Search PubMed⌕ Search
Home/Archive/PubMed

PubMed · 16447385

Using context to build semantics.

Peter J Kwantes
View original source ↗Export metadata ↓

Abstract

Latent semantic analysis (LSA) is a model of knowledge representation for words. It works by applying dimension reduction to local co-occurrence data from a large collection of documents after performing singular value decomposition on it. When the reduction is applied, the system forms condensed representations for the words that incorporate higher order associations. The higher order associations are primarily responsible for any semantic similarity between words in LSA. In this article, a memory model is described that creates semantic representations for words that are similar in form to those created by LSA. However, instead of applying dimension reduction, the model builds the representations by using a retrieval mechanism from a well-known account of episodic memory.

Explore related subjects

Association LearningComputer SimulationHumansMental RecallModels, StatisticalSemanticsVerbal LearningJournal ArticleResearch Support, Non-U.S. Gov't

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Peter J Kwantes. 2005. Using context to build semantics.. https://doi.org/10.3758/bf03196761

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed2006

Stimulus and response repetition effects in the detection of sounds: evidence of obligatory retrieval and use of a prior event.

Three experiments were performed to examine the extent to which the time required to detect a probe sound is determined by the acoustic characteristics of a preceding prime sound and by the nature of the response made to the prime. The results of Experiment 1 revealed that the effect of frequency repetition was facilitative when a response was made to the occurrence of both the prime and the probe, but was inhibitory when a response was made only to the probe. In contrast, there was no significant effect of location repetition when responses were made to both prime and probe but there was a strongly inhibitory effect when a response was required only to the probe. These results suggest that the acoustic characteristics of the prime along with information about any response made to it, are used in preparing and executing a response to the probe. Experiment 2 was designed to disrupt the temporal link between prime and probe as a way to discourage such retrieval. The results of the experiment revealed that although a response was required to both the prime and probe, no significant effect of frequency repetition was apparent but a robust inhibitory effect of location repetition emerged. In the third experiment repetition effects both within a trial (i.e., prime to probe) and between trials (i.e., probe to prime) were examined. A facilitative effect of frequency repetition was apparent when the interval between sounds was brief, and an inhibitory effect of location repetition was apparent when the interval between sounds was relatively lengthy. Taken together these results suggest that even simple perceptual judgments may be influenced by retrieval of information about a previous processing episode and that effects of frequency repetition and location repetition may be mediated by different mechanisms.

Association Learning↗
PubMed2006

Bayesian associative learning.

The Bayesian approach to belief updating is par excellence a theory of learning--a theory of how beliefs should be revised in the light of new evidence. Yet despite its considerable utility as a framework for understanding cognition, it has not been prominent in theorizing about elementary learning processes in animals and humans. New research by Kruschke seems set to change this and to offer a bridge between rational and empirically descriptive accounts of learning.

Association Learning↗
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↗