Gaze and informativeness during category learning: Evidence for an inverse relation

In what follows, we explore the general relationship between eye gaze during a category learning task and the information conveyed by each member of the learned category. To understand the nature of this relationship empirically, we used eye tracking during a novel object classification paradigm. Results suggest that the average fixation time per object during learning is inversely proportional to the amount of information that object conveys about its category. This inverse relationship may seem counterintuitive; however, objects that have a high-information value are inherently more representative of their category. Therefore, their generality captures the essence of the category structure relative to less representative objects. As such, it takes relatively less time to process these objects than their less informative companions. We use a general information measure referred to as representational information theory (Vigo, 2011a, 2013a) to articulate and interpret the results from our experiment and compare its predictions to those of three models of prototypicality.     

Evidence for a procedural-learning-based system in perceptual category learning

The consistency of the mapping from category to response location was investigated to test the hypothesis that abstract category labels are learned by the hypothesis testing system to solve rule-based tasks, whereas response position is learned by the procedural-learning system to solve informationintegration tasks. Accuracy rates were examined to isolate global performance deficits, and modelbased analyses were performed to identify the types of response strategies used by observers. A-B training (consistent mapping) led to more accurate responding relative to yes-no training (variable mapping) in the information-integration category learning task. Model-based analyses indicated that the yes-no accuracy decline was due to an increase in the use of rule-based strategies to solve the information-integration task. Yes-no training had no effect on the accuracy of responding or distribution of best-fitting models relative to A-B training in the rule-based category learning tasks. These results both provide support for a multiple-systems approach to category learning in which one system is procedural-learning-based and argue against the validity of single-system approaches.     

Executive attention and task switching in category learning: Evidence for stimulus-dependent representation

One class of multiple-system models of category learning posits that within a single category-learning task people can learn to utilize different systems with different category representations to classify different stimuli. This is referred to as stimulus-dependent representation (SDR). The use of SDR implies that learners switch from subtask to subtask as trials demand. Thus, the use of SDR can be assessed via slowed response times, following a representation switch. Additionally, the use of SDR requires control of executive attention to keep inactive representations from interfering with the current response. Subjects were given a category learning task composed of one- and two-dimensional substructures. Control of executive attention was measured using a working memory capacity (WMC) task. Subjects most likely to be using SDR showed greater slowing of responses following a substructure switch and a greater correlation between learning performance and WMC. These results provide support for the principle of SDR in category learning and the reliance of SDR on executive attention.     

Prototype-distortion category learning: a two-phase learning process across a distributed network

This paper reviews a body of work conducted in our laboratory that applies functional magnetic resonance imaging (fMRI) to better understand the biological response and change that occurs during prototype-distortion learning. We review results from two experiments (Little, Klein, Shobat, McClure, & Thulborn, 2004; Little & Thulborn, 2005) that provide support for increasing neuronal efficiency by way of a two-stage model that includes an initial period of recruitment of tissue across a distributed network that is followed by a period of increasing specialization with decreasing volume across the same network. Across the two studies, participants learned to classify patterns of random-dot distortions (Posner & Keele, 1968) into categories. At four points across this learning process subjects underwent examination by fMRI using a category-matching task. A large-scale network, altered across the protocol, was identified to include the frontal eye fields, both inferior and superior parietal lobules, and visual cortex. As behavioral performance increased, the volume of activation within these regions first increased and later in the protocol decreased. Based on our review of this work we propose that: (i) category learning is reflected as specialization of the same network initially implicated to complete the novel task, and (ii) this network encompasses regions not previously reported to be affected by prototype-distortion learning.     

Acquired distinctiveness and equivalence in human discrimination learning: Evidence for an attentional process

In a first stage of training, participants learned to associate four visual cues (two different colors and two different shapes) with verbal labels. For Group S, one label was applied to both colors and another to both shapes; for Group D, one label was applied to one color and one shape, and the other label to the other cues. When subsequently required to learn a task in which a given motor response was required to one of the colors and one of the shapes, and a different response to the other color and the other shape, Group D learned more readily than Group S. The task was designed so that the associations formed during the first stage of training could not generate differential transfer to the second stage. The results are consistent, however, with the proposal that training in which similar cues are followed by different outcomes will engage a learning process that boosts the attention paid to features that distinguish these cues.     

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.     

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.     

Judgments of learning: Evidence for a two-stage process

Three experiments tested the hypothesis that people make judgments of learning (JOLs) by attempting to retrieve the target first. If this were the whole story, then the reaction time (RT) functions for making JOLs with no special instructions would parallel those found when people are told to first attempt retrieval and then make a JOL. In the present data, monotonic functions, showing an increase in RT with decreasing JOL, were found when people were instructed to retrieve covertly or overtly and then make a JOL, as would be expected if retrieval fluency entirely determined JOLs. However, the functions for making uninstructed JOLs were different: Low JOLs were made quickly, not slowly, and the curves were inverted U shapes, rather than linear. Furthermore, people's memory performance was somewhat better, especially on low-JOL items, when they were instructed to first retrieve as opposed to when they were told only to make JOLs. To account for these data, we propose a two-stage model of JOLs, with the first stage occurring prior to attempted retrieval.     

The case for implicit category learning

This article evaluates the evidence regarding the claim that people can learn a novel category implicitly--that is, by an implicit memory system that is qualitatively different from an explicit system. The evidence that is considered is based on the prototype extraction task, in which participants are first exposed to a set of category exemplars under incidental learning instructions and are then required to categorize novel test items. Knowlton and Squire (1993) first reported that memory-impaired patients performed normally on the prototype extraction task while being impaired on a comparable recognition task. Several studies have replicated these results, but other articles have criticized the evidence for implicit category learning on both methodological and theoretical grounds. In this article, we consider five of these criticisms-for example, that the normal performance of the patients is due to intact working memory mechanisms (see, e.g., Palmeri & Flannery, 1999) or to the lesser cognitive demands of prototype extraction rather than recognition (e.g., Nosofsky & Zaki, 1998). For each of the five criticisms, we offer counterevidence that supports implicit category learning.     

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.     

The effects of concurrent task interference on category learning: evidence for multiple category learning systems

Participants learned simple and complex category structures under typical single-task conditions and when performing a simultaneous numerical Stroop task. In the simple categorization tasks, each set of contrasting categories was separated by a unidimensional explicit rule, whereas the complex tasks required integrating information from three stimulus dimensions and resulted in implicit rules that were difficult to verbalize. The concurrent Stroop task dramatically impaired learning of the simple explicit rules, but did not significantly delay learning of the complex implicit rules. These results support the hypothesis that category learning is mediated by multiple learning systems.     

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.     

Stimulus modality interacts with category structure in perceptual category learning

Two experiments were conducted that examined information integration and rule-based category learning, using stimuli that contained auditory and visual information. The results suggest that it is easier to perceptually integrate information within these sensory modalities than across modalities. Conversely, it is easier to perform a disjunctive rule-based task when information comes from different sensory modalities, rather than from the same modality. Quantitative model-based analyses suggested that the information integration deficit for across-modality stimulus dimensions was due to an increase in the use of hypothesis-testing strategies to solve the task and to an increase in random responding. The modeling also suggested that the across-modality advantage for disjunctive, rule-based category learning was due to a greater reliance on disjunctive hypothesis-testing strategies, as opposed to unidimensional hypothesis-testing strategies and random responding.     

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.     

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.     

Perceptual learning in visual category acquisition

The attribute structure of a set of dot patterns was studied by having subjects segment (parse) the dots of each pattern into parts or subunits by drawing circles around groups of dots from each pattern. These parsing data were obtained for subjects who had no prior experience with the patterns and for subjects who had previously learned to identify the patterns as members of one of four categories. Analyses of the parsing data indicated that category learning increased the salience of large subunits that were similar in orientation for patterns that were members of the same category. This evidence for perceptual learning was obtained even when the category training procedure required learning to identify the patterns individually, suggesting that attribute abstraction and item learning are not incompatible. It was also obtained without an increase in overall intersubject agreement. The latter result led us to question the usefulness of intersubject agreement as an index of category knowledge.     

Dual-task interference in perceptual category learning

The effect of a working-memory-demanding dual task on perceptual category learning was investigated. In Experiment 1, participants learned unidimensional rule-based or information integration category structures. In Experiment 2, participants learned a conjunctive rule-based category structure. In Experiment 1, unidimensional rule-based category learning was disrupted more by the dual working memory task than was information integration category learning. In addition, rule-based category learning differed qualitatively from information integration category learning in yielding a bimodal, rather than a normal, distribution of scores. Experiment 2 showed that rule-based learning can be disrupted by a dual working memory task even when both dimensions are relevant for optimal categorization. The results support the notion of at least two systems of category learning a hypothesis-testing system that seeks verbalizable rules and relies on working memory and selective attention, and an implicit system that is procedural-learning based and is essentially automatic.