DIMENSIONAL INTERPRETATION AND CONFIGURATION INVARIANCE IN MULTIDIMENSIONAL SCALING: AN EMPIRICAL STUDY

This paper is concerned with: a ) the invariance of nonmetric multi- dimensional scaling soludions over changes in stimulus domain; and b) pro- cedures for assisting the researcher in interpreting the axes of multidimen- sional scaling configurations. The stimuli consisted of the names of 17 popular brands of automobiles. Similarities and preference judgments were obtained from 37 subjects, divided into two groups of approximately equal size. Each group received 11 of the 17 stimuli. A core set of 6 stimuli was common over both groups. Each group also rated the 11 car models on 20 semantic differential scales. Results of the study suggested that the interpoint distances of bhe core set of 5 stimuli remained stable over subject groups and differences in stimulus set composition. Semantic differential ratings were used to find directions in the configurational space (obtained from overall similarity judg- ments) whose projections were maximally correlated with the outside (property) vectors. Stimulus configurations developed from the semantic differential space were highly congruent with those found by the analysis of direct similarity judgments. However, attempts to develop stimulus configurations by the "unfolding" of preference data alone did not lead to configurations which closely matched those found by the analysis of direct similarities data.     

Visual similarity is stronger than semantic similarity in guiding visual search for numbers

Using a visual search task, we explored how behavior is influenced by both visual and semantic information. We recorded participants’ eye movements as they searched for a single target number in a search array of single-digit numbers (0–9). We examined the probability of fixating the various distractors as a function of two key dimensions: the visual similarity between the target and each distractor, and the semantic similarity (i.e., the numerical distance) between the target and each distractor. Visual similarity estimates were obtained using multidimensional scaling based on the independent observer similarity ratings. A linear mixed-effects model demonstrated that both visual and semantic similarity influenced the probability that distractors would be fixated. However, the visual similarity effect was substantially larger than the semantic similarity effect. We close by discussing the potential value of using this novel methodological approach and the implications for both simple and complex visual search displays.     

Mapping the similarity space of paintings: Image statistics and visual perception

What makes two images look similar? Here we test the hypothesis that perceived similarity of artwork is related to basic image statistics to which the early visual system is attuned. In two experiments, we employ multidimensional scaling (MDS) analysis of paired-image similarity ratings from observers for paintings. Two sets of images, classified as “landscapes” and “portraits/still-life”, are tested separately. For the landscapes, we find that one of the first two MDS scales of similarity is strongly correlated with a basic greyscale image statistic, whereas the other dimension can be accounted for by a semantic variable (representation of people). For portrait/still-life, the first two MDS scales of similarity are most highly correlated with semantic variables. Linear combinations of statistical and nonstatistical features achieve improved predictive values for the first two MDS scales for both sets. The statistics that play the largest role in shaping similarity judgements in our tests are the activity fraction measure of sparseness and the log-log slope of the spatial frequency amplitude spectrum. We discuss these results in the context of scene perception and in terms of efficient coding of statistical regularities in scenes.     

Semantic differential studies of children's language development: Some methodological problems

A semantic differential study of development of antonym meanings revealed methodological problems (e.g., concept-scale interaction and developmental changes in scale-checking style) not fully considered in previous semantic differential investigations of child language development. Twenty adjectives were rated on 10 scales by subjects from grades K to 5 (approximately 6 through 11 years old) and adults. Analyses by means, polarity ratings, and average sums of squared differences failed to yield consistent developmental trends. Subjects' choices of scale ratings did show clear age-related differences: Grades K and 1 chose extreme ratings almost exclusively; grades 2 through 4 showed an increase in neutral choices and decreases in extreme ratings; grades 4 and 5 and adults chose extreme ratings least often, neutral choices most often, and more intermediate ratings than any younger subjects. Results show that scale choice is a crucial factor in designing semantic differential studies with children, there is a low correlation between mean scores and ratings given by individual children, and shifts in scale-checking style should be considered when interpreting apparent developmental changes in children's semantic differential ratings.     

Effect of scaling method on perception of textiles

This study examined the effect of scaling methods on tactual ratings of fabrics. Four tactile attributes, namely, stiffness, thickness, roughness, and warmth, of 11 test fabrics with a wide variety of tactual qualities were evaluated by 75 college students enrolled in textile courses. The tactual ratings of all four attributes evaluated by touch and sight and touch were significantly correlated. The tactual evaluation made on a 7-point semantic differential scale on each test fabric was compared with tactual ranking by paired-comparison method. Significant correlations were found between two rankings of four attributes.     

How useful are corpus-based methods for extrapolating psycholinguistic variables?

Subjective ratings for age of acquisition, concreteness, affective valence, and many other variables are an important element of psycholinguistic research. However, even for well-studied languages, ratings usually cover just a small part of the vocabulary. A possible solution involves using corpora to build a semantic similarity space and to apply machine learning techniques to extrapolate existing ratings to previously unrated words. We conduct a systematic comparison of two extrapolation techniques: k-nearest neighbours, and random forest, in combination with semantic spaces built using latent semantic analysis, topic model, a hyperspace analogue to language (HAL)-like model, and a skip-gram model. A variant of the k-nearest neighbours method used with skip-gram word vectors gives the most accurate predictions but the random forest method has an advantage of being able to easily incorporate additional predictors. We evaluate the usefulness of the methods by exploring how much of the human performance in a lexical decision task can be explained by extrapolated ratings for age of acquisition and how precisely we can assign words to discrete categories based on extrapolated ratings. We find that at least some of the extrapolation methods may introduce artefacts to the data and produce results that could lead to different conclusions that would be reached based on the human ratings. From a practical point of view, the usefulness of ratings extrapolated with the described methods may be limited.     

An analysis of generative representation systems

Three ideas are basic to generative theory: (a) Subjects are assumed to attend to the relations among stimuli, extracting the transformations relating pairs of stimuli; (b) the set of abstracted transformations is decomposed or reduced to an elementary set of generators; (c) subjects use the elementary generators as the basis for judging similarity. The purpose of this paper is to illustrate these ideas with an experiment in which subjects were asked to rate the similarity between stimulus pairs. The stimulus materials consisted of the permutations of a 4-item pattern with the properties of a dihedral group which insured the existence of sets of elementary transformations. Three analytic techniques were used to determine the generator set of transformations abstracted by subjects. The first analysis consisted of a monotonic regression between dissimilarity ratings and the number of elementary generators of a given permutation. The residual variance of this monotone regression, suitably normalized, was used as a quantitative goodness-of-fit measure. For the stochastic analysis, cumulative distributions of dissimilarity ratings were obtained for permutations requiring one, two, or three generators. The idea was that permutations requiring fewer generators should be associated with distributions of lower dissimilarity values (higher similarity scores) as compared to permutations predicted to be transformationally more complex. The final analysis, a multidimensional scaling of dissimilarity ratings, converted subjects' ratings into spatial structures to determine whether individual subjects' ratings exhibited the predicted spatial arrangement. The monotone regression and stochastic analyses abstracted similar generator sets for individual subjects, some of which provided perfect fits to the data. Although the scaling analysis yielded similar estimates of generators, for some subjects, transformations with the same number of generators yielded unequal “cognitive” distances resulting in some-what deformed spatial structures for these subjects. It was concluded that the results generally supported a generative model as an approximation to subjects' representations of interstimulus relationships.     

Dimensions of the group environment scale

Student teachers in yearlong mental health consultation groups completed Moos's Group Environment Scale (GES) and three semantic differential ratings of their satisfaction with the group experience. Factor analysis of the GES yielded four dimensions, the first two corresponding to Moos's hypothesized dimensions of Relationship and System Maintenance/System Change. Intercorrelations of factor composite scores with the differential ratings provided concurrent validity evidence for the factored dimensions. Results were interpreted as supporting a multidimensional rather than unidimensional structure for the GES.     

Language encodes geographical information

Population counts and longitude and latitude coordinates were estimated for the 50 largest cities in the United States by computational linguistic techniques and by human participants. The mathematical technique Latent Semantic Analysis applied to newspaper texts produced similarity ratings between the 50 cities that allowed for a multidimensional scaling (MDS) of these cities. MDS coordinates correlated with the actual longitude and latitude of these cities, showing that cities that are located together share similar semantic contexts. This finding was replicated using a first-order co-occurrence algorithm. The computational estimates of geographical location as well as population were akin to human estimates. These findings show that language encodes geographical information that language users in turn may use in their understanding of language and the world.     

Mental representations of affect knowledge

The circumplex structure derived from similarity ratings of affect words is assumed to be a conceptual representation of affect anchored in semantic knowledge. Recently, it been suggested that this structure is not based on semantic knowledge at all, but may instead reflect a type of episodic knowledge: The degree to which emotions covary in everyday life. In two experience-sampling studies, we compared the semantic and the episodic hypotheses by comparing participants' similarity ratings to the observed covariations in their own affective experience computed from their momentary reports. In Study 2, participants also provided estimates of the degree to which their emotions covaried. Evidence from both studies indicate that similarity judgements are related both to semantic and episodic information, indicating that a pure episodic account of similarity ratings, and the mental representation of affect that they reflect, is untenable.     

Subjective responses to alcoholic and non-alcoholic beverages by alcoholic respondents

The semantic differential ratings of real alcohol has been used as a pre-post measure of the effectiveness of behavioral treatment of alcoholism. This attitudinal measure has yielded predominantly negative results. Evidence has been presented that the semantic differential used to rate alcoholic and non-alcoholic slides was not unifactorial. The present study tested whether the semantic differential ratings of real alcohol were unifactorial. Eighty alcoholics tasted two alcoholic and two non-alcoholic beverages and rated them on 20 adjectives derived from 10 supposedly bipolar adjective pairs. For the alcoholic drinks there was some concept scale interaction and definite evidence for a splitting of adjective pairs into two evaluative factors, one of which was mixed with a potency factor. The negative findings of previous research may not reflect a lack of attitudinal change, but a methodological error.     

Differentiation of affective and denotative meaning systems and their influence in personality ratings.

The present study presents an empirical example of the dichotomy of affective and denotative meaning systems and their influence on individual differences in personality ratings. The three-mode factor analytic technique with a newly developed transformation methodology for the scale mode was applied to data collected by Hogenraad from 50 French-speaking Belgians, rating 40 personality concepts against 40 semantic differential scales. Results indicated that three affective dimensions (evaluation, potency, and activity) proved to be dominant in the indigenous factor structure of personality impressions and that three dimensions in the "other" space, orthogonal to affect, are clearly interpretable denotative semantic features of personalities. Three idealized individual differences on interactions of these two meaning systems with four concept factors were highlighted by the final rotated inner core matrix. The present methodology along with the semantic differential technique and three-mode factor analysis can be applied to various types of subjects and/or concept domains for better understanding of intra- and intercultural differences.     

A multidimensional scaling and semantic differential technique study of the perception of environmental settings

Abstract.— Multidimensional scaling is compared with the semantic differential technique (SDT) as methods for dimensional analysis of the perception of environmental settings. In one experiment employing as stimuli SDT scales used in previous experiments a General Evaluation , a Social Status , a Potency , a Closedness , and an Affection factor were found for both methods. A Unity and a Complexity factor merged into a single factor, an O riginality factor was only found for multidimensional scaling. In a second experiment employing as stimuli perspective drawings of urban and suburban settings, the results differed in that the General Evaluation , the Social Status , the Closedness , and the Unity/Complexity factors were found for the SDT, and only the first two for multidimensional scaling. Further experiments with other stimuli and methods of representation are needed in order to draw firm conclusions regarding the correspondence between SDT and multidimensional scaling results within the area under consideration.     

THE SEMANTICS OF FACIAL EXPRESSIONS AND THE PREDICTION OF THE MEANINGS OF STEREOSCOPICALLY FUSED FACIAL EXPRESSIONS

Two experiments were concerned with scaling affective meanings of posed facial expressions using a specially devised form of semantic differential and with determining the extent to which the meaning of stereoscopically fused facial expressions could be predicted. Both experiments provided evidence for at least three dimensions. High reliability for scaling the affective meanings of facial expressions was demonstrated. Two methods were employed for predicting the meaning of a composite expression. The results indicated that the predictions based on the congruity principle do no better than those based on a simple algebraic mean of the two components.     

More than E,P and A in semantic differential scales: AN ANSWER TO QUESTIONS RAISED BY S. T. M. LANE1

The present study demonstrates the separation of affective and denotative meaning systems in the semantic space by a new factor analytic technique. The data used were collected in Brazil and Hungary following standard pan-cultural procedures (Osgood et al., 1975) for semantic differential ratings on 100 heterogeneous concepts against 60 indigenous scales elicited from teen-age males in the respective language/culture communities. Results from both cultures provided evidence for dominance of affective Evaluation, Potency and Activity (E, P, A) in the indigenous factor structures. The dimensions in the “other” (Denotation) space, orthogonal to Affect, are also clearly interpretable affect-free semantic features of concepts. The simultaneous salience of these two semantic systems for most scales highlights the importance of cultural meanings of scales for each indigenous culture under study.     

Perceptual dimensions of infant cry signals: a semantic differential analysis

In the present study the Semantic Differential technique was used to examine perceived characteristics of infant cry sounds. 24 cries (6 each of pain, hunger, birth, and "pleasure") were rated by 39 mothers on 50 scales. A factor analysis of the scale ratings uncovered three main factors, labelled Affect, Potency, and Evaluation, respectively. 5 "factorially-pure" scales were selected to represent each factor. A hierarchical cluster analysis of the mean factor-scale ratings uncovered 3 major cry clusters which essentially represented 3 cry-type groups, hunger, pleasure, and pain/birth. A close correspondence was found between this set of clusters and another set generated independently from cry-recognition data, indicating that the semantic differential factor-scales effectively discriminated perceptually distinct cries. This procedure adequately separates the physical (acoustic) properties of signals from their listener-perceived (auditory) ones. This distinction is not possible with studies which depend upon multivariate techniques and casts doubt on their findings.     

Semantic distance norms computed from an electronic dictionary (WordNet)

WordNet, an electronic dictionary (or lexical database), is a valuable resource for computational and cognitive scientists. Recent work on the computing of semantic distances among nodes (synsets) in WordNet has made it possible to build a large database of semantic distances for use in selecting word pairs for psychological research. The database now contains nearly 50,000 pairs of words that have values for semantic distance, associative strength, and similarity based on co-occurrence. Semantic distance was found to correlate weakly with these other measures but to correlate more strongly with another measure of semantic relatedness, featural similarity. Hierarchical clustering analysis suggested that the knowledge structure underlying semantic distance is similar in gross form to that underlying featural similarity. In experiments in which semantic similarity ratings were used, human participants were able to discriminate semantic distance. Thus, semantic distance as derived from WordNet appears distinct from other measures of word pair relatedness and is psychologically functional. This database may be downloaded fromwww.psychonomic.org/archive/.     

Comparison of perceptual dimensions uncovered from infant-cry signals using the method of pair-comparisons and the semantic differential

Since infant cry sounds may provide caregivers with information concerning an infant's needs, several studies have examined the perceptual features of cry signals. This tradition is followed in the present study, which utilises two techniques; the method of pair-comparisons (coupled to INDSCAL) and the Semantic Differential (SD). In each case 24 infant cries (six each of pain, hunger, pleasure and birth) are used as stimuli. Data were collected from mothers; 39 rated each cry on 50 SD scales, and 26 provided ratings of similarity for every possible pairwise combination of cries. The similarity of the perceptual spaces described by the two techniques is demonstrated by a canonical analysis which compared the SD factor matrix with the INDSCAL dimension-weight matrix. Two of three possible common dimensions were uncovered. This similarity is also indicated by a standard multiple regression analysis by which the INDSCAL dimensions were described in terms of the SD scales. However, it appears the SD emphasised emotional characteristics of the cry signals whereas INDSCAL emphasised physical attributes.