Training redundant artificial neural networks: Imposing biology on technology

One biological principle that is often overlooked in the design of artificial neural networks (ANNs) is redundancy. Redundancy is the replication of processes within the brain. This paper examines the effects of redundancy on learning in ANNs when given either a function-approximation task or a pattern-classification task. The function-approximation task simulated a robotic arm reaching toward an object in two-dimensional space, and the pattern-classification task was detecting parity. Results indicated that redundant ANNs learned the pattern-classification problem much faster, and converge on a solution 100% of the time, whereas standard ANNs sometimes failed to learn the problem. Furthermore, when overall network error is considered, redundant ANNs were significantly more accurate than standard ANNs in performing the function-approximation task. These results are discussed in terms of the relevance of redundancy to the performance of ANNs in general, and the relevance of redundancy in biological systems in particular.     

Measurement of directional lever response reaction time with the Commodore 64

Routines are described for timing directional lever responses with the Commodore 64 micro-computer. Millisecond reaction timing is carried out with on-board hardware clocks, and lever responses are detected by monitoring the position of a joystick interfaced to a controller port. In a demonstration program, a machine code subroutine is used to measure the reaction times of lever responses in a spatial relations task. In addition, modifications to the demonstration program are suggested to adapt it for use with other tasks.     

Concurrent and predictive electrodermal correlates of symptomatology in recent-onset schizophrenic patients.

Electrodermal activity and symptomatology were interrelated in a group of 56 male and 13 female recent-onset schizophrenic patients. Electrodermal activity was indexed by the frequency of nonspecific skin conductances responses and the number of trials to habituation of the skin conductance orienting response. Symptomatology was assessed by the Brief Psychiatric Rating Scale (BPRS) on two separate test occasions. The first test occasion was during the inpatient period when psychotic symptoms were prevalent and medications were variable. The second test occasion was several months later during an outpatient period when symptoms were stabilized and medications held constant. Electrodermal activity was positively and significantly related to a number of symptoms in male patients, most reliably the BPRS factors Activation and Hostility/Suspiciousness. These relationships were most consistent during the outpatient period. Of particular theoretical interest, greater electrodermal activity during the inpatient period was associated with greater outpatient psychopathology. The results suggest that heightened inpatient electrodermal activity is predictive of poor short-term symptomatic recovery in recent-onset, acute, male schizophrenic patients.     

Lateral asymmetries in electrodermal responses to nonattended stimuli: a reply to Walker and Ceci

Walker and Ceci (1983) pose a number of interesting and potentially important criticisms and alternative explanations regarding the laterality finding and hypothesis of Dawson and Schell (1982). The present evaluation finds each of the criticisms and alternative explanations to be inadequate. We then distinguish between two alternative interpretations of the Dawson and Schell hypothesis, one based on inherent functional properties of the two hemispheres and the other based on the notion that each hemisphere is a partially independent pool of processing resources. Walker and Ceci's fundamental objection seems to apply only to the first interpretation.     

The role of language familiarity in voice identification

Four experiments examined the effects of language characteristics on voice identification. In Experiment 1, monolingual English listeners identified bilinguals' voices much better when they spoke English than when they spoke German. The opposite outcome was found in Experiment 2, in which the listeners were monolingual in German. In Experiment 3, monolingual English listeners also showed better voice identification when bilinguals spoke a familiar language (English) than when they spoke an unfamiliar one (Spanish). However, English-Spanish bilinguals hearing the same voices showed a different pattern, with the English-Spanish difference being statistically eliminated. Finally, Experiment 4 demonstrated that, for English-dominant listeners, voice recognition deteriorates systematically as the passage being spoken is made less similar to English by rearranging words, rearranging syllables, and reversing normal text. Taken together, the four experiments confirm that language familiarity plays an important role in voice identification.     

Scaling of apparent accentedness by magnitude estimation and sensory modality matching

In each of three experiments, 24 students judged the accentedness present in the speech of eight Spanish-English bilinguals.Ss gave magnitude estimations and also squeezed a hand dynamometer to indicate the amount of accentedness in the reading of an English passage by each of the speakers. There was significant agreement amongSs regarding the speech samples with each scaling method, and interscale agreement was good. Power functions fitted to the data had exponents falling in the range expected from earlier psychophysical studies. Scale values correlated significantly with the frequency of accented pronunciations by the speakers as judged by two independent judges. The use of these scaling methods for future research on linguistic features of accent and on the relation between accent and language attitudes is discussed.     

Sensory-modality opinion scales for individual subjects

In two sessions each, 16 subjects were asked to match apparent handgrip force and apparent duration in a counterbalanced order to the pronounceability of 12 trigrams in order to assess sensory modality opinion scales for individuals. Scales for individuals, like those for group data, were consistent across tasks and showed reliability across sessions. Subjects displayed strong agreement on the relative pronounceabilities of the trigrams according to Kendall’s test for concordance. A significant intersession correlation for subjects’ multimodality exponents indicated reliable differences in these exponents among individuals. Comparisons with psychophysical scales were made and discussed.     

A further consideration in the application of an analysis-of-variance model for the intrasubject replication design

It is argued that the analysis-of-variance model is inappropriate for assessing treatment effects in single-subject designs. In particular, such designs are demonstrated to violate the crucial assumption concerning the statistical independence of observations. Alternative methods of data analysis are suggested.     

Individual differences in power functions for a 1-week intersession interval

Two experiments examined correlations of the power function exponents of individual Ss obtained in each of two sessions. Half the Ss for any task performed second sessions immediately after the first, the other half after a week’s delay. In Experiment I, groups of 16 Ss gave magnitude estimations of apparent area, or else of area and loudness. In Experiment II, groups of 16 Ss made cross-modality matches of apparent time duration to area. Significant correlations in all cases indicated consistent and persisting S differences in exponents. The results are related to the findings of other studies of such individual differences.     

EFFECTS OF ECOLOGY AND SUBJECTIVE CULTURE ON INDIVIDUAL TRADITIONAL-MODERN ATTITUDE CHANGE,ACHIEVEMENT MOTIVATION,AND POTENTIAL FOR ECONOMIC DEVELOPMENT IN THE JAPANESE AND ESKIMO SOCIETIES1

L'auteur propose un modèle théorique permettant une analyse détaillée des potentialités à long terme que possède une société donnée dans le domaine du développement économique. Les variables indépendantes comprennent (a) l'écologie et le système de subsistance qu'elle impose, (b) le type plus ou moins sévère de socialisation, (c) la stratification sociale et la distribution de l'autorité et (e) la nature de l'influence et des contacts modernes. Les effets de ces facteurs sont observés sur les variables dépendantes suivantes: (a) l'attitude moderniste, (b) la tendance dominatrice, (c) la motivation pour le succès et (d) le degré de tolérance. Ce modèle a déjà trouvé une confirmation partielle dans plusieurs cultures par l'étude de la susceptibilité à la modification des attitudes traditionnelles. Il est ici appliqué à deux sociétés différant nettement sous certains des aspects cités: les Japonais et les Esquimaux d'Alaska.