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.     

Open and closed loops: A computational approach to attention and consciousness

Within recent years, researchers have proposed the independence of attention and consciousness on both empirical and conceptual grounds. However, the elusive nature of these constructs complicates progress in the investigation of their interaction. We present a framework within which we conceptualize attention and consciousness in computational terms. Here, the concepts are consi-dered as large-scale, functionally and structurally different processes, embedded in a biologically inspired architecture, spanning the full arc from stimulus to response. Our architecture assumes a general independence of attention and consciousness, but supposes strong interactions. Furthermore, it addresses the developmental aspect, stressing that these functions have to gradually develop through learning.     

Talking nets: a multiagent connectionist approach to communication and trust between individuals

A multiagent connectionist model is proposed that consists of a collection of individual recurrent networks that communicate with each other and, as such, is a network of networks. The individual recurrent networks simulate the process of information uptake, integration, and memorization within individual agents, and the communication of beliefs and opinions between agents is propagated along connections between the individual networks. A crucial aspect in belief updating based on information from other agents is the trust in the information provided. In the model, trust is determined by the consistency with the receiving agents' existing beliefs and results in changes of the connections between individual networks, called trust weights. These weights lead to a selective propagation and thus to the filtering out of less reliable information, and they implement H. P. Grice's (1975) maxims of quality and quantity in communication. The unique contribution of communicative mechanisms beyond intrapersonal processing of individual networks was explored in simulations of key phenomena involving persuasive communication and polarization, lexical acquisition, spreading of stereotypes and rumors, and a lack of sharing unique information in group decisions.     

Graded semantic and phonological similarity effects in priming: evidence for a distributed connectionist approach to morphology

A considerable body of empirical and theoretical research suggests that morphological structure governs the representation of words in memory and that many words are decomposed into morphological components in processing. The authors investigated an alternative approach in which morphology arises from the interaction of semantic and phonological codes. A series of cross-modal lexical decision experiments shows that the magnitude of priming reflects the degree of semantic and phonological overlap between words. Crucially, moderately similar items produce intermediate facilitation (e.g., lately-late). This pattern is observed for word pairs exhibiting different types of morphological relationships, including suffixed-stem (e.g., teacher-teach), suffixed-suffixed (e.g., saintly-sainthood), and prefixed-stem pairs (preheat-heat). The results can be understood in terms of connectionist models that use distributed representations rather than discrete morphemes.     

Discussion on the reverberatory model of short-term memory: a computational approach

Up to now a number of models have been proposed to underlie memory formation in the central nervous system. Two of these models are the reverberatory circuit model and the other one the self-feedback loop model. This paper considers these two models regarding their ability to preserve neural activity and to hold information. In the self-feedback loop model, the activity level of the loop output is computed regarding the short lasting initial input. In the reverberatory circuit model, the activity levels of the proposed two-layer network outputs were computed regarding the short lasting initial inputs of the network. In the self-feedback loop model, the activity level of the loop output changes with each reverberation until it reaches a specific limit and then remains at that level. In the reverberatory circuit model, the activity levels of the proposed two-layer network outputs display an oscillatory behavior. These models can preserve the input activity, but they change its level with each reverberation. Information carried by a single neuron is related to its activity level. Therefore these models change the information during the reverberation. Short-term memory must hold the information for a certain period of time, so these models cannot be proposed to underlie short-term memory formation.     

Identifying neural correlates of consciousness: the state space approach

This article sketches an idealized strategy for the identification of neural correlates of consciousness. The proposed strategy is based on a state space approach originating from the analysis of dynamical systems. The article then focuses on one constituent of consciousness, phenomenal awareness. Several rudimentary requirements for the identification of neural correlates of phenomenal awareness are suggested. These requirements are related to empirical data on selective attention, on completely intrinsic selection and on globally unconscious states. As an example, neuroscientific findings on synchronized gamma activity are categorized according to these requirements.     

Whole word morphologizer: expanding the word-based lexicon: a nonstochastic computational approach

Whole Word Morphologizer is a small computer implementation of word-based morphology. The program automatically identifies morphological relations in a small word-based lexicon, literally learning its morphology, and uses the knowledge it acquires to generate new words. It is based on a model of the mental lexicon in which all entries are whole, entire, fully fledged words and relies solely on basic cognitive principles (differentiation and generalization) for the automatic acquisition of morphological relations and the population of the lexicon.     

Doing without schema hierarchies: a recurrent connectionist approach to normal and impaired routine sequential action

In everyday tasks, selecting actions in the proper sequence requires a continuously updated representation of temporal context. Previous models have addressed this problem by positing a hierarchy of processing units, mirroring the roughly hierarchical structure of naturalistic tasks themselves. The present study considers an alternative framework, in which the representation of context depends on recurrent connections within a network mapping from environmental inputs to actions. The ability of this approach to account for human performance was evaluated by applying it, through simulation, to a specific everyday task. The resulting model learned to deal flexibly with a complex set of sequencing constraints, encoding contextual information at multiple time scales within a single, distributed internal representation. Degrading this representation led to errors resembling those observed both in everyday behavior and in apraxia. Analysis of the model's function yielded numerous predictions relevant to both normal and apraxic performance.     

Statistical learning of phonetic categories: insights from a computational approach

Recent evidence ( Maye, Werker & Gerken, 2002 ) suggests that statistical learning may be an important mechanism for the acquisition of phonetic categories in the infant's native language. We examined the sufficiency of this hypothesis and its implications for development by implementing a statistical learning mechanism in a computational model based on a mixture of Gaussians (MOG) architecture. Statistical learning alone was found to be insufficient for phonetic category learning – an additional competition mechanism was required in order for the categories in the input to be successfully learnt. When competition was added to the MOG architecture, this class of models successfully accounted for developmental enhancement and loss of sensitivity to phonetic contrasts. Moreover, the MOG with competition model was used to explore a potentially important distributional property of early speech categories – sparseness – in which portions of the space between phonetic categories are unmapped. Sparseness was found in all successful models and quickly emerged during development even when the initial parameters favoured continuous representations with no gaps. The implications of these models for phonetic category learning in infants are discussed.     

Forgetting curves: implications for connectionist models

Forgetting in long-term memory, as measured in a recall or a recognition test, is faster for items encoded more recently than for items encoded earlier. Data on forgetting curves fit a power function well. In contrast, many connectionist models predict either exponential decay or completely flat forgetting curves. This paper suggests a connectionist model to account for power-function forgetting curves by using bounded weights and by generating the learning rates from a monotonically decreasing function. The bounded weights introduce exponential forgetting in each weight and a power-function forgetting results when weights with different learning rates are averaged. It is argued that these assumptions are biologically reasonable. Therefore power-function forgetting curves are a property that may be expected from biological networks. The model has an analytic solution, which is a good approximation of a power function displaced one lag in time. This function fits better than any of the 105 suggested two-parameter forgetting-curve functions when tested on the most precise recognition memory data set collected by. Unlike the power-function normally used, the suggested function is defined at lag zero. Several functions for generating learning rates with a finite integral yield power-function forgetting curves; however, the type of function influences the rate of forgetting. It is shown that power-function forgetting curves cannot be accounted for by variability in performance between subjects because it requires a distribution of performance that is not found in empirical data. An extension of the model accounts for intersecting forgetting curves found in massed and spaced repetitions. The model can also be extended to account for a faster forgetting rate in item recognition (IR) compared to associative recognition in short but not long retention intervals.     

SLAM: a connectionist model for attention in visual selection tasks

SLAM, the SeLective Attention Model, performs visual selective attention tasks, an analysis of which shows that two processes, object and attribute selection, are both necessary and sufficient. It is based upon the McClelland and Rumelhart (1981) model for visual word recognition, with the addition of a response selection and evaluation mechanism. The responses may be correct or incorrect and, in particular conditions, SLAM may not make a response at all. Moreover, it allows for the generation of specific responses in time. SLAM's main characteristics are parallelism restricted by competition within modules, heterarchical processing in a hierarchical structure, and generation of responses as a result of relaxation given the conjoint constraints of stimulation, object, and attribute selection. The model is considered to represent an individual subject performing filtering tasks and demonstrates appropriate selective behavior. It is also tested quantitatively using a single tentative set of model parameters. The study reports simulations of four different filtering experiments, modeling response latencies, and error proportions. Specifications are made to take account of instructions, previous trials, and the effect of a barmarker cue and of asynchronies in stimulus and cue onsets. The model is then extended in order to provide simulations of a number of Stroop experiments, which can be regarded as filtering tasks with nonequivalent stimuli. The extension required for Stroop simulations is the addition of direct connections between compatible stimulus and response aspects. The direct connections do not affect the simulation of simpler filtering tasks. A variety of different experiments carried out by different authors is simulated. The model is discussed in terms of how modular architecture and the interaction of excitation and inhibition generate facilitation or inhibition of response latencies.     

The brain basis of a "consciousness monitor": scientific and medical significance.

Surgical patients under anesthesia can wake up unpredictably and be exposed to intense, traumatic pain. Current medical techniques cannot maintain depth of anesthesia at a perfectly stable and safe level; the depth of unconsciousness may change from moment to moment. Without an effective consciousness monitor anesthesiologists may not be able to adjust dosages in time to protect patients from pain. An estimated 40,000 to 200,000 midoperative awakenings may occur in the United States annually. E. R. John and coauthors present the scientific basis of a practical "consciousness monitor" in two articles. One article is empirical and shows widespread and consistent electrical field changes across subjects and anesthetic agents as soon as consciousness is lost; these changes reverse when consciousness is regained afterward. These findings form the basis of a surgical consciousness monitor that recently received approval from the U.S. Food and Drug Administration. This may be the first practical application of research on the brain basis of consciousness. The other John article suggests theoretical explanations at three levels, a neurophysiological account of anesthesia, a neural dynamic account of conscious and unconscious states, and an integrative field theory. Of these, the neurophysiology is the best understood. Neural dynamics is evolving rapidly, with several alternative points of view. The field theory sketched here is the most novel and controversial.     

Personality and mental time travel: a differential approach to autonoetic consciousness

Recent research on autonoetic consciousness indicates that the ability to remember the past and the ability to project oneself into the future are closely related. The purpose of the present study was to confirm this proposition by examining whether the relationship observed between personality and episodic memory could be extended to episodic future thinking and, more generally, to investigate the influence of personality traits on self-information processing in the past and in the future. Results show that Neuroticism and Harm Avoidance predict more negative past memories and future projections. Other personality dimensions exhibit a more limited influence on mental time travel (MTT). Therefore, our study provide an additional evidence to the idea that MTT into the past and into the future rely on a common set of processes by which past experiences are used to envision the future.     

The functional role of consciousness: A phenomenological approach

In this paper, a theoretical account of the functional role of consciousness in the cognitive system of normal subjects is developed. The account is based upon an approach to consciousness that is drawn from the phenomenological tradition. On this approach, consciousness is essentially peripheral self-awareness, in a sense to be duly explained. It will be argued that the functional role of consciousness, so construed, is to provide the subject with just enough information about her ongoing experience to make it possible for her to easily obtain as much information as she may need. The argument for this account of consciousness' functional role will proceed in three main stages. First, the phenomenological approach to consciousness as peripheral self-awareness will be expounded and endorsed. Second, an account of the functional role of peripheral perceptual awareness will be offered. Finally, the account of the functional role of peripheral self-awareness will be obtained by straightforward extension from the functional role of peripheral perceptual awareness. This revised version was published online in June 2006 with corrections to the Cover Date.     

TRACX: a recognition-based connectionist framework for sequence segmentation and chunk extraction

Individuals of all ages extract structure from the sequences of patterns they encounter in their environment, an ability that is at the very heart of cognition. Exactly what underlies this ability has been the subject of much debate over the years. A novel mechanism, implicit chunk recognition (ICR), is proposed for sequence segmentation and chunk extraction. The mechanism relies on the recognition of previously encountered subsequences (chunks) in the input rather than on the prediction of upcoming items in the input sequence. A connectionist autoassociator model of ICR, truncated recursive autoassociative chunk extractor (TRACX), is presented in which chunks are extracted by means of truncated recursion. The performance and robustness of the model is demonstrated in a series of 9 simulations of empirical data, covering a wide range of phenomena from the infant statistical learning and adult implicit learning literatures, as well as 2 simulations demonstrating the model's ability to generalize to new input and to develop internal representations whose structure reflects that of the items in the input sequence. TRACX outperforms PARSER (Perruchet & Vintner, 1998) and the simple recurrent network (SRN, Cleeremans & McClelland, 1991) in matching human sequence segmentation on existing data. A new study is presented exploring 8-month-olds' use of backward transitional probabilities to segment auditory sequences.     

A connectionist approach to the organization and continuity of working models of attachment

Most research on adult attachment dynamics has been conducted under the assumption that working models are generalized cognitive-motivational structures that are highly stable and activated in a wide array of circumstances. Recent research, however, suggests that people develop attachment representations that are relationship specific, leading them to hold distinct working models in different kinds of relationships. The objective of this article is to outline a connectionist approach to the working model construct that has the potential to explain how global and relationship-specific working models are organized within the same mental system and how different learning environments can support continuity in those representations over time.     

Learning to perceive object unity: a connectionist account

To explore questions of how human infants begin to perceive partly occluded objects, we devised two connectionist models of perceptual development. The models were endowed with an existing ability to detect several kinds of visual information that have been found important in infants’ and adults’ perception of object unity (motion, co‐motion, common motion, relatability, parallelism, texture and T‐junctions). They were then presented with stimuli consisting of either one or two objects and an occluding screen. The models’ task was to determine whether the object or objects were joined when such a percept was ambiguous, after specified amounts of training with events in which a subset of possible visual information was provided. The model that was trained in an enriched environment achieved superior levels of performance and was able to generalize veridical percepts to a wide range of novel stimuli. Implications for perceptual development in humans, current theories of development and origins of knowledge are discussed.     

Identification and integration of sensory modalities: Neural basis and relation to consciousness

A key question in studying consciousness is how neural operations in the brain can identify streams of sensory input as belonging to distinct modalities, which contributes to the representation of qualitatively different experiences. The basis for identification of modalities is proposed to be constituted by self-organized comparative operations across a network of unimodal and multimodal sensory areas. However, such network interactions alone cannot answer the question how sensory feature detectors collectively account for an integrated, yet phenomenally differentiated experiential content. This problem turns out to be different from, although related to, the binding problem. It is proposed that the neural correlate of an enriched, multimodal experience is constituted by the attractor state of a dynamic associative network. Within this network, unimodal and multimodal sensory maps continuously interact to influence each other’s attractor state, so that a feature change in one modality results in a fast re-coding of feature information in another modality. In this scheme, feature detection is coded by firing-rate, whereas firing phase codes relational aspects.