What you learn is what you see: using eye movements to study infant cross-situational word learning

Recent studies show that both adults and young children possess powerful statistical learning capabilities to solve the word-to-world mapping problem. However, the underlying mechanisms that make statistical learning possible and powerful are not yet known. With the goal of providing new insights into this issue, the research reported in this paper used an eye tracker to record the moment-by-moment eye movement data of 14-month-old babies in statistical learning tasks. Various measures are applied to such fine-grained temporal data, such as looking duration and shift rate (the number of shifts in gaze from one visual object to the other) trial by trial, showing different eye movement patterns between strong and weak statistical learners. Moreover, an information-theoretic measure is developed and applied to gaze data to quantify the degree of learning uncertainty trial by trial. Next, a simple associative statistical learning model is applied to eye movement data and these simulation results are compared with empirical results from young children, showing strong correlations between these two. This suggests that an associative learning mechanism with selective attention can provide a cognitively plausible model of cross-situational statistical learning. The work represents the first steps in using eye movement data to infer underlying real-time processes in statistical word learning.     

The Suitability of an Inhibitory Learning Approach in Exposure When Habituation Fails: A Clinical Application to Misophonia

Recent findings have led to a reconceptualization of the mechanisms that account for the efficacy of exposure-based treatments. Termed the “inhibitory learning model,” this approach emphasizes new learning when confronted with previously avoided stimuli rather than merely the cessation of fear or aversive emotional responding. In this paper, we propose the applicability of the inhibitory learning model for conditions and contexts in which simple exposure does not produce habituation. We illustrate this application from an in-progress randomized controlled treatment trial for adults with misophonia. Misophonia is a condition marked by strong aversive reactions to specific sounds. It is a difficult to treat and understudied syndrome. All participants in the trial received exposure, either before or after a stress management module of treatment. Exposure treatment emphasized altered expectancies for the target sounds as well as deliberate practice in hearing sounds on the individually developed hierarchy. Inhibitory learning strategies were employed to increase treatment adherence and commitment, shape patient behavior during exposures, manufacture negative prediction errors, increase perceived control over reactions, and promote learning that generalized to functional improvements. The findings are discussed in the context of future applications of the inhibitory learning model for psychopathology associated with avoidance.     

Metacognitive monitoring during category learning: how success affects future behaviour

The purpose of this study was to see how people perceive their own learning during a category learning task, and whether their perceptions matched their performance. In two experiments, participants were asked to learn natural categories, of both high and low variability, and make category learning judgements (CLJs). Variability was manipulated by varying the number of exemplars and the number of times each exemplar was presented within each category. Experiment 1 showed that participants were generally overconfident in their knowledge of low variability families, suggesting that they considered repetition to be more useful for learning than it actually was. Also, a correct trial, for a particular category, was more likely to occur if the previous trial was correct. CLJs had the largest increase when a trial was correct following an incorrect trial and the largest decrease when an incorrect trial followed a correct trial. Experiment 2 replicated these results, but also demonstrated that global CLJ ratings showed the same bias towards repetition. These results indicate that we generally identify success as being the biggest determinant of learning, but do not always recognise cues, such as variability, that enhance learning.     

Influence of manipulated category knowledge on prototype classification and recognition

The recognition and classification of category members was explored, following a variable number of learning trials. In Experiment 1, subjects received 1 or 9 learning trials, followed by a recognition-then-classification test containing old, new, prototype, and foil patterns. In Experiment 2, subjects received 1, 6, or 12 trials, and made either classification or recognition judgments. In each experiment, classification accuracy for all item types was at near-chance -performance after a single trial but steadily increased with increased learning trials. On the transfer test, oldness judgments were highest for the category prototype after a single trial. However, with increased learning trials, oldness judgments increased for old instances and decreased for the category prototype and new instances. We suggest that false recognition of the category prototype, especially after a single learning trial, need not reflect an abstraction process. We discuss the possibility that an abstracted prototype may emerge with additional learning as an unfamiliar, ideal point.     

Effects of test anxiety on motor learning: The processing efficiency hypothesis

Abstract

The hypothesis that, under test (stress) conditions, high trait-anxious individuals invest extra processing resources to reach motor learning levels similar to those of low-anxious individuals was investigated. A transfer procedure (practice trial, rest interval, and transfer trial) was employed to analyse motor learning effects. In Experiment 1, a secondary task (click detection) was used to measure the extra processing resources expenditure on motor learning during the practice trial. Neither motor performance in the transfer trial nor secondary task performance differed as a function of test condition or anxiety trait. A critical analysis of the secondary task paradigm suggested that it was not sensitive enough to reflect the amount of processing capacity expended on the primary motor tasks. In Experiment 2, subjects performed a loading task (verbal shadowing of sentences) concurrently with the motor tasks during the practice trial to prevent extra resources from being allocated to motor learning. Results showed that anxiety trait was associated to performance time increments on the high-demanding motor tasks in the transfer trial. A plausible explanation is that anxiety impairs learning because it reduces the efficiency of processing. An alternative interpretation, concerning an ability deficit of high trait-anxious individuals, is discussed.     

Auditory Model Enhances Relative-Timing Learning

In recent experiments (Q. Lai, C. H. Shea, & M. Little, 2000; C. H. Shea, G. Wulf, J. Park, & B. Gaunt, 2001), auditory models were found to be effective in enhancing relative-timing performance and learning in constant practice conditions. In the present experiment, the authors examined (a) whether an auditory model also enhances relative-timing learning in blocked and random practice conditions and (b) whether experience with the auditory model enhances participants' ability to produce the response by using different effector sequences. Participants (N = 48) were randomly assigned to 1 of 4 acquisition conditions in which an auditory model was or was not present and the practice schedule was blocked or random. In the acquisition phase, all participants alternately pressed 2 keys on a keyboard in an attempt to match 5 goal intervals. In the auditory model conditions, a sequence of tones was played before each acquisition trial. The tones represented the correct timing sequence for that trial. One retention and 3 effector-transfer tests without feedback were administered about 24 hr after the completion of acquisition. The results indicated that the auditory model enhanced relative-timing performance and learning independently of practice schedule. In addition, the auditory model enhanced relative timing on the effector-transfer tests, but absolute-timing performance and learning were not affected by the auditory model. Thus, relative timing was found to be effector independent but absolute timing was not.     

Studies in observational paired-associate learning: Informational,social, and individual difference variables

Four experiments examined the relative importance of informational (proportion of correct responses and kinds of errors emitted by a model), social (model competency, sex of model, video vs. audio taped model), and individual difference (sex of subject, grade) variables in observational paired-associate learning. In Experiments I–III, vicarious subjects received cycles of study-model-test trials, while direct subjects were given the same sequences with intervening test or stimulus familiarization trials. In Experiment IV, vicarious subjects received cycles of study-test-model-test trials, while direct subjects received the cycles with a test trial replacing the model trial. No confirmation was provided on test and model trials. Whereas the effects attributable to social and individual difference variables were generally negligible, mere accuracy of the model's responses repeatedly covaried with performance on the last test trial of each cycle. Conditional analyses established that (1) vicarious facilitation is comparable across cycles and localized in items responded to incorrectly on immediately preceding test trials, and (2) observers learn fewer incorrect than correct model responses. Vicarious groups performed at reliably higher levels than direct subjects on model correct but not incorrect items. The results strongly suggest a close correspondence between direct and vicarious verbal learning principles and mechanisms.     

Sequential congruency effects in implicit sequence learning

We deal with situations incongruent with our automatic response tendencies much better right after having done so on a previous trial than after having reacted to a congruent trial. The nature of the mechanisms responsible for these sequential congruency effects is currently a hot topic of debate. According to the conflict monitoring model these effects depend on the adjustment of control triggered by the detection of conflict on the preceding situation. We tested whether these conflict monitoring processes can operate implicitly in an implicit learning procedure, modulating the expression of knowledge of which participants are not aware. We reanalyze recently published data, and present an experiment with a probabilistic sequence learning procedure, both showing consistent effects of implicit sequence learning. Despite being implicit, the expression of learning was reduced or completely eliminated right after trials incongruent with the learned sequence, thus showing that sequential congruency effects can be obtained even when the source of congruency itself remains implicit.     

Multiple imitation mechanisms in children

Four studies using a computerized paradigm investigated whether children's imitation performance is content-specific and to what extent dependent on other cognitive processes such as trial-and-error learning, recall, and observational learning. Experiment 1 showed that 3-year-olds could successfully imitate what we call novel cognitive rules (e.g., first → second → third), which involved responding to 3 different pictures whose spatial configuration varied randomly from trial to trial. However, these same children failed to imitate what we call novel motor-spatial rules (e.g., up → down → right), which involved responding to 3 identical pictures that remained in a fixed spatial configuration from trial to trial. Experiment 2 showed that this dissociation was not due to a general difficulty in encoding motor-spatial content, as children successfully recalled, following a 30-s delay, a new motor-spatial sequence that had been learned by trial and error. Experiment 3 replicated these results and further demonstrated that 3-year-olds can infer a novel motor-spatial sequence following observation of a partially correct and partially incorrect response-a dissociation between imitation and observational learning (or emulation learning). Finally, Experiment 4 presented 3-year-olds with "familiar" motor-spatial sequences that involved making a linear response (e.g., left → middle → right) as well as "novel" motor-spatial sequences (e.g., right → up → down) used in Experiments 1-3 that were nonlinear and always involved a change in direction. Children had no difficulty imitating familiar motor-spatial sequences but again failed to imitate novel motor-spatial sequences. These results suggest that there may be multiple, dissociable imitation learning mechanisms that are content-specific. More importantly, the development of these imitation systems appears to be independent of the operations of other cognitive systems, including trial and error learning, recall, and observational learning.     

Computer simulations of the Rescorla-Wagner and Pearce-Hall models in conditioning and contingency judgment

A Visual BASIC program, running under Windows 3.1, simulated the predictions of the Rescorla—Wagner (Rescorla & Wagner, 1972) and the Pearce-Hall (Pearce & Hall, 1980) models and compared them to the normative contingency coefficient ΔP (Jenkins & Ward, 1965). The simulations can be applied to a variety of phenomena in human contingency judgment as well as learning and conditioning. Possible simulations include acquisition and extinction of excitatory and inhibitory conditioning, latent inhibition, blocking and overshadowing, or any other associative learning involving two single predictors, their compound, a contextual stimulus, and an outcome. The Pearce-Hall model has never been computerized before. In addition, unique features of this software include extensive use of the graphic user interface, context-sensitive help, verification of trial combinations, toggling of the contextual stimulus from ever-present to mutually exclusive with the discrete predictors, data entry via contingency tables or specifications trial by trial, single or batch randomizations of trial order, and specification of initial values. The associative simulator is both a powerful scientific instrument and a user-friendly teaching aid.     

Flexible shaping: How learning in small steps helps

Humans and animals can perform much more complex tasks than they can acquire using pure trial and error learning. This gap is filled by teaching. One important method of instruction is shaping, in which a teacher decomposes a complete task into sub-components, thereby providing an easier path to learning. Despite its importance, shaping has not been substantially studied in the context of computational modeling of cognitive learning. Here we study the shaping of a hierarchical working memory task using an abstract neural network model as the target learner. Shaping significantly boosts the speed of acquisition of the task compared with conventional training, to a degree that increases with the temporal complexity of the task. Further, it leads to internal representations that are more robust to task manipulations such as reversals. We use the model to investigate some of the elements of successful shaping.     

The Effects of Response Repetitions on Sight Word Acquisition for Students with Mild Disabilities

Single response learning trials, defined in terms of the three-term contingency model, were compared with repeated response learning trials. The purpose of Study 1 was to assess the effectiveness and efficiency of single and repeated response learning trials on sight word remediation. In Study 1, one single response learning trial having one response opportunity per word was compared to one repeated response learning trial having five response opportunities. In Study 2, three single response trials were compared to three repeated response trials. In both studies, the training conditions were compared using an alternating treatments design. Measuring the cumulative number of sight words mastered showed the training conditions equally effective. However, measuring (a) the mean duration of training sessions, (b) the cumulative frequency of words mastered per session duration, (c) the mean number of training sessions per word required for mastery, and (d) the training time per mastered word showed that the single response condition was more efficient. Discussion focuses on the need to include training data when determining best practices for teaching.     

Conditional visuo-motor learning and dimension reduction

Conditional visuo-motor learning consists in learning by trial and error to associate visual cues with correct motor responses, that have no direct link. Converging evidence supports the role of a large brain network in this type of learning, including the prefrontal and the premotor cortex, the basal ganglia (BG) and the hippocampus. In this paper we focus on the role of a major structure of the BG, the striatum. We first present behavioral results and electrophysiological data recorded from this structure in monkeys engaged in learning new visuo-motor associations. Visual stimuli were presented on a video screen and the animals had to learn, by trial and error, to select the correct movement of a joystick, in order to receive a liquid reward. Behavioral results revealed that the monkeys used a sequential strategy, whereby they learned the associations one by one although they were presented randomly. Human subjects, tested on the same task, also used a sequential strategy. Neuronal recordings in monkeys revealed learning-related modulations of neural activity in the striatum. We then present a mathematical model inspired by viability theory developed to implement the use of strategies during learning. This model complements existing models of the BG based on reinforcement learning (RL), which do not take into account the use of strategies to reduce the dimension of the learning space.     

Three experiences affecting spatial discrimination learning by ambulatory children

Children 1 and 3 years of age were given a two-choice spatial discrimination task. They were placed in front of a barrier and encouraged to walk around it; one route around the barrier was blocked. Three experiments studied the effects of different types of experience with this spatial problem. The first examined trial and error experience, in which the child was allowed to walk the route he or she had chosen. Although most children readily learned to choose the open route, the pattern of choices indicated that 3-year-olds were more likely to achieve a learning criterion after an error than 1-year-olds. A second experiment looked at the effects of showing the children the layout of the problem prior to trial and error experience. The 3-year-olds were again more likely than the 1-year-olds to achieve criterion after an error; furthermore, they were likely to know the correct route on the first trial. The third experiment let the children watch their parents take the correct route. This procedure eliminated age differences in performance. The initial choices of the 1-year-olds indicated that they were likely to know the correct route, and if they did not, they were just as likely to learn from their mistakes as the 3-year-olds. The data were interpreted within a mathematical model of learning. Based upon these analyses, we propose two developmental trends. One involves a growth in sensitivity to the consequences of a choice of route. The other is a progression from social learning to more independent and ideational methods of spatial problem solving.     

Resource allocation decision making in an uncertain environment

This experiment investigated how individuals learn to allocate limited resources across competing activities in order to maximize their objective when the form of the objective function is uncertain. On each of 25 trials, subjects selected an allocation policy and received a corresponding profit. The objective function that was used to assign profits to allocations was unknown initially and was learned through trial by trial outcome feedback. A total of 144 subjects were randomly assigned to nine groups constructed from a 3 (objective function form) by 3 (amount of error variability) factorial design. Several measures of learning were analyzed including (a) the distance between the current and optimal allocation policy as a function of training, (b) the magnitude of change in allocation as a function of past profits, and (c) the direction of change in allocations as a function of the direction of change in past profits. The results provide evidence for an integrated approach to learning — a functional learning process based on global information about the shape of the objective function and a hill-climbing learning process based on local information about trial to trial improvements in profit.     

Task switching: a PDP model

When subjects switch between a pair of stimulus-response tasks, reaction time is slower on trial N if a different task was performed on trial N - 1. We present a parallel distributed processing (PDP) model that simulates this effect when subjects switch between word reading and color naming in response to Stroop stimuli. Reaction time on "switch trials" can be slowed by an extended response selection process which results from (a) persisting, inappropriate states of activation and inhibition of task-controlling representations; and (b) associative learning, which allows stimuli to evoke tasks sets with which they have recently been associated (as proposed by Allport & Wylie, 2000). The model provides a good fit to a large body of empirical data, including findings which have been seen as problematic for this explanation of switch costs, and shows similar behavior when the parameters are set to random values, supporting Allport and Wylie's proposal.     

The sensitivity of judgment-of-learning resolution to past test performance,new learning,and forgetting

When people judge their learning of items across study–test trials, their accuracy in discriminating between learned and unlearned items improves on the second trial. We examined the source of this improvement by estimating the contribution of three factors—memory for past test performance (MPT), new learning, and forgetting—to accuracy on trial 2. In Experiment 1, during an initial trial, participants studied paired associates, made a judgment of learning (JOL) for each one, and were tested. During the second trial, we manipulated two variables: when the JOL was made (either immediately before or after studying an item) and whether participants were told the outcome of the initial recall attempt on trial 1. In Experiment 2, the same procedure was used with a 1-week retention interval between study and test on trial 2. In both experiments, JOL resolution was higher on trial 2 than on trial 1. Fine-grained analyses of JOL magnitude and decomposition of resolution supported several conclusions. First, MPT contributed the most to boosts in JOL magnitude and improvements in resolution across trials. Second, JOLs and subsequent resolution were sensitive to new learning and forgetting, but only when participants’ judgments were made after study. Thus, JOLs appear to integrate information from multiple factors, and these factors jointly contribute to JOL resolution.     

Visuospatial working memory in school-aged children exposed in utero to cocaine.

OBJECTIVE: Among the neurocognitive impairments reported as associated with prenatal cocaine exposure, slower response time, and less efficient learning in school-aged children are common to findings from several laboratories. This study presents performance data on a spatial working memory task in 75 prenatally cocaine exposed (CE) and 55 nondrug-exposed (NDE) 8- to 10-year-old children. METHODS: Children were administered a novel neuropsychological measure of immediate- and short-term memory for visuospatial information, the Groton Maze Learning Test (GMLT), a computer-based hidden maze learning test that consists of a "timed chase test" (a simple measure of visuomotor speed), eight learning trials followed by a delayed recall trial after an 8-minute delay and a reverse learning trial. Performance is expressed as correct moves per second and number of errors per trial. RESULTS: Across all trials, the cocaine-exposed group showed significantly slower correct moves per second and made significantly more errors. There were no significant main effects for amounts of alcohol, tobacco, or marijuana exposure. After an 8-minute delay and compared to the eighth trial, cocaine-exposed children showed less consolidation in learning compared to nonexposed children. When asked to complete the maze in reverse, cocaine-exposed children showed a greater decrement in performance (decreased correct moves per second and increased errors) compared to the eighth learning trial. CONCLUSIONS: Children exposed in utero to cocaine exhibit a possible impairment in procedural learning and diminished efficiency in creating and accessing an internal spatial map to master the hidden maze.     

Derivation of learning process statistics for a general markov model

A Markov learning model may be stated in the form of a transition matrix, starting vector, and response probability vector. Utilizing these and some general properties of absorbing Markov chains, general expressions are derived for several statistics of the learning process which can be applied to any model of this form. Included are derivations for the mean learning curve, number of total errors, trial numbers of the first success and the last error, and the number of error runs. As an illustration, all derivations are worked out for the simple two-state one-element model.