Dissimilarity Judgments Research Articles

An analysis of the effects of structural parameter variations on the auditory perception of environmental noises transmitted through a simulated window

The present paper concerns the influence of the independent variation of some structural parameters on the auditory perception of environmental noises transmitted through a window. The pane of glass in its frame was modelled as a thin baffled plate with viscoelastic boundary conditions; transmitted noises were then synthesized by convolving a binaurally recorded environmental noise to different calculated impulse responses of the plate involving so many different values of the structural parameters (structural damping factor and mounting conditions). Stimuli were pairwise submitted to a jury of subjects who were asked to give dissimilarity and preference judgments. Analysis first allowed to identify the relevant auditory attributes that were likely used by the subjects in their differentiation task, and also focused on drawing up a preference ranking of the transmitted noises in order to provide recommendations for window structural modifications that could improve acoustic comfort in inner spaces. On this point, depending on the structural parameter, recommendations appear to be more or less consensual in relation to the outdoor sound context.

Effects of Training on the Acoustic–Phonetic Representation of Synthetic Speech

Investigate training-related changes in acoustic-phonetic representation of consonants produced by a text-to-speech (TTS) computer speech synthesizer. Forty-eight adult listeners were trained to better recognize words produced by a TTS system. Nine additional untrained participants served as controls. Before and after training, participants were tested on consonant recognition and made pairwise judgments of consonant dissimilarity for subsequent multidimensional scaling (MDS) analysis. Word recognition training significantly improved performance on consonant identification, although listeners never received specific training on phoneme recognition. Data from 31 participants showing clear evidence of learning (improvement>or=10 percentage points) were further investigated using MDS and analysis of confusion matrices. Results show that training altered listeners' treatment of particular acoustic cues, resulting in both increased within-class similarity and between-class distinctiveness. Some changes were consistent with current models of perceptual learning, but others were not. Training caused listeners to interpret the acoustic properties of synthetic speech more like those of natural speech, in a manner consistent with a flexible-feature model of perceptual learning. Further research is necessary to refine these conclusions and to investigate their applicability to other training-related changes in intelligibility (e.g., associated with learning to better understand dysarthric speech or foreign accents).

Gender and homophily: Differences in male and female association in Blau space

Homophily, the tendency for similar individuals to associate, is one of the most robust findings in social science. Despite this robustness, we have less information about how personal characteristics relate to differences in the strength of homophily. Nor do we know much about the impact of personal characteristics on judgments of relative dissimilarity. The present study compares the strength of age, religious, and educational homophily for male and female non-kin ties using network data from the 1985 General Social Survey. It also compares the patterning of ties among dissimilar alters for both sexes. The results of this exploratory effort indicate that males and females are almost equally homophilous, although religious homophily exerts a stronger influence on females than males. Males and females do, however, differ in their tendency to associate with certain types of dissimilar alters. Education is essentially uniform for both sexes, religious difference is more important for females than males, and those over sixty or under thirty are less different from the middle categories of age for females than for males. The results suggest that males are able to bridge larger areas of social space in their non-kin interpersonal networks and likely accumulate greater social capital as a consequence.

Texture perception through direct and indirect touch: An analysis of perceptual space for tactile textures in two modes of exploration

Considerable information about the texture of objects can be perceived remotely through a probe. It is not clear, however, how texture perception with a probe compares with texture perception with the bare finger. Here we investigate the perception of a variety of textured surfaces encountered daily (e.g., corduroy, paper, and rubber) using the two scanning modes—direct touch through the finger and indirect touch through a probe held in the hand—in two tasks. In the first task, subjects rated the overall pair-wise dissimilarity of the textures. In the second task, subjects rated each texture along three continua, namely, perceived roughness, hardness, and stickiness of the surfaces, shown previously as the primary dimensions of texture perception in direct touch. From the dissimilarity judgment experiment, we found that the texture percept is similar though not identical in the two scanning modes. From the adjective rating experiments, we found that while roughness ratings are similar, hardness and stickiness ratings tend to differ between scanning conditions. These differences between the two modes of scanning are apparent in perceptual space for tactile textures based on multidimensional scaling (MDS) analysis. Finally, we demonstrate that three physical quantities, vibratory power, compliance, and friction carry roughness, hardness, and stickiness information, predicting perceived dissimilarity of texture pairs with indirect touch. Given that different types of texture information are processed by separate groups of neurons across direct and indirect touch, we propose that the neural mechanisms underlying texture perception differ between scanning modes.

Dissimilarity and the Classification of Male Singing Voices

Traditionally, timbre has been defined as that perceptual attribute that differentiates two sounds when pitch and loudness are equal, and thus is a measure of dissimilarity. By such a definition, each voice possesses a set of timbres, and the ability to identify any voice or voice category across different pitch-loudness-vowel combinations must be due to an ability to "link" these timbres by abstracting the "timbre transformation," the manner in which timbre subtly changes across pitch and loudness for a specific voice or voice category. Using stimuli produced across the singing range by singers from different voice categories, this study sought to examine how timbre and pitch interact in the perception of dissimilarity in male singing voices. This study also investigated whether or not listener experience affects the perception of timbre as a function of pitch. The resulting multidimensional scaling (MDS) representations showed that for all stimuli and listeners, dimension 1 correlated with pitch, while dimension 2 correlated with spectral centroid and separated vocal stimuli into the categories baritone and tenor. Dimension 3 appeared highly idiosyncratic depending on the nature of the stimuli and on the experience of the listener. Inexperienced listeners appeared to rely more heavily on pitch in making dissimilarity judgments than did experienced listeners. The resulting MDS representations of dissimilarity across pitch provide a glimpse of the timbre transformation of voice categories across pitch.

An Information Theory Analysis of Visual Complexity and Dissimilarity

The subjective complexity of a computer-generated bitmap image can be measured by magnitude estimation scaling, and its objective complexity can be measured by its compressed file size. There is a high correlation between these measures of subjective and objective complexity over a large set of marine electronic chart and radar images. The subjective dissimilarity of a pair of bitmap images can be predicted from subjective and objective measures of the complexity of each image, and from the subjective and objective complexity of the image produced by overlaying the two simple images. In addition, the subjective complexity of the image produced by overlaying two simple images can be predicted from the subjective complexity of the simple images and the subjective dissimilarity of the image pair. The results of the experiments that generated these complexity and dissimilarity judgments are consistent with a theory, outlined here, that treats objective and subjective measures of image complexity and dissimilarity as vectors in Euclidean space.

Multidimensional Scaling of Schematically Represented Faces Based on Dissimilarity Estimates and Evoked Potentials of Differences Amplitudes

This study researches the input of the cerebral occipital and temporal cortex in the analysis of facial configuration and expressive characteristics. Analysis is based on the construction of a spherical model for the differentiation of schematically presented faces with quantitatively altering curvature of the mouth and brows. The model is designed using the method of multidimensional scaling of the dissimilarity judgments between stimuli (faces) and the amplitude of evoked potentials of differences (EPD) between abrupt stimulus changes recorded from the occipital and posterior temporal cortex. Analysis of the structure of the spherical model of facial differentiation depending on the electrode site and the latency of the EPD component within the duration of 120-240 ms has demonstrated that the activity of the occipital and posterior temporal cortex of the right hemisphere is associated with the emotional characteristics of the presented face, whereas facial configuration is reflected in the activation of both posterior temporal cortex and the occipital cortex of the left hemisphere. At all electrode sites maximum information of the emotional expression and configuration is represented in inter-peak amplitude P120-N180. With increasing latency there is increased distortion of the structure of differences in the spherical model of schematically presented faces, which is interpreted as an attenuation of electrical activity associated with the analysis of the emotional expression, which occurs more rapidly than configuration analysis.

Acoustic correlates of timbre space dimensions: A confirmatory study using synthetic tones

Timbre spaces represent the organization of perceptual distances, as measured with dissimilarity ratings, among tones equated for pitch, loudness, and perceived duration. A number of potential acoustic correlates of timbre-space dimensions have been proposed in the psychoacoustic literature, including attack time, spectral centroid, spectral flux, and spectrum fine structure. The experiments reported here were designed as direct tests of the perceptual relevance of these acoustical parameters for timbre dissimilarity judgments. Listeners presented with carefully controlled synthetic tones use attack time, spectral centroid, and spectrum fine structure in dissimilarity rating experiments. These parameters thus appear as major determinants of timbre. However, spectral flux appears as a less salient timbre parameter, its salience depending on the number of other dimensions varying concurrently in the stimulus set. Dissimilarity ratings were analyzed with two different multidimensional scaling models (CLASCAL and CONSCAL), the latter providing psychophysical functions constrained by the physical parameters. Their complementarity is discussed.

Clustering with qualitative information

We consider the problem of clustering a collection of elements based on pairwise judgments of similarity and dissimilarity. Bansal et al. (in: Proceedings of 43rd FOCS, 2002, pp. 238–247) cast the problem thus: given a graph G whose edges are labeled “ + ” (similar) or “ - ” (dissimilar), partition the vertices into clusters so that the number of pairs correctly (resp., incorrectly) classified with respect to the input labeling is maximized (resp., minimized). It is worthwhile studying both complete graphs, in which every edge is labeled, and general graphs, in which some input edges might not have labels. We answer several questions left open by Bansal et al. (2002) and provide a sound overview of clustering with qualitative information. Specifically, we demonstrate a factor 4 approximation for minimization on complete graphs, and a factor O ( log n ) approximation for general graphs. For the maximization version, a PTAS for complete graphs was shown by Bansal et al. (2002), we give a factor 0.7664 approximation for general graphs, noting that a PTAS is unlikely by proving APX-hardness. We also prove the APX-hardness of minimization on complete graphs.

The psychomechanics of simulated sound sources: material properties of impacted bars.

Sound can convey information about the materials composing an object that are often not directly available to the visual system. Material and geometric properties of synthesized impacted bars with a tube resonator were varied, their perceptual structure was inferred from multidimensional scaling of dissimilarity judgments, and the psychophysical relations between the two were quantified. Constant cross-section bars varying in mass density and viscoelastic damping coefficient were synthesized with a physical model in experiment 1. A two-dimensional perceptual space resulted, and the dimensions were correlated with the mechanical parameters after applying a power-law transformation. Variable cross-section bars varying in length and viscoelastic damping coefficient were synthesized in experiment 2 with two sets of lengths creating high- and low-pitched bars. In the low-pitched bars, there was a coupling between the bar and the resonator that modified the decay characteristics. Perceptual dimensions again corresponded to the mechanical parameters. A set of potential temporal, spectral, and spectrotemporal correlates of the auditory representation were derived from the signal. The dimensions related to mass density and bar length were correlated with the frequency of the lowest partial and are related to pitch perception. The correlate most likely to represent the viscoelastic damping coefficient across all three stimulus sets is a linear combination of a decay constant derived from the temporal envelope and the spectral center of gravity derived from a cochlear representation of the signal. These results attest to the perceptual salience of energy-loss phenomena in sound source behavior.

Dissimilarity and the Classification of Female Singing Voices: A Preliminary Study

Summary: Traditionally, timbre has been defined as that perceptual attribute that differentiates two sounds when pitch and loudness are equal and thus is a measure of dissimilarity. By such a definition, each voice possesses a set of timbres, and the identity of any voice or voice category across different pitch–loudness–vowel combinations must be due to an abstraction of the pattern of timbre transformation. Using stimuli produced across the singing range by singers from different voice categories, this study sought to examine how timbre and pitch interact in the perception of dissimilarity. This study also investigated whether listener experience affects the perception of timbre as a function of pitch. The resulting multidimensional scaling (MDS) representations showed that for all stimuli and listeners, dimension 1 correlated with pitch, whereas dimension 2 correlated with spectral centroid and separated vocal stimuli into the categories mezzo-soprano and soprano. Dimension 3 appeared highly idiosyncratic depending on the nature of the stimuli and on the experience of the listener. Inexperienced listeners appeared to rely more heavily on pitch in making dissimilarity judgments than did experienced listeners. The resulting MDS representations of dissimilarity across pitch provide a glimpse of the timbre transformation of voice categories across pitch.

Avoiding the dangers of averaging across subjects when using multidimensional scaling

Ashby, Maddox and Lee (Psychological Science, 5 (3) 144) argue that it can be inappropriate to fit multidimensional scaling (MDS) models to similarity or dissimilarity data that have been averaged across subjects. They demonstrate that the averaging process tends to make dissimilarity data more amenable to metric representations, and conduct a simulation study showing that noisy data generated using one distance metric, when averaged, may be better fit using a different distance metric. This paper argues that a Bayesian measure of MDS models has the potential to address these difficulties, because it takes into account data-fit, the number of dimensions used by an MDS representation, and the precision of the data. A method of analysis based on the Bayesian measure is demonstrated through two simulation studies with accompanying theoretical analysis. In the first study, it is shown that the Bayesian analysis rejects those MDS models showing better fit to averaged data using the incorrect distance metric, while accepting those that use the correct metric. In the second study, different groups of simulated ‘subjects’ are assumed to use different underlying configurations. In this case, the Bayesian analysis rejects MDS representations where a significant proportion of subjects use different configurations, or when their dissimilarity judgments contain significant amounts of noise. It is concluded that the Bayesian analysis provides a simple and principled means for systematically accepting and rejecting MDS models derived from averaged data.

Judging the urgency of nonvocal auditory warning signals: perceptual and cognitive processes.

Previous research has identified acoustic properties modulating the perceived urgency of alarms. The authors conducted 3 experiments using a multidimensional approach in which participants made acoustic dissimilarity judgments and urgency dissimilarity judgments for pairs of sequences. Experiment 1 confirmed the validity of acoustic parameters in urgency perception of experimental alarms. Experiment 2 confirmed the role of these acoustic parameters with real alarms but suggested the importance of additional factors. Experiment 3 compared the relative degrees of urgency of alarms from Experiments 1 and 2, highlighting the role of both sequence structure and associated mental representation. The authors conclude that the design of alarms should not be based exclusively on acoustic factors but should also take into consideration the acquisition of an appropriate mental representation.

DISSIMILARITY JUDGMENTS IN RELATION TO TEMPORAL AND SPATIAL FACTORS FOR THE SOUND FIELD IN AN EXISTING HALL

In relation to the temporal and spatial factors of sound fields, dissimilarity judgments for different source locations on a stage were performed. This study is based on the model of the auditory–brain system, which consists of the autocorrelation and crosscorrelation mechanisms for sound signals arriving at two ears and the specialization of human cerebral hemispheres. There are three temporal factors (τ 1,φ 1 and τ e) extracted from the autocorrelation function and four spatial factors (LL,IACC,τ IACC and W IACC) from the interaural crosscorrelation function of binaural signals. In addition to these temporal and spatial factors, the orthogonal factors of the subjective preference for sound field (Δt 1 and T sub) were taken into account. The psychological distance between sound fields of different source locations on the stage were calculated by using these temporal, spatial and orthogonal factors of sound fields. Using these distances and their linear combination, dissimilarity can be calculated. Results of multivariable analysis show that the calculated scale values of dissimilarity agree well with the measured scale values.

A confirmatory analysis of four acoustic correlates of timbre space

Exploratory multidimensional scaling studies of musical instrument timbres generally yield two- to four-dimensional perceptual spaces. Acoustic parameters have been derived that correlate moderately to highly with the perceptual dimensions. In a confirmatory study, two three-dimensional sets of synthetic, harmonic sounds equalized for fundamental frequency, loudness, and perceived duration were designed. The first two dimensions corresponded to attack time and spectral centroid in both sound sets. The third dimension corresponded to spectral flux (variation of the spectral centroid over time) in the first set and to the energy ratio of odd to even harmonics in the second set. Group analyses of dissimilarity judgments for all pairs of sounds homogeneously distributed in each space revealed a two-dimensional solution for the first set and a three-dimensional solution for the second set. Log attack time and spectral centroid were confirmed as perceptual dimensions in both solutions. The even/odd energy ratio was confirmed as a third dimension in the second set. Spectral flux was not confirmed in the first set, suggesting that this parameter should be re-examined. Analyses of individual data sets tested for differences across listeners in the mapping of acoustic parameters to perceptual dimensions. [Work supported by the CTI program of the CNRS.]

Mixed Tree and Spacial Representations of Dissimilarity Judgments

Whereas much previous research, focusing on the comparative fit at the aggregate level, has shown that either tree or spatial representations of dissimilarity judgments may be appropriate, we investigate whether there exist classes of subjects differing in the extent to which dissimilarity judgments are better represented by additive tree or spatial multidimensional scaling models. We develop a mixture model for the analysis of dissimilarity data that entails both representation and measurement model components. The latter involves distributional assumptions on the measurement error and enables estimation by maximum likelihood. The former component allows dissimilarity judgments to be represented either by an additive tree structure or by a spatial configuration, or a mixture of both. To investigate the appropriateness of additive tree versus spatial representations, the model is applied to twenty empirical data sets. We compare the fit of our model with that of aggregate additive tree and spatial models, as well as with mixtures of additive trees and mixtures of spatial configurations, respectively. We formulate some empirical generalizations on the relative importance of tree versus spatial structures in representing dissimilarity judgments.

The labeled dissimilarity scale: a metric of perceptual dissimilarity.

Fundamental to the concept of psychological distance is the idea that confusability allows discovery of the perceptual relationships between objects, which provides understanding of the underlying principles that govern the functioning of a system. Thus, judgments of dissimilarity (conceptually proportional to the inverse of confusability) may provide insight into the elusive underlying quality-coding mechanisms in that sensory system. In the present experiments, a labeled dissimilarity scale (LDS) that reflects the magnitude of odorant dissimilarity was developed in a fashion similar to that reported by Green (Green, Shaffer, & Gilmore, 1993). This scale was produced by rating the perceptual intensity implied by adverbs describing different levels of dissimilarity, and then attaching those descriptors to appropriate locations on a numerical scale. The usefulness of the scale was demonstrated by its ability to produce visual color space with ratings of dissimilarity of Munsell color chips. The stability and reliability of the LDS was evaluated by comparing it with the traditional scaling technique of magnitude estimation (ME). It was found that the scales produced similar ratings of odorant dissimilarity and showed a similar susceptibility to the effects of contrast convergence. However, the coefficients of variation of dissimilarities rated with ME were much higher than those produced with the LDS. The subjects also dealt with the LDS without the anxiety that usually accompanies first-time users of ME. The LDS provided stable ratings of odorant dissimilarity and preserved the inferred ratio scale properties of ME.

A Comparison of Multidimensional Scaling Methods for Perceptual Mapping

Multidimensional scaling has been applied to a wide range of marketing problems, in particular to perceptual mapping based on dissimilarity judgments. The introduction of methods based on the maximum likelihood principle is one of the most important developments. In this article, the authors compare the three available Maximum Likelihood Multidimensional Scaling (MLMDS) methods, namely, MULTISCALE, MAXSCAL, and PROSCAL, and the traditional multidimensional scaling (MDS) method KYST in a Monte Carlo study with 243 synthetic data sets. The MLMDS methods outperform KYST with respect to recovering the perceptual maps. MAXSCAL recovers the true distances between brands somewhat better than MULTISCALE, which is somewhat better than PROSCAL. With regard to distance recovery, the MLMDS methods are quite robust to violations of distributional assumptions. The decision criteria for selecting the number of dimensions are less robust to distributional violations. The results support the use of Consistent Akaike Information Criterion for the selection of the number of dimensions. The authors recommend that dissimilarity judgments be collected on interval scales or on ordinal scales with a substantial number of scale values. The authors discuss implications of the results for the design and analysis of perceptual mapping studies.

Multidimensional scaling of musical timbre constrained by physical parameters

A new multidimensional scaling technique [S. Winsberg and G. De Soete, Br. J. Math. Stat. Psychol. 50, 55–72 (1997)] is applied to the analysis of dissimilarity judgments on musical timbres in both group and individual data. This technique constrains the resulting spatial model such that the order of items along a given perceptual dimension preserves their order along a previously established physical dimension. The fit between perceptual and physical dimensions is achieved with spline functions and yields what may be interpreted as the auditory transform of the physical dimension needed to obtain the perceptual one. A reanalysis of ten timbre spaces from the literature shows that this kind of model does not work as well on group data as it does on individual data due to differences in the nature of underlying dimensions and in the form of the auditory transforms for different listeners. Further, an analysis of individual data sheds light on the reasons why higher dimensions in published timbre spaces are so often difficult to interpret.

Perceptual analysis of synthesized struck bars

This study is part of a project seeking to understand musical timbre and perceptual phenomena of the sound environment. It concerns the sensitivity of listeners to sounds emitted by simple vibrating structures, such as xylophone bars. Previous studies of timbre have characterized this attribute in terms of a small number of perceptual dimensions of an analytic nature (attack time, spectral centroid, spectral flux). Other studies have suggested a return to the physical characteristics of the sound sources themselves in order to understand their influence on timbre perception. Sounds were synthesized with a computer model of the vibrating bar developed at ENST that allows precise control of physical parameters. Listeners made dissimilarity judgments on all pairs of different sounds. Multidimensional scaling analyses of these data yield a perceptual space with dimensions that correspond strongly to the varied physical parameters (material density and the first term of internal damping) as well as to salient auditory dimensions (pitch for the former and a combination of spectral centroid and decay rate for the latter). These results argue in favor of a privileged processing of source characteristics by the auditory system.