Competing strategies in categorization: expediency and resistance to knowledge restructuring

The authors investigated people's ability to restructure their knowledge when additional information about a categorization task is revealed. In 2 experiments, people first learned to rely on a fairly accurate (but imperfect) predictor. At various points in training, a complex relationship between 2 other predictors was revealed in a schematic diagram that could support perfect performance. In Experiment 1, people adopted the complex strategy when it was revealed at the outset but were unable to restructure their knowledge after the expedient predictor had been learned. In Experiment 2, expedient knowledge persisted even with an adaptive display. The persistence of expedient knowledge is explained by associative blocking of potential alternative cues. A 3rd experiment analyzed the strategies people use with and without the diagram. The study confirmed that the diagram, when presented at the outset, significantly alters people's approach to the task.     

Restructuring partitioned knowledge: the role of recoordination in category learning

Knowledge restructuring refers to changes in the strategy with which people solve a given problem. Two types of knowledge restructuring are supported by existing category learning models. The first is a relearning process, which involves incremental updating of knowledge as learning progresses. The second is a recoordination process, which involves novel changes in the way existing knowledge is applied to the task. Whereas relearning is supported by both single- and multiple-module models of category learning, only multiple-module models support recoordination. To date, only relearning has been directly supported empirically. We report two category learning experiments that provide direct evidence of recoordination. People can fluidly alternate between different categorization strategies, and moreover, can reinstate an old strategy even after prolonged use of an alternative. The knowledge restructuring data are not well fit by a single-module model (ALCOVE). By contrast, a multiple-module model (ATRIUM) quantitatively accounts for recoordination. Low-level changes in the distribution of dimensional attention are shown to subsequently affect how ATRIUM coordinates its modular knowledge. We argue that learning about complex tasks occurs at the level of the partial knowledge elements used to generate a response strategy.     

Perspectives of probabilistic inferences: Reinforcement learning and an adaptive network compared

The assumption that people possess a strategy repertoire for inferences has been raised repeatedly. The strategy selection learning theory specifies how people select strategies from this repertoire. The theory assumes that individuals select strategies proportional to their subjective expectations of how well the strategies solve particular problems; such expectations are assumed to be updated by reinforcement learning. The theory is compared with an adaptive network model that assumes people make inferences by integrating information according to a connectionist network. The network's weights are modified by error correction learning. The theories were tested against each other in 2 experimental studies. Study 1 showed that people substantially improved their inferences through feedback, which was appropriately predicted by the strategy selection learning theory. Study 2 examined a dynamic environment in which the strategies' performances changed. In this situation a quick adaptation to the new situation was not observed; rather, individuals got stuck on the strategy they had successfully applied previously. This "inertia effect" was most strongly predicted by the strategy selection learning theory.     

The Impact of Goal Specificity on Strategy Use and the Acquisition of Problem Structure

Theories of skill acquisition have made radically different predictions about the role of general problem-solving methods in acquiring rules that promote effective transfer to new problems. Under one view, methods that focus on reaching specific goals, such as means-ends analysis, are assumed to provide the basis for efficient knowledge compilation (Anderson, 1987), whereas under an alternative view such methods are believed to disrupt rule induction (Sweller, 1988). We suggest that the role of general methods in learning varies with both the specificity of the problem solver's goal and the systematicity of the strategies used for testing hypotheses about rules. In the absence of a specific goal people are more likely to use a rule-induction learning strategy, whereas provision of a specific goal fosters use of difference reduction, which tends to be a non-rule-induction strategy. We performed two experiments to investigate the impact of goal specificity and systematicity of rule-induction strategies in learning and transfer within a complex dynamic system. The results of Experiment 1 indicated that during free exploration of a problem space, greater learning occurred if participants adopted more systematic strategies for rule induction, and that participants come to favor such strategies. Experiment 2 revealed that participants who were provided with a specific goal performed well on the initial problem but were impaired on a transfer test using a similar problem with a different goal. Instruction on a systematic rule-induction strategy facilitated solution for both the initial and transfer problems, but participants' use of this strategy declined if they had a specific goal. Our results support Sweller's (1988) proposal that general problemsolving methods applied to a specific goal foster acquisition of knowledge about an isolated solution path but do not provide an effective way of learning the overall structure of a problem space. We interpret these results in terms of dualspace theories of search through problem space.     

Why Higher Working Memory Capacity May Help You Learn: Sampling,Search, and Degrees of Approximation

Algorithms for approximate Bayesian inference, such as those based on sampling (i.e., Monte Carlo methods), provide a natural source of models of how people may deal with uncertainty with limited cognitive resources. Here, we consider the idea that individual differences in working memory capacity (WMC) may be usefully modeled in terms of the number of samples, or “particles,” available to perform inference. To test this idea, we focus on two recent experiments that report positive associations between WMC and two distinct aspects of categorization performance: the ability to learn novel categories, and the ability to switch between different categorization strategies (“knowledge restructuring”). In favor of the idea of modeling WMC as a number of particles, we show that a single model can reproduce both experimental results by varying the number of particles—increasing the number of particles leads to both faster category learning and improved strategy‐switching. Furthermore, when we fit the model to individual participants, we found a positive association between WMC and best‐fit number of particles for strategy switching. However, no association between WMC and best‐fit number of particles was found for category learning. These results are discussed in the context of the general challenge of disentangling the contributions of different potential sources of behavioral variability.     

Attention and working memory capacity: insights from blocking, highlighting, and knowledge restructuring

The concept of attention is central to theorizing in learning as well as in working memory. However, research to date has yet to establish how attention as construed in one domain maps onto the other. We investigate two manifestations of attention in category- and cue-learning to examine whether they might provide common ground between learning and working memory. Experiment 1 examined blocking and highlighting effects in an associative learning paradigm, which are widely thought to be attentionally mediated. No relationship between attentional performance indicators and working memory capacity (WMC) was observed, despite the fact that WMC was strongly associated with overall learning performance. Experiment 2 used a knowledge restructuring paradigm, which is known to require recoordination of partial category knowledge using representational attention. We found that the extent to which people successfully recoordinated their knowledge was related to WMC. The results illustrate a link between WMC and representational-but not dimensional-attention in category learning.     

Sensitivity to Shared Information in Social Learning

Social learning has been shown to be an evolutionarily adaptive strategy, but it can be implemented via many different cognitive mechanisms. The adaptive advantage of social learning depends crucially on the ability of each learner to obtain relevant and accurate information from informants. The source of informants’ knowledge is a particularly important cue for evaluating advice from multiple informants; if the informants share the source of their information or have obtained their information from each other, then their testimony is statistically dependent and may be less reliable than testimony from informants who do not share information. In this study, we use a Bayesian model to determine how rational learners should incorporate the effects of shared information when learning from other people, conducting three experiments that examine whether human learners behave similarly. We find that people are sensitive to a number of different patterns of dependency, supporting the use of a sophisticated strategy for social learning that goes beyond copying the majority, and broadening the situations in which social learning is likely to be an adaptive strategy.     

Ad hoc category restructuring

Participants learned to classify seemingly arbitrary words into categories that also corresponded to ad hoc categories (see, e.g., Barsalou, 1983). By adapting experimental mechanisms previously used to study knowledge restructuring in perceptual categorization, we provide a novel account of how experimental and preexperimental knowledge interact. Participants were told of the existence of the ad hoc categories either at the beginning or the end of training. When the ad hoc labels were revealed at the end of training, participants switched from categorization based on experimental learning to categorization based on preexperimental knowledge in some, but not all, circumstances. Important mediators of the extent of that switch were the amount of performance error experienced during prior learning and whether or not prior knowledge was in conflict with experimental learning. We present a computational model of the trade-off between preexperimental knowledge and experimental learning that accounts for the main results.     

Error discounting in probabilistic category learning

The assumption in some current theories of probabilistic categorization is that people gradually attenuate their learning in response to unavoidable error. However, existing evidence for this error discounting is sparse and open to alternative interpretations. We report 2 probabilistic-categorization experiments in which we investigated error discounting by shifting feedback probabilities to new values after different amounts of training. In both experiments, responding gradually became less responsive to errors, and learning was slowed for some time after the feedback shift. Both results were indicative of error discounting. Quantitative modeling of the data revealed that adding a mechanism for error discounting significantly improved the fits of an exemplar-based and a rule-based associative learning model, as well as of a recency-based model of categorization. We conclude that error discounting is an important component of probabilistic learning.     

The potential role of state-dependent learning in cognitive therapy with spider phobics

Investigated the potential role of state-dependent learning in cognitive therapy with spider phobics. It was hypothesized that one strategy for increasing the effectiveness of cognitive therapy for anxiety would be to induce anxiety in subjects while they were learning the cognitive restructuring principles (i.e., during treatment). Thirty-two undergraduate volunteers were randomly assigned to one of six experimental conditions. Subjects in the five treatment conditions (the sixth was a no-treatment control condition) met for three sessions of cognitive restructuring,in vivo exposure, or a facts lecture (i.e., the placebo group). Results indicated that when working with spider phobics, three sessions of cognitive restructuring,in vivo exposure, or a facts lecture resulted in equal effectiveness immediately following treatment, but are more effective than no-treatment at all.     

The dynamics of learning and allocation of study time to a region of proximal learning

In contrast to the dominant discrepancy reduction model, which favors the most difficult items, people, given free choice, devoted most time to medium-difficulty items and studied the easiest items first. When study time was experimentally manipulated, best performance resulted when most time was given to the medium-difficulty items. Empirically determined information uptake functions revealed steep initial learning for easy items with little subsequent increase. For medium-difficulty items, initial gains were smaller but more sustained, suggesting that the strategy people had used, when given free choice, was largely appropriate. On the basis of the information uptake functions, a negative spacing effect was predicted and observed in the final experiment. Overall, the results favored the region of proximal learning framework.     

Leave Her out of It: Person‐Presentation of Strategies is Harmful for Transfer

A common practice in textbooks is to introduce concepts or strategies in association with specific people. This practice aligns with research suggesting that using “real‐world” contexts in textbooks increases students’ motivation and engagement. However, other research suggests this practice may interfere with transfer by distracting students or leading them to tie new knowledge too closely to the original learning context. The current study investigates the effects on learning and transfer of connecting mathematics strategies to specific people. A total of 180 college students were presented with an example of a problem‐solving strategy that was either linked with a specific person (e.g., “Juan's strategy”) or presented without a person. Students who saw the example without a person were more likely to correctly transfer the novel strategy to new problems than students who saw the example presented with a person. These findings are the first evidence that using people to present new strategies is harmful for learning and transfer.     

Learning the form of causal relationships using hierarchical bayesian models

People learn quickly when reasoning about causal relationships, making inferences from limited data and avoiding spurious inferences. Efficient learning depends on abstract knowledge, which is often domain or context specific, and much of it must be learned. While such knowledge effects are well documented, little is known about exactly how we acquire knowledge that constrains learning. This work focuses on knowledge of the functional form of causal relationships; there are many kinds of relationships that can apply between causes and their effects, and knowledge of the form such a relationship takes is important in order to quickly identify the real causes of an observed effect. We developed a hierarchical Bayesian model of the acquisition of knowledge of the functional form of causal relationships and tested it in five experimental studies, considering disjunctive and conjunctive relationships, failure rates, and cross-domain effects. The Bayesian model accurately predicted human judgments and outperformed several alternative models.     

Can unconscious knowledge allow control in sequence learning?

This paper investigates the conscious status of both the knowledge that an item is legal (judgment knowledge) and the knowledge of why it is legal (structural knowledge) in sequence learning. We compared ability to control use of knowledge (Process Dissociation Procedure) with stated awareness of the knowledge (subjective measures) as measures of the conscious status of knowledge. Experiment 1 showed that when people could control use of judgment knowledge they were indeed conscious of having that knowledge according to their own statements. Yet Experiment 2 showed that people could exert such control over the use of judgment knowledge when claiming they had no structural knowledge: i.e. conscious judgment knowledge could be based on unconscious structural knowledge. Further implicit learning research should be clear over whether judgment or structural knowledge is claimed to be unconscious as the two dissociate in sequence learning.     

Explicit feedback maintains implicit knowledge

The role of feedback was investigated with respect to conscious and unconscious knowledge acquired during artificial grammar learning (AGL). After incidental learning of training sequences, participants classified further sequences in terms of grammaticality and reported their decision strategy with or without explicit veridical feedback. Sequences that disobeyed the learning structure conformed to an alternative structure. Feedback led to an increase in the amount of reported conscious knowledge of structure (derived rules and recollections) but did not increase its accuracy. Conversely, feedback maintained the accuracy of unconscious knowledge of structure (intuition or familiarity-based responses) which otherwise degraded. Results support a dual-process account of AGL. They suggest that implicit learning of the to-be-rejected structure at test contaminates familiarity-based classifications whereas feedback allows competing familiarity signals to be contextualised, which is incompatible with theories that consider familiarity context-insensitive.     

Blocking in category learning

Many theories of category learning assume that learning is driven by a need to minimize classification error. When there is no classification error, therefore, learning of individual features should be negligible. The authors tested this hypothesis by conducting three category-learning experiments adapted from an associative learning blocking paradigm. Contrary to an error-driven account of learning, participants learned a wide range of information when they learned about categories, and blocking effects were difficult to obtain. Conversely, when participants learned to predict an outcome in a task with the same formal structure and materials, blocking effects were robust and followed the predictions of error-driven learning. The authors discuss their findings in relation to models of category learning and the usefulness of category knowledge in the environment.     

Comparing Multiple Paths to Mastery: What is Learned?

Contemporary theories of learning postulate one or at most a small number of different learning mechanisms. However, people are capable of mastering a given task through qualitatively different learning paths such as learning by instruction and learning by doing. We hypothesize that the knowledge acquired through such alternative paths differs with respect to the level of abstraction and the balance between declarative and procedural knowledge. In a laboratory experiment we investigated what was learned about patterned letter sequences via either direct instruction in the relevant patterns or practice in solving letter-sequence extrapolation problems. Results showed that both types of learning led to mastery of the target task as measured by accuracy performance. However, behavioral differences emerged in how participants applied their knowledge. Participants given instruction showed more variability in the types of strategies they used to articulate their knowledge as well as longer solution times for generating the action implications of that knowledge as compared to the participants given practice. Results are discussed regarding the implications for transfer, generalization, and procedural application. Learning theories that claim generality should be tested against cross-scenario phenomena, not just parametric variations of a single learning scenario.     

Lifelong Learning: A Pacification of ‘Know How’

A tendency of previous studies of lifelong learning to focus on learning and learning subjectivities may have led to an underestimation of potential effects in terms of a system of knowledge constitutive processes that operates powerfully to shape our societies. In this paper we explore lifelong learning and practices in the construction of knowledge at the point where a new relationship is being attempted between university courses and workplaces through programmes for learning. Drawing from Foucault and others we argue a strategic relation between discourses of lifelong learning and knowledge practices in such locations. Discourses of lifelong learning appear to support the reaching out of disciplinary practices into the workplace where theoretical knowledge is combined with knowledge derived from work experience, as a new form of knowledge that has use value. Discipline as a modality of power appears reconfigured and multiplied in new power-knowledge constellations which aim to subdue the desire and power of know how. Rather than lifelong learning as learning apparatus and strategy in the promotion of a will to learn as has been suggested elsewhere, we offer an alternative account. Here the promotion of a will to learn articulates with the will to knowledge in part through discourses of lifelong learning. Practices of knowledge constitution support the pacification of know how through its reconfiguration as knowledge that can be codified and mobilized for economic innovation.     

Counting strategies and semantic analysis as applied to class inclusion

This study examined strategic and semantic aspects of the answers given by preschool children to class inclusion problems. The Piagetian logical model for class inclusion was contrasted with an alternative, problem processing model in three experiments. A major component of the alternative model is an enumeration strategy which is advantageous for learning reliable counting skills. The counting strategy was found to explain the inclusion errors of young children better than did the logic of the task. It was also found that young children understand the semantics of inclusion but are unable to coordinate their semantic knowledge with their counting strategy. Methodologically, one of the experiments suggested a fruitful extension of task analysis (Simon, 1969) to experimental design.     

Regulatory fit effects in a choice task

This article examines the interface between motivation and choice. In category learning, a regulatory fit has been shown to increase exploration of alternative response strategies even when exploration is suboptimal. In the present study, promotion- and prevention-focused subjects performed a choice task that required them to choose from one of two decks of cards on each trial. They either gained or lost points with each draw. In Experiment 1, optimal performance required an exploratory response pattern that entailed sampling from a deck that initially appeared disadvantageous but ultimately became advantageous. In Experiment 2, optimal performance required an exploitative response pattern. A softmax action selection model that includes an exploitation parameter was applied to each subject’s data from both experiments and revealed greater exploration of alternative strategies for people with a regulatory fit. This response strategy was optimal in Experiment 1 and led to superior performance, but was suboptimal in Experiment 2 and led to inferior performance.