Kevin Burns

Bayesian inference in disputed authorship: A case study of cognitive errors and a new system for decision support

Bayesian inference provides a formal framework for assessing the odds of hypotheses in light of evidence. This makes Bayesian inference applicable to a wide range of diagnostic challenges in the field of chance discovery, including the problem of disputed authorship that arises in electronic commerce, counter-terrorism and other forensic applications. For example, when two documents are so similar that one is likely to be a hoax written from the other, the question is: Which document is most likely the source and which document is most likely the hoax? Here I review a Bayesian study of disputed authorship performed by a biblical scholar, and I show that the scholar makes critical errors with respect to several issues, namely: Causal Basis, Likelihood Judgment and Conditional Dependency. The scholars errors are important because they have a large effect on his conclusions and because similar errors often occur when people, both experts and novices, are faced with the challenges of Bayesian inference. As a practical solution, I introduce a graphical system designed to help prevent the observed errors. I discuss how this decision support system applies more generally to any problem of Bayesian inference, and how it differs from the graphical models of Bayesian Networks.

Atoms of EVE′: A Bayesian basis for esthetic analysis of style in sketching

At its root level, style is actually an esthetic agreement between people. The question is, how can esthetic agreements be modeled and measured in artificial intelligence? This paper offers a formal theory called EVE′ and applies it to a novel test bed of dynamic drawings that combine features of music and sketching. The theory provides mathematical measures of expectations, violations, and explanations, which are argued to be the atomic components of the esthetic experience. The approach employs Bayesian methods to extend information measures proposed in other research. In particular, it is shown that information theory is useful at an entropic level to measure expectations (E) of signals and violations (V) of expectations, but that Bayesian theory is needed at a semantic level to measure explanations (E′) of meaning for the signals. The entropic and semantic measures are then combined in further measures of tension and pleasure at an esthetic level that is actually style.

Mental Models of Line Drawings

Mental models are internal representations of world structure, used to accomplish cognitive tasks. I postulate specific representations (of objects and images) and associated context (of world and view) for mental models of line drawings. I then analyze the representations and context to predict specific perceptual modes, including the relative strengths of these modes. The predicted modes are supported by a well-known example [from Rock, 1983 The Logic of Perception (Cambridge, MA: MIT Press)] where object perception changes with image orientation.

The Structure of Style

Style is a fundamental and ubiquitous aspect of the human experience: Everyone instantly and constantly assesses people and things according to their individual styles, academics establish careers by researching musical, artistic, or architectural styles, and entire industries maintain themselves by continuously creating and marketing new styles. Yet what exactly style is and how it works are elusive: We certainly know it when we see it, but there is no shared and clear understanding of the diverse phenomena that we call style. The Structure of Style explores this issue from a computational viewpoint, in terms of how information is represented, organized, and transformed in the production and perception of different styles. New computational techniques are now making it possible to model the role of style in the creation of and response to human artifactsand therefore to develop software systems that directly make use of style in useful ways. Argamon, Burns, and Dubnov organize the research they have collected in this book according to the three roles that computation can play in stylistics. The first section of the book, Production, provides conceptual foundations by describing computer systems that create artifactsmusical pieces, texts, artworksin different styles. The second section, Perception, explains methods for analyzing different styles and gleaning useful information, viewing style as a form of communication. The final section, Interaction, deals with reciprocal interaction between style producers and perceivers, in areas such as interactive media, improvised musical accompaniment, and game playing. The Structure of Style is written for researchers and practitioners in areas including information retrieval, computer art and music, digital humanities, computational linguistics, and artificial intelligence, who can all benefit from this comprehensive overview and in-depth description of current research in this active interdisciplinary field.

EVE′s energy in aesthetic experience: a Bayesian basis for haiku humour

EVE′ is a mathematical model of aesthetic experience, founded on Bayesian probability and grounded in cognitive psychology. The model addresses three stages of cognitive processing: starting with expectations (E) of signals, which produce pleasure; followed by violations (V) of expectations, which produce tension; and ending with explanations (E ′) of meaning, which produce pleasure-prime. This creates a tradeoff to be optimized in the design of artworks, because an audience can only obtain pleasure-prime from E ′ at the expense of pleasure from E. A total measure of aesthetic pleasure X is derived as the product of two terms, X = Y * Z, where Y is a measure of entropy in violations, and Z is a measure of how completely entropy is converted to energy in explanations. The model is applied to the art of humour in haiku form, by evaluating one poem in detail and by generating additional poems as exemplars.

Computing the creativeness of amusing advertisements: A Bayesian model of Burma-Shave's muse

Abstract How do humans judge the creativeness of an artwork or other artifact? This article suggests that such judgments are based on the pleasures of an aesthetic experience, which can be modeled as a mathematical product of psychological arousal and appraisal. The arousal stems from surprise, and is computed as a marginal entropy using information theory. The appraisal assigns meaning, by which the surprise is resolved, and is computed as a posterior probability using Bayesian theory. This model is tested by obtaining human ratings of surprise, meaning, and creativeness for artifacts in a domain of advertising design. The empirical results show that humans do judge creativeness as a product of surprise and meaning, consistent with the computational model of arousal and appraisal. Implications of the model are discussed with respect to advancing artificial intelligence in the arts as well as improving the computational evaluation of creativity in engineering and design.

Bayesian Boxes: A Colored Calculator for Picturing Posteriors

The need for Bayesian inference arises in military intelligence, medical diagnosis and many other practical applications. The problem is that human inferences are generally conservative by Bayesian standards, i.e., people fail to extract all the certainty they should from the data they are given. Here I present a diagram called ”Bayesian Boxes” designed to correct conservatism. The diagram uses colored lines and boxes to illustrate the Bayesian posterior and the underlying principle. Compared to other diagrams, Bayesian Boxes is novel in illustrating the conceptual features (e.g., hypotheses and evidence) and computational structure (e.g., products and ratio) of Bayesian inference.

Style in Poker

Style is the cognitive basis for behavior in game play. This is because mental limits force human beings to act based on reduced rule-sets, which in game parlance are called styles, rather than exhaustive enumeration of options, which in game theory are called strategies. This paper explores the computational underpinnings of style in poker, by analyzing three versions of a two-player game ranging from very simple to rather complex, using theoretical analyses and deterministic calculations. The results show that simple styles derived from commonsense reasoning often closely approximate the Nash equilibrium strategies. Moreover, styles often outperform Nash equilibrium strategies against sub-optimal strategies, and some styles are seen to be nearly maximally super-optimal - i.e., almost equivalent to a player who is perfectly Bayesian. This is an important finding with respect to the practical tradeoff between effort and winnings, because the computational implementation of styles is trivial compared to that of strategies

Fun in Slots

People play games for fun. Yet we are lacking a fundamental understanding of what fun is and how fun works in games and other media. For example, why do thousands of people spend millions of dollars playing slot machines, especially when most know they will lose money in the long run? To answer this question, The author presents an aesthetic analysis of slot play using a Bayesian-information approach. The finding is that fun in slots can be seen as arising from a difference in information gained from good versus bad outcomes. This difference is modeled by marginal entropies and the result is a measure of fun in slot play, showing for what range of payoff probabilities slots are fun and at what probability they are most fun. The approach is extended to games of skill and the same Bayesian-information theory is used to derive computational measures of fun in these games

Strategic Style in Pared-Down Poker

This chapter deals with the manner of making diagnoses and decisions, called strategic style, in a gambling game called Pared-down Poker. The approach treats style as a mental mode in which choices are constrained by expected utilities. The focus is on two classes of utility, i.e., money and effort, and how cognitive styles compare to normative strategies in optimizing these utilities. The insights are applied to real-world concerns like managing the war against terror networks and assessing the risks of system failures. After “Introducing the Interactions” involved in playing poker, the contents are arranged in four sections, as follows. “Underpinnings of Utility” outlines four classes of utility and highlights the differences between them: economic utility (money), ergonomic utility (effort), informatic utility (knowledge), and aesthetic utility (pleasure). “Inference and Investment” dissects the cognitive challenges of playing poker and relates them to real-world situations of business and war, where the key tasks are inference (of cards in poker, or strength in war) and investment (of chips in poker, or force in war) to maximize expected utility. “Strategies and Styles” presents normative (optimal) approaches to inference and investment, and compares them to cognitive heuristics by which people play poker–-focusing on Bayesian methods and how they differ from human styles. The normative strategy is then pitted against cognitive styles in head-to-head tournaments, and tournaments are also held between different styles. The results show that style is ergonomically efficient and economically effective, i.e., style is smart. “Applying the Analysis” explores how style spaces, of the sort used to model individual behavior in Pared-down Poker, might also be applied to real-world problems where organizations evolve in terror networks and accidents arise from system failures.