Abductive reasoning in neural-symbolic systems

Abduction is or subsumes a process of inference. It entertains possible hypotheses and it chooses hypotheses for further scrutiny. There is a large literature on various aspects of non-symbolic, subconscious abduction. There is also a very active research community working on the symbolic (logical) characterisation of abduction, which typically treats it as a form of hypothetico-deductive reasoning. In this paper we start to bridge the gap between the symbolic and sub-symbolic approaches to abduction. We are interested in benefiting from developments made by each community. In particular, we are interested in the ability of non-symbolic systems (neural networks) to learn from experience using efficient algorithms and to perform massively parallel computations of alternative abductive explanations. At the same time, we would like to benefit from the rigour and semantic clarity of symbolic logic. We present two approaches to dealing with abduction in neural networks. One of them uses Connectionist Modal Logic and a translation of Horn clauses into modal clauses to come up with a neural network ensemble that computes abductive explanations in a top-down fashion. The other combines neural-symbolic systems and abductive logic programming and proposes a neural architecture which performs a more systematic, bottom-up computation of alternative abductive explanations. Both approaches employ standard neural network architectures which are already known to be highly effective in practical learning applications. Differently from previous work in the area, our aim is to promote the integration of reasoning and learning in a way that the neural network provides the machinery for cognitive computation, inductive learning and hypothetical reasoning, while logic provides the rigour and explanation capability to the systems, facilitating the interaction with the outside world. Although it is left as future work to determine whether the structure of one of the proposed approaches is more amenable to learning than the other, we hope to have contributed to the development of the area by approaching it from the perspective of symbolic and sub-symbolic integration.

Investigating long-term semantic priming of middle- and low-familiarity category exemplars

The authors examined long-term semantic priming of middle- and low-familiarity category exemplar targets as a result of participants' studying multiple middle- and low-familiarity exemplar primes within a category. The authors used a lexical decision task to test the hypothesis that semantic encoding of middle- and low-familiarity category exemplar primes causes long-term priming of middle- and low-familiarity category exemplar targets from the same category. Participants were 12 undergraduate students who judged semantic characteristics of category exemplar primes at study and at test and who provided lexical decisions on both primes and targets in a paradigm in which at least 6 intervening items separated primes and targets. Results supported the hypothesis by showing long-term semantic priming of middle- and low-familiarity category exemplar targets. The authors discuss results in terms of theories of semantic priming.     

Does Wechsler Intelligence Scale administration and scoring proficiency improve during assessment training?

Studies have found that Wechsler scale administration and scoring proficiency is not easily attained during graduate training. These findings may be related to methodological issues. Using a single-group repeated measures design, this study documents statistically significant, though modest, error reduction on the WAIS-III and WISC-III during a graduate course in assessment. The study design does not permit the isolation of training factors related to error reduction, or assessment of whether error reduction is a function of mere practice. However, the results do indicate that previous study findings of no or inconsistent improvement in scoring proficiency may have been the result of methodological factors. Implications for teaching individual intelligence testing and further research are discussed.     

A threat in the computer: the race implicit association test as a stereotype threat experience

Three experiments test whether the threat of appearing racist leads White participants to perform worse on the race Implicit Association Test (IAT) and whether self-affirmation can protect from this threat. Experiments 1 and 2 suggest that White participants show a stereotype threat effect when completing the race IAT, leading to stronger pro-White scores when the test is believed to be diagnostic of racism. This effect increases for domain-identified (highly motivated to control prejudice) participants (Experiment 2). In Experiment 3, self-affirmation inoculates participants against stereotype threat while taking the race IAT. These findings have methodological implications for use of the race IAT and theoretical implications concerning the malleability of automatic prejudice and the potential interpersonal effects of the fear of appearing racist.     

Child pornography offenses are a valid diagnostic indicator of pedophilia

This study investigated whether being charged with a child pornography offense is a valid diagnostic indicator of pedophilia, as represented by an index of phallometrically assessed sexual arousal to children. The sample of 685 male patients was referred between 1995 and 2004 for a sexological assessment of their sexual interests and behavior. As a group, child pornography offenders showed greater sexual arousal to children than to adults and differed from groups of sex offenders against children, sex offenders against adults, and general sexology patients. The results suggest child pornography offending is a stronger diagnostic indicator of pedophilia than is sexually offending against child victims. Theoretical and clinical implications are discussed.     

Phobic anxiety in 11 nations. Part I: Dimensional constancy of the five-factor model

The Fear Survey Schedule-III (FSS-III) was administered to a total of 5491 students in Australia, East Germany, Great Britain, Greece, Guatemala, Hungary, Italy, Japan, Spain, Sweden, and Venezuela, and submitted to the multiple group method of confirmatory analysis (MGM) in order to determine the cross-national dimensional constancy of the five-factor model of self-assessed fears originally established in Dutch, British, and Canadian samples. The model comprises fears of bodily injury-illness-death, agoraphobic fears, social fears, fears of sexual and aggressive scenes, and harmless animals fears. Close correspondence between the factors was demonstrated across national samples. In each country, the corresponding scales were internally consistent, were intercorrelated at magnitudes comparable to those yielded in the original samples, and yielded (in 93% of the total number of 55 comparisons) sex differences in line with the usual finding (higher scores for females). In each country, the relatively largest sex differences were obtained on harmless animals fears. The organization of self-assessed fears is sufficiently similar across nations to warrant the use of the same weight matrix (scoring key) for the FSS-III in the different countries and to make cross-national comparisons feasible. This opens the way to further studies that attempt to predict (on an a priori basis) cross-national variations in fear levels with dimensions of national cultures.     

The simpler syntax hypothesis

What roles do syntax and semantics have in the grammar of a language? What are the consequences of these roles for syntactic structure, and why does it matter? We sketch the Simpler Syntax Hypothesis, which holds that much of the explanatory role attributed to syntax in contemporary linguistics is properly the responsibility of semantics. This rebalancing permits broader coverage of empirical linguistic phenomena and promises a tighter integration of linguistic theory into the cognitive scientific enterprise. We suggest that the general perspective of the Simpler Syntax Hypothesis is well suited to approaching language processing and language evolution, and to computational applications that draw upon linguistic insights.     

The faculty of language: what's special about it?

We examine the question of which aspects of language are uniquely human and uniquely linguistic in light of recent suggestions by Hauser, Chomsky, and Fitch that the only such aspect is syntactic recursion, the rest of language being either specific to humans but not to language (e.g. words and concepts) or not specific to humans (e.g. speech perception). We find the hypothesis problematic. It ignores the many aspects of grammar that are not recursive, such as phonology, morphology, case, agreement, and many properties of words. It is inconsistent with the anatomy and neural control of the human vocal tract. And it is weakened by experiments suggesting that speech perception cannot be reduced to primate audition, that word learning cannot be reduced to fact learning, and that at least one gene involved in speech and language was evolutionarily selected in the human lineage but is not specific to recursion. The recursion-only claim, we suggest, is motivated by Chomsky's recent approach to syntax, the Minimalist Program, which de-emphasizes the same aspects of language. The approach, however, is sufficiently problematic that it cannot be used to support claims about evolution. We contest related arguments that language is not an adaptation, namely that it is "perfect," non-redundant, unusable in any partial form, and badly designed for communication. The hypothesis that language is a complex adaptation for communication which evolved piecemeal avoids all these problems.     

Concurrent validity of the Wechsler abbreviated scale of intelligence and the Kaufman brief intelligence test among psychiatric inpatients

This study examined the convergent and discriminant validity of the Wechsler Abbreviated Scale of Intelligence and the Kaufman Brief Intelligence Test for 85 psychiatric inpatients. The correlation between the WASI Full Scale and K-BIT Composite IQ scores was significant (r = .89, p < .001). Multitrait-multimethod analysis of the subtest scores showed that the K-BIT had higher internal consistency for its two subtests but, therefore, less differentiation of cognitive functioning than the brief Wechsler scale, as would be expected due to the larger number and diversity of the latter subtests. Correlations among the Wechsler scale subtests were lower than among those for the K-BIT, so the former may tap different cognitive functions and yield more clinically useful information than the latter. This brief Wechsler scale appears to be a valid screening measure of verbal, performance. and general intellectual ability for use with an inpatient psychiatric population when considerations of the setting or patient preclude administration of a longer measure of intellectual ability.