9.15) • general single-channel model • with short ISI (< 100 ms), both responses are emitted together (grouping) and both are delayed • sometimes RT2 suffers a PRP delay even when the ISI is greater than RT1  feedback • The Decision Complexity Advantage • the most restricting limit in human performance relates to the absolute # of decisions/sec rather than the # of bits/sec (cf. bandwidth) – the frequency of decisions and their complexity do not trade off reciprocally  decision complexity advantage • some fundamental limit to the central-processing or decision-making rate, independent of decision complexity, that limits the speed of other stages of processing – 2.5 decisions/sec for decisions of the simplest possible kind • Pacing • the circumstances under which the operator proceeds from one stimulus to the next

  9. dichotomous dimension • force-paced schedule – constant interval, ISI, independent of the operator’s response • self-paced schedule – response-stimulus interval (RSI), depend on the latency of the operator’s response • continuous dimension – defines the value of the timing parameters (RSI, ISI) • Response Factors • Response Complexity • increased complexity requires more monitoring of the response -- sometimes delay • Response Feedback • two effects on performance, depending on the sensory modality • delays, distortions, or elimination of the intrinsic feedback (the perceived sound of one’s voice or the visualization of one’s moving hand) – substantial deficits in performance • less serious distortion of extrinsic feedback (the click of a depressed key or the appearance of a visual letter on a screen – delay or distortion of feedback can be harmful • Response Repetition • response is slowed by its repetition • single trial RT paradigm (repetition is good) • eliminates the repeated engagement of the response selection stage – time saved • reaction time relatively long (200 – 300 msec) compared to typing • typing -- response is slowed by its repetition • high speed in typing  separate responses selected without engaging a higher-level decision process  no longer any benefit to repetitions by bypassing  advantageous to employ separate muscle groups for successive responses (advantage for alternation)

  10. Preview and Transcription • the class of transcription tasks (e.g., typing, reading aloud, and musical sight reading) allow the operator to make use of preview, lag, and parallel processing • more than one stimulus displayed at a time (preview is available)  lag the response behind perception  perception and response are occurring in parallel  preview (seeing into the future) or lag (responding behind the present) • possible in self-paced (typing) or force paced (oral translation) tasks • maintain a running “buffer” memory of encoded stimuli that have not yet been executed as responses • benefits of lag • allowance for variability • allowance for chunking • Allowance for Variability • a steady stream of output at a constant rate can proceed even if input encoding is temporarily slowed • Allowance for Chunking • evidence in typing that inputs are encoded in chunks  letters in each chunk processed more or less in parallel  the output must be serial • encoding, buffer storage, response – proceed in parallel with little mutual interference, and are even time shared with a fourth mental activity, the monitoring of errors in response

  11. Use of Preview • preview helps performance • benefits of preview • making available more advance information • giving the operator an opportunity to perceive chunks  not related to the semantic level of processing but chunk-sized units to be processed in parallel • the benefits of chunking are primarily perceptual and may be seen in storage but not in response