Multiattribute Decision Making in Context: A Dynamic Neural Network Methodology

A theoretical structure for multiattribute decision making is presented, based on a dynamical system for interactions in a neural network incorporating affective and rational variables. This enables modeling of problems that elude two prevailing economic decision theories: subjective expected utility theory and prospect theory. The network is unlike some that fit economic data by choosing optimal weights or coefficients within a predetermined mathematical framework. Rather, the framework itself is based on principles used elsewhere to model many other cognitive and behavioral data, in a manner approximating how humans perform behavioral functions. Different, interconnected modules within the network encode (a) attributes of objects among which choices are made, (b) object categories, (c) and goals of the decision maker. An example is utilized to simulate the actual consumer choice between old and new versions of Coca-Cola. Potential applications are also discussed to market decisions involving negotiations between participants, such as international petroleum traders.     

Confidence in judgments based on incomplete information: An investigation using both hypothetical and real gambles

This study assessed how confidence in judgments is affected by the need to make inferences about missing information. Subjects indicated their likelihood of taking each of a series of gambles based on both probability and payoff information or only one of these sources of information. They also rated their confidence in each likelihood judgment. Subjects in the Explicit Inference condition were asked to explicitly estimate the values of missing information before making their responses while subjects in the Implicit Inference condition were not. The manner in which probability information was framed was also manipulated. Experiment 1 employed hypothetical gambles and Experiment 2 employed gambles with real money. Expressed likelihood of taking gambles was higher when probability was phrased in terms of '% chance of winning' rather than '% chance of losing', but this difference was somewhat less with real gambles than with hypothetical gambles. Confidence ratings in each experiment were actually higher on incomplete information trials than on complete information trials in the Explicit Inference condition. Results were related to the general issue of confidence in judgments.     

Repairing the Microsoft BASIC RND function

Difficulties with the Microsoft BASIC function for generating random numbers (RND) are cited and new problems reported. Fixes are suggested that allow use of some of the speed of the original function and that produce greatly improved pseudorandom numbers. A published generator is combined with the RND function as a compromise solution.     

Apple slide tachistoscope

We outline a BASIC program and subsidiary machine language timing routines that enable the Apple II series of computers to be employed as control devices for the tachistoscopic presentation of 35-mm slides and the collection of response latencies. The program is functional for experiments that require pictorial stimuli to be presented in up to three separate fields, with precise display times, precise interstimulus and intertriai intervals, and the collection of manual or vocal reaction times with millisecond accuracy. The major advantages of this system are that initial setup costs are substantially lower than those for other similar systems (e.g., Loftus, Gillispie. Tigre, & Nelson, 1984) and that user customization of the program may be accomplished by anyone possessing rudimentary knowledge of the BASIC programming language.     

Seeing through CORNEA

For useful comments on an early draft of this paper, I wish to thank Andrew Cortens, William Hasker, Richard Swinburne and Mark Webb. I am especially grateful to William Alston, Frances Howard-Snyder and John O'Leary-Hawthorne for extended discussion on several pertinent matters.     

Information Load: A Test of an Inverted-U Hypothesis with Hourly and Salaried Employees

We surveyed 714 hourly and 516 salaried employees of a midwestern telephone company to test the effects of information load on work-related outcomes. Using curvilinear regression analyses, we found support for our hypothesis that employees are less satisfied with outcomes as the load of information deviates positively or negatively from some level. We also predicted, and found, that this quadratic function was more prevalent in the hourly group than in the salaried group. Implications, future research directions, and limitations of the present study are discussed.     

Regular and irregular inflection in the acquisition of German noun plurals.

In this paper we study the acquisition of German noun plurals in relation to the question of how children represent regular and irregular inflection. Pinker and Prince (1992) have demonstrated several dissociations between regular and irregular inflection in the English past tense system. However, in English, the default status of -ed is confounded with its high frequency; therefore inflectional systems other than English past tense formation must be examined. The noun plural system in German is particularly interesting, because most nouns have irregular plurals in German and the regular (default) plural is less frequent than several of the irregular plurals. Thus it is unclear how a language learner determines whether German even has a regular plural, and if so what form it takes. Based on longitudinal data from impaired and unimpaired monolingual German-speaking children, we find a striking, statistically significant correlation: plural affixes that are used in overregularizations, namely -n or -s, are left out within compounds. This correlation shows that even impaired children are sensitive to the distinction between regular and irregular morphology. We propose a linguistic analysis of the correlation in terms of Kiparsky's (1982, 1985) level-ordering model plus an additional ordering condition on affixes: default (regular) affixes cannot serve as input to compounding processes.     

Parallel distributed processing and neuropsychology: A neural network model of Wisconsin card sorting and verbal fluency

Neural networks can be used as a tool in the explanation of neuropsychological data. Using the Hebbian Learning Rule and other such principles as competition and modifiable interlevel feedback, researchers have successfully modeled a widely used neuropsychological test, the Wisconsin Card Sorting Test. One of these models is reviewed here and extended to a qualitative analysis of how verbal fluency might be modeled, which demonstrates the importance of accounting for the attentional components of both tests. Difficulties remain in programming sequential cognitive processes within a parallel distributed processing (PDP) framework and integrating exceedingly complex neuropsychological tests such as Proverbs. PDP neural network methodology offers neuropsychologists co-validation procedures within narrowly defined areas of reliability and validity.