Visualization of learning in multilayer perceptron networks using principal component analysis

This paper is concerned with the use of scientific visualization methods for the analysis of feedforward neural networks (NNs). Inevitably, the kinds of data associated with the design and implementation of neural networks are of very high dimensionality, presenting a major challenge for visualization. A method is described using the well-known statistical technique of principal component analysis (PCA). This is found to be an effective and useful method of visualizing the learning trajectories of many learning algorithms such as back-propagation and can also be used to provide insight into the learning process and the nature of the error surface.

Local-global processing in early-onset schizophrenia: Evidence for an impairment in shifting the spatial scale of attention

In this study we report the results of two experiments on visual attention conducted with patients with early-onset schizophrenia. These experiments investigated the effect of irrelevant spatial-scale information upon the processing of relevant spatial-scale information, and the ability to shift the spatial scale of attention, across consecutive trials, between different levels of the hierarchical stimulus. Twelve patients with early-onset schizophrenia and matched controls performed local-global tasks under: (1) directed attention conditions with a consistency manipulation and (2) divided-attention conditions. In the directed-attention paradigm, the early-onset patients exhibited the normal patterns of global advantage and interference, and were not unduly affected by the consistency manipulation. Under divided-attention conditions, however, the early-onset patients exhibited a local-processing deficit. The source of this local processing deficit lay in the prolonged reaction time to local targets, when these had been preceded by a global target, but not when preceded by a local target. These findings suggest an impaired ability to shift the spatial scale of attention from a global to a local spatial scale in early-onset schizophrenia. (C) 2003 Elsevier Science (USA). All rights reserved.

Effects of alcohol exposure during early life on neuron numbers in the rat hippocampus. I. Hilus neurons and granule cells

We previously showed that 16-day-old rats exposed to a relatively high dose of ethanol at 10-15 postnatal days of age have fewer neurons in the hilus region of the hippocampus compared with controls. Dentate gyrus granule cell numbers, however, showed no statistically significant changes attributable to the ethanol treatment. It is possible that some of the changes in brain morphology, brought about as a result of the exposure to ethanol during early life, may not be manifested until later in life. This question has been further addressed in an extension to our previous study. Wistar rats were exposed to a relatively high daily dose of ethanol on postnatal days 10-15 by placement in a chamber containing ethanol vapour, for 3 h/day. The blood ethanol concentration was found to be similar to430 mg/dl at the end of the period of exposure. Groups of ethanol-treated (ET), separation control (SC), and mother-reared control (MRC) rats were anaesthetised and killed either at 16 or 30 days of age by perfusion with phosphate-buffered 2.5% glutaraldehyde. The Cavalieri principle and the physical disector methods were used to estimate, respectively, the regional volumes and neuron cell numerical densities in the hilus and granule cell regions of the dentate gyrus. The total numbers of neurons in the hilus region and granule cell layer were computed from these estimates. It was found that 16-day-old animals had 398,000-441,000 granule cells, irrespective of group. The numbers of granule cells increased such that by 30 days of age, rats had 487,000-525,500 granule cells. However, there were no significant differences between ethanol-treated rats and their age-matched controls in granule cell numbers. In contrast, ethanol-treated rats had slightly but significantly fewer neurons in the hilus region than did control animals at 16 days of age, but not at 30 days of age. Therefore, it appears that a short period of ethanol exposure during early life can have effects on neuron numbers of some hippocampal neurons, but not others. The effects on hilar neuron numbers, observed as a result of such short periods of ethanol treatment, appeared to be transitory. (C) 2003 Wiley-Liss, Inc.

Constitutively Active Mutant D816VKit Induces Megakayocyte and Mast Cell Differentiation of Early Haemopoietic Cells from Murine Foetal Liver

Mutations of Kit at position D816 have been implicated in mastocytosis, acute myeloid leukaemia and germ cell tumours. Expression of this mutant Kit in cell lines results in factor-independent growth, differentiation and increased survival in vitro and tumourigenicity in vivo. Mutant D816VKit and wild-type Kit were expressed in murine primary haemopoietic cells and grown in stem cell factor (SCF) or the absence of factors. Expression of D816VKit did not lead to transformation as assessed by a colony assay, but resulted in enhanced differentiation of cells when compared to control cells. D816VKit induced an increase in the number of cells differentiating along the megakaryocyte lineage in the absence of factors. SCF had an added effect with an increase in differentiation of mast cells. Expression of wild-type Kit in the presence of SCF also failed to cause transformation and induced differentiation of mast cells and megakaryocytes. We conclude that constitutive expression of D816VKit in primary haemopoietic cells is not a sufficient transforming stimulus but leads to the survival and maturation of cells whose phenotype is influenced by the presence of SCF. (C) 2003 Elsevier Science Ltd. All rights reserved.

Effects of settler size and density on early post-settlement survival of Ciona intestinalis in the field

Effects of variation in larval quality on post-metamorphic performance in marine invertebrates are increasingly apparent. Recently, it has been shown that variation in offspring size can also strongly affect post-settlement survival, but variation in environmental conditions can mediate this effect. The quality of habitat into which marine invertebrate larvae settle can vary markedly, and 1 influence on quality is the number of conspecifics present. We tested the effects of settler size and settler density on early (1 wk after settlement) post-settlement survival in the field for the solitary ascidian Ciona intestinalis. Larger settlers survived better than smaller settlers, within and among groups of siblings. Increases in the density of settlers decreased survival, but the density-dependent effects were much stronger for smaller settlers. We suggest that larger settlers are better able to cope with intra-specific competition because they have greater energetic reserves or a greater capacity to feed than smaller settlers.

Is affective learning really resistant to extinction? Stimulus ratings and affective priming give different answers

Affective learning, the learning of likes and dislikes, is proposed to differ from signal learning, the learning of relationships between events. However, affective learning research varies in the methodology used, and in addition, researchers concerned primarily with affective learning tend to use different paradigms from those concerned with signal learning. The current research used an affective priming task in addition to verbal ratings to assess changes in the valence of neutral geometric shapes in an aversive differential conditioning procedure. After acquisition, affective learning was present as indexed by ratings and affective priming, whereas after extinction, affective learning remained significant only in the ratings. This study suggests that different measures of affective learning may be differentially sensitive to valence, which has implications for studies that employ verbal ratings as the sole measure of affective learning. Moreover, there is no evidence from the current study that affective learning differs from signal learning.

Attenuation of perceptual asymmetries in patients with early-onset schizophrenia: Evidence in favour of reduced hemispheric differentiation in schizophrenia?

Lateral biases in visual perception have been demonstrated in normal individuals and in patients with unilateral brain lesions. It has been suggested that the absence of structural and functional asymmetries in schizophrenia could be due to a failure in lateralisation that may be most pronounced in those patients whose illness onset is at an early age. Here we examined lateral biases in patients with schizophrenia of an early onset (N = 21) and a late onset.(N = 19), and their respective age-matched control groups, using the greyscales task, a sensitive measure of asymmetries in visual processing. The stimuli consisted of two rectangles, one above the other, shaded in opposite directions and matched overall for darkness. Participants judged which of the two rectangles looked darker overall. Previous studies using this task in healthy participants have reported a reliable bias, such that the rectangle with the darker end on the left is selected preferentially. Whereas the late-onset patients in this study exhibited a perceptual bias of similar direction and magnitude to that of controls, this was not the case for the early-onset patients, who exhibited significantly less bias than their control group. The reduced perceptual bias seen in the early-onset group, but not the late-onset group, suggests an attenuation of right hemisphere mechanisms dedicated to processing vistiospatial information. The attenuated perceptual asymmetry in the early-onset group only may be consistent with the view that (i) an earlier illness onset reflects a greater loss of hemispheric differentiation and (ii) reduced functional asymmetries in the early-onset group are a manifestation of a failure to allocate functions to one or the other hemisphere.

Mutations in the DLG3 gene cause nonsyndromic X-linked mental retardation

We have identified truncating mutations in the human DLG3 ( neuroendocrine dlg) gene in 4 of 329 families with moderate to severe X-linked mental retardation. DLG3 encodes synapse-associated protein 102 (SAP102), a member of the membrane-associated guanylate kinase protein family. Neuronal SAP102 is expressed during early brain development and is localized to the postsynaptic density of excitatory synapses. It is composed of three amino-terminal PDZ domains, an src homology domain, and a carboxyl-terminal guanylate kinase domain. The PDZ domains interact directly with the NR2 subunits of the NMDA glutamate receptor and with other proteins responsible for NMDA receptor localization, immobilization, and signaling. The mutations identified in this study all introduce premature stop codons within or before the third PDZ domain, and it is likely that this impairs the ability of SAP102 to interact with the NMDA receptor and/or other proteins involved in downstream NMDA receptor signaling pathways. NMDA receptors have been implicated in the induction of certain forms of synaptic plasticity, such as long-term potentiation and long-term depression, and these changes in synaptic efficacy have been proposed as neural mechanisms underlying memory and learning. The disruption of NMDA receptor targeting or signaling, as a result of the loss of SAP102, may lead to altered synaptic plasticity and may explain the intellectual impairment observed in individuals with DLG3 mutations.

The early sperm gets the good egg: mating order effects in free spawners

Mating order can have important consequences for the fertilization success of males whose ejaculates compete to fertilize a clutch of eggs. Despite an excellent body of literature on mating-order effects in many animals, they have rarely been considered in marine free-spawning invertebrates, where both sexes release gametes into the water column. In this study, we show that in such organisms, mating order can have profound repercussions for male reproductive success. Using in vitro fertilization for two species of sea urchin we found that the 'fertilization history' of a clutch of eggs strongly influenced the size distribution of unfertilized eggs, and consequently the likelihood that they will be fertilized. Males that had first access to a batch of eggs enjoyed elevated fertilization success because they had privileged access to the largest and therefore most readily fertilizable eggs within a clutch. By contrast, when a male's sperm were exposed to a batch of unfertilized eggs left over from a previous mating event, fertilization rates were reduced, owing to smaller eggs remaining in egg clutches previously exposed to sperm. Because of this size-dependent fertilization, the fertilization history of eggs also strongly influenced the size distribution of offspring, with first-spawning males producing larger, and therefore fitter, offspring. These findings suggest that when there is variation in egg size, mating order will influence not only the quantity but also the quality of offspring sired by competing males.

Extinction of affective learning: No evidence for dual process accounts of human learning

The acquisition and extinction of affective valence to neutral geometrical shape conditional stimuli was investigated in three experiments. Experiment 1 employed a differential conditioning procedure with aversive shock USs. Differential electrodermal responding was evident during acquisition and lost during extinction. As indexed by verbal ratings, the CS1 acquired negative valence during acquisition,which was reduced after extinction. Affective priming, a reaction time based demand free measure of stimulus valence, failed to provide evidence for affective learning. Experiment 2 employed pictures of happy and angry faces as USs.Valence ratings after acquisitionweremore positive for theCS paired with happy faces (CS-H) and less positive for the CS paired with angry faces (CS-A) than during baseline. Extinction training reduced the extent of acquired valence significantly for both CSs, however, ratings of the CS-A remained different from baseline. Affective priming confirmed these results yielding differences between CS-A and CS-H after acquisition for pleasant and unpleasant targets, but for pleasant targets only after extinction. Experiment 3 replicated the design of Experiment 2, but presented the US pictures backwardly masked. Neither rating nor affective priming measures yielded any evidence for affective learning. The present results confirm across two different experimental procedures that, contrary to predictions from dual process accounts of human learning, affective learning is subject to extinction.