Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “representation learning”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

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

At least 937 records · Page 52Linked to original sources
PubMed1990

Predicate-argument structure as a link between linguistic and nonlinguistic representations.

We present a study wherein a severe Broca's aphasic patient was trained to learn symbols representing both pure transitive and dative predicates--predicates differing in argument structure--in a visually based artificial language (c-ViC). We found a decrease in performance when two symbols, rather than one, were used to depict these "verbs." However, this decrease in performance was more pronounced for symbols representing pure transitive verbs--those that allow only one argument structure--than for symbols representing dative verbs--those that allow two different argument structures. Also, dative "verbs" yielded better performance when they were inserted in more complex, three-argument "sentences" than when they were inserted in two-argument "sentences." The opposite pattern was found for pure transitives. These results are discussed in terms of our claim that argument structure serves as a point of connection between linguistic information and non-linguistic visual information and in terms of the possibility that argument structure entries are shaped by the form in which visual information is parsed.

Aged↗
PubMed1996

Transient impairment of cholinergic function in the rat insular cortex disrupts the encoding of taste in conditioned taste aversion.

The muscarinic antagonist scopolamine blocks conditioned taste aversion (CTA) when microinjected bilaterally into the rat insular cortex shortly before the exposure of the rat to a novel taste (the conditioned stimulus, CS) in CTA training. Scopolamine has no effect when microinjected shortly after the exposure to the novel taste or shortly before the application of the malaise-inducing agent (unconditioned stimulus, UCS). Scopolamine does not affect sensory, motor and retrieval mechanisms required for performing the CTA task, and does not block CTA when injected into another cortical area. The effect of scopolamine is independent of the taste used as CS. Furthermore, microinjection of scopolamine into the insular cortex shortly before the pre-exposure to a new taste in a latent inhibition paradigm, impairs the attenuation of CTA by that pre-exposure. Other muscarinic antagonists, pirenzepine and AF DX-116, have an effect similar to that of scopolamine. Comparison of the dose-dependency curves of the muscarinic antagonists suggests a predominant role in CTA for M2 subtype receptors. Carbachol, a muscarinic agonist, also impairs the encoding of taste in the insular cortex, but the results are confounded by the ability of that ligand to induce seizures. Our findings suggest that cholinergic neuromodulation participates in processing the CS in the gustatory cortex in CTA, either by encoding novelty at the cellular level, or by instructing the neural circuits to store the novel taste representation.

Animals↗
PubMed1999

Achieving visual object constancy across plane rotation and depth rotation.

Visual object constancy is the ability to recognise an object from its image despite variation in the image when the object is viewed from different angles. I describe research which probes the human visual system's ability to achieve object constancy across plane rotation and depth rotation. I focus on the ecologically important case of recognising familiar objects, although the recognition of novel objects is also discussed. Cognitive neuropsychological studies of patients with specific deficits in achieving object constancy are reviewed, in addition to studies which test neurally intact subjects. In certain cases, the recognition of invariant features allows objects to be recognised irrespective of the view depicted, particularly if small, distinctive sets of objects are presented repeatedly. In contrast, in most situations, recognition is sensitive to both the view in-plane and in-depth from which an object is depicted. This result suggests that multiple, view-specific, stored representations of familiar objects are accessed in everyday, entry-level visual recognition, or that transformations such as mental rotation or interpolation are used to transform between retinal images of objects and view-specific, stored representations.

Attention↗
PubMed1984

Processing category terms in context: instantiation and the structure of semantic categories.

The present study sought to determine if semantic categories processed in context are encoded as particular exemplars. In Experiment 1 we replicated previous results on an extended and modified set of stimuli by showing that when subjects read sentences containing a category term in a context designed to bias encoding toward an atypical exemplar, the atypical exemplar serves as a better retrieval cue than a typical exemplar. In Experiment 2 we tested whether these cued-recall results were due to processes operating at encoding or retrieval. The pattern of semantic interference obtained in a modified Stroop paradigm clearly contradicted the position that readers routinely encode general terms as examples, or "instantiations." In particular, there was significant color-naming interference when typical exemplars served as targets even when preceded by sentences designed to bias encoding toward an atypical exemplar. No significant color-naming interference was generated to atypical exemplars. Experiment 3 ruled out the possibility that differences between the cued-recall and Stroop results in the first two experiments were due to encoding strategy differences. It was concluded that assigning a referent to a category term is not a routine activity in sentence encoding and that processing category terms entails activation of summary representations.

Adult↗
PubMed1992

The memory representation of motor skills: a test of schema theory.

Currently, a popular model for the central representation of motor skills is embodied in Schmidt's schema theory of discrete motor skill learning (Schmidt, 1975). Two experiments are reported here that contrast predictions from a schema abstraction model that is the basis for schema theory with those from an exemplar-based model of motor skill memory representation. In both experiments, subjects performed 300 trials per day of three variations of a three-segment timing task over 4 days of acquisition. The subjects then either immediately transferred to four novel variations of the same task (Experiment 1) that varied in degree of similarity to the exemplars experienced during acquisition; or performed two novel and two previously produced exemplars, following 24-h and 1-week retention intervals (Experiment 2). The results indicated that novel task transfer was not affected by the degree of similarity between the acquisition and transfer exemplars, and that there was no advantage for a previously produced exemplar over a novel exemplar after either a 24-hr or 1 week retention interval. Also, in both experiments, a consistent pattern of bias in responding was noted for novel task transfer and retention. These results are indicative of a schema abstraction model of memory representation for motor skills.

Journal Article↗
PubMed1983

Effects of lateral reversal on recognition memory for photographs of faces.

Recognition memory for photographs of faces, which were initially shown in the normal orientation, then tested either normally (unchanged) or laterally reversed (changed), was examined in four experiments involving different experimental designs (between and within group) and different methods of testing (yes-no and forced-choice). Although the overall effect of the transformation was not as powerful as others (e.g. hair style change) which have been applied to faces, and even failed to attain statistical significance in a fifth experiment in which all the photographs were initially reversed at presentation then tested unchanged or changed back to their normal orientation, the main findings were as follows: (a) subjects recognized fewer reversed (changed) than normal (unchanged) photographs whether or not they were informed of the transformation; (b) the adverse effect of reversal occurred on faces looking straight ahead (full-faces) and, to a slightly lesser extent, on those looking to the left of the observer (left-lookers); (c) left-lookers and right-lookers (the left-lookers initially reversed in Expt 5) were less well recognized than full-faces; and (d) subjects had difficulty identifying orientation, where accuracy fell almost to chance level. These results are taken as support for a feature (rather than Gestalt) model of facial recognition in which the two sides of the face are differentiated in its memory representation.

Adolescent↗
PubMed2002

The organization of visual object representations: a connectionist model of effects of lesions in perirhinal cortex.

We have developed a simple connectionist model based on the idea that perirhinal cortex has properties similar to other regions in the ventral visual stream, or 'what' pathway. The model is based on the assumption that representations in the ventral visual stream are organized hierarchically, such that representations of simple features of objects are stored in caudal regions of the ventral visual stream, and representations of the conjunctions of these features are stored in more rostral regions. We propose that a function of these feature conjunction representations is to help to resolve 'feature ambiguity', a property of visual discrimination problems that can emerge when features of an object predict a given outcome (e.g. reward) when part of one object, but predict a different outcome when part of another object. Several recently reported effects of lesions of perirhinal cortex in monkeys have provided key insights into the functions of this region. In the present study these effects were simulated by comparing the performance of connectionist networks before and after removal of a layer of units corresponding to perirhinal cortex. The results of these simulations suggest that effects of lesions in perirhinal cortex on visual discrimination may be due not to the impairment of a specific type of learning or memory, such as declarative or procedural, but to compromising the representations of visual stimuli. Furthermore, we propose that attempting to classify perirhinal cortex function as either 'perceptual' or 'mnemonic' may be misguided, as it seems unlikely that these broad constructs will map neatly onto anatomically defined regions of the brain.

Animals↗
PubMed2006

Neural markers of categorization in 6-month-old infants.

Little is known of the neural processes that underlie concept-formation abilities in human infants. We investigated category-learning processes in infants both by using a common behavioral measure and by recording the brain's electrical activity (event-related potentials, or ERPs). ERPs were recorded while 6-month-olds viewed cat images during training, followed by novel cat images interspersed with novel dog images during test. The data indicate that distinct neural signals correspond with learning of a category presented during familiarization, preferential responding to a novel category, and representation of category exemplars at multiple levels of inclusiveness. The results suggest that fundamental components of the neural architecture supporting object categorization are functional within the first half-year of postnatal life, before infants acquire language and young children engage in formal learning of semantic categories. The findings are discussed in terms of their implications for models of category learning and development.

Arousal↗
PubMed2000

Optic flow helps humans learn to navigate through synthetic environments.

Self-movement through an environment generates optic flow, a potential source of heading information. But it is not certain that optic flow is sufficient to support navigation, particularly navigation along complex, multi-legged paths. To address this question, we studied human participants who navigated synthetic environments with and without salient optic flow. Participants used a keyboard to control realistic simulation of self-movement through computer-rendered, synthetic environments. Because these environments comprised series of identically textured virtual corridors and intersections, participants had to build up some mental representation of the environment in order to perform. The impact of optic flow on learning was examined in two experiments. In experiment 1, participants learned to navigate multiple T-junction mazes with and without accompanying optic flow. Optic flow promoted faster learning, mainly by preventing disorientation and backtracking in the maze. In experiment 2, participants found their way around a virtual city-block environment, experiencing two different kinds of optic flow as they went. By varying the rate at which the display was updated, we created optic flow that was either fluid or choppy. Here, fluid optic flow (as compared with choppy optic flow) enabled participants to locate a remembered target position more accurately. When other cues are unavailable, optic flow can be a significant aid in wayfinding. Among other things, optic flow can facilitate path integration, which involves updating a mental representation of place by combining the trajectories of previously travelled paths [corrected].

Adolescent↗
PubMedDate not supplied

Unifying cost and information in information-theoretic competitive learning.

In this paper, we introduce costs into the framework of information maximization and try to maximize the ratio of information to its associated cost. We have shown that competitive learning is realized by maximizing mutual information between input patterns and competitive units. One shortcoming of the method is that maximizing information does not necessarily produce representations faithful to input patterns. Information maximizing primarily focuses on some parts of input patterns that are used to distinguish between patterns. Therefore, we introduce the cost, which represents average distance between input patterns and connection weights. By minimizing the cost, final connection weights reflect input patterns well. We applied the method to a political data analysis, a voting attitude problem and a Wisconsin cancer problem. Experimental results confirmed that, when the cost was introduced, representations faithful to input patterns were obtained. In addition, improved generalization performance was obtained within a relatively short learning time.

Algorithms↗
PubMed2007

Associative representational plasticity in the auditory cortex: a synthesis of two disciplines.

Historically, sensory systems have been largely ignored as potential loci of information storage in the neurobiology of learning and memory. They continued to be relegated to the role of "sensory analyzers" despite consistent findings of associatively induced enhancement of responses in primary sensory cortices to behaviorally important signal stimuli, such as conditioned stimuli (CS), during classical conditioning. This disregard may have been promoted by the fact that the brain was interrogated using only one or two stimuli, e.g., a CS(+) sometimes with a CS(-), providing little insight into the specificity of neural plasticity. This review describes a novel approach that synthesizes the basic experimental designs of the experimental psychology of learning with that of sensory neurophysiology. By probing the brain with a large stimulus set before and after learning, this unified method has revealed that associative processes produce highly specific changes in the receptive fields of cells in the primary auditory cortex (A1). This associative representational plasticity (ARP) selectively facilitates responses to tonal CSs at the expense of other frequencies, producing tuning shifts toward and to the CS and expanded representation of CS frequencies in the tonotopic map of A1. ARPs have the major characteristics of associative memory: They are highly specific, discriminative, rapidly acquired, exhibit consolidation over hours and days, and can be retained indefinitely. Evidence to date suggests that ARPs encode the level of acquired behavioral importance of stimuli. The nucleus basalis cholinergic system is sufficient both for the induction of ARPs and the induction of specific auditory memory. Investigation of ARPs has attracted workers with diverse backgrounds, often resulting in behavioral approaches that yield data that are difficult to interpret. The advantages of studying associative representational plasticity are emphasized, as is the need for greater behavioral sophistication.

Animals↗
PubMed1985

A program for machine learning of counting criteria: empirical induction of logic-based classification rules.

A program has been developed which derives classification rules from empirical observations and expresses these rules in a knowledge representation format called 'counting criteria'. Decision rules derived in this format are often more comprehensible than rules derived by existing machine learning programs such as AQ11. Use of the program is illustrated by the inference of discrimination criteria for certain types of bacteria based upon their biochemical characteristics. The program may be useful for the conceptual analysis of data and for the automatic generation of prototype knowledge bases for expert systems.

Artificial Intelligence↗
PubMedDate not supplied

The neural consequences of repetition: clinical implications of a learning hypothesis.

Repetitive strain injuries (RSIs) are difficult to treat. Some individuals with RSIs may ultimately develop chronic pain syndromes or movement problems like focal hand dystonia (FDh), a disorder of motor control manifested in a specific context during skilled, hand tasks. This paper reports on the results of four neuroplasticity studies suggesting that repetitive hand opening and closing can lead to motor control problems, measurable somatosensory changes, and problems in graphesthesia and stereognosis. The experiments support a learning hypothesis for the origin of severe RSIs, particularly FDh. This degradation in the sensory representation of the hand may not only explain the therapeutic challenge of returning these patients to work, but also provide a foundation for developing more effective physical rehabilitation strategies. Implications and conjectures for the applications of this learning hypothesis to conditions of chronic pain are also discussed.

Adult↗
PubMed2001

Competitive Hebbian learning and the hippocampal place cell system: modeling the interaction of visual and path integration cues.

The hippocampus has long been thought essential for implementing a cognitive map of the environment. However, almost 30 years since place cells were found in rodent hippocampal field CA1, it is still unclear how such an allocentric representation arises from an ego-centrically perceived world. By means of a competitive Hebbian learning rule responsible for coding visual and path integration cues, our model is able to explain the diversity of place cell responses observed in a large set of electrophysiological experiments with a single fixed set of parameters. Experiments included changes observed in place fields due to exploration of a new environment, darkness, retrosplenial cortex inactivation, and removal, rotation, and permutation of landmarks. To code for visual cues for each landmark, we defined two perceptual schemas representing landmark bearing and distance information over a linear array of cells. The information conveyed by the perceptual schemas is further processed through a network of adaptive layers which ultimately modulate the resulting activity of our simulated place cells. In path integration terms, our system is able to dynamically remap a bump of activity coding for the displacement of the animal in relation to an environmental anchor. We hypothesize that path integration information is computed in the rodent posterior parietal cortex and conveyed to the hippocampus where, together with visual information, it modulates place cell activity. The resulting network yields a more direct treatment of partial remapping of place fields than other models. In so doing, it makes new predictions regarding the nature of the interaction between visual and path integration cues during new learning and when the system is challenged with environmental changes.

Animals↗
PubMed2005

Interpreting hippocampal function as recoding and forecasting.

A model of hippocampal function, centered on region CA3, reproduces many of the cognitive and behavioral functions ascribed to the hippocampus. Where there is precise stimulus control and detailed quantitative data, this model reproduces the quantitative behavioral results. Underlying the model is a recoding conjecture of hippocampal computational function. The expanded conjecture includes a special role for randomization and, as recoding progresses with experience, the occurrence of sequence learning and sequence compression. These functions support the putative higher-order hippocampal function, i.e. production of representations readable by a linear decoder and suitable for both neocortical storage and forecasting. Simulations confirm the critical importance of randomly driven recoding and the neurocognitive relevance of sequence learning and compression. Two forms of sequence compression exist, on-line and off-line compression: both are conjectured to support neocortical encoding of context and declarative memory as described by .

Cognition↗
PubMed2003

Music and learning-induced cortical plasticity.

Auditory stimuli are encoded by frequency-tuned neurons in the auditory cortex. There are a number of tonotopic maps, indicating that there are multiple representations, as in a mosaic. However, the cortical organization is not fixed due to the brain's capacity to adapt to current requirements of the environment. Several experiments on cerebral cortical organization in musicians demonstrate an astonishing plasticity. We used the MEG technique in a number of studies to investigate the changes that occur in the human auditory cortex when a skill is acquired, such as when learning to play a musical instrument. We found enlarged cortical representation of tones of the musical scale as compared to pure tones in skilled musicians. Enlargement was correlated with the age at which musicians began to practice. We also investigated cortical representations for notes of different timbre (violin and trumpet) and found that they are enhanced in violinists and trumpeters, preferentially for the timbre of the instrument on which the musician was trained. In recent studies we extended these findings in three ways. First, we show that we can use MEG to measure the effects of relatively short-term laboratory training involving learning to perceive virtual instead of spectral pitch and that the switch to perceiving virtual pitch is manifested in the gamma band frequency. Second, we show that there is cross-modal plasticity in that when the lips of trumpet players are stimulated (trumpet players assess their auditory performance by monitoring the position and pressure of their lips touching the mouthpiece of their instrument) at the same time as a trumpet tone, activation in the somatosensory cortex is increased more than it is during the sum of the separate lip and trumpet tone stimulation. Third, we show that musicians' automatic encoding and discrimination of pitch contour and interval information in melodies are specifically enhanced compared to those in nonmusicians in that musicians show larger functional mismatch negativity (MMNm) responses to occasional changes in melodic contour or interval, but that the two groups show similar MMNm responses to changes in the frequency of a pure tone.

Auditory Cortex↗
PubMed2005

Motor learning by observing.

Learning complex motor behaviors like riding a bicycle or swinging a golf club is based on acquiring neural representations of the mechanical requirements of movement (e.g., coordinating muscle forces to control the club). Here we provide evidence that mechanisms matching observation and action facilitate motor learning. Subjects who observed a video depicting another person learning to reach in a novel mechanical environment (imposed by a robot arm) performed better when later tested in the same environment than subjects who observed similar movements but no learning; moreover, subjects who observed learning of a different environment performed worse. We show that this effect is not based on conscious strategies but instead depends on the implicit engagement of neural systems for movement planning and control.

Adult↗
PubMed1997

Decision boundaries in one-dimensional categorization.

Decision-boundary theories of categorization are often difficult to distinguish from exemplar-based theories of categorization. The authors developed a version of the decision-boundary theory, called the single-cutoff model, that can be distinguished from the exemplar theory. The authors present 2 experiments that test this decision-boundary model. The results of both experiments point strongly to the absence of single cutoff in most participants, and no participant displayed use of the optimal boundary. The range of nonoptimal solutions shown by individual participants was accounted for by an exemplar-based adaptive-learning model. When combined with the results of previous research, this suggests that a comprehensive model of categorization must involve both rules and exemplars, and possibly other representations as well.

Decision Making↗