No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
Die Sinnesorgane als Informationsempfänger
Abstract
Hatten wir bisher den Weg der informationstragenden Signale ausschließlich im Bereich der physikalischen Übertragungsmedien verfolgt, so wollen wir nun das Schicksal der Signale beim empfangsseitigen Kommunikationspartner, dem Perzipienten, betrachten, d. h. im psychophysiologischen Bereich.
Ss can discriminate phonemes presented singly and in random order. Ss discriminated better between speech sounds to which they have attached different phonemic labels than between sounds which they normally put in the same phoneme class. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Um gleich am Anfang der Behandlung der Strukturfrage der Sinnesmannigfaltigkeiten die Kantische Lehre vom Verstand hervorzuheben, ist das in späterer Zeit mit dem Namen der Wahrnehmung bezeichnete in der obigen Überschrift mit der Bezeichnung der Anschauung benannt. Es ist nämlich meines Erachtens unmöglich, eine Strukturlehre aufzubauen ohne der von Kant gegebenen Grundlage, d. h. ohne Berücksichtigung der Struktur des Verstandes. Die Struktur der Wahrnehmung, der Anschauung und diejenige der Begrifflichkeit bilden beide zusammen die Struktur des Verstandes, welcher Anschauung und Begriff ist. In gewöhnlicher Wortsprache kann die Struktur des Verstandes in Kürze folgendermaßen wiedergegeben werden (s. hierzu Reenpää 1952). Anschauung ist Anschauung a priori, Zeit- und Raumanschauung, und Anschauung a posteriori, Anschauungsqualität und -intensität. Außer diesen einstelligen Anschauungsobjekten haben wir die zwei- und mehrstelligen Anschauungsobjekte, von denen hier dasjenige der Anschauungsgleichheit, insbesondere die Gleichzeitigkeit zu berücksichtigen ist. Alle Anschauung ist im „Jetzt”, aktuell; was nicht aktual zeitlich ist, gehört dem Begrifflichen zu.
Die Informationstheorie hat die Aufgabe, Aussagen über die Leistungsfähigkeit einer gegebenen Übertragungsanlage zu machen und ihre Ausnutzung durch die tatsächlich anfallende Menge von Nachrichten zu beurteilen, d. h. die Leistungsfähigkeit einer Anlage zu vergleichen mit der Leistung der Nachrichtenquelle, deren Signale sie übertragen soll [40, 31].
Difference‐limens for the discrimination of intermittence were obtained at ten interruption rates of random noise between 1 and 320 cps. The duty cycle of the intermittent noise was 0.5 and the sound pressure level was 75 db. Data from four listeners give average deviations from the standard frequency (ΔF) of 0.02 to 9.74 cps. The graph relating ΔF and F from 1 to 320 cps shows ΔF to be a monotonically increasing function of F. When the data are plotted with logarithmic coordinates, two functions are revealed, suggesting a change in the method of frequency discrimination at about 5 cps. This change appears due to the listener's ability to count the noise bursts from 1 to 5 cps, but not above about 5 cps. The relationship between 5 and 320 cps may be described by a parabolic function of the form y = ax b . Below 5 cps, the form of the relationship cannot be determined from the data available. The relative difference‐limen, ΔF/F, lies between 0.008 and 0.03, and graphic integration of 1/ΔF yields 560 just noticeable differences in the range studied. These data show difference‐limens for flutter that are smaller than those previously reported.
Measurements of frequency discrimination for single damped waves are reported for frequencies between 200 and 5000 cps, and for values of the damping σ between 0 and 500 sec−1 (σ is the reciprocal of the time constant of the envelope of the damped wave). There is an increase in the difference limen for frequency as the damping increases, or as the effective duration decreases. The increase in the difference limed is accompanied by a loss in the pitch character of the stimulus, and there is usually a sharp deterioration in discrimination above a certain value of damping, i.e., above a certain band width of the stimulus. The results are related to existing data on the perception of short sinusoidal stimuli.
The evaluation of quality of color reproduction poses many complex problems. Optimum reproduction needs to be identified. Since it depends upon the limitations of the reproduction process, as well as upon human vision and judgment, optimum reproduction will probably have to be determined for each process separately. The program is to vary the production controls in systematic manners, measure the resulting color reproduction in the best way known (e.g., the I.C.I. method at the present time), submit the reproductions to visual judgment, and study the judgment data in comparison with the measurements in order to find significant correlations. The growing experience of such studies of color photography is suggested as a guide. Preliminary estimates of optimum reproduction and of seriousness of deviations may be based tentatively on results of studies of noticeability of color differences and on fragmentary results of studies of color photography. These estimates can be improved as various parts of the program are carried out.
The electromagnetic signals used in communication are subject to the general laws of radiation. One obtains a complete representation of a signal by dividing the time-frequency plane into cells of unit area and associating with every cell a “ladder” of distinguishable steps in signal intensity. The steps are determined by Einstein's law of energy fluctuation, involving both waves and photons.
The rectification of random narrow-band noise in the presence of a C.W. or modulated signal is analysed. The detector considered is of the type providing a rectified voltage which is a function of the instantaneous amplitude of the input wave. The smoothing present in the detector circuit is such that the high-frequency components of the applied wave are removed but the low-frequency variations of its envelope are faithfully transmitted.The mean (or D.C.) component of the rectified voltage and its r.m.s. fluctuation about that mean (or L.F. noise output) are calculated as a function of the input signal/noise ratio and particular attention is paid to linear and square-law detectors.The output signal/noise ratio for an amplitude-modulated signal input and for the audible beat reception of a C.W. signal are calculated.The spectrum of the L.F. noise output from a detector supplied with random noise is shown to be closely similar for linear and square-law detectors.The discrimination of a weak signal in noise is shown to be not critically dependent upon the law oi the detector, and in particular the difference for a linear and a square-law detector is negligible.The effect of receiver bandwidth, meter time-constant and integration time in improving the discernment of a weak signal is considered.
These experiments are designed to test the following hypothesis. The rate of the temporal integration of energy in the ear (at threshold) is dependent on the width of the frequency band of the energy to be integrated. Duration is exactly equivalent to intensity only when all the energy to be integrated is in a narrow band of frequencies. The hypothesis tested by taking advantage of the spectral distribution of energy in short tones. As a tone becomes very short, the effective band width of the energy increases. The band width of energy is essentially defined by the reciprocal of the duration of the tone. Thus as the duration of a tone decreases, not only does the total energy in that tone decrease, but the band width of energy also increases. The intensity threshold, then, has to be increased (as duration is decreased) to compensate for both effects if the hypothesis is correct. The results are in line with the predictions of the hypothesis. The width of the band necessary for maximum integration is also related to frequency and the width of critical bands.
Fletcher has proposed the use of a logarithmic frequency scale such that the frequency level equals the number of octaves, tones, or semitones that a given frequency lies above a reference frequency of 16.35 cycles/sec., a frequency which is in the neighborhood of that producing the lowest pitch audible to the average ear. The merits of such a scale are here briefly discussed, and arguments are presented in favor of this choice of reference frequency. Using frequency level as a count of octaves or semitones from the reference C0, a rational system of subscript notation follows logically for the designation of musical tones without the aid of staff notation. In addition to certain conveniences such as uniformity of characters and simplicity of subscripts (the eight C's of the piano, for example, are represented by C1 to C8) this method shows by a glance at the subscript the frequency level of a given tone counted in octaves from the reference C0 = 16.352 cycles/sec. From middle C4, frequency 261.63 cycles/sec., the interval is four octaves to the reference frequency, so that below C4 there are roughly four octaves of audible sound. Various subdivisions of the octave are considered in light of their ease of calculation and significance, and the semitone, including its hundredth part, the cent, is shown to be suitable. Consequently, for general use in which a unit smaller than the octave is necessary it is recommended that frequency level counted in semitones from the reference frequency be employed.
Noise interrupted at a steady rate has essentially the same spectrum over the range of frequencies transduced by the earphone as does continuous noise. The frequency corresponding to the rate of interruption is not intensified in the spectrum. Consequently, the ability of listeners to respond differentially to the rate of interruption cannot be explained on the basis of a simple resonance theory of hearing. The point at which an interrupted noise becomes indistinguishable from a continuous noise depends upon the rate of interruption, the sound‐time fraction, and the intensity of the noise. For a sound‐time fraction of 0.5, the presence of interruptions can be detected at rates well above 1000 per second. Differential sensitivity to changes in the rate of interruption (with a sound‐time fraction of 0.5) is poor above 250 interruptions per second. Also at these high rates the listener loses his ability to match the frequency of a pure tone to the rate of interruption. Presumably the ability to perceive interruptions in a random noise depends upon the synchronous firing of the fibers in the auditory nerve. This hypothesis is supported by the correspondence between auditory sensitivity to changes in the rate of interruption of a noise and the tactual sensitivity to changes in the frequency of a vibrating pressure applied to the skin.
Sommaire L’audiométrie objective telle que nous la pratiquons relève des relations qui lient le circuit auditionphonation. En effet,
toute perturbation sur ce circuit est immédiatement détectable par les troubles qu’elle entraîne soit dans le rythme, soit
dans le timbre.
From the audible amplitude and frequency modulation levels, both of pure tones and of sections of white noise, the elements of the sound changes in the hearing surfaces are combined.By means of a model which has as basis the perception of a change in sound intensity of 1 dB of a frequency group with a mean time constant of 20 ms., the audibility of the amplitude modulation, the frequency modulation and that of phase differences, as well as the masking is clarified.The maximum perceptible information is estimated and its application to syllable-intelligibility illustrated.
Results of psychoacoustic experiments are used to estimate the greatest precision necessary in quantizing narrow band‐width data specifying the vowel sounds of speech.
An exploratory experiment to determine just discriminable differences in the intensity of a synthetic vowel sound is described. Results of AB listening tests performed with intra‐sample time spacings of 0.5, 1.0, and 1.5 seconds are given. The median value of the DL for vowel intensity obtained from the tests is 1.5 db, and the DL is never less than 1.2 db. Tentative data on the time error for a vowel sound are noted and compared to similar data for pure tones.
Measurements were made on a sample of vowel utterances, by male talkers, of the band widths of the first three formants. It was found that the band width was essentially constant and independent of the particular vowel. The mean values for bars 1, 2, and 3 were 130, 100, and 185 cps. respectively. Ten percent of the 300 band widths measured were less than 90 cps and ten percent greater than 260 cps.
Calculations of the informational capacity of the human ear are made by computing the number of discriminable soundpatterns per second, and applying the Shannon information theory. A maximum of 104 bits/sec transmission found. This is compared with the capacity of existing auditory channels and recording media, and with the rate of actual informationperception from speech and music. It is shown that a capacity of upwards from 5 × 104 bits/sec, depending on the informational match to the ear, is necessary for high fidelity transmission or recording. It is also shown that the brain can utilize less than 1 percent of the information transmitted by the ear. Finally, an average capacity of about 0.3 bit/sec, or of 40 tones/sec, is calculated for an individual cochlear fiber.
An uncertainty principle in acoustics, arising wholly from classical views, is presented. This principle is that Δν.Δt∼1, where ν is the intrinsic frequency of an acoustic signal and Δt is its time duration. Applying this principle one finds that it is consistent with experiments on the change in frequency in the vibrato and the failure to detect it by ear, with recorded tests on minimum perceptible differences in frequency, and with the minimal time for tone perception. The problems suggested by the principle are: (1) variations in Δt and Δν by an artificial vibrato with aural observations of detectable Δν, (2) redetermination of minimum perceptible differences in frequency as dependent upon Δt and (3) an examination of Δt required for tone perception with varied values of Δν
Summary Tones between 15 and 200 kcycles are audible when applied by bone-conduction. The resulting sensation is to be observed not over the whole surface of the head and neck, but only over many points, of which early 60 have been investigated in cases of normally hearing persons of both sexes and middle age. The sensation is that of a very high tone and does not change according to frequency. The highest limit of hearing may lie near 200 kcycles. The reasons are discussed why hearing in this “ultrasonic” range seems to be a real one, which can be understood by observations of
Stimulus variables influencing the fusion of interrupted white noise have been re-examined in a single trained observer using repetition rate as the independent variable. Critical auditory decay times have been calculated for all observed fusion thresholds under the assumption that the cue for the perception of minimal flutter is a decay of one detectable step in the loudness of the interrupted signal during the off interval. These critical decay times are satisfactorily described by the value 120 milliseconds (independent of intensity) when the burst duration in the interrupted signal is longer than a critical value (5 milliseconds for this listener), but for shorter burst durations the decay times are much lower and exhibit a relationship to burst duration and intensity. A shift in the population of neural units activated under the latter conditions is postulated as an explanation.
The range of individual differences in flutter fusion thresholds under standard conditions has been examined by the use of a tape recorded test requiring repeated discriminations of interrupted from continuous noise. From a sample of 46 normal individuals a population of threshold scores may be postulated having a mean critical fusion frequency of 82.1 interruptions per second with a standard deviation of 19.4 when the interrupted noise has a sound-time fraction of 0.90 and an intensity of 70 db re 0.0002 dyne/cm2.
Recent data on pitch discrimination are shown to agree reasonably well with the values that would be predicted on the assumption that jnd's for pitch are subjectively equal when measured in mels. The facts relating to diplacusis and the estimation of musical intervals suggest that the difference limen may be expected to vary irregularly with frequency. A further argument is made that Kock's theoretical explanation of pitch discrimination is invalid.
Discrimination thresholds for the detection of a change in the sound level of a tone were obtained under five experimental procedures. These procedures differed primarily in terms of the presence or absence of an objective comparison signal and in terms of the stability of the test conditions under examination. Under comparable conditions, discrimination in the absence of an objective comparison signal is only slightly less acute than in the presence of such a signal. On the other hand, relatively large increments in the detection thresholds are associated with increases in the instability of the testing conditions, especially over long discrimination intervals. The results are examined in terms of molar concepts more pertinent to the listener than to the ear.
Synthetic vowel sounds, formed by electrical excitation of resonant circuits by a regular series of impulses, have been used as stimuli in a group of psychophysical experiments. The experiments include measurements of just noticeable differences (JNDs) in formant frequency and in band width for the synthetic sounds with one and two resonances, and with resonance band widths comparable to those found in measuredvowel spectra. Average JNDs in formant frequency and band width at 1000 cps are 17 cps and 25 cps, respectively. The results are used to establish a frequency scale for the representation of vowelformants. Equal distances on this scale correspond to an equal number of JNDs in formant frequency. The proposed scale is approximated more closely by a linear frequency scale than by a mel scale. The approximate number of JNDs between pairs of vowels (arranged on a scale from front to back vowels) is evaluated, and is in the range 15–20 on the average. In general, the data indicate that frequency analysis of vowels using a filter band width as narrow as 45 cps provides more information than the perceptual processes can resolve.
Thresholds for the detection of a decrement in noise intensity between repeated bursts of noise were determined as a function of the duration of the interval between successive bursts. The results indicate a critical duration (55 milliseconds) between successive noise bursts: (1) above which, the differential threshold is constant and independent of the interval between successive bursts and (2) below which, the differential threshold increases proportionately as the interval between successive bursts decreases. Since an equivalent critical interval has previously been obtained by several different independent measures of auditory persistence, the observed deterioration of differential sensitivity is interpreted in terms of the overlap or addition of auditory persistence with the direct effects of stimulation.
The usefulness of logarithms in the measurement of many of the stimuli to which human beings are sensitive is almost too obvious to need argument. Three reasons are commonly given to justify the practice:
1. The intensity ranges of the physical stimuli are enormous—energy ranges of trillions to one are involved in vision and hearing.
2. To a rough approximation, discrimination follows a law of relativity: the just detectable increment in a stimulus is proportional to the magnitude of the stimulus (Weber's law). Hence, to the extent that Weber's law holds, the logarithmic difference that is just detectable is constant.
3. According to Fechner's law, the subjective magnitude of a sensation is supposed to be proportional to the logarithm of the magnitude of the stimulus.
The ratio of the minimum perceptible increment in sound intensity to the total intensity, DeltaEE, which is called the differential sensitivity of the ear, was measured as a function of frequency and intensity. Measurements were made over practically the entire range of frequencies and intensities for which the ear is capable of sensation. The method used was that of beating tones, this method giving the simplest transition from one intensity to another. The source of sound was a special moving coil telephone receiver having very little distortion, actuated by alternating currents from vacuum tube oscillators. Observations were made on twelve male observers. Average curves show that at any frequency, DeltaEE is practically constant for intensitites greater than 106 times the threshold intensity; near the auditory threshold DeltaEE increases. Weber's law holds above this intensity, the value of DeltaEE=constant lying between 0.05 and 0.15 depending on the frequency. As a function of frequency DeltaEE is a minimum at about 2500 c.p.s., the minimum being more sharply defined at low sound intensities than it is at high. This frequency corresponds to the region of greatest absolute sensitivity of the ear. Analytical expressions are given [Eqs. (2), (3), (4) and (5)] which represent DeltaEE, within the error of observation, as a function of frequency and intensity. Using these equations it is calculated that at about 1300 c.p.s. the ear can distinguish 370 separate tones between the threshold of audition and the threshold of feeling.
Citation
PARRY MOON and DOMINA EBERLE SPENCER, "A Metric for Colorspace," J. Opt. Soc. Am. 33, 260-266 (1943)
http://www.opticsinfobase.org/josa/abstract.cfm?URI=josa-33-5-260
"Sensitivity to changes in the intensity of a random noise was determined over a wide range of intensities. The just detectable increment in the intensity of the noise is of the same order of magnitude as the just detectable increment in the intensity of pure tones. For intensities more than 30 db above the threshold of hearing for noise the size in decibels of the increment which can be heard 50 per cent of the time is approximately constant (0.41 db)… . Functions which describe intensity discrimination also describe the masking by white noise of pure tones and of speech. It is argued, therefore, that the determination of differential sensitivity to intensity is a special case of the more general masking experiment. The loudness of the noise was also determined, and just noticeable differences are shown to be unequal in subjective magnitude." (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Auditory thresholds determined by the intensity of the sound, the "minimum audible field," and pressure amplitude at the observer's ear drum, and the "minimum audible pressure" are reported for frequencies from 60 to 15,000 c.p.s. The former measurements yield somewhat higher values, due largely to wave motion in the ear canal and diffraction caused by the head. An erroneous formula for the binaural field on p. 299 is corrected in vol. 5, 1933, 60. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
The first part of this chapter discusses the fundamental characteristics of sound vibrations. The remaining sections discuss differential thresholds, masking and fatigue, combinations of tones, the subjective attributes of complex sounds, temporal effects in hearing, sound localization, and other binaural effects, all primarily from the point of view of the stimulus. 175-item bibliography. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
During recent years a program of work has been in progress at the National Physical Laboratory with the objective of providing improved data on several aspects of subjective acoustics. In connection with standards for audiometry, measurements have been made of the threshold of hearing for pure tones by earphone listening, and these have since been extended to the case of listening in free field. Latterly a redetermination of the equal-loudness relations for pure tones has been completed aimed at resolving discrepancies between former determinations and providing an improved basis for the establishment of a standard set of contours. These results apply to a large team of otologically normal observers, and cover the range from 25–15 000 cps and up to 130 db in sound pressure level. The results of this investigation enable the equivalent loudness of any pure tone to be expressed by simple formulas with coefficients varying smoothly with frequency. Considerable attention has also been devoted to the determination of the loudness scale, i.e., the function relating loudness level in phons to the magnitude of the loudness sensation expressed on the sone scale. An assessment of the experimental evidence has led to the formulation of a simple relation which appears adequate for practical purposes in noise measurement. Investigations are continuing on the determination of loudness levels of complex sounds from their objective spectra.
The distribution of an O's absolute judgments, in which he identifies a stimulus as having a particular value, may indicate how much information he obtained about which of several alternative stimuli occurred at a particular time. The amount of information conveyed to O can be measured in
bits. This measure may give an estimate of the minimum number of stimulus categories which will transmit the maximum amount of information. A technique is described for constructing a scale of equal discriminability to select the stimuli for maximum information transmission. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Entsprechend dem Schroteffekt der Elektronen mu man einen Schroteffekt der Photonen erwarten, d. h. ein Strom von Photonen mu bei konstant gehaltenen ueren Bedingungen statistische Schwankungseffekte zeigen, die bei kleinen Photonenzahlen groe prozentische Werte erreichen mssen. In den vorliegenden Untersuchungen wird gezeigt, da das dunkeladaptierte Auge eine so groe Empfindlichkeit besitzt, da es Lichtblitze, bestehend aus 40 bis 90 Photonen, schon wahrnehmen kann. Die dann zu erwartenden statistischen Schwankungen liegen gerade an der Grenze des Unterscheidungsvermgens des Auges, so da die Mglichkeit besteht, da gewisse Helligkeitsschwankungen, die man beim Beobachten schwchster Lichtquellen wahrnimmt, mindestens zum Teil auf dem Schroteffekt der Photonen beruhen.
Three phenomenological models are considered from which can be constructed a macroscopic statistical description of varieties of electronic noise, of which shot, thermal, and Barkhausen noise, electron multiplier and precipitation noise, clutter, ignition, and impulsive random noise in general are representative examples. The models examined are (I) non‐overlapping, periodic noise waves, common in pulse‐time modulation and other communication schemes, where the amplitude, phase, duration, and epoch in a period interval are subject to statistical variations; (II), non‐overlapping, nonperiodic disturbances, encountered in servo‐mechanism operation and keyed‐carrier communication techniques, for instance, which are like (I), but lack the basic periodic structure; and (III), poisson noise, consisting of the superposition of independent, randomly occurring elementary impulses. Much of the electronic noise mentioned above belongs to this more comprehensive type, where overlapping of the basic pulses is the characteristic feature. Because all (second‐order) moments are required in general for the analysis of noise in nonlinear systems, the attempt is made here to determine explicitly on the basis of the appropriate model the second‐order probability density W 2 in the important stationary cases. For noise of types (I) and (II) this appears impractical except in the simplest cases: only the lower order moments prove tractable. However, for poisson noise (III) an explicit treatment is possible for impulsive random noise, nearly normal random noise, and for the limiting, normal random cases (of which shot and thermal noise are examples). In Part I the main features of the models (I–III) are discussed, and the general probability density W l (X 1 , t 1 ; …; X s , t s ) in the nonstationary instances is formally constructed. In Part II, the distribution-
density for nearly normal random noise is given, the first and second (second‐order) moments of the various distributions are determined, and from these in turn are found the spectral distribution of the energy in the random waves.
In this section we use the representations of the noise currents given in section 2.8 to derive some statistical properties of I(t). The first six sections are concerned with the probability distribution of I(t) and of its zeros and maxima. Sections 3.7 and 3.8 are concerned with the statistical properties of the envelope of I(t). Fluctuations of integrals involving I2(t) are discussed in section 3.9. The probability distribution of a sine wave plus a noise current is given in 3.10 and in 3.11 an alternative method of deriving the results of Part III is mentioned. Prof. Uhlenbeck has pointed out that much of the material in this Part is closely connected with the theory of Markoff processes. Also S. Chandrasekhar has written a review of a class of physical problems which is related, in a general way, to the present subject.22
Experiments have been conducted to determine difference limens (DL's) for vowelformant frequency. The DL's are obtained from quality judgments on synthetic vowel sounds. The results indicate the maximum accuracy necessary in analyzing the formant structure of spoken vowels and in synthesizing the sounds from the resulting formant data.