Category Learning Experiments Research Articles

Exploration, Distributed Attention, and Development of Category Learning

Category learning is a crucial aspect of cognition that involves organizing entities into equivalence classes. Whereas adults tend to focus on category-relevant features, young children often distribute attention between relevant and irrelevant ones. The reasons for children’s distributed attention are not fully understood. In two category-learning experiments with adults and with children aged 4, 5, and 6 (N = 201), we examined potential drivers of distributed attention, including (a) immature filtering of distractors and (b) the general tendency for exploration or broad information sampling. By eliminating distractor competition, we reduced filtering demands. Despite identifying the features critical for accurate categorization, children, regardless of their categorization performance, continued sampling more information than was necessary. These results indicate that the tendency to sample information extensively contributes to distributed attention in young children. We identify candidate drivers of this tendency that need to be examined in future research.

The spacing effect in remote information-integration category learning.

The present experiments examined whether the temporal distribution of procedural category learning experiences would impact learning outcomes. Participants completed the remote category learning experiments on a smartphone in one of two learning conditions: massed or distributed. Consistent with expectations, distributed learners in both experiments reached higher accuracy levels than massed learners. In Experiment 1 the effect disappeared after accounting for reaction time differences, suggesting that it was driven by attentional mechanisms. In Experiment 2, the spacing advantage was only present for previously studied items during a post-learning test, suggesting a role of consolidation. In both experiments, it seems likely that temporal spacing helped participants discover the optimal information-integration categorization strategy. These results suggest that adult category learning is facilitated by temporal spacing. Future work may further explore the effects of temporal and contextual distinctiveness of learning experiences on category learning outcomes.

The Boolean Language of Thought is recoverable from learning data

According to the Language of Thought Hypothesis (LoTH), an influential account in philosophy and cognitive science, human cognition is underlain by symbolic reasoning in a formal language. In this account, concepts are expressions in a Language of Thought, deduction is syntactic manipulation in this language, and learning is an inference of expressions in this language from data. This picture raises the question of what LoT humans have, and how to infer it from behavior. In this paper, we pave the way towards answering this question, by approaching a more fundamental question: to what extent is it possible in principle to recover the human LoT from experimental data? To answer this question, we focus on the fragment of LoT that is concerned with representing Boolean categories and simulate the recovery of the Boolean LoT from category learning experiments. Our findings show that in principle the vast majority of Boolean LoTs can be accurately recovered from experimental data. However, we find that this crucially depends on the employed experimental design. Moreover, we find evidence that LoTs with fewer operators can be recovered from category learning data faster.

The Forward Effect of Delayed Judgments of Learning Is Influenced by Difficulty in Memory and Category Learning.

Delayed judgment of learning (JOL) is a widely used metacognitive monitoring strategy that can also enhance learning outcomes. However, the potential benefits of delayed JOL on subsequent learning of new material, known as the forward effect of delayed JOL, and its stability and underlying mechanisms have yet to be fully explored. In this study, we investigated the forward effect of delayed JOL using previously unexamined word pair materials and explored the boundary conditions of this effect by manipulating the difficulty of the materials. We also examined this effect within the context of category learning. Our findings demonstrate that delayed JOL significantly enhanced the retention of new information (Experiment 1A), while the forward effect of the delayed JOL occurred only for material with a certain degree of difficulty rather than for easy material (Experiment 1B). These findings were extended and replicated using category learning (Experiment 2). These results suggest that delayed JOL can be used as a preparation strategy for subsequent learning, particularly when faced with challenging materials. Our study provides novel insights into the potential benefits and limitations of delayed JOL and contributes to our understanding of the underlying mechanisms that govern metacognitive monitoring and learning strategies.

Comparing virtual reality, desktop-based 3D, and 2D versions of a category learning experiment.

Virtual reality (VR) has seen increasing application in cognitive psychology in recent years. There is some debate about the impact of VR on both learning outcomes and on patterns of information access behaviors. In this study we compare performance on a category learning task between three groups: one presented with three-dimensional (3D) stimuli while immersed in the HTC Vive VR system (n = 26), another presented with the same 3D stimuli while using a flat-screen desktop computer (n = 26), and a third presented with a two-dimensional projection of the stimuli on a desktop computer while their eye movements were tracked (n = 8). In the VR and 3D conditions, features of the object to be categorized had to be revealed by rotating the object. In the eye tracking control condition (2D), all object features were visible, and participants' gaze was tracked as they examined each feature. Over 240 trials we measured accuracy, reaction times, attentional optimization, time spent on feedback, fixation durations, and fixation counts for each participant as they learned to correctly categorize the stimuli. In the VR condition, participants had increased fixation counts compared to the 3D and 2D conditions. Reaction times for the 2D condition were significantly faster and fixation durations were lower compared to the VR and 3D conditions. We found no significant differences in learning accuracy between the VR, 3D, and 2D conditions. We discuss implications for both researchers interested in using VR to study cognition, and VR developers hoping to use non-VR research to guide their designs and applications.

On the importance of feedback for categorization: Revisiting category learning experiments using an adaptive filter model.

Associative accounts of category learning have been, for the most part, abandoned in favor of cognitive explanations (e.g., similarity, explicit rules). In the current work, we implement an Adaptive Linear Filter (ALF) closely related to the Rescorla and Wagner learning rule, and use it to tackle three learning tasks that pose challenges to an associative view of category learning. Across three computational simulations, we show that the ALF is in fact able to make the predictions that seemed problematic. Notably, in our simulations we use exactly the same model and specifications, attesting to the generality of our account. We discuss the consequences of our findings for the category learning literature. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

LAG-1: A dynamic, integrative model of learning, attention, and gaze.

It is clear that learning and attention interact, but it is an ongoing challenge to integrate their psychological and neurophysiological descriptions. Here we introduce LAG-1, a dynamic neural field model of learning, attention and gaze, that we fit to human learning and eye-movement data from two category learning experiments. LAG-1 comprises three control systems: one for visuospatial attention, one for saccadic timing and control, and one for category learning. The model is able to extract a kind of information gain from pairwise differences in simple associations between visual features and categories. Providing this gain as a reentrant signal with bottom-up visual information, and in top-down spatial priority, appropriately influences the initiation of saccades. LAG-1 provides a moment-by-moment simulation of the interactions of learning and gaze, and thus simultaneously produces phenomena on many timescales, from the duration of saccades and gaze fixations, to the response times for trials, to the slow optimization of attention toward task relevant information across a whole experiment. With only three free parameters (learning rate, trial impatience, and fixation impatience) LAG-1 produces qualitatively correct fits for learning, behavioural timing and eye movement measures, and also for previously unmodelled empirical phenomena (e.g., fixation orders showing stimulus-specific attention, and decreasing fixation counts during feedback). Because LAG-1 is built to capture attention and gaze generally, we demonstrate how it can be applied to other phenomena of visual cognition such as the free viewing of visual stimuli, visual search, and covert attention.

Computational Cognitive models of Categorization: Predictions under Conditions of Classification Uncertainty

In the category learning literature, similarity models have monopolized a good deal of research. The prototype and exemplar models are both based on the idea that people represent the structure of categories and category instances in the physical world in a mental similarity space. However, evidence for these models comes mainly from paradigms that provide subjects with deterministic feedback (i.e., exemplars belong to their corresponding categories with probability = 1). There is evidence that results obtained with deterministic feedback paradigms may not generalize well under probabilistic feedback conditions (i.e., where exemplars belong to their corresponding categories with probability less than 1). In this current work, we also suggest that probabilistic feedback may better reflect natural conditions, which is another important reason to pursue probabilistic feedback research. Thus, in the current work we set up a category learning experiment with probabilistic feedback and use it to evaluate different models. In addition to the two similarity models discussed above, we also use an associationist model that does not rely on the similarity construct. To compare our three models, we rely on computational modeling, which is a standard way of model comparison in cognitive psychology. Our results show that our associationist model outperforms similarity models on all our model evaluation measures. After presenting our results, we discuss why the similarity-based models fail, and also suggest some future lines of research that are possible using probabilistic feedback procedures.

Supplemental Material for On the Importance of Feedback for Categorization: Revisiting Category Learning Experiments Using an Adaptive Filter Model

Supplemental Material for On the Importance of Feedback for Categorization: Revisiting Category Learning Experiments Using an Adaptive Filter Model

Machine learning identifies the dynamics and influencing factors in an auditory category learning experiment

Human learning is one of the main topics in psychology and cognitive neuroscience. The analysis of experimental data, e.g. from category learning experiments, is a major challenge due to confounding factors related to perceptual processing, feedback value, response selection, as well as inter-individual differences in learning progress due to differing strategies or skills. We use machine learning to investigate (Q1) how participants of an auditory category-learning experiment evolve towards learning, (Q2) how participant performance saturates and (Q3) how early we can differentiate whether a participant has learned the categories or not. We found that a Gaussian Mixture Model describes well the evolution of participant performance and serves as basis for identifying influencing factors of task configuration (Q1). We found early saturation trends (Q2) and that CatBoost, an advanced classification algorithm, can separate between participants who learned the categories and those who did not, well before the end of the learning session, without much degradation of separation quality (Q3). Our results show that machine learning can model participant dynamics, identify influencing factors of task design and performance trends. This will help to improve computational models of auditory category learning and define suitable time points for interventions into learning, e.g. by tutorial systems.

Is there an exemplar theory of concepts?

It is common to describe two main theories of concepts: prototype theories, which rely on some form of summary description of a category, and exemplar theories, which claim that concepts are represented as remembered category instances. This article reviews a number of important phenomena in the psychology of concepts, arguing that they have no proposed exemplar explanation. In some of these cases, it is difficult to see how an exemplar theory would be adequate. The article concludes that exemplars are certainly important in some categorization judgments and in category-learning experiments, but that there is no exemplar theory of human concepts in a broad sense.

Shaping relations: Exploiting relational features for visuospatial priming.

Although relational reasoning has been described as a process at the heart of human cognition, the exact character of relational representations remains an open debate. Symbolic-connectionist models of relational cognition suggest that relations are structured representations, but that they are ultimately grounded in feature sets; thus, they predict that activating those features can affect the trajectory of the relational reasoning process. The present work points out that such models do not necessarily specify what those features are though, and endeavors to show that spatial information is likely a part of it. To this end, it presents 2 experiments that used visuospatial priming to affect the course of relational reasoning. The first is a relational category-learning experiment in which this type of priming was shown to affect which spatial relation was learned when multiple were possible. The second used crossmapping analogy problems, paired with this same type of priming, to show that visuospatial cues can make participants more likely to map analogs based on relational roles, even with short presentation times. (PsycINFO Database Record

Procedural learning during declarative control.

There is now abundant evidence that human learning and memory are governed by multiple systems. As a result, research is now turning to the next question of how these putative systems interact. For instance, how is overall control of behavior coordinated, and does learning occur independently within systems regardless of what system is in control? Behavioral, neuroimaging, and neuroscience data are somewhat mixed with respect to these questions. Human neuroimaging and animal lesion studies suggest independent learning and are mostly agnostic with respect to control. Human behavioral studies suggest active inhibition of behavioral output but have little to say regarding learning. The results of two perceptual category-learning experiments are described that strongly suggest that procedural learning does occur while the explicit system is in control of behavior and that this learning might be just as good as if the procedural system was controlling the response. These results are consistent with the idea that declarative memory systems inhibit the ability of the procedural system to access motor output systems but do not prevent procedural learning.

How experimental trial context affects perceptual categorization

To understand object categorization, participants are tested in experiments often quite different from how people experience object categories in the real world. Learning and knowledge of categories is measured in discrete experimental trials, those trials may or may not provide feedback, trials appear one after another, after some fixed inter-trial interval, with hundreds of trials in a row, within experimental blocks with some structure dictated by the experimental design. In the real world, outside of certain educational and vocational contexts, opportunities to learn and use categories are intermixed over time with a whole multitude of intervening experiences. It is clear from any elementary understanding of human cognition that sequential effects matter, yet this understanding is often ignored, and categorization trials are often instead treated as independent events, immune to local trial context. In this perspective, we use some of our work to illustrate some of the consequences of the fact that categorization experiments have a particular trial structure. Experimental trial context can affect performance in category learning and categorization experiments in ways that can profoundly affect theoretical conclusions.

Illusory correlations in preschoolers

People form illusory correlations (ICs) such that they will at times perceive a relationship between variables even when no such relationship exists. Three experiments with undergraduates (N=20), 5-year-olds (N=71), and 3-year-olds (N=52) explored the origins of ICs: When, and under what conditions, does the disposition to form ICs emerge? Results showed that, like adults, 5-year-olds formed ICs for large samples of evidence (Experiment 1), but 5-year-olds did not do so for smaller samples (Experiment 2). Three- and 5-year-olds formed ICs for small samples when the evidence was presented in a way that promoted an attention shift from one category to the other during category learning (Experiment 3). These results suggest ICs are due, in large part, to limitations in processing capacity and biases in attentional learning mechanisms. The results provide insights regarding the conditions that influence children's ability to detect and use regularities in the available input and thus have implications for understanding their performance on a broad range of cognitive tasks.

Learning-Induced Changes in Attentional Allocation during Categorization: A Sizable Catalog of Attention Change as Measured by Eye Movements

Learning how to allocate attention properly is essential for success at many categorization tasks. Advances in our understanding of learned attention are stymied by a chicken-and-egg problem: there are no theoretical accounts of learned attention that predict patterns of eye movements, making data collection difficult to justify, and there are not enough datasets to support the development of a rich theory of learned attention. The present work addresses this by reporting five measures relating to the overt allocation of attention across 10 category learning experiments: accuracy, probability of fixating irrelevant information, number of fixations to category features, the amount of change in the allocation of attention (using a new measure called Time Proportion Shift - TIPS), and a measure of the relationship between attention change and erroneous responses. Using these measures, the data suggest that eye-movements are not substantially connected to error in most cases and that aggregate trial-by-trial attention change is generally stable across a number of changing task variables. The data presented here provide a target for computational models that aim to account for changes in overt attentional behaviors across learning.

Visual category learning.

Visual categories group together different objects as the same kinds of thing. We review a selection of research on how visual categories are learned. We begin with a guide to visual category learning experiments, describing a space of common manipulations of objects, categories, and methods used in the category learning literature. We open with a guide to these details in part because throughout our review we highlight how methodological details can sometimes loom large in theoretical discussions of visual category learning, how variations in methodological details can significantly affect our understanding of visual category learning, and how manipulations of methodological details can affect how visual categories are learned. We review a number of core theories of visual category learning, specifically those theories instantiated as computational models, highlighting just some of the experimental results that help distinguish between competing models. We examine behavioral and neural evidence for single versus multiple representational systems for visual category learning. We briefly discuss how visual category learning influences visual perception, describing empirical and brain imaging results that show how learning to categorize objects can influence how those objects are represented and perceived. We close with work that can potentially impact translation, describing recent experiments that explicitly manipulate key methodological details of category learning procedures with the goal of optimizing visual category learning. WIREs Cogn Sci 2014, 5:75-94. doi: 10.1002/wcs.1268 CONFLICT OF INTEREST: The authors have declared no conflicts of interest for this article. For further resources related to this article, please visit the WIREs website.

Extracting prototypes from exemplars What can corpus data tell us about concept representation?

AbstractOver the past four decades, two distinct alternatives have emerged to rule-based models of how linguistic categories are stored and represented as cognitive structures, namely the prototype and exemplar theories. Although these models were initially thought to be mutually exclusive, shifts from one mechanism to the other have been observed in category learning experiments, bringing the models closer together. In this paper we implement a technique akin to varying abstraction modelling, that assumes intermediate abstraction processes to underlie category representations and categorization decisions; we do so using familiar statistical techniques such as regression and clustering that track frequency distributions in input. With this model we simulate, on the basis of actual usage of Russian try verbs and Finnish think verbs as observed in corpora, how prototypes for near-synonymous verbs could be formed from concrete exemplars at different levels of abstraction.In so doing, we take a closer look at the cognitive linguistic flirtation with multiple categorization theories, suggesting three improvements anchored in the fact that cognitive linguistics is a usage-based theory of language. Firstly, we show that language provides support for considering single prototype and full exemplar models as opposite ends along a continuum of abstraction. Secondly, we present a methodology that simulates how prototypes can be obtained from exemplars at more than one level of abstraction in a systematic and verifiable way. And thirdly, we illustrate our claims on the basis of work on verbs, denoting intangible events that are neither stable in nor independent of time and express relational concepts; this implies that verbs are more susceptible to their meanings being influenced by the concepts they relate.

Learning time-varying categories

Many kinds of objects and events in our world have a strongly time-dependent quality. However, most theories about concepts and categories either are insensitive to variation over time or treat it as a nuisance factor that produces irrational order effects during learning. In this article, we present two category learning experiments in which we explored peoples' ability to learn categories whose structure is strongly time-dependent. We suggest that order effects in categorization may in part reflect a sensitivity to changing environments, and that understanding dynamically changing concepts is an important part of developing a full account of human categorization.

The hazards of explanation: Overgeneralization in the face of exceptions.

Seeking explanations is central to science, education, and everyday thinking, and prompting learners to explain is often beneficial. Nonetheless, in 2 category learning experiments across artifact and social domains, we demonstrate that the very properties of explanation that support learning can impair learning by fostering overgeneralizations. We find that explaining encourages learners to seek broad patterns, hindering learning when patterns involve exceptions. By revealing how effects of explanation depend on the structure of what is being learned, these experiments simultaneously demonstrate the hazards of explaining and provide evidence for why explaining is so often beneficial. For better or for worse, explaining recruits the remarkable human capacity to seek underlying patterns that go beyond individual observations.