Effects of perceivability, similarity, and bias--论文代写范文精选

它本质上是难以察觉的,或因为它非常类似于其他字母,或者因为观察人士不愿给出正确的响应。因此,这些因素之间的关系,甚至实际上是否所有独立的理论概念,仍然是一个悬而未决的问题。下面的paper代写范文进行详述。

Introduction 
One of the landmark artifacts of western culture is a common writing system based on the Latin alphabet. The importance of the Latin alphabet has inspired researchers since the early days of modern psychology and visual science to investigate how letters are identified, and to characterize the similarity structure of the alphabet (Cattell, 1886; Javal, 1881). Over the past 130 years numerous researchers have studied the alphabet, and have attributed performance in letter identification and rating tasks to factors such as letter perceivability, letter similarity, and response biases. However, it is often challenging to distinguish the effects of these factors, and thus difficult to establish the psychological validity and independence of each individual factor. For example, a letter may be identified poorly because it is inherently difficult to perceive, or because it is highly similar to other letters in the alphabet, or because observers are reluctant to give the correct response. Thus, the relationship between these factors, and indeed whether they are all even independent theoretical concepts, remains an open question.

The purpose of this article is to look at the factors that historically have been used to account for letter identification accuracy, and to propose a model by which these factors can be estimated. Specifically, we will examine past research on the identification and confusion of the alphabet, in order to (1) identify the contexts in which the alphabet has been studied in the past, (2) establish the psychological meanings of perceivability, similarity, and response bias, and (3) identify a method and model for estimating the three factors simultaneously from two experiments we will also report. We will begin by discussing a comprehensive review of this research on the alphabet.

Overview of prior research motivations and theoretical constructs 
Most previous behavioral research on the alphabet has focused on describing the perceivability, confusability, or similarity space of the letters. By and large, studies can be characterized by three primary motivations: (1) applied attempts to make written text more comprehensible or allow learners to acquire reading skills more easily; (2) empirical research aimed at understanding the visual system; and (3) theoretical research attempting to characterize or model how letters are represented by the visual or cognitive system. Many early researchers were concerned with identifying typefaces, fonts, and letters that were more or less legible, with the aim of improving printing and typesetting. For example, Javal (1881), Helmholtz's students Cattell (1886) and Sanford (1888), Roethlein (1912), and Tinker (1928) all attempted to rank letters in their order of legibility, and also identified letter pairs that were especially confusable in order to allow faster reading and less error-prone communication. 

Cattell (1886), Javal (1881), and Sanford (1888) each made suggestions about how to modify some letters to be more distinguishable and readable. One of the most substantial efforts aimed at improving the legibility of typeset text was made by Ovink (1938), who published a book describing in detail the errors and confusions produced for letters and numbers of eleven different fonts, including detailed recommendations for how each letter should be formed to improve its legibility. Other early applied research was concerned with ophthalmological tests (including Javal, 1881, as well as Hartridge & Owen, 1922 and Banister, 1927). Similar applied research has continued in more recent years: Bell (1967), Gupta, Geyer, and Maalouf (1983) and van Nes (1983) each have dealt with practical modern applications of font face and letter confusions. 

Not surprisingly, because much of this research attempted to identify font faces that were more or less easy to read, the primary psychological construct adopted by these researchers was akin to perceivability (although researchers often used the term legibility). In addition, many of these researchers also noted when letters were especially confusable because of visual similarity. For example, Roethlein (1912) reported the rank order of letter legibility, implying that perceivability is an inherent aspect of the form of the letter, but also reported common confusions, implying similarity was an additional factor. 

Despite the obvious practical applications for this type of research, by far the most common motivation for collecting letter similarity matrices has been to understand aspects of the perceptual system. Early researchers performed detailed psychophysical studies into the limits of letter perceivability with respect to numerous secondary variables (e.g., presentation time: Sanford, 1888; distance and size: Korte, 1923, Sanford, 1888; peripheral eccentricity: Dockeray & Pillsbury, 1910), adopting techniques that continue to be used today. Later researchers have attempted to use similarity and confusion matrices to understand other aspects of visual perception, such as representation and configurality (e.g., McGraw, Rehling, & Goldstone, 1994). The interest in studying the alphabet has even generalized beyond investigations of visual perception to include studies of tactile perception (Craig, 1979; Loomis, 1974), learning (Popp, 1964), choice behavior (Townsend, 1971a,b) and other relevant psychological phenomena. Such studies have often attempted to verify or test models of perceptual decision making. These models frequently included a description of the visual features used to represent letters, which in turn have produced similarity matrices of their own. 

Occasionally, these theoretic similarity matrices have been published, albeit sometimes in the form of a representational feature set that can be used to represent all letters (Geyer & DeWald, 1973, Gibson, 1969). These models also began to introduce response bias as a quantifiable measure (e.g., Townsend, 1971b). But typically, response bias was coupled only with letter-pair similarity to account for data patterns, abandoning the notion of perceivability. The almost universal presence of similarity-based confusions typically make a perceivability-bias model untenable prima facie, and because (for many models and experimental methods) perceivability is completely constrained once the entire similarity structure has been defined, perceivability has been viewed as redundant. In this view, perceivability is equated with a letter's mean similarity to the rest of the alphabet. 

Other theoretical measures of letter similarity have been developed that were not directly based on theories or models of the visual system, but rather examined the physical images representing the letters. For example, some researchers have used simple methods of letter congruency or overlap (e.g., Dunn-Rankin, Leton, & Shelton, 1968; Gibson, 1969) to measure letter similarity, whereas others have developed more elaborate techniques relying on Fourier decomposition (Blommaert, 1988; Coffin, 1978; Gervais, Harvey, & Roberts, 1984). These methods rarely make any commitments about biases or perceivability, and focus on producing objective measures of letter similarity. They offer the potential for validating novel alternative theories of visual letter perception, as they produce fundamentally different similarity spaces for letters. For example, overlap measures are perhaps most consistent with the hypothesis of Bouma (1971), who advocated the importance of letter and word shape (formal implementations of which have more recently been explored by Latecki, Lakaemper, & Wolter, 2005). 

Overlap methods, as well as Fourier methods, are consistent with global-to-local encoding hypotheses (e.g., Dawson & Harshman, 1986; Navon, 1977), and both of these strategies differ from the more dominent bottom-up featurecoding approach. In addition to objective similarity measures, recent work by Pelli, Burns, Farell, and Moore-Page (2006) and others (e.g., Majaj, Pelli, Kurshan, & Palomares, 2002) has reintroduced complexity measures that can be applied to individual letters, and thus may provide similar objective measures of perceivability. To our knowledge, such metrics have not been reported for entire alphabets, although Pelli reported summary measures across font faces.

51Due网站原创范文除特殊说明外一切图文著作权归51Due所有；未经51Due官方授权谢绝任何用途转载或刊发于媒体。如发生侵犯著作权现象，51Due保留一切法律追诉权。
更多论文代写范文欢迎访问我们主页 www.51due.com 当然有论文代写需求可以和我们24小时在线客服 QQ:800020041 联系交流。-X