Matthew J. Dry

Exemplar by feature applicability matrices and other Dutch normative data for semantic concepts

Features are at the core of many empirical and modeling endeavors in the study of semantic concepts. This article is concerned with the delineation of features that are important in natural language concepts and the use of these features in the study of semantic concept representation. The results of a feature generation task in which the exemplars and labels of 15 semantic categories served as cues are described. The importance of the generated features was assessed by tallying the frequency with which they were generated and by obtaining judgments of their relevance. The generated attributes also featured in extensive exemplar by feature applicability matrices covering the 15 different categories, as well as two large semantic domains (that of animals and artifacts). For all exemplars of the 15 semantic categories, typicality ratings, goodness ratings, goodness rank order, generation frequency, exemplar associative strength, category associative strength, estimated age of acquisition, word frequency, familiarity ratings, imageability ratings, and pairwise similarity ratings are described as well. By making these data easily available to other researchers in the field, we hope to provide ample opportunities for continued investigations into the nature of semantic concept representation. These data may be downloaded from the Psychonomic Society’s Archive of Norms, Stimuli, and Data, www.psychonomic.org/archive.

Dose-related effects of alcohol on cognitive functioning

We assessed the suitability of six applied tests of cognitive functioning to provide a single marker for dose-related alcohol intoxication. Numerous studies have demonstrated that alcohol has a deleterious effect on specific areas of cognitive processing but few have compared the effects of alcohol across a wide range of different cognitive processes. Adult participants (N = 56, 32 males, 24 females aged 18–45 years) were randomized to control or alcohol treatments within a mixed design experiment involving multiple-dosages at approximately one hour intervals (attained mean blood alcohol concentrations (BACs) of 0.00, 0.048, 0.082 and 0.10%), employing a battery of six psychometric tests; the Useful Field of View test (UFOV; processing speed together with directed attention); the Self-Ordered Pointing Task (SOPT; working memory); Inspection Time (IT; speed of processing independent from motor responding); the Traveling Salesperson Problem (TSP; strategic optimization); the Sustained Attention to Response Task (SART; vigilance, response inhibition and psychomotor function); and the Trail-Making Test (TMT; cognitive flexibility and psychomotor function). Results demonstrated that impairment is not uniform across different domains of cognitive processing and that both the size of the alcohol effect and the magnitude of effect change across different dose levels are quantitatively different for different cognitive processes. Only IT met the criteria for a marker for wide-spread application: reliable dose-related decline in a basic process as a function of rising BAC level and easy to use non-invasive task properties.

Human Performance on Visually Presented Traveling Salesperson Problems with Varying Numbers of Nodes.

We investigated the properties of the distribution of human solution times for Traveling Salesperson Problems (TSPs) with increasing numbers of nodes. New experimental data are presented that measure solution times for carefully chosen representative problems with 10, 20, . . . 120 nodes. We compared the solution times predicted by the convex hull procedure proposed by MacGregor and Ormerod (1996), the hierarchical approach of Graham, Joshi, and Pizlo (2000), and by five algorithms drawn from the artificial intelligence and operations research literature. The most likely polynomial model for describing the relationship between mean solution time and the size of a TSP is linear or near-linear over the range of problem sizes tested, supporting the earlier finding of Graham et al. (2000). We argue the properties of the solution time distributions place strong constraints on the development of detailed models of human performance for TSPs, and provide some evaluation of previously proposed models in light of our findings.

The roles of the convex hull and the number of potential intersections in performance on visually presented traveling salesperson problems

The planar Euclidean version of the traveling salesperson problem requires finding the shortest tour through a two-dimensional array of points. MacGregor and Ormerod (1996) have suggested that people solve such problems by using a global-to-local perceptual organizing process based on the convex hull of the array. We review evidence for and against this idea, before considering an alternative, local-to-global perceptual process, based on the rapid automatic identification of nearest neighbors. We compare these approaches in an experiment in which the effects of number of convex hull points and number of potential intersections on solution performance are measured. Performance worsened with more points on the convex hull and with fewer potential intersections. A measure of response uncertainty was unaffected by the number of convex hull points but increased with fewer potential intersections. We discuss a possible interpretation of these results in terms of a hierarchical solution process based on linking nearest neighbor clusters.

Decision Making and Confidence Given Uncertain Advice

We study human decision making in a simple forced-choice task that manipulates the frequency and accuracy of available information. Empirically, we find that people make decisions consistent with the advice provided, but that their subjective confidence in their decisions shows 2 interesting properties. First, peoples confidence does not depend solely on the accuracy of the advice. Rather, confidence seems to be influenced by both the frequency and accuracy of the advice. Second, people are less confident in their guessed decisions when they have to make relatively more of them. Theoretically, we develop and evaluate a type of sequential sampling process model-known as a self-regulating accumulator-that accounts for both decision making and confidence. The model captures the regularities in peoples behavior with interpretable parameter values, and we show its ability to fit the data is not due to excessive model complexity. Using the model, we draw conclusions about some properties of human reasoning under uncertainty.

Features of graded category structure: Generalizing the family resemblance and polymorphous concept models

Many real-world categories contain graded structure: certain category members are rated as more typical or representative of the category than others. Research has shown that this graded structure can be well predicted by the degree of commonality across the feature sets of category members. We demonstrate that two prominent feature-based models of graded structure, the family resemblance (Rosch & Mervis, 1975) and polymorphous concept models (Hampton, 1979), can be generalized via the contrast model (Tversky, 1977) to include both common and distinctive feature information, and apply the models to the prediction of typicality in 11 semantic categories. The results indicate that both types of feature information play a role in the prediction of typicality, with common features weighted more heavily for within-category predictions, and distinctive features weighted more heavily for contrast-category predictions. The same pattern of results was found in additional analyses employing rated goodness and exemplar generation frequency. It is suggested that these findings provide insight into the processes underlying category formation and representation.

Similar, But Not the Same: A Comparison of the Utility of Directly-rated and Feature-based Similarity Measures for Generating Spatial Models of Conceptual Data

Spatial models are employed to represent conceptual data in a wide range of fields within psychological research. In order to generate spatial models, it is necessary to first obtain empirical similarity data. A number of methods are available for collecting these data, but little effort has been made to compare their relative utility. In this article, we compare directly rated and five feature-based similarity data types in regard to their ability to be adequately represented by a spatial model (representational goodness of fit), and the ability of the representations to predict three external empirical variables (predictive validity). The results indicate that the representational goodness of fit of the feature-based similarities is noticeably superior to the directly rated similarities, and that the predictions of representations derived from common feature similarity data are substantially more likely than the predictions of all of the alternative representations. It is suggested that these findings are highly relevant to researchers employing spatial models to represent conceptual data, given that direct pairwise ratings have generally been considered the “gold standard” means of obtaining empirical similarities.

The aesthetic appeal of minimal structures: Judging the attractiveness of solutions to traveling salesperson problems

Ormerod and Chronicle (1999) reported that optimal solutions to traveling salesperson problems were judged to be aesthetically more pleasing than poorer solutions and that solutions with more convex hull nodes were rated as better figures. To test these conclusions, solution regularity and the number of potential intersections were held constant, whereas solution optimality, the number of internal nodes, and the number of nearest neighbors in each solution were varied factorially. The results did not support the view that the convex hull is an important determinant of figural attractiveness. Also, in contrast to the findings of Ormerod and Chronicle, there were consistent individual differences. Participants appeared to be divided as to whether the most attractive figure enclosed a given area within a perimeter of minimum or maximum length. It is concluded that future research in this area cannot afford to focus exclusively on group performance measures.

Uncovering Contrast Categories in Categorization With a Probabilistic Threshold Model

A contrast category effect on categorization occurs when the decision to apply a category term to an entity not only involves a comparison between the entity and the target category but is also influenced by a comparison of the entity with 1 or more alternative categories from the same domain as the target. Establishing a contrast category effect on categorization in natural language categories has proven to be laborious, especially when the categories concerned are natural kinds situated at the superordinate level of abstraction. We conducted 3 studies with these categories to look for an influence on categorization of both similarity to the target category and similarity to a contrast category. The results are analyzed with a probabilistic threshold model that assumes categorization decisions arise from the placement of threshold criteria by individual categorizers along a single scale that holds the experimental stimuli. The stimulis positions along the scale are shown to be influenced by similarity to both target and contrast. These findings suggest that the prevalence of contrast category effects on categorization might have been underestimated. Additional analyses demonstrate how the proposed model can be employed in future studies to systematically investigate the origins of contrast category effects on categorization.

Clustering, Randomness, and Regularity: Spatial Distributions and Human Performance on the Traveling Salesperson Problem and Minimum Spanning Tree Problem.

We investigated human performance on the Euclidean Traveling Salesperson Problem (TSP) and Euclidean Minimum Spanning Tree Problem (MST-P) in regards to a factor that has previously received little attention within the literature: the spatial distributions of TSP and MST-P stimuli. First, we describe a method for quantifying the relative degree of clustering, randomness or regularity within point distributions. We then review evidence suggesting this factor might influence human performance on the two problem types. Following this we report an experiment in which the participants were asked to solve TSP and MST-P test stimuli that had been generated to be either highly clustered, random, or highly regular. The results indicate that for both the TSP and MST-P the participants tended to produce better quality solutions when the stimuli were highly clustered compared to random, and similarly, better quality solutions for random compared to highly regular stimuli. It is suggested that these results provide support for the ideas that human solv- ers attend to salient clusters of nodes when solving these problems, and that a similar process (or series of processes) may underlie human performance on these two tasks.