Influence of dorsal hippocampal lesions and MMP inhibitors on spontaneous recovery following a habituation/classical conditioning head–shake task

The present investigation combined a classical conditioning paradigm with a head–shake response (HSR) habituation task in order to evaluate the importance of dorsal hippocampal neural plasticity to spontaneous recovery. In the first experiment animals exhibited rapid HSR habituation (air stimulus to the ear) and an 85% level of spontaneous recovery following a 24 h inter-session interval. The addition of a brief tone prior to the air stimulus produced a similar pattern of habituation during the first session, but the level of spontaneous recovery was reduced (44%) during Session II. In a second experiment dorsal hippocampal lesioned rats placed on this tone/HSR paradigm responded with an 87% level of spontaneous recovery during Session II; while neocortex lesioned control rats indicated significantly reduced levels of spontaneous recovery (55%). In a third experiment bilateral injections of a general MMP inhibitor, FN-439, into the dorsal hippocampus resulted in high levels of spontaneous recovery (81%); while control rats injected with artificial cerebrospinal fluid displayed a significant attenuation of spontaneous recovery (45%). Finally, animals bilaterally injected with a specific MMP-3 inhibitor into the dorsal hippocampus indicated very similar results to those obtained following FN-439 injection. These findings indicate that animals prepared with dorsal hippocampal lesions, or injections with an MMP inhibitor, revealed an impaired association between the tone and air stimulus thus maximum spontaneous recovery was present 24 h later. Thus, it appears that the dorsal hippocampus influences habituation by conserving responses and reducing spontaneous recovery when a temporally contingent signaling cue is present.     

Age-related differences in the involvement of the prefrontal cortex in attentional control

We investigated the relative involvement of cortical regions supporting attentional control in older and younger adults during performance on a modified version of the Stroop task. Participants were exposed to two different types of incongruent trials. One of these, an incongruent-ineligible condition, produces conflict at the non-response level, while the second, an incongruent-eligible condition, produces conflict at both non-response and response levels of information processing. Greater attentional control is needed to perform the incongruent-eligible condition compared to other conditions. We examined the cortical recruitment associated with this task in an event-related functional magnetic resonance imaging paradigm in 25 older and 25 younger adults. Our results indicated that while younger adults demonstrated an increase in the activation of cortical regions responsible for maintaining attentional control in response to increased levels of conflict, such sensitivity and flexibility of the cortical regions to increased attentional control demands was absent in older adults. These results suggest a limitation in older adults’ capabilities for flexibly recruiting the attentional network in response to increasing attentional demands.     

Similarity and rules United: similarity- and rule-based processing in a single neural network

A central controversy in cognitive science concerns the roles of rules versus similarity. To gain some leverage on this problem, we propose that rule- versus similarity-based processes can be characterized as extremes in a multidimensional space that is composed of at least two dimensions: the number of features (Pothos, 2005) and the physical presence of features. The transition of similarity- to rule-based processing is conceptualized as a transition in this space. To illustrate this, we show how a neural network model uses input features (and in this sense produces similarity-based responses) when it has a low learning rate or in the early phases of training, but it switches to using self-generated, more abstract features (and in this sense produces rule-based responses) when it has a higher learning rate or is in the later phases of training. Relations with categorization and the psychology of learning are pointed out.     

Acquiring contextualized concepts: a connectionist approach

Conceptual knowledge is acquired through recurrent experiences, by extracting statistical regularities at different levels of granularity. At a fine level, patterns of feature co-occurrence are categorized into objects. At a coarser level, patterns of concept co-occurrence are categorized into contexts. We present and test CONCAT, a connectionist model that simultaneously learns to categorize objects and contexts. The model contains two hierarchically organized CALM modules (Murre, Phaf, & Wolters, 1992). The first module, the Object Module, forms object representations based on co-occurrences between features. These representations are used as input for the second module, the Context Module, which categorizes contexts based on object co-occurrences. Feedback connections from the Context Module to the Object Module send activation from the active context to those objects that frequently occur within this context. We demonstrate that context feedback contributes to the successful categorization of objects, especially when bottom-up feature information is degraded or ambiguous.     

Is it worth generating rules from neural network ensembles?

Although many authors generated comprehensible models from individual networks, much less work has been done in the explanation of ensembles. DIMLP is a special neural network model from which rules are generated at the level of a single network and also at the level of an ensemble of networks. We applied ensembles of 25 DIMLP networks to several datasets of the public domain and a classification problem related to post-translational modifications of proteins. For the classification problems of the public domain, the average predictive accuracy of rulesets extracted from ensembles of neural networks was significantly better than the average predictive accuracy of rulesets generated from ensembles of decision trees. By varying the architectures of DIMLP networks we found that the average predictive accuracy of rules, as well as their complexity were quite stable. The comparison to other rule extraction techniques applied to neural networks showed that rules generated from DIMLP ensembles gave very good results. In the last problem related to bioinformatics, the best result obtained by ensembles of DIMLP networks was also significantly better than the best result obtained by ensembles of decision trees. Thus, although neural networks take much longer to train than decision trees and also rules are generated at a greater computational cost (however, still polynomial), at least for several classification problems it was worth using neural network ensembles, as extracted rules were more accurate, on average. The DIMLP software is available for PC-Linux under http://us.expasy.org/people/Guido.Bologna.html.     

On the Poverty of Scientism, or: The Ineluctable Roughness of Rationality

If there is one rationality there must be a plurality of them. This conclusion follows, I argue, partly from the extreme and ineradicable vagueness of the fundamental concepts that every would-be rational explanation must presuppose. Logicistic/scientistic assaults on this vagueness are doomed to fail partly because they are unable to acknowledge the imaginative dimension of rational thought. Being limited to the play of "outward appearances," scientific investigations are also dependent on "inward imaginings" on their speculative side. The upshot is that schools of philosophy should be characterized by the kind of imaginary they adopt rather than by their logics. In which case, every attempt to get and tell something right about the world is bound to incorporate mythopoeic elements in its explanations.     

Neurophysiology and neuroanatomy of smooth pursuit in humans

Smooth pursuit eye movements enable us to focus our eyes on moving objects by utilizing well-established mechanisms of visual motion processing, sensorimotor transformation and cognition. Novel smooth pursuit tasks and quantitative measurement techniques can help unravel the different smooth pursuit components and complex neural systems involved in its control. The maintenance of smooth pursuit is driven by a combination of the prediction of target velocity and visual feedback about performance quality, thus a combination of retinal and extraretinal information that has to be integrated in various networks. Different models of smooth pursuit with specific in- and output parameters have been developed for a better understanding of the underlying neurophysiological mechanisms and to make quantitative predictions that can be tested in experiments. Functional brain imaging and neurophysiological studies have defined motion sensitive visual area V5, frontal (FEF) and supplementary (SEF) eye fields as core cortical smooth pursuit regions. In addition, a dense neural network is involved in the adjustment of an optimal smooth pursuit response by integrating also extraretinal information. These networks facilitate interaction of the smooth pursuit system with multiple other visual and non-visual sensorimotor systems on the cortical and subcortical level. Future studies with fMRI advanced techniques (e.g., event-related fMRI) promise to provide an insight into how smooth pursuit eye movements are linked to specific brain activation. Applying this approach to neurological and also mental illness can reveal distinct disturbances within neural networks being present in these disorders and also the impact of medication on this circuitry.