Don Trumbo

The organization of skilled response

Abstract A series of experiments concerned with the ways in which responses become organized is discussed. The principle parameter in most studies was stimulus coherence, and tracking tasks were used as a vehicle because graded responses permit detailed and fine-grained analyses. Both spatial and temporal coherence are used in response organization, and type of response strategy varys with degree of stimulus coherence. The effects of secondary tasks, sequence length and task coding were also examined.

INFLUENCE OF KNOWLEDGE OF RESULTS ON PERFORMANCE IN A MONITORING TASK

Vigilance performance was tested in six independent groups with a modified version of the Mackworth Clock Test. The groups constituted a 3 × 2 design with (a) no knowledge of results, observer-presented knowledge of results, and machine-presented knowledge of results as one dimension, and (b) physical presence or absence of an observer in the test cubicle as the other variable. All groups were retested without knowledge of results one and seven days after the initial test. On the initial test day, groups with no knowledge of results and machine-presented knowledge of results showed the typical decrement function throughout the session. In contrast, Ss receiving verbal knowledge of results from the experimenter showed less decrement and a significantly higher over-all performance. The superiority of the groups receiving the verbal report persisted on the two subsequent test days despite the fact that all extrinsic knowledge of results was withdrawn on these tests. Observer-presented knowledge facilitated performance regardless of the physical presence or absence of the observer in the test cubicle.

A Versatile Electronic Tracking Apparatus (VETA)

An electronic tracking and scoring system which gives access to relatively complete response data is described. Input, output, error and acceleration data are recorded on magnetic tape and may either be read out for visual inspection or read directly into analog computers and correlators for further analysis. An important feature of the system is that it readily permits the programming of various complex wave form inputs of either regular or irregular patterns in sequences of any length. Type or pattern of input may be varied between trials. The system is highly automated and easily calibrated.

Number of Alternatives and Sequence Length in Acquisition of a Step-Function Tracking Task

The roles of two task parameters, sequence length (N) and number of alternatives in the population from which the sequence was drawn (K), were examined in a tracking task. The tasks were irregular step-function inputs wherein N was defined as the number of targets (steps) in a repeating sequence and K was defined as the number of alternative target positions. N and K were varied independently in a 3 × 3 factorial design with 9 Ss per cell. The major findings were that tracking performance, as measured by integrated error scores, is affected by increases in N, but not proportionately, while neither K nor the N × K interaction was significant. Relations of the results to verbal learning data are discussed.

Rate and predictability in rate-tracking tasks.

Abstract The constant rate of target displacement and the coherence of the track were varied in two factorial experiments. Both experiments involved pursuit tracking; the second also included compensatory tracking conditions. The results of both experiments showed significant increases in the criterion integrated absolute error scores with increases in rate and with decreases in coherence. The superiority of the pursuit mode increased with rate, but the mode × coherence interaction was not significant, indicating that both modes benefited about equally from input coherence. No significant losses in performance occurred after a one month retention interval. Results are compared with prior data from step-function tracking.

Analog computer methods for scoring continuous performance records

Methods for scoring continuous records of tracking performance with analog signal correlator and frequency analysis systems are described. An index of lead-lag obtained from the correlator output is compared with discrete lead-lag scores obtained by hand-scoring oscillographic records for three tasks which differ in amount of task coherence. The results indicate relatively high agreement between the two scoring methods and support the use of the correlator as the more efficient method. Sample data from the analog frequency analysis system are compared for operators with high and low integrated error scores. The results indicate consistently greater power in the response than in the input at the fundamental for both good and poor Ss, indicating a tendency to overshoot the target with the primary movement, but relatively more power in the initial odd harmonics for the better Ss. Ratios of target spectra to response spectra provide transfer functions for the human operator.

On Response-Response Compatibility

This study was designed as a first step in the investigation of response-response compatibility effects, a compatibility phenomenon hypothesized by Fitts and his associates. A two-hand cranking task was used in which S could, by manipulating two cranks, adjust a mechanical stylus to any one of one hundred points on a ten-by-ten matrix. The four possible combinations of clockwise and counterclockwise movements of the two cranks made up four response conditions. Three different stimulus types (codes) were used: digital, pictorial, and instructional. The stimuli, projected onto the rear side of a translucent screen located directly in front of S, indicated the point on the matrix to which S was to move the stylus. The target series used for each of the three stimulus types required the use of the four response conditions with equal frequency. It was found that speed of both total adjustment and initial response was dependent upon the particular stimulus type used, but not dependent upon response conditions. In contrast, number of correct initial movements was dependent upon response conditions but not stimulus types. The latter was, however, a function of response sets rather than compatibility. No evidence of significant interaction was found. The present findings fail to support the hypotheses of Fitts and his associates and time and motion principles which also, in effect, contend that R-R compatibility effects exist. Furthermore, the existence of response sets with no corresponding effect on the over-all efficiency of the system suggests that sets and compatibility are two different and not necessarily interacting phenomena.