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 1,729 records · Page 96Linked to original sources
PubMed1983

R L Gregory and others: the wrong picture of the picture theory of perception.

It is the position of R L Gregory and other cognitive theorists that perceptual knowledge conceived as an inner picture leads to an infinite regress, but that perceptual knowledge conceived as an abstract or coded representation does not. It is argued here that this view is mistaken. All inner representations, whether pictorial or abstract, lead to the regress because all representations, inner or outer, require interpretation, and hence an interpreter. The problem will not disappear, furthermore, by formalizing the representation because rule-following is not equivalent to interpretation. The regress can only be avoided if the whole organism is made the interpreter, and representations are given their appropriate place: in the external world, not inside heads.

Brain↗
PubMed1992

Cooperative representation of visual borders.

Recently it has been reported that the visual cortical cells which are engaged in cooperative coding of global stimulus features, display synchrony in their firing rates when both are stimulated. Alternative models identify global stimulus features with the course spatial scales of the image. Versions of the Munsterberg or Café Wall illusions which differ in their low spatial frequency content were used to show that in all cases it was the high spatial frequencies in the image which determined the strength and direction of these illusions. Since cells responsive to high spatial frequencies have small receptive fields, cooperative coding must be involved in the representation of long borders in the image.

Attention↗
PubMed2002

Differential transfer benefits of increased practice for constant, blocked, and serial practice schedules.

The effects of practice schedule and amount of practice on the development of the generalized motor program (GMP) and on parameter estimation were investigated. Participants (N = 108) practiced the same relative timing but different absolute durations of a multisegment timing task. Practice schedules (constant, blocked, or serial) were crossed with amounts of practice (low and high). Inclusion of a constant practice condition allowed the authors to investigate the variability of practice prediction. Participants practiced the same proportional durations in a serial or a blocked schedule, which enabled the authors to examine contextual interference. A constant practice schedule enhanced GMP performance when task parameters remained the same, but varied practice schedules were beneficial when task parameters changed. A serial as opposed to a blocked practice schedule was superior when the performance of a task governed by a different GMP was required. Increased practice led to a consolidated task representation that was unavailable for updating.

Humans↗
PubMed1991

Semantic inhibition of ignored words during a figure classification task.

Two experiments investigated inhibitory processes in visual spatial attention. In particular, factors influencing the suppression of ignored visual stimuli were investigated. Subjects responded to geometric shapes while distractors were presented in the periphery. Distractors consisted of a single word, a pair of unrelated words, or a single word paired with a string of non-linguistic symbols. Semantic processing of the ignored words was measured with a subsequent lexical decision task. Test probes presented after the prime displays revealed suppression effects for words semantically related to a previously ignored word, but only for conditions in which two distractor words were presented. Suppression was not observed when the prime consisted of a single word or a single word paired with non-linguistic symbols. In Experiment 2 two different time delays between the onset of the primes and the onset of the test probes were used. At the shorter interval facilitatory priming was observed, while at longer intervals suppression was observed. The facilitation-suppression pattern suggests that ignored items are recognized before being suppressed. In summary, the results suggest the following: (1) that selectively attending does not restrict ignored items from gaining access to their semantic representations, and (2) that inhibition is an important process in determining the focus of attention. These results are discussed within a selective attention framework in which all items in a display gain access to memory representations, and attention to selected items causes competing items to be inhibited.

Attention↗
PubMed2006

How color enhances visual memory for natural scenes.

We offer a framework for understanding how color operates to improve visual memory for images of the natural environment, and we present an extensive data set that quantifies the contribution of color in the encoding and recognition phases. Using a continuous recognition task with colored and monochrome gray-scale images of natural scenes at short exposure durations, we found that color enhances recognition memory by conferring an advantage during encoding and by strengthening the encoding-specificity effect. Furthermore, because the pattern of performance was similar at all exposure durations, and because form and color are processed in different areas of cortex, the results imply that color must be bound as an integral part of the representation at the earliest stages of processing.

Adult↗
PubMed2003

Encoding predictive reward value in human amygdala and orbitofrontal cortex.

Adaptive behavior is optimized in organisms that maintain flexible representations of the value of sensory-predictive cues. To identify central representations of predictive reward value in humans, we used reinforcer devaluation while measuring neural activity with functional magnetic resonance imaging. We presented two arbitrary visual stimuli, both before and after olfactory devaluation, in a paradigm of appetitive conditioning. In amygdala and orbitofrontal cortex, responses evoked by a predictive target stimulus were decreased after devaluation, whereas responses to the nondevalued stimulus were maintained. Thus, differential activity in amygdala and orbitofrontal cortex encodes the current value of reward representations accessible to predictive cues.

Adult↗
PubMed2005

Beyond memory, navigation, and inhibition: behavioral evidence for hippocampus-dependent cognitive coordination in the rat.

Injecting tetrodotoxin (TTX) into one hippocampus impaired avoidance of a place defined by distal cues while rats were on a slowly rotating arena. The impairment could be explained by a deficit in memory, navigation, or behavioral inhibition. Here, we show that the TTX injection abolished the ability of rats to organize place-avoidance behavior specifically when distal room and local arena cues were continuously dissociated. The results provide evidence that injecting TTX into one hippocampus specifically impaired the coordination of representations that support organized behavior because of the following: (1) rats normally coordinate separate room and arena avoidance memories; (2) the TTX injection spared spatial, relational, and representational memory, navigation, and behavioral inhibition; and (3) the TTX-induced impairment of place avoidance depended on the need to coordinate representations of local and distal stimuli.

Anesthetics, Local↗
PubMed2006

The role of perceptually represented structure in analogical problem solving.

Current models of analogical reasoning assume that representations of source examples and target problems occur in an amodal format--that is, a representation whose construction and processing are independent of activity in the perceptual and motor cortices of the brain. We examined the possible use of kinesthetic information--perceptual structures associated with the sensation of space and force--in the representation of source examples and target problems. Participants who recreated a source story while acting out the key elements were more likely to access the story when later working on the target problem than were participants who only verbally recreated the story or who verbally recreated it as well as sketched it. We argue that enactment made kinesthetic and spatial features more salient in participants' source story representations and that this aided performance. These results suggest that current models of analogical reasoning might be improved by including perceptual information as part of their representational schemes.

Humans↗
PubMed1998

Prime time advertisements: repetition priming from faces seen on subject recruitment posters.

Repetition priming is defined as a gain in item recognition after previous exposure. Repetition priming of face recognition has been shown to last for several months, despite contamination by everyday exposure to both experimental and control faces in the interval. Here we show that gains in face recognition in the laboratory are found from faces initially seen in a rather different context--on subject recruitment posters, even when the advertisements make no specific mention of experiments involving face recognition. The priming was greatest when identical pictures were shown in the posters and in the test phase, although different views of faces did give significant priming in one study. Follow-up studies revealed poor explicit memory for the faces shown on the posters. The results of these experiments are used to develop a model in which repetition priming reflects the process of updating representations of familiar faces.

Adult↗
PubMed2004

Spatiotemporal analysis of neuromagnetic activation associated with mirror reading.

Our previous report has confirmed that visually presented words can elicit neuromagnetic activities from the visual cortex, angular gyrus and Broca's area. It is not clear how the words are visuospatially transformed and recognized by the human brain. Mirror reading is characterized by reading which runs in the opposite direction to normal reading, with reversals of letters. It would be very interesting to find out if there are any neuromagnetic differences between the mirror reading and normal reading. Four right-handed healthy adults have been studied with a whole cortex Magnetoencephalography (MEG) system. The stimuli consisted of eight normal oriented words and eight inverted exclamation mark section "reversed words" (mirror-image of the words). All stimuli were randomly presented on the screen in front of the subjects using DirectX. MEG data were analyzed using both single dipole modeling and synthetic aperture magnetometry (SAM). Four responses to the reversed words were identified in all four subjects. In comparison to the normally oriented words, the reversed words elicited a stronger response at a latency of 248+/-6 ms. SAM results indicated that the reversed words invoked strong activations in the left and right parietal cortices but the normally oriented words did not. The mirror reading elicited one magnetic response which is different from that of the normal reading. The difference between the mirror reading and the normal reading in terms of neuromagnetic activation may reflect the development of novel representations for reversed words.

Adult↗
PubMed1992

Object exploration and reactions to spatial and nonspatial changes in hooded rats following damage to parietal cortex or hippocampal formation.

Hooded rats with bilateral lesions of the anterior part of the hippocampal formation (HIP), anterior region of the posterior parietal cortex (APC), or posterior region of the posterior parietal cortex (PPC) were compared with controls for their exploration of 5 objects in an open field, habituation of locomotion and object investigation, and response to spatial and nonspatial change. First, all groups displayed habituation of both locomotor and exploratory activity. Second, controls selectively reexplored displaced objects, and APC-lesioned rats reexplored all objects, whereas PPC- and HIP-lesioned rats failed to react to the spatial change. Third, a novel object induced reexploration in all groups. The results are consistent with the roles of the HIP and PPC in spatial information processing. Moreover, the APC and PPC are involved in attentional effortful processing and visuospatial information processing necessary for spatial representation, respectively.

Animals↗
PubMed2006

Heterarchical reinforcement-learning model for integration of multiple cortico-striatal loops: fMRI examination in stimulus-action-reward association learning.

The brain's most difficult computation in decision-making learning is searching for essential information related to rewards among vast multimodal inputs and then integrating it into beneficial behaviors. Contextual cues consisting of limbic, cognitive, visual, auditory, somatosensory, and motor signals need to be associated with both rewards and actions by utilizing an internal representation such as reward prediction and reward prediction error. Previous studies have suggested that a suitable brain structure for such integration is the neural circuitry associated with multiple cortico-striatal loops. However, computational exploration still remains into how the information in and around these multiple closed loops can be shared and transferred. Here, we propose a "heterarchical reinforcement learning" model, where reward prediction made by more limbic and cognitive loops is propagated to motor loops by spiral projections between the striatum and substantia nigra, assisted by cortical projections to the pedunculopontine tegmental nucleus, which sends excitatory input to the substantia nigra. The model makes several fMRI-testable predictions of brain activity during stimulus-action-reward association learning. The caudate nucleus and the cognitive cortical areas are correlated with reward prediction error, while the putamen and motor-related areas are correlated with stimulus-action-dependent reward prediction. Furthermore, a heterogeneous activity pattern within the striatum is predicted depending on learning difficulty, i.e., the anterior medial caudate nucleus will be correlated more with reward prediction error when learning becomes difficult, while the posterior putamen will be correlated more with stimulus-action-dependent reward prediction in easy learning. Our fMRI results revealed that different cortico-striatal loops are operating, as suggested by the proposed model.

Association Learning↗
PubMed2003

Language learning and innateness: some implications of Compounds Research.

In noun compounds in English, the modifying noun may be singular (mouse-eater) or an irregularly inflected plural (mice-eater), but regularly inflected plurals are dispreferred (*rats-eater). This phenomenon has been taken as strong evidence for dual-mechanism theories of lexical representations, which hold that regular (rule-governed) and irregular (exception) items are generated by qualitatively different and innately specified mechanisms. Using corpus analyses, behavioral studies, and computational modeling, we show that the rule-versus-exceptions approach makes a number of incorrect predictions. We propose a new account in which the acceptability of modifiers is determined by a constraint satisfaction process modulated by semantic, phonological, and other factors. The constraints are acquired by the child via general purpose learning algorithms, based on noun compounds and other constructions in the input. The account obviates the regular/irregular dichotomy while simultaneously providing a superior account of the data.

Child↗
PubMed1999

Perceptual learning by preschool children using stimuli with varying proportions of common elements.

The effect of preexposure to visual stimuli on subsequent discrimination was studied in 17 girls and 23 boys, ages 53 to 61 mo., using a test in which a sample stimulus had to be identified from a set of comparison stimuli. Preexposure significantly facilitated discrimination. This facilitation was more noticeable between stimuli which shared a large proportion of common elements. The results were consistent with an associative theory of the representation of stimuli suggested by McLaren, Kaye, and Mackintosh in 1989, and implications are considered for teaching.

Association↗
PubMed1995

Studies of the limbic comparator: limbic circuit training-induced unit activity and avoidance behavior in rabbits with anterior dorsal thalamic lesions.

Multiunit activity of cingulate cortex and the anterior ventral (AV) thalamic nucleus was recorded as rabbits learned to avoid a shock by locomoting in response to a tone (CS+) and to ignore a nonpredictive tone (CS-). Rabbits with anterior dorsal (AD) thalamic lesions avoided shock more often than controls during the first training session and the first session of extinction training given after the completion of acquisition. Training-induced neuronal changes in cingulate cortex and in components of the AV nucleus were lost in the rabbits with lesions. These effects were comparable to previously observed effects of subicular lesions. It is proposed that interactions of AD thalamic, subicular, and cingulate cortical neurons yield a stable mnemonic representation of the associative significance of the CS. The representation is used in a comparator circuit, which inhibits behavior when unexpected events occur.

Acoustic Stimulation↗
PubMed1995

Coding of hierarchical versus linear pattern structure in rats and humans.

In 3 experiments, rats and humans learned serial patterns composed of 24, 30, or 36 items. Patterns had a 2-, 3-, or 4-level hierarchical rule structure. In Experiments 1 and 2, patterns had either perfect hierarchical structure or 2 modified chunks that violated hierarchical structure, thus producing linear structure (i.e., nonhierarchical structure). For both rats and humans, pattern structure predicted pattern learning difficulty and also the nature and relative frequency of errors. Both treated chunks that were inconsistent with hierarchical structure as violation chunks, that is, they made errors that reflected their "tendency to regularize the perception of an irregular pattern" (F. Restle & B. L. Burnside, 1972). The results support the view that rats can abstract and encode a representation of multilevel hierarchical structure in serial patterns in much the same way as humans do in analogous tasks.

Animals↗
PubMed2000

Effects of US devaluation on win-stay and win-shift radial maze performance in rats.

Previous studies have shown double dissociations between win-stay and win-shift radial maze learning in terms of their underlying neural substrates. To examine the content of the associations formed in the two tasks, the authors devalued the food unconditioned stimulus (US) by taste aversion to differentiate stimulus-stimulus(CS-US) and stimulus-response (CS-CR) learning. US devaluation was performed in rats that were over- or undertrained on the win-stay task. Devaluation substantially reduced food consumption on the maze but failed to disrupt choice accuracy, regardless of the amount of training. Devaluation did not affect latency in overtrained rats but did increase latency in undertrained rats. In the win-shift task, devaluation caused rats to reject the reinforcer, yet they continued to accurately win-shift, but with significantly longer latencies (Experiment 3). The results suggest that an S-R association may mediate performance after extended win-stay training. In contrast, a US representation appears to be recalled during early win-stay and win-shift performance.

Animals↗
PubMed1999

Metabolic engineering from a cybernetic perspective. 1. Theoretical preliminaries

The theoretical basis of a cybernetic metabolic network design and analysis framework, which has been subsequently successfully applied to predict system response to genetic alteration, is presented. This conceptual methodology consists of three main branches, namely, a model realization framework, a representation of genetic alteration, and lastly, a metabolic design component. These concepts are introduced as a series of postulates that describe the basic tenets of the approach. Each branch is discussed in turn, starting with the cybernetic representation of arbitrarily complex metabolic networks. A set of postulates is put forth that affords the modular construction of cybernetic models of metabolic networks using as a base a library of elementary pathways. This is followed by a discussion of the representation of genetic alterations within the cybernetic framework. It is postulated that the objective of the base network and the altered system are identical (at least on the time scale required for the organism to "learn" new objectives). This implies, with respect to resource allocation, that the base network and its genetically altered counterpart may still be treated as optimal systems; however, the set of competing physiological choices open to the altered network expands or contracts depending upon the nature of the genetic perturbation. Lastly, to add a predictive design aspect to the methodology, we present a set of postulates that outline the application of metabolic control analysis to cybernetic model systems. We postulate that sensitivity coefficients computed from a cybernetic model, although still local in scope, have the added benefit of a systematic representation of regulatory function as described by the cybernetic variables. Thus, information gained from sensitivity measurements stemming from a cybernetic model include the explicit input of metabolic regulation, a component that is lacking in a purely kinetic representation of metabolic function. The sensitivity results can then be employed to develop qualitative strategies for the rational alteration of metabolic function, which can be evaluated by simulation of an appropriately modified cybernetic model of the base network.

Journal Article↗