C. Douglas Creelman

PEST: Efficient Estimates on Probability Functions

An adaptive procedure for rapid and efficient psychophysical testing is described. PEST (Parameter Estimation by Sequential Testing) was designed with maximally efficient trial‐by‐trial sequential decisions at each stimulus level, in a sequence which tends to converge on a selected target level. An appendix introduces an approach to measuring test efficiency as applied to psychophysical testing problems.

Human discrimination of auditory duration

A series of experiments measured human ability to discriminate between durations of auditory signals presented in a noise background. Independent variables were the signal voltage, the “base” duration T, and the increment duration ΔT. Separate experiments assessed the effect of each of these on discrimination. A decision‐theoretical model is presented, based on a “counting mechanism,” which operates on impulses generated over the relevant durations. The source of these impulses is assumed to be random. Limitations on performance come from uncertainty regarding the end points of the time interval and from limited memory. The decision processes underlying the model are presented as a general theory of duration discrimination.

Detection of Signals of Uncertain Frequency

Theoretical models for frequency sensitivity in human observers are discussed. One decision procedure for a multiple‐filter model is considered in some detail as a general model for decision situations in which each available response is tied to more than one of the possible signal alternatives. Two experiments were conducted in an attempt to choose between a sweeping‐filter model and a multiple‐filter model. Detection in a two‐alternative forced‐choice experiment in which the signal could be one of two possible signals was measured over a range of frequency separations. The data yield further support for a multiple‐filter model.

Detection, Discrimination, and the Loudness of Short Tones

Two experiments were conducted, with the same observers in each. A method of adjustment was first used to measure the relative loudness of short tonal signals at four different durations and four intensity levels. The second experiment measured the detectability of each duration which had been used in the first, and scales were constructed on the basis of the measurements. It was shown that the relative loudness of a short tone (the level to which it will be set to sound equally loud) can be predicted from the detectability scale. The relationship is an extremely simple one: the equal‐loudness settings yield signals of equal detectability.

Some Pitfalls in Adaptive Testing: Comments on “Temporal Integration and Periodicity Pitch” [R. A. Campbell and S. A. Counter, J. Acoust. Soc. Amer. 45, 691–693 (1969)[

Campbell and Counter (1969) measured “thresholds” for various short durations of low‐frequency sinusoidal signals. The adaptive psychophysical procedure they used is shown to yield uninterpretable results; values they report can represent a wide range of performance levels. In this note, we take Campbells and Counters study as an example of the methodological traps in seemingly straightforward testing methods.

Detection of Complex Signals as a Function of Signal Bandwidth and Duration

A study is reported which examined the efficiency of human observers in the detection of a stimulus waveform which is a train of damped sinusoids. The signal duration and degree of damping (or spectral bandwidth) were varied, with the energy of the signal held constant. Increased bandwidth is shown to decrease detection of relatively long duration signals. A negligible effect of bandwidth was found for short durations. Detection of an amplitude difference, in contrast to detection of the presence of weak signal in noise, is shown to be relatively unaffected either by signal duration or bandwidth over the range of values studied.

Loudness and Detectability Scaling of Bandpass‐Filtered Short Tones

Two observers adjusted gated sinusoids, passed through a bandpass filter and presented in a noise background, to equal the loudness of unfiltered signals. Bandpass‐filtering decreased loudness relative to unfiltered signals when duration was less than about 100 msec. Detectability scales were constructed for 5‐msec filtered and unfiltered signals presented in continuous noise, and the scales showed superior performance at the signal amplitudes used in the matching portion of the experiment for the unfiltered signals. The scale discrepancy was about 3 dB, whereas the loudness difference was about 5 dB. Some theoretical implications of the data are discussed.

Simultaneous Adaptive Threshold Estimation with Frequency Uncertainty

An adaptive threshold tracking procedure (PEST) was used to adjust the amplitude of each of an ensemble of possible sinusoidal signals. The data were thresholds for each alternative in the context of the ensemble. This procedure avoids some of the problems in strategy for the observer and other problems in theoretical interpretation for the experimenter. Separate conditions involved variation in the number of alternative signals in the ensemble (3–33) and the spacing between lowest and highest member of the ensemble (50 Hz–5000 Hz). Within limits, prior findings are confirmed; loss in sensitivity due to uncertainty reaches a maximum of 3–5 dB with all but very narrow ranges of frequency, compared to detection of a fixed, known frequency.

Central Periodicity Pitch

Perceptual phenomena are reported that are observed when interrupted noise is presented to one ear and either correlated or uncorrelated continuous noise is presented contralaterally. Also described are effects observed when the contralateral stimulus is also interrupted noise. These and related results support the conclusion that there exists a central mechanism for the perception of periodicity pitch and require that existing models of central neutral processes of pitch perception and binaural fusion be modified.

Loudness Matching, Detection, and Discrimination of Band‐Limited Short Tones

With a method of adjustment procedure, observers equated hated 1000‐cps sinewave signals, one of which was passed, after hating, through a narrow‐band filter. Filtered signals were found to be less loud than unfiltered signals of the same duration. This effect depends on the duration of the signals; at 2 msec, filtered signals are roughly 6 dB below unfiltered signals in loudness. The loudness discrepancy diappears at durations greater than 50 msec. Signal‐detection and amplitude‐discrimination measures were made separately for filtered and unfiltered signals of 5‐msec duration. Amplitude discrimination seems to be superior for unfiltered signals as compared to filtered signals, although the discrepancy is small and hardly sufficient to account for the loudness differences obtained.