Context-dependent savings in procedural category learning

Environmental context can have a profound influence on the efficacy of intervention protocols designed to eliminate undesirable behaviors. This is clearly seen in drug rehabilitation clinics where patients often relapse soon after leaving the context of the treatment facility. A similar pattern is commonly observed in controlled laboratory studies of context-dependent savings in instrumental conditioning, where simply placing an animal back into the original conditioning chamber can renew an extinguished instrumental response. Surprisingly, context-dependent savings in human procedural learning has not been carefully examined in the laboratory. Here, we provide the first known empirical demonstration of context-dependent savings in a perceptual categorization task known to recruit procedural learning. We also present a computational account of these savings using a biologically detailed model in which a key role is played by cholinergic interneurons in the striatum.     

From conditioning to category learning: an adaptive network model

We used adaptive network theory to extend the Rescorla-Wagner (1972) least mean squares (LMS) model of associative learning to phenomena of human learning and judgment. In three experiments subjects learned to categorize hypothetical patients with particular symptom patterns as having certain diseases. When one disease is far more likely than another, the model predicts that subjects will substantially overestimate the diagnosticity of the more valid symptom for the rare disease. The results of Experiments 1 and 2 provide clear support for this prediction in contradistinction to predictions from probability matching, exemplar retrieval, or simple prototype learning models. Experiment 3 contrasted the adaptive network model with one predicting pattern-probability matching when patients always had four symptoms (chosen from four opponent pairs) rather than the presence or absence of each of four symptoms, as in Experiment 1. The results again support the Rescorla-Wagner LMS learning rule as embedded within an adaptive network model.     

Analyzing the RULEX model of category learning

Recent approaches to human category learning have often (re)invoked the notion of systematic search for good rules. The RULEX model of category learning is emblematic of this renewed interest in rule-based categorization, and is able to account for crucial findings previously thought to provide evidence in favor of prototype or exemplar models. However, a major difficulty in comparing RULEX to other models is that RULEX is framed in terms of a stochastic search process, with no analytic expressions available for its predictions. The result is that RULEX predictions can only be found through time consuming simulations, making model-fitting very difficult, and all but prohibiting more detailed investigations of the model. To remedy this problem, this paper describes an algorithmic method of calculating RULEX predictions that does not rely on numerical simulation, and yields some insight into the behavior of the model itself.     

A model of probabilistic category learning.

A new connectionist model (named RASHNL) accounts for many "irrational" phenomena found in nonmetric multiple-cue probability learning, wherein people learn to utilize a number of discrete-valued cues that are partially valid indicators of categorical outcomes. Phenomena accounted for include cue competition, effects of cue salience, utilization of configural information, decreased learning when information is introduced after a delay, and effects of base rates. Experiments 1 and 2 replicate previous experiments on cue competition and cue salience, and fits of the model provide parameter values for making qualitatively correct predictions for many other situations. The model also makes 2 new predictions, confirmed in Experiments 3 and 4. The model formalizes 3 explanatory principles: rapidly shifting attention with learned shifts, decreasing learning rates, and graded similarity in exemplar representation.     

Learning myopia: an adaptive recency effect in category learning

Recency effects (REs) have been well established in memory and probability learning paradigms but have received little attention in category earning research. Extant categorization models predict REs to be unaffected by learning, whereas a functional interpretation of REs, suggested by results in other domains, predicts that people are able to learn sequential dependencies and incorporate this information into their responses. These contrasting predictions were tested in 2 experiments involving a classification task in which outcome sequences were autocorrelated. Experiment 1 showed that reliance on recent outcomes adapts to the structure of the task, in contrast to models' predictions. Experiment 2 provided constraints on how sequential information is learned and suggested possible extensions to current models to account for this learning.     

Abstraction and model evaluation in category learning

Thirty previously published data sets, from seminal category learning tasks, are reanalyzed using the varying abstraction model (VAM). Unlike a prototype-versus-exemplar analysis, which focuses on extreme levels of abstraction only, a VAM analysis also considers the possibility of partial abstraction. Whereas most data sets support no abstraction when only the extreme possibilities are considered, we show that evidence for abstraction can be provided using the broader view on abstraction provided by the VAM. The present results generalize earlier demonstrations of partial abstraction (Vanpaemel & Storms, 2008), in which only a small number of data sets was analyzed. Following the dominant modus operandi in category learning research, Vanpaemel and Storms evaluated the models on their best fit, a practice known to ignore the complexity of the models under consideration. In the present study, in contrast, model evaluation not only relies on the maximal likelihood, but also on the marginal likelihood, which is sensitive to model complexity. Finally, using a large recovery study, it is demonstrated that, across the 30 data sets, complexity differences between the models in the VAM family are small. This indicates that a (computationally challenging) complexity-sensitive model evaluation method is uncalled for, and that the use of a (computationally straightforward) complexity-insensitive model evaluation method is justified.     

An associative model of adaptive inference for learning word-referent mappings

People can learn word–referent pairs over a short series of individually ambiguous situations containing multiple words and referents (Yu & Smith, 2007, Cognition 106: 1558–1568). Cross-situational statistical learning relies on the repeated co-occurrence of words with their intended referents, but simple co-occurrence counts cannot explain the findings. Mutual exclusivity (ME: an assumption of one-to-one mappings) can reduce ambiguity by leveraging prior experience to restrict the number of word–referent pairings considered but can also block learning of non-one-to-one mappings. The present study first trained learners on one-to-one mappings with varying numbers of repetitions. In late training, a new set of word–referent pairs were introduced alongside pretrained pairs; each pretrained pair consistently appeared with a new pair. Results indicate that (1) learners quickly infer new pairs in late training on the basis of their knowledge of pretrained pairs, exhibiting ME; and (2) learners also adaptively relax the ME bias and learn two-to-two mappings involving both pretrained and new words and objects. We present an associative model that accounts for both results using competing familiarity and uncertainty biases.     

SUSTAIN: a network model of category learning

SUSTAIN (Supervised and Unsupervised STratified Adaptive Incremental Network) is a model of how humans learn categories from examples. SUSTAIN initially assumes a simple category structure. If simple solutions prove inadequate and SUSTAIN is confronted with a surprising event (e.g., it is told that a bat is a mammal instead of a bird), SUSTAIN recruits an additional cluster to represent the surprising event. Newly recruited clusters are available to explain future events and can themselves evolve into prototypes-attractors-rules. SUSTAIN's discovery of category substructure is affected not only by the structure of the world but by the nature of the learning task and the learner's goals. SUSTAIN successfully extends category learning models to studies of inference learning, unsupervised learning, category construction, and contexts in which identification learning is faster than classification learning.     

Expanding the search for a linear separability constraint on category learning

Formal models of categorization make different predictions about the theoretical importance of linear separability. Prior research, most of which has failed to find support for a linear separability constraint on category learning, has been conducted using tasks that involve learning two categories with a small number of members. The present experiment used four categories with three or nine patterns per category that were either linearly separable or not linearly separable. With overall category structure equivalent across category types, the linearly separable categories were found to be easier to learn than the not linearly separable categories. An analysis of individual participants' data showed that there were more participants operating under a linear separability constraint when learning large categories than when learning small ones. Formal modeling showed that an exemplar model could not account for many of these data. These results are taken to support the existence of multiple processes in categorization.     

A knowledge-resonance (KRES) model of category learning

This article introduces a connectionist model of category learning that takes into account the prior knowledge that people bring to new learning situations. In contrast to connectionist learning models that assume a feedforward network and learn by the delta rule or backpropagation, this model, the knowledge-resonance model, or KRES, employs a recurrent network with bidirectional symmetric connection whose weights are updated according to a contrastive Hebbian learning rule. We demonstrate that when prior knowledge is represented in the network, KRES accounts for a considerable range of empirical results regarding the effects of prior knowledge on category learning, including (1) the accelerated learning that occurs in the presence of knowledge, (2) the better learning in the presence of knowledge of category features that are not related to prior knowledge, (3) the reinterpretation of features with ambiguous interpretations in light of error-corrective feedback, and (4) the unlearning of prior knowledge when that knowledge is inappropriate in the context of a particular category.     

Types of semantic category and stimulus modality in procedural learning: implications for models of semantic memory

The present work attempts to determine whether procedural learning of a semantic categorisation task is influenced by the type of semantic category of the stimuli (biological and non-biological elements). It is also an attempt to determine the effect of the stimulus presentation modality on the categorisation task. A semantic categorisation task (4 series of 40 stimuli) was administered to 256 participants (128 classifying pictures, and 128 classifying words). Biological categories were responded to faster than non-biological ones although there were no significant differences between the interaction of the category type and the stimulus presentation modality. Reaction times progressively decreased with practice. However, the initial differences disappeared when subjects were trained. The way that current models account for these investigation findings is discussed. In addition, it is suggested that there is an attentional bias in favor of biological elements, which disappears when presumably less relevant elements become more relevant as a function of the task characteristics.     

ALCOVE: an exemplar-based connectionist model of category learning.

ALCOVE (attention learning covering map) is a connectionist model of category learning that incorporates an exemplar-based representation (Medin & Schaffer, 1978; Nosofsky, 1986) with error-driven learning (Gluck & Bower, 1988; Rumelhart, Hinton, & Williams, 1986). Alcove selectively attends to relevant stimulus dimensions, is sensitive to correlated dimensions, can account for a form of base-rate neglect, does not suffer catastrophic forgetting, and can exhibit 3-stage (U-shaped) learning of high-frequency exceptions to rules, whereas such effects are not easily accounted for by models using other combinations of representation and learning method.     

The divergent autoencoder (DIVA) model of category learning

A novel theoretical approach to human category learning is proposed in which categories are represented as coordinated statistical models of the properties of the members. Key elements of the account are learning to recode inputs as task-constrained principle components and evaluating category membership in terms of model fit-that is, the fidelity of the reconstruction after recoding and decoding the stimulus. The approach is implemented as a computational model called DIVA (for DIVergent Autoencoder), an artificial neural network that uses reconstructive learning to solve N-way classification tasks. DIVA shows good qualitative fits to benchmark human learning data and provides a compelling theoretical alternative to established models.     

Extending the ALCOVE model of category learning to featural stimulus domains

The ALCOVE model of category learning, despite its considerable success in accounting for human performance across a wide range of empirical tasks, is limited by its reliance on spatial stimulus representations. Some stimulus domains are better suited to featural representation, characterizing stimuli in terms of the presence or absence of discrete features, rather than as points in a multidimensional space. We report on empirical data measuring human categorization performance across a featural stimulus domain and show that ALCOVE is unable to capture fundamental qualitative aspects of this performance. In response, a featural version of the ALCOVE model is developed, replacing the spatial stimulus representations that are usually generated by multidimensional scaling with featural representations generated by additive clustering. We demonstrate that this featural version of ALCOVE is able to capture human performance where the spatial model failed, explaining the difference in terms of the contrasting representational assumptions made by the two approaches. Finally, we discuss ways in which the ALCOVE categorization model might be extended further to use “hybrid” representational structures combining spatial and featural components.     

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.     

Unitization during category learning

Five experiments explored the question of whether new perceptual units can be developed if they are useful for a category learning task, and if so, what the constraints on this unitization process are. During category learning, participants were required to attend either a single component or a conjunction of 5 components. Consistent with unitization, the conjunctive task became much easier with practice; this improvement was not found for the single-component task or for conjunctive tasks in which the components could not be unitized. Influences of component organization (Experiment 1), component contiguity (Experiment 2), component proximity (Experiment 3), and number of components (Experiment 4) on practice effects were found. Deconvolved response times (Experiment 5) showed that prolonged practice yielded faster responses than predicted by an analytic model that integrates evidence from independently perceived components.     

On learning complex procedural knowledge

Lewicki, Czyzewska, and Hoffman (1987) demonstrated learning without awareness in a visual search task. Rules determined target location on every seventh trial on the basis of target locations in the preceding six trials. Learning was demonstrated by negative transfer effects when the rules were changed. When questioned afterwards, the subjects could not describe the rules and denied awareness of them. This experiment was designed to replicate that of Lewicki et al. and to test several hypotheses about this apparent learning without awareness. Transfer conditions were included to determine whether rule learning was primarily perceptual or motor. The present assessment of awareness was based on an objective definition of awareness, rather than a subjective definition as in Lewicki et al.'s study. Their effect was replicated, and the transfer conditions revealed that learning relied on perceptual aspects of the task. The objective measure of awareness provided further evidence that subjects were unaware of the rules.     

Cognitive changes in conjunctive rule-based category learning: An ERP approach

When learning rule-based categories, sufficient cognitive resources are needed to test hypotheses, maintain the currently active rule in working memory, update rules after feedback, and to select a new rule if necessary. Prior research has demonstrated that conjunctive rules are more complex than unidimensional rules and place greater demands on executive functions like working memory. In our study, event-related potentials (ERPs) were recorded while participants performed a conjunctive rule-based category learning task with trial-by-trial feedback. In line with prior research, correct categorization responses resulted in a larger stimulus-locked late positive complex compared to incorrect responses, possibly indexing the updating of rule information in memory. Incorrect trials elicited a pronounced feedback-locked P300 elicited which suggested a disconnect between perception, and the rule-based strategy. We also examined the differential processing of stimuli that were able to be correctly classified by the suboptimal single-dimensional rule (“easy” stimuli) versus those that could only be correctly classified by the optimal, conjunctive rule (“difficult” stimuli). Among strong learners, a larger, late positive slow wave emerged for difficult compared with easy stimuli, suggesting differential processing of category items even though strong learners performed well on the conjunctive category set. Overall, the findings suggest that ERP combined with computational modelling can be used to better understand the cognitive processes involved in rule-based category learning.     

Array models for category learning

A family of models for category learning is developed, all members being based on a common memory array but differing in memory access and decision processes. Within this framework, fully controlled comparisons of exemplar-similarity, feature-frequency, and prototype models reveal isomorphism between models of different types under some conditions but empirically testable differences under others. It is shown that current exemplar-memory models, in which categorization judgments are based on similarities of perceived and remembered category exemplars, can be interpreted as generalized likelihood models but can be modified in a simple way to yield pure similarity models. Distance-based exemplar models are formulated that provide means of investigating issues concerning deterministic versus probabilistic decision rules and links between categorization and properties of perceptual dimensions. Other theoretical issues discussed include aspects of similarity, the role of memory storage versus computation in category judgments, and the limits of applicability of array models.