Concept learning and the intradimensional spread of the relevant values of the instances II. Effects of extreme values and position of class border

Abstract.— Thirty subjects took part in a concept-learning-like experiment with instances. the values of which varied continuously intradimensionally, which in turn resulted in continuously varyicg intradimensional differences between instances belonging to different classes. The results were construed as supporting the hypothesis that with subjects attending selectively to specific before-the-eyes characteristics of the instances, one hypotheses testing routine may be described as follows: Step 1: The subjects select and test intradimensional differences for relevance in the order of their relative sizes, and Step 2: on finding some such differences presumably relevant proceed to hypothesize midpoints or other easily registerable points on the involved dimension as class borders. This routine was thought to integrate various lines of thought within concept learning dealing with the effects on learning of intradimensional value separation, "obviousness" and relative frequencies of relevant attributes. It was also thought to represent a memory structure, and its relevance to some reasonings within memory research was pointed out.     

Concept learning and the intradimensional spread of the relevant values of the instances I. Effects of intraclass variability

Abstract.— The effect on concept learning of the intra-class variability of the relevant values of the instances was investigated. 20 subjects took part, and two ranges of the concerned intraclass variability, one about twice the other, were used. Depending upon the learning strategy used by the subjects the smaller intraclass variability was shown to have a limited facilitating effect on concept learning. Some facilitation thus occurred when strategies involved a rather unselective attention to the appearance characteristics of the instances but not when they involved selective focusing on these characteristics, or on elements of relevant knowledge established pre-experimentally. The results were discussed in relation both to relevant S-R-conceptions and information processing models of concept learning. It was concluded that automatic generalization as described in S-R-theories might have to be integrated into the information processing models for some categories of learning.     

Awareness, conditioning, and information processing in complex learning situations

Lindahl, Maj-Britt. Awareness, conditioning, and information processing in complex learning situations. Scand. J. Psychol., 1973, 14, 121–130.-Learning without awareness was studied with a method involving three phases for each of 2. 10 subjects: (1) a problem solving phase, (2) a phase of applying the solution found in (1), and (3) a test phase. In (2) the solution was applied to the classification of 50 successively presented instances. These were in one condition sampled so that a perceptually easily discriminable basis of classification obtained that was different from and redundant to the one of the solution of phase (1). The results of the test phase showed that this redundant basis of classification influenced the performance of the subjects in this condition despite unawareness of it according to verbal reports. The phenomenon was interpreted as a case of learning without awareness, describable in conditioning terms, that influenced performance. The limitations of this type of learning were discussed, and its relationship to an aware information processing type of learning was outlined.     

Immediate reproduction of temporal information under four cognitive strategies

The main purpose was to determine the retention characteristics of temporal information when subjects experienced time under a retention interval of immediate reproduction and various cognitive strategies for time estimation. Four levels of cognitive strategy were used, viz., conscious, mental counting, counting aloud without auditory cues, and counting aloud with auditory cues. The latter three cognitive strategies were experimenter-defined, time-aiding techniques. Subjects were instructed to refrain from employing time-aiding techniques under a conscious cognitive strategy for time estimation. Visual durations of 1, 2, and 4 sec. were estimated by 12 subjects under the method of reproduction. Two measures of performance were computed, viz., variable and constant errors. The general conclusions were: (a) the effectiveness of mental counting, counting aloud without auditory cues, and counting aloud with auditory cues as cognitive strategies over conscious cognitive strategy in terms of variability depends on the duration used, and (b) in terms of accuracy and variability an increase in the number of cues under time-aiding techniques does not necessarily produce better performance.     

How do people solve the "weather prediction" task?: individual variability in strategies for probabilistic category learning

Probabilistic category learning is often assumed to be an incrementally learned cognitive skill, dependent on nondeclarative memory systems. One paradigm in particular, the weather prediction task, has been used in over half a dozen neuropsychological and neuroimaging studies to date. Because of the growing interest in using this task and others like it as behavioral tools for studying the cognitive neuroscience of cognitive skill learning, it becomes especially important to understand how subjects solve this kind of task and whether all subjects learn it in the same way. We present here new experimental and theoretical analyses of the weather prediction task that indicate that there are at least three different strategies that describe how subjects learn this task. (1) An optimal multi-cue strategy, in which they respond to each pattern on the basis of associations of all four cues with each outcome; (2) a one-cue strategy, in which they respond on the basis of presence or absence of a single cue, disregarding all other cues; or (3) a singleton strategy, in which they learn only about the four patterns that have only one cue present and all others absent. This variability in how subjects approach this task may have important implications for interpreting how different brain regions are involved in probabilistic category learning.     

Search strategies in decision making: the success of “success”

Examination of search strategies has tended to focus on choices determined by decision makers' personal preferences among relevant cues, and not on learning cue‐criterion relationships. We present an empirical and rational analysis of cue search for environments with objective criteria. In such environments, cues can be evaluated on the basis of three properties: validity (the probability that a cue identifies the correct choice if cue values differ between alternatives); discrimination rate (the proportion of occasions on which a cue has differing values); and success (the expected proportion of correct choices when only that cue can be used). Our experiments show that though there is a high degree of individual variability, success is a key determinant of search. Furthermore, a rational analysis demonstrates why success‐directed search is the most adaptive strategy in many circumstances. Copyright © 2004 John Wiley & Sons, Ltd.     

Selective attention and the breadth of learning: A developmental study

Nursery school and second-grade subjects were trained on an optional intradimensional/extradimensional shift task with (1) no overtraining, (2) overtraining on the initial problem only, (3) overtraining on the shift problem only, or (4) overtraining on both the initial and the shift problems. Predictions concerning the effects of age and training conditions on the type of solution and the breadth of learning for the shift problem were derived from selective attention theory. However, the results were not consistent with the one-look assumption of such models. Instead, a multiple-look theory in which the breadth of attention varies with task demands seems most tenable.     

Where Has Our Theory Gone? Learning Theory and Intentional Intervention

Reflecting on the principle that every effective counseling intervention works to the extent that it incorporates sound learning principles, this article emphasizes intentional selection of intervention strategies with reference to 4 learning models: association, reinforcement, cognition‐perceptual, and cognition‐rational/linguistic. Three contexts—developmental, social, and spiritual—are described that require modification of intervention strategy, and a case example is given.     

Induction of a principle

Two experiments are reported in which student subjects attempted to discover a principle obtaining among pairs of numbers and letters. In the first experiment, subjects were more successful when they were free to select whatever number-letter pairs they wished than if they were restricted in whole or in part to pairs specified by the experimenter. In the second experiment, subjects who did discover the principle were compared to those who did not. Successful subjects were shown to be slightly more systematic in their approach to the task, to work at a faster pace, to write down more positive instances, and to have a much stronger tendency to vary only one variable of the task at a time.

Wason (1960) reported a study in which subjects tried to discover a principle applying to instances each of which consisted of three numbers. He discovered that subjects mostly showed enumerative rather than eliminative induction, i.e. they made little use of instances that would have enabled them to eliminate wrong hypotheses. Wetherick (1962) found that by modifying Wason's procedure in certain ways, enumerative behaviour was reduced and the subject's chance of discovering the principle was increased. Elimination of hypotheses remained infrequent.

In the Wason study, the subject was free to write down, for each instance, any three numbers he wished, i.e. his choice of instances was not restricted in any way. This may account in part for the behaviour Wason observed. In the first experiment to be reported here, the subject is restricted, in various degrees, in his choice of instances, to determine whether the principle is discovered more (or less) readily under such restriction.     

Strategies in probabilistic categorization: results from a new way of analyzing performance

The "Weather Prediction" task is a widely used task for investigating probabilistic category learning, in which various cues are probabilistically (but not perfectly) predictive of class membership. This means that a given combination of cues sometimes belongs to one class and sometimes to another. Prior studies showed that subjects can improve their performance with training, and that there is considerable individual variation in the strategies subjects use to approach this task. Here, we discuss a recently introduced analysis of probabilistic categorization, which attempts to identify the strategy followed by a participant. Monte Carlo simulations show that the analysis can, indeed, reliably identify such a strategy if it is used, and can identify switches from one strategy to another. Analysis of data from normal young adults shows that the fitted strategy can predict subsequent responses. Moreover, learning is shown to be highly nonlinear in probabilistic categorization. Analysis of performance of patients with dense memory impairments due to hippocampal damage shows that although these patients can change strategies, they are as likely to fall back to an inferior strategy as to move to more optimal ones.     

Practice variability promotes an external focus of attention and enhances motor skill learning

Variability in practice has been shown to enhance motor skill learning. Benefits of practice variability have been attributed to motor schema formation (variable versus constant practice), or more effortful information processing (random versus blocked practice). We hypothesized that, among other mechanisms, greater practice variability might promote an external focus of attention on the intended movement effect, while less variability would be more conducive to a less effective internal focus on body movements. In Experiment 1, the learning of a throwing task was enhanced by variable versus constant practice, and variable group participants reported focusing more on the distance to the target (external focus), while constant group participants focused more on their posture (internal focus). In Experiment 2, golf putting was learned more effectively with a random compared with a blocked practice schedule. Furthermore, random group learners reported using a more effective distal external focus (i.e., distance to the target) to a greater extent, whereas blocked group participants used a less effective proximal focus (i.e., putter) more often. While attentional focus was assessed through questionnaires in the first two experiments, learners in Experiment 3 were asked to report their current attentional focus at any time during practice. Again, the learning of a throwing task was more effective after random relative to blocked practice. Also, random practice learners reported using more external focus cues, while in blocked practice participants used more internal focus cues. The findings suggest that the attentional foci induced by different practice schedules might be at least partially responsible for the learning differences.     

Extradimensional transfer in the easy-to-hard effect

In three experiments analyzing determinants of the easy-to-hard effect, pigeons acquired a hard discrimination after training on other problems. Intradimensional pretraining resulted in immediate transfer to the hard discrimination. Extradimensional pretraining consistently did not produce immediate transfer but did facilitate learning rate. In Experiment 1, the compounding of cues from an easy extradimentsional discrimination with those from the hard discrimination resulted in the former overshadowing the latter. When different types of extradimensional discriminations were introduced in Experiments 2 and 3, the degree of transfer was not proportional to the similarity in incidental background cues across problems. The findings indicate that in the easy-to-hard effect: (a) intra- and extradimensional mechanisms jointly contribute to the development of stimulus control, (b) intradimensional transfer is more consistent with the gradient-interaction model than the selective attention model, and (c) extradimensional transfer is better accounted for by the construct of general attentiveness rather than by the neutralization of background cues.     

The influence of cue unreliability on judgment in a multiple cue probability learning task

The effect of cue reliability was tested in a two-cue multiple-cue probability learning (MCPL) task. Subjects were 42 undergraduates. The cue validities were.78 and.37. Unreliability in a cue was defined as variability in multiple observations of that cue on a given trial. That variability was provided by adding random errors to the true value of the cue. One group was given consistent cues (i.e., the more valid cue was the more reliable cue), one group was given inconsistent cues, and the control group was given true scores. Cue reliability did not affect subjects' consistency or achievement, but did interact with the presence of outcome feedback. The subjective weights showed that subjects thought they weighted the more reliable cue more heavily regardless of the cue validities or their actual cue weightings. There were wide individual differences in performance, illustrating the need for an idiographic-statistical approach in studying decision making.     

Persistence of the spacing effect in free recall under varying incidental learning conditions

In three incidental learning experiments, an attempt was made to eliminate the processing deficit under massed presentation that is assumed to be responsible for the spacing effect in free recall, according to the attenuation of attention hypothesis. This was to be accomplished in Experiment I by requiring subjects to attend specifically to the total exposure duration of each item and in Experiments II and III by requiring subjects to rate the successive occurrences of repeated items on different semantic rating scales. The results of the three experiments consistently showed that these manipulations were ineffective in eliminating the spacing effect. Subsidiary analyses indicated that the activities involved in doing the semantic rating tasks do not provide direct access to retrieval cues useful for subsequent recall. Instead, it appears that, in order to perform the semantic rating tasks reliably, subjects must compare the to-be-rated item with previously rated items, and this comparison process may serve as the source of retrieval cues for subsequent recall.     

STAGES IN CONCEPT FORMATION

Previous studies revealed a pre-functional stage of concept learning, in that the relevant stimulus attribute aquired perceptual distinctiveness at a stage of learning when subjects as yet were unable to sort instances into appropriate categories. This early stage of perceptual learning prior to behavioral manifestation could not be identified under the novel conditions introduced in the present experiment: sensory predifferentiation training and an extra request to verbalize the concept.     

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.