Transitive choices by a simple, fully connected, backpropagation neural network: implications for the comparative study of transitive inference.

In search of the minimal requirements for transitive reasoning, a simple neural network was trained and tested on the non-verbal version of the conventional "five-term-series task" – a paradigm used with human adults, children and a variety of non-human species. The transitive performance of the network was analogous in several aspects to that reported for children and animals. The three effects usually associated with transitive choices i.e. "symbolic distance", "lexical marking" and "end-anchor", were also clearly shown by the neural network. In a second experiment, where the training conditions were manipulated, the network failed to match the behavioural pattern reported for human adults in the test following an ordered presentation of the premises. However, it mimicked young children's performance when tested with a novel comparison term. Although we do not intend to suggest a new model of transitive inference, we conclude, in line with other authors, that a simple error-correcting rule can generate transitive behaviour similar to the choice pattern of children and animals in the binary form of the five-term-series task without requiring high-order logical or paralogical abilities. The analysis of the training history and of the final internal structure of the network reveals the associative strategy employed. However, our results indicate that the scope of the associative strategy used by the network might be limited. The extent to which the conventional five-term-series task, in absence of appropriate manipulations of training and testing conditions, is suitable to detect cognitive differences across species is also discussed on the basis of our results. Accepted after revision: 29 May 2001 Electronic Publication     

When Hyperpropositions Meet ...

With each proposition P we associate a set of proposition (a hyperproposition) which determines the order in which one may retreat from accepting P, if one cannot fully hold on to P. We first describe the structure of hyperpropositions. Then we describe two operations on propositions, subtraction and merge, which can be modelled in terms of hyperpropositions. Subtraction is an operation that takes away part of the content of a proposition. Merge is an operation that determines the maximal consistent content of two propositions considered jointly. The merge operation gives rise to an inference relation which is, in a certain sense, optimally paraconsistent.     

Deductivism Within Pragma-Dialectics

The present paper elaborates a deductivist account of natural language argu-ment in the context of pragma-dialectics. It reviews earlier debates, criticizes some standard misconceptions in the literature, and argues that the identification and analysis of deductive argument schemes can be the basis of a compelling theory of argumentative discourse.     

Discovering the Alphabetic Principle: A Lost Opportunity for Educational Reform

We review the recently published report by the Committee on the Prevention of Reading Difficulties in Young Children from a behavior analytic perspective. The authors prescribe what children should learn to be successful readers based on the correlates of reading and the developmental progression of reading errors. The report offers no discussion of theoretical models that can be used by teachers and parents to reform reading instruction or specific procedures that have been shown to be effective at promoting various reading skills. We discuss these omissions in terms of the level of research included in the review, the authors' structural approach to synthesizing this research, and the constructivist orientation common to many early childhood educators. The paper concludes by identifying models and strategies from the behavior analysis literature that can be used to guide instruction and offers a modest suggestion for future reviews of reading research.     

Training requirements and retention characteristics of serial list organization by macaque monkeys

This work evaluated the prospect that organizational accounts of the retention of list information by monkeys might be an artifact of familiarity with conditional relationships. Seven sophisticated macaques were trained on four five-item lists. Each acquisition selectively excluded one of the internal conditional pairs of the typical four-problem sequence (AB,BC,CD,DE) that defines a five-item serially ordered list. Then, all possible novel pairings and the trained pairs appeared together in a test. After this, the previously omitted pair was trained and animals were retested. On all tasks, initial tests revealed little organization and much intersubject variability of characteristic choice strategies, but subsequent inclusion of all four conditional pairs always yielded organized serial choice. On both the four-problem tests and in a later retention, errors were directly related to interitem distance between the objects paired on test trials. These results helped to specify the conditions required for demonstration of non-human primate analogs of transitivity, and showed that even sophisticated monkeys organize information in retention only if they know all interitem relationships. Received: 7 October 1998 / Accepted after revision: 10 October 1999     

Bayesian estimation and testing of structural equation models

The Gibbs sampler can be used to obtain samples of arbitrary size from the posterior distribution over the parameters of a structural equation model (SEM) given covariance data and a prior distribution over the parameters. Point estimates, standard deviations and interval estimates for the parameters can be computed from these samples. If the prior distribution over the parameters is uninformative, the posterior is proportional to the likelihood, and asymptotically the inferences based on the Gibbs sample are the same as those based on the maximum likelihood solution, for example, output from LISREL or EQS. In small samples, however, the likelihood surface is not Gaussian and in some cases contains local maxima. Nevertheless, the Gibbs sample comes from the correct posterior distribution over the parameters regardless of the sample size and the shape of the likelihood surface. With an informative prior distribution over the parameters, the posterior can be used to make inferences about the parameters underidentified models, as we illustrate on a simple errors-in-variables model.We thank David Spiegelhalter for suggesting applying the Gibbs sampler to structural equation models to the first author at a 1994 workshop in Wiesbaden. We thank Ulf Böckenholt, Chris Meek, Marijtje van Duijn, Clark Glymour, Ivo Molenaar, Steve Klepper, Thomas Richardson, Teddy Seidenfeld, and Tom Snijders for helpful discussions, mathematical advice, and critiques of earlier drafts of this paper.     

Can you tickle yourself if you swap bodies with someone else?

The effect of the body transfer illusion on the perceived strength of self- and externally-generated “tickle” sensations was investigated. As expected, externally generated movement produced significantly higher ratings of tickliness than those associated with self-generated movements. Surprisingly, the body transfer illusion had no influence on the ratings of tickliness, suggesting that highly surprising, and therefore hard to predict, experiences of body image and first-person perspective do not abolish the attenuation of tickle sensations. In addition, evidence was found that a version of the rubber hand illusion exists within the body transfer illusion. We situate our findings within the larger debate over sensory attenuation: (1) there is an attenuation of prediction errors that depends upon the context in which sensory input is predicted (i.e., efference copy), and (2) sensory attenuation is a necessary consequence of self-generated movement irrespective of context (i.e., active inference). The results support the notion of active inference.     

Inference and coherence in causal-based artifact categorization

In four experiments, we tested conditions under which artifact concepts support inference and coherence in causal categorization. In all four experiments, participants categorized scenarios in which we systematically varied information about artifacts’ associated design history, physical structure, user intention, user action and functional outcome, and where each property could be specified as intact, compromised or not observed. Consistently across experiments, when participants received complete information (i.e., when all properties were observed), they categorized based on individual properties and did not show evidence of using coherence to categorize. In contrast, when the state of some property was not observed, participants gave evidence of using available information to infer the state of the unobserved property, which increased the value of the available information for categorization. Our data offers answers to longstanding questions regarding artifact categorization, such as whether there are underlying causal models for artifacts, which properties are part of them, whether design history is an artifact’s causal essence, and whether physical appearance or functional outcome is the most central artifact property.