PubMed Health⌕ Search

SEARCH · PubMed Health

Results for “Contrastive learning”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 397 records · Page 22Linked to original sources
PubMed2004

Shape discrimination by wasps (Paravespula germanica) at the food source: generalization among various types of contrast.

Wasps ( Paravespula germanica) were trained and tested at an artificial feeding site, using convex shapes that produced colour contrast, luminance contrast, or motion contrast against the background. With each of the three types of contrast, we tested the wasps' capacity to discriminate the learned shape from novel shapes. In addition, in each experiment we tested the wasps' capability to recognize the learned shape when it offered a different type of contrast than that it had during the training. With the coloured shapes, a side-glance at the colour discrimination performance of the wasps was possible in addition. Wasps are found to discriminate between a variety of convex shapes regardless of the type of contrast that they produce against the background. Mainly, they discriminate the learned shape from novel shapes even if the colour of the shapes or the type of contrast they produce against the background is altered in the test. Thus, wasps generalize the learned shape from one colour to another, as well as between colour contrast, luminance contrast, and motion contrast.

Animals↗
PubMed1994

Maudsley reactive and non-reactive rats differ in exploratory behavior but not in learning.

The Maudsley reactive (MR) and Maudsley non-reactive (MNR) inbred rat strains were created as an animal model of anxiety, based on open field behavior. We wished to determine whether previously described characteristic open field behavior and learning deficits in conditioned avoidance could be generalized to other paradigms of exploratory behavior (such as the staircase test) and learning (such as the T-maze swim test). As with other open field paradigms, the MR rats showed stable and large differences in the staircase test, compared with the MNR rats. In contrast, large and stable learning differences in conditioned avoidance and T-maze swim tests were not observed. We conclude that the MR rats exhibit exploratory behavior that is inhibited in a wide variety of paradigms, but learning deficits are not major characteristics of the phenotype. The inhibited exploratory behavior is a stable characteristic with large inter-strain differences, making measures of open field behavior suitable for quantitative trait loci analysis to determine the genes which explain these large and stable strain differences. Identification of these genes could provide clues to the genetic origins of some human anxiety disorders.

Animals↗
PubMed2004

Dissecting the pathological effects of human Abeta40 and Abeta42 in Drosophila: a potential model for Alzheimer's disease.

Accumulation of amyloid-beta (Abeta) peptides in the brain has been suggested to be the primary event in sequential progression of Alzheimer's disease (AD). Here, we use Drosophila to examine whether expression of either the human Abeta40 or Abeta42 peptide in the Drosophila brain can induce pathological phenotypes resembling AD. The expression of Abeta42 led to the formation of diffused amyloid deposits, age-dependent learning defects, and extensive neurodegeneration. In contrast, expression of Abeta40 caused only age-dependent learning defects but did not lead to the formation of amyloid deposits or neurodegeneration. These results strongly suggest that accumulation of Abeta42 in the brain is sufficient to cause behavioral deficits and neurodegeneration. Moreover, Drosophila may serve as a model for facilitating the understanding of molecular mechanisms underlying Abeta toxicity and the discovery of novel therapeutic targets for AD.

Aging↗
PubMed2005

Consideration of ergonomic aspects in the development of a new endoscopic navigation system.

During the development of new navigation systems, the ergonomic aspects of the accompanying software are seldom taken into account. The VN system is a navigation system working with real and previously stored neuroendoscopic images. The latter enable the module to execute virtual back movement, e.g. in case of bleeding. Several other modules are offered. The aim of the study was the ergonomic analysis of different modules and their learning curves in cadaveric heads. The endoscope was navigated by the neurosurgeon while another person operated the computer. A total of 128 experiments were performed with two software versions. When the landmark tracking module was used, a real learning curve could be observed. By contrast, testing the measurement module did not produce a learning curve. A significant reduction of the time required by the three modules investigated could be observed with software version 2. The module for virtual back movement works best with a minimum time of 20 s for image storage. During machine-human interactions ergonomic software use is important, especially if operative procedures are performed.

Clinical Competence↗
PubMed2005

Trace eyeblink conditioning requires the hippocampus but not autophosphorylation of alphaCaMKII in mice.

Little is known about signaling mechanisms underlying temporal associative learning. Here, we show that mice with a targeted point mutation that prevents autophosphorylation of alphaCaMKII (alphaCaMKII(T286A)) learn trace eyeblink conditioning normally. This forms a sharp contrast to the severely impaired spatial learning in the water maze and contextual fear conditioning observed in alphaCaMKII(T286A) mutants. Importantly, hippocampal lesions impaired trace eyeblink conditioning in alphaCaMKII(T286A) mice, suggesting a potential role of hippocampal alphaCaMKII-independent mechanisms. These results indicate that hippocampal signaling mechanisms that underlie temporal associative learning as assessed by trace eyeblink conditioning may differ from those of spatial and contextual learning.

Animals↗
PubMed2003

Learning-induced neural plasticity associated with improved identification performance after training of a difficult second-language phonetic contrast.

Adult native Japanese speakers have difficulty perceiving the English /r-l/ phonetic contrast even after years of exposure. However, after extensive perceptual identification training, long-lasting improvement in identification performance can be attained. This fMRI study investigates localized changes in brain activity associated with 1 month of extensive feedback-based perceptual identification training by native Japanese speakers learning the English /r-l/ phonetic contrast. Before and after training, separate functional brain imaging sessions were conducted for identification of the English /r-l/ contrast (difficult for Japanese speakers), /b-g/ contrast (easy), and /b-v/ contrast (difficult), in which signal-correlated noise served as the reference control condition. Neural plasticity, denoted by exclusive enhancement in brain activity for the /r-l/ contrast, does not involve only reorganization in brain regions concerned with acoustic-phonetic processing (superior and medial temporal areas) but also the recruitment of additional bilateral cortical (supramarginal gyrus, planum temporale, Broca's area, premotor cortex, supplementary motor area) and subcortical regions (cerebellum, basal ganglia, substantia nigra) involved with auditory-articulatory (perceptual-motor) mappings related to verbal speech processing and learning. Contrary to what one may expect, brain activity for perception of a difficult contrast does not come to resemble that of an easy contrast as learning proceeds. Rather, the results support the hypothesis that improved identification performance may be due to the acquisition of auditory-articulatory mappings allowing for perception to be made in reference to potential action.

Brain↗
PubMed1993

Recovery of frontal cortex-mediated visual behaviors following neurotrophic rescue of axotomized neurons in medial frontal cortex.

Unilateral lesions extending across the boundary region of visual and parietal cortex in adult rats result in the death of 20-35% of neurons in layers II-III of the caudal third of medial frontal cortex ipsilaterally, a neuron population labeled with 3H-thymidine on the 19th day of gestation (E19). Additionally, there is a consistent 15% loss of these labeled neurons in an area between 50% and 60% of the distance along the caudal-rostral extent of medial frontal cortex, an area that may function analogously to the frontal eye field of primates. All of these neurons are rescued from axotomy-induced death by delivering into the posterior cortex lesion cavity for 2 weeks a macromolecular fraction of culture medium conditioned by embryonic primordia of the frontal-occipital pathway (CM). Moreover, the rescue is apparently permanent, with normal numbers of these neurons present in CM animals 6-7 weeks after the neurotrophic factor is no longer being supplied exogenously. Behaviorally, control operates receiving a similarly prepared fraction of unconditioned medium are significantly impaired in the number of trials needed to learn two visual discrimination tasks. This deficit is attributable in part to a bias in erroneous responses to the side contralateral to the lesion. The error bias reflects a failure to inhibit repeated incorrect responding contralaterally. In contrast, the CM animals learn both visual tasks in a normal number of trials and have no contralateral error bias. Rather, all CM animals have an contralateral error bias. Rather, all CM animals have an ipsilateral error bias (interpreted as an unmasking of the contralateral neglect expected after a parietal cortex lesion).(ABSTRACT TRUNCATED AT 250 WORDS)

Animals↗
PubMed2004

Learning in a unidimensional absolute identification task.

We tested whether there is long-term learning in the absolute identification of line lengths. Line lengths are unidimensional stimuli, and there is a common belief that learning of these stimuli quickly reaches a low-level asymptote of about seven items and progresses no more. We show that this is not the case. Our participants served in a 1.5-h session each day for over a week. Although they did not achieve perfect performance, they continued to improve day by day throughout the week and eventually learned to distinguish between 12 and 20 line lengths. These results are in contrast to common characterizations of learning in absolute identification tasks with unidimensional stimuli. We suggest that this learning reflects improvement in short-term processing.

Cognition↗
PubMed1994

Acquisition of vowel contrasts in Dutch.

The aim of this research is to disentangle the contributions of anatomical development of the vocal tract and articulatory learning to the acquisition of vowel contrasts. Four groups of subjects were selected: normal 2-year-olds, normal 4-year-olds, 4-year-olds with a delay in phonological development of approximately 2 years, and adults. Elicited renditions of three Dutch vowels, /u/, /a/, and /a/ were acoustically analysed. The F1 and F2 values of the two vowels constituting a large contrast (/a/ versus /u/) were significantly different in all subject groups. On the other hand, the extent to which the spectra of /a/ and /a/ differed appeared to be affected by age. The 2-year-olds, in contrast to the normal 4-year-olds and adults, produced nondistinctive F2s for these two vowels. The results from the phonologically delayed 4-year-olds paralleled those of the 2-year-olds, which suggests that the acquisition of subtle vowel contrasts is primarily dependent on articulatory learning, rather than the purported anatomical development of the vocal tract.

Adult↗
PubMed1997

Subchronic treatment increases the duration of the cognitive enhancement induced by metrifonate.

The study compared the efficacy of acute versus chronic metrifonate treatment to improve initial and reversal learning of the water maze spatial navigation task in medial septal-lesioned rats. Acute oral administration of 30 mg/kg metrifonate at 30 min, but not at 150 or 360 min, before training improved the initial acquisition of the water maze task. In contrast, improvement of initial learning performance of medial septal-lesioned rats pretreated for 21 days with metrifonate was observed irrespective of the timing of metrifonate treatment relative to behavioral testing. Reversal learning was assessed after a four-day wash-out period. No drug treatment was administered during this part of the study. All the medial septal-lesioned rats that had received only acute treatment with metrifonate during the initial learning stage were now as impaired as vehicle treated medial septal-lesioned rats. However, the group subchronically pretreated with metrifonate performed better than the vehicle-treated medial septal-lesioned controls. These results indicate that both acute and subchronic treatment with metrifonate can facilitate spatial learning in medial septal-lesioned rats and the transient nature of this beneficial effect after single acute administration is transformed into a long-lasting improvement by subchronic treatment.

Animals↗
PubMed2003

Differential effects of learning on neurogenesis: learning increases or decreases the number of newly born cells depending on their birth date.

The hippocampal formation, to which new neurons are added on a daily basis throughout life, is important in spatial learning. Surviving de novo produced cells integrate into the functional circuitry, where they can influence both normal and pathological behaviors. In this study, we examined the effect of the water-maze (a hippocampal-dependent spatial task) on neurogenesis. Learning in this task can be divided into two phases, an early phase during which performance improves rapidly, and a late phase during which asymptotic levels of performance are reached. Here we demonstrate that the late phase of learning has a multifaceted effect on neurogenesis depending on the birth date of new neurons. The number of newly born cells increased contingently with the late phase and a large proportion of these cells survived for at least 4 weeks and differentiated into neurons. In contrast, late-phase learning decreased the number of newly born cells produced during the early phase. This decline in neurogenesis was positively correlated with performance in the water-maze. Thus, rats with the highest de novo cell number were less able to acquire and use spatial information than those with low numbers of new cells. These results show that learning has a complex effect on hippocampal neurogenesis, and reveals a novel mechanism through which neurogenesis may influence normal and pathological behaviors.

Aging↗
PubMed2003

Effects of redundant and nonredundant bimodal sensory stimulation on heart rate in bobwhite quail embryos.

Research with both animal embryos and human infants has provided evidence that information presented redundantly and in temporal synchrony across sensory modalities (intersensory redundancy) can guide selective attention, perceptual learning, and memory during early development. How this facilitation is achieved remains relatively unexamined. This study examined the effects of redundant versus nonredundant bimodal stimulation on a measure of physiological arousal (heart rate) in bobwhite quail embryos. Results show that quail embryos exposed to concurrent but nonredundant auditory and visual stimulation during the late stages of incubation exhibit significantly elevated heart rates following stimulus exposure and during stimulus reexposure when compared to embryos exposed to redundant and synchronous audiovisual stimulation, unimodal auditory stimulation, or no supplemental prenatal sensory stimulation. These findings indicate a functional distinction between redundant and nonredundant bimodal stimulation during early development and suggest that nonredundant bimodal stimulation during the prenatal period can raise arousal levels, thereby potentially interfering with the attentional capacities and perceptual learning of bobwhite quail. In contrast, intersensory redundancy appears to foster arousal levels that facilitate selective attention and perceptual learning during prenatal development.

Age Factors↗
PubMed1988

From conditioning to category learning: an adaptive network model.

We used adaptive network theory to extend the Rescorla-Wagner (1972) least mean squares (LMS) model of associative learning to phenomena of human learning and judgment. In three experiments subjects learned to categorize hypothetical patients with particular symptom patterns as having certain diseases. When one disease is far more likely than another, the model predicts that subjects will substantially overestimate the diagnosticity of the more valid symptom for the rare disease. The results of Experiments 1 and 2 provide clear support for this prediction in contradistinction to predictions from probability matching, exemplar retrieval, or simple prototype learning models. Experiment 3 contrasted the adaptive network model with one predicting pattern-probability matching when patients always had four symptoms (chosen from four opponent pairs) rather than the presence or absence of each of four symptoms, as in Experiment 1. The results again support the Rescorla-Wagner LMS learning rule as embedded within an adaptive network model.

Association Learning↗
PubMed1998

Increased thresholds for long-term potentiation and contextual learning in mice lacking the NMDA-type glutamate receptor epsilon1 subunit.

The NMDA-type glutamate receptor (GluR) channel, composed of the GluRepsilon and GluRzeta subunits, plays a key role in synaptic plasticity in the CNS. The mutant mice lacking the GluRepsilon1 subunit exhibited a reduction in hippocampal long-term potentiation (LTP), but a stronger tetanic stimulation restored the impairment and the saturation level of LTP was unaltered. These results suggest an increase of threshold for LTP induction in the GluRepsilon1 mutant mice. After a series of backcrosses we established a GluRepsilon1 mutant mouse line with a 99.99% pure C57BL/6 genetic background. The performance of the mutant mice in tone- and context-dependent fear conditioning tests was comparable with that of the wild-type mice. However, a significant difference in the extent of contextual learning became apparent when the chamber exposure time before footshock was shortened. Furthermore, there was a significant difference in freezing responses immediately after footshock on the conditioning day between the wild-type and mutant mice, and the difference was not restored by longer chamber exposure in contrast to the contextual learning on the next day of the conditioning. These results suggest that the GluRepsilon1 subunit of the NMDA receptor channel is a determinant of thresholds for both hippocampal LTP and contextual learning and plays differential roles in two forms of contextual fear memories.

Animals↗
PubMed2005

Lessons learned and unanswered questions about sexual assault perpetration.

The most important lesson learned about interpersonal violence in the past 20 years is the astonishingly high prevalence of sexual assault in American society. The extensiveness of unreported sexual assault has been repeatedly documented through the use of self-report data from well-constructed surveys of victims and perpetrators. In contrast, little has been learned about how to effectively reduce rates of sexual assault perpetration. Theoretically derived universal prevention programs targeted at adolescents are sorely needed.

Aggression↗
PubMed2002

Slow feature analysis: unsupervised learning of invariances.

Invariant features of temporally varying signals are useful for analysis and classification. Slow feature analysis (SFA) is a new method for learning invariant or slowly varying features from a vectorial input signal. It is based on a nonlinear expansion of the input signal and application of principal component analysis to this expanded signal and its time derivative. It is guaranteed to find the optimal solution within a family of functions directly and can learn to extract a large number of decorrelated features, which are ordered by their degree of invariance. SFA can be applied hierarchically to process high-dimensional input signals and extract complex features. SFA is applied first to complex cell tuning properties based on simple cell output, including disparity and motion. Then more complicated input-output functions are learned by repeated application of SFA. Finally, a hierarchical network of SFA modules is presented as a simple model of the visual system. The same unstructured network can learn translation, size, rotation, contrast, or, to a lesser degree, illumination invariance for one-dimensional objects, depending on only the training stimulus. Surprisingly, only a few training objects suffice to achieve good generalization to new objects. The generated representation is suitable for object recognition. Performance degrades if the network is trained to learn multiple invariances simultaneously.

Algorithms↗