Neural dynamics underlying the acquisition of distinct auditory category structures.

Despite the multidimensional and temporally fleeting nature of auditory signals we quickly learn to assign novel sounds to behaviorally relevant categories. The neural systems underlying the learning and representation of novel auditory categories are far from understood. Current models argue for a rigid specialization of hierarchically organized core regions that are fine-tuned to extracting and mapping relevant auditory dimensions to meaningful categories. Scaffolded within a dual-learning systems approach, we test a competing hypothesis: the spatial and temporal dynamics of emerging auditory-category representations are not driven by the underlying dimensions but are constrained by category structure and learning strategies. To test these competing models, we used functional Magnetic Resonance Imaging (fMRI) to assess representational dynamics during the feedback-based acquisition of novel non-speech auditory categories with identical dimensions but differing category structures: rule-based (RB) categories, hypothesized to involve an explicit sound-to-rule mapping network, and information integration (II) based categories, involving pre-decisional integration of dimensions via a procedural-based sound-to-reward mapping network. Adults were assigned to either the RB (n = 30, 19 females) or II (n = 30, 22 females) learning tasks. Despite similar behavioral learning accuracies, learning strategies derived from computational modeling and involvements of corticostriatal systems during feedback processing differed across tasks. Spatiotemporal multivariate representational similarity analysis revealed an emerging representation within an auditory sensory-motor pathway exclusively for the II learning task, prominently involving the superior temporal gyrus (STG), inferior frontal gyrus (IFG), and posterior precentral gyrus. In contrast, the RB learning task yielded distributed neural representations within regions involved in cognitive-control and attentional processes that emerged at different time points of learning. Our results unequivocally demonstrate that auditory learners' neural systems are highly flexible and show distinct spatial and temporal patterns that are not dimension-specific but reflect underlying category structures and learning strategies.

The impact of training methodology and representation on rule-based categorization: An fMRI study.

Hélie, Shamloo, & Ell (2017) showed that regular classification learning instructions (A/B) promote between-category knowledge in rule-based categorization whereas conceptual learning instructions (YES/NO) promote learning within-category knowledge with the same categories. Here we explore how these tasks affect brain activity using fMRI. Participants learned two sets of two categories. Computational models were fit to the behavioral data to determine the type of knowledge learned by each participant. fMRI contrasts were computed to compare BOLD signal between the tasks and between the types of knowledge. The results show that participants in the YES/NO task had more activity in the pre-supplementary motor area, prefrontal cortex, and the angular/supramarginal gyrus. These brain areas are related to working memory and part of the dorsal attention network, which showed increased task-based functional connectivity with the medial temporal lobes. In contrast, participants in the A/B task had more activity in the thalamus and caudate. These results suggest that participants in the YES/NO task used bivalent rules and may have treated each contextual question as a separate task, switching task each time the question changed. Activity in the A/B condition was more consistent with participants applying direct Stimulus → Response rules. With regards to knowledge representation, there was a large shared network of brain areas, but participants learning between-category information showed additional posterior parietal activity, which may be related to the inhibition of incorrect motor programs.

The impact of training methodology and category structure on the formation of new categories from existing knowledge.

Categorization decisions are made thousands of times every day, and a typical adult knows tens of thousands of categories. It is thus relatively rare that adults learn new categories without somehow reorganizing pre-existing knowledge. Yet, most perceptual categorization research has investigated the ability to learn new categories without considering they relation to existing knowledge. In this article, we test the ability of young adults to merge already known categories into new categories as a function of training methodology and category structures using two experiments. Experiment 1 tests participants' ability to merge rule-based or information-integration categories that are either contiguous, semi-contiguous, or non-contiguous in perceptual space using a classification paradigm. Experiment 2 is similar Experiment 1 but uses a YES/NO learning paradigm instead. The results of both experiments suggest a strong effect of the contiguity of the merged categories in perceptual space that depends on the type of category representation that is learned. The type of category representation that is learned, in turn, depends on a complex interaction of the category structures and training task. We conclude by discussing the relevance of these results for categorization outside the laboratory.

Trial-by-trial identification of categorization strategy using iterative decision-bound modeling.

Identifying the strategy that participants use in laboratory experiments is crucial in interpreting the results of behavioral experiments. This article introduces a new modeling procedure called iterative decision-bound modeling (iDBM), which iteratively fits decision-bound models to the trial-by-trial responses generated from single participants in perceptual categorization experiments. The goals of iDBM are to identify: (1) all response strategies used by a participant, (2) changes in response strategy, and (3) the trial number at which each change occurs. The new method is validated by testing its ability to identify the response strategies used in noisy simulated data. The benchmark simulation results show that iDBM is able to detect and identify strategy switches during an experiment and accurately estimate the trial number at which the strategy change occurs in low to moderate noise conditions. The new method is then used to reanalyze data from Ell and Ashby (2006). Applying iDBM revealed that increasing category overlap in an information-integration category learning task increased the proportion of participants who abandoned explicit rules, and reduced the number of training trials needed to abandon rules in favor of a procedural strategy. Finally, we discuss new research questions made possible through iDBM.

Criterion learning in rule-based categorization: simulation of neural mechanism and new data.

In perceptual categorization, rule selection consists of selecting one or several stimulus-dimensions to be used to categorize the stimuli (e.g., categorize lines according to their length). Once a rule has been selected, criterion learning consists of defining how stimuli will be grouped using the selected dimension(s) (e.g., if the selected rule is line length, define 'long' and 'short'). Very little is known about the neuroscience of criterion learning, and most existing computational models do not provide a biological mechanism for this process. In this article, we introduce a new model of rule learning called Heterosynaptic Inhibitory Criterion Learning (HICL). HICL includes a biologically-based explanation of criterion learning, and we use new category-learning data to test key aspects of the model. In HICL, rule selective cells in prefrontal cortex modulate stimulus-response associations using pre-synaptic inhibition. Criterion learning is implemented by a new type of heterosynaptic error-driven Hebbian learning at inhibitory synapses that uses feedback to drive cell activation above/below thresholds representing ionic gating mechanisms. The model is used to account for new human categorization data from two experiments showing that: (1) changing rule criterion on a given dimension is easier if irrelevant dimensions are also changing (Experiment 1), and (2) showing that changing the relevant rule dimension and learning a new criterion is more difficult, but also facilitated by a change in the irrelevant dimension (Experiment 2). We conclude with a discussion of some of HICL's implications for future research on rule learning.

Is pressure stressful? The impact of pressure on the stress response and category learning.

We examined the basic question of whether pressure is stressful. We proposed that when examining the role of stress or pressure in cognitive performance, it is important to consider the type of pressure, the stress response, and the aspect of cognition assessed. In Experiment 1, outcome pressure was not experienced as stressful but did lead to impaired performance on a rule-based (RB) category-learning task, but not on a more procedural information-integration (II) task. In Experiment 2, the addition of monitoring pressure resulted in a modest stress response to combined pressure and impairment on both tasks. Across experiments, higher stress appraisals were associated with decreased performance on the RB, but not on the II, task. In turn, higher stress reactivity (i.e., heart rate) was associated with enhanced performance on the II, but not on the RB, task. This work represents an initial step toward integrating the stress cognition and pressure cognition literatures and suggests that integrating these fields may require consideration of the type of pressure, the stress response, and the cognitive system mediating performance.

Targeted training of the decision rule benefits rule-guided behavior in Parkinson's disease.

The impact of Parkinson's disease (PD) on rule-guided behavior has received considerable attention in cognitive neuroscience. The majority of research has used PD as a model of dysfunction in frontostriatal networks, but very few attempts have been made to investigate the possibility of adapting common experimental techniques in an effort to identify the conditions that are most likely to facilitate successful performance. The present study investigated a targeted training paradigm designed to facilitate rule learning and application using rule-based categorization as a model task. Participants received targeted training in which there was no selective-attention demand (i.e., stimuli varied along a single, relevant dimension) or nontargeted training in which there was selective-attention demand (i.e., stimuli varied along a relevant dimension as well as an irrelevant dimension). Following training, all participants were tested on a rule-based task with selective-attention demand. During the test phase, PD patients who received targeted training performed similarly to control participants and outperformed patients who did not receive targeted training. As a preliminary test of the generalizability of the benefit of targeted training, a subset of the PD patients were tested on the Wisconsin card sorting task (WCST). PD patients who received targeted training outperformed PD patients who did not receive targeted training on several WCST performance measures. These data further characterize the contribution of frontostriatal circuitry to rule-guided behavior. Importantly, these data also suggest that PD patient impairment, on selective-attention-demanding tasks of rule-guided behavior, is not inevitable and highlight the potential benefit of targeted training.

Learning and generalization of within-category representations in a rule-based category structure.

The task requirements during the course of category learning are critical for promoting within-category representations (e.g., correlational structure of the categories). Recent data suggest that for unidimensional rule-based structures, only inference training promotes the learning of within-category representations, and generalization across tasks is limited. It is unclear if this is a general feature of rule-based structures, or a limitation of unidimensional rule-based structures. The present work reports the results of three experiments further investigating this issue using an exclusive-or rule-based structure where successful performance depends upon attending to two stimulus dimensions. Participants were trained using classification or inference and were tested using inference. For both the classification and inference training conditions, within-category representations were learned and could be generalized at test (i.e., from classification to inference) and this result was dependent upon a congruence between local and global regions of the stimulus space. These data further support the idea that the task requirements during learning (i.e., a need to attend to multiple stimulus dimensions) are critical determinants of the category representations that are learned and the utility of these representations for supporting generalization in novel situations.

Cerebellar pathology does not impair performance on identification or categorization tasks.

In comparison to the basal ganglia, prefrontal cortex, and medial temporal lobes, the cerebellum has been absent from recent research on the neural substrates of categorization and identification, two prominent tasks in the learning and memory literature. To investigate the contribution of the cerebellum to these tasks, we tested patients with cerebellar pathology (seven with bilateral degeneration, six with unilateral lesions, and two with midline damage) on rule-based and information-integration categorization tasks and an identification task. In rule-based tasks, it is assumed that participants learn the categories through an explicit reasoning process. In information-integration tasks, optimal performance requires the integration of information from multiple stimulus dimensions, and participants are typically unaware of the decision strategy. The identification task, in contrast, required participants to learn arbitrary, color-word associations. The cerebellar patients performed similar to matched controls on all three tasks and performance did not vary with the extent of cerebellar pathology. Although the interpretation of these null results requires caution, these data contribute to the current debate on cerebellar contributions to cognition by providing boundary conditions on understanding the neural substrates of categorization and identification, and help define the functional domain of the cerebellum in learning and memory.

The effect of training methodology on knowledge representation in categorization.

Category representations can be broadly classified as containing within-category information or between-category information. Although such representational differences can have a profound impact on decision-making, relatively little is known about the factors contributing to the development and generalizability of different types of category representations. These issues are addressed by investigating the impact of training methodology and category structures using a traditional empirical approach as well as the novel adaptation of computational modeling techniques from the machine learning literature. Experiment 1 focused on rule-based (RB) category structures thought to promote between-category representations. Participants learned two sets of two categories during training and were subsequently tested on a novel categorization problem using the training categories. Classification training resulted in a bias toward between-category representations whereas concept training resulted in a bias toward within-category representations. Experiment 2 focused on information-integration (II) category structures thought to promote within-category representations. With II structures, there was a bias toward within-category representations regardless of training methodology. Furthermore, in both experiments, computational modeling suggests that only within-category representations could support generalization during the test phase. These data suggest that within-category representations may be dominant and more robust for supporting the reconfiguration of current knowledge to support generalization.

The impact of category structure and training methodology on learning and generalizing within-category representations.

When interacting with categories, representations focused on within-category relationships are often learned, but the conditions promoting within-category representations and their generalizability are unclear. We report the results of three experiments investigating the impact of category structure and training methodology on the learning and generalization of within-category representations (i.e., correlational structure). Participants were trained on either rule-based or information-integration structures using classification (Is the stimulus a member of Category A or Category B?), concept (e.g., Is the stimulus a member of Category A, Yes or No?), or inference (infer the missing component of the stimulus from a given category) and then tested on either an inference task (Experiments 1 and 2) or a classification task (Experiment 3). For the information-integration structure, within-category representations were consistently learned, could be generalized to novel stimuli, and could be generalized to support inference at test. For the rule-based structure, extended inference training resulted in generalization to novel stimuli (Experiment 2) and inference training resulted in generalization to classification (Experiment 3). These data help to clarify the conditions under which within-category representations can be learned. Moreover, these results make an important contribution in highlighting the impact of category structure and training methodology on the generalization of categorical knowledge.

Focal putamen lesions impair learning in rule-based, but not information-integration categorization tasks.

Previous research on the role of the basal ganglia in category learning has focused on patients with Parkinson's and Huntington's disease, neurodegenerative diseases frequently accompanied by additional cortical pathology. The goal of the present study was to extend this work to patients with basal ganglia lesions due to stroke, asking if similar changes in performance would be observed in patients with more focal pathology. Patients with basal ganglia lesions centered in the putamen (6 left side, 1 right side) were tested on rule-based and information-integration visual categorization tasks. In rule-based tasks, it is assumed that participants can learn the category structures through an explicit reasoning process. In information-integration tasks, optimal performance requires the integration of information from two or more stimulus components, and participants are typically unaware of the category rules. Consistent with previous studies involving patients with degenerative disorders of the basal ganglia, the stroke patients were impaired on the rule-based task, and quantitative, model-based analyses indicate that this deficit was due to the inefficient application of decision strategies. In contrast, the patients were unimpaired on the information-integration task. This pattern of results provides converging evidence supporting a role of the basal ganglia and, in particular, the putamen in rule-based category learning.

One Giant Leap for Categorizers: One Small Step for Categorization Theory.

We explore humans' rule-based category learning using analytic approaches that highlight their psychological transitions during learning. These approaches confirm that humans show qualitatively sudden psychological transitions during rule learning. These transitions contribute to the theoretical literature contrasting single vs. multiple category-learning systems, because they seem to reveal a distinctive learning process of explicit rule discovery. A complete psychology of categorization must describe this learning process, too. Yet extensive formal-modeling analyses confirm that a wide range of current (gradient-descent) models cannot reproduce these transitions, including influential rule-based models (e.g., COVIS) and exemplar models (e.g., ALCOVE). It is an important theoretical conclusion that existing models cannot explain humans' rule-based category learning. The problem these models have is the incremental algorithm by which learning is simulated. Humans descend no gradient in rule-based tasks. Very different formal-modeling systems will be required to explain humans' psychology in these tasks. An important next step will be to build a new generation of models that can do so.

Learning robust cortico-cortical associations with the basal ganglia: an integrative review.

This article focuses on the interaction between the basal ganglia (BG) and prefrontal cortex (PFC). The BG are a group of nuclei at the base of the forebrain that are highly connected with cortex. A century of research suggests that the role of the BG is not exclusively motor, and that the BG also play an important role in learning and memory. In this review article, we argue that one important role of the BG is to train connections between posterior cortical areas and frontal cortical regions that are responsible for automatic behavior after extensive training. According to this view, one effect of BG trial-and-error learning is to activate the correct frontal areas shortly after posterior associative cortex activation, thus allowing for Hebbian learning of robust, fast, and efficient cortico-cortical processing. This hypothesized process is general, and the content of the learned associations depends on the specific areas involved (e.g., associations involving premotor areas would be more closely related to behavior than associations involving the PFC). We review experiments aimed at pinpointing the function of the BG and the frontal cortex and show that these results are consistent with the view that the BG is a general purpose trainer for cortico-cortical connections. We conclude with a discussion of some implications of the integrative framework and how this can help better understand the role of the BG in many different tasks.

Category learning deficits in Parkinson's disease.

Sixteen patients with Parkinson's disease (PD), 15 older controls (OCs), and 109 younger controls (YCs) were compared in 2 category-learning tasks. Participants attempted to assign colored geometric figures to 1 of 2 categories. In rule-based tasks, category membership was defined by an explicit rule that was easy to verbalize, whereas in information-integration tasks, there was no salient verbal rule and accuracy was maximized only if information from 3 stimulus components was integrated at some predecisional stage. The YCs performed the best on both tasks. The PD patients were highly impaired compared with the OCs, in the rule-based categorization task but were not different from the OCs in the information-integration task. These results support the hypothesis that learning in these 2 tasks is mediated by functionally separate systems.

The impact of category separation on unsupervised categorization.

Most previous research on unsupervised categorization has used unconstrained tasks in which no instructions are provided about the underlying category structure or in which the stimuli are not clustered into categories. Few studies have investigated constrained tasks in which the goal is to learn predefined stimulus clusters in the absence of feedback. These studies have generally reported good performance when the stimulus clusters could be separated by a one-dimensional rule. In the present study, we investigated the limits of this ability. Results suggest that even when two stimulus clusters are as widely separated, as in previous studies, performance is poor if within-category variance on the relevant dimension is nonnegligible. In fact, under these conditions, many participants failed even to identify the single relevant stimulus dimension. This poor performance is generally incompatible with all current models of unsupervised category learning.

Unsupervised category learning with integral-dimension stimuli.

Despite the recent surge in research on unsupervised category learning, the majority of studies have focused on unconstrained tasks in which no instructions are provided about the underlying category structure. Relatively little research has focused on constrained tasks in which the goal is to learn predefined stimulus clusters in the absence of feedback. The few studies that have addressed this issue have focused almost exclusively on stimuli for which it is relatively easy to attend selectively to the component dimensions (i.e., separable dimensions). In the present study, we investigated the ability of participants to learn categories constructed from stimuli for which it is difficult, if not impossible, to attend selectively to the component dimensions (i.e., integral dimensions). The experiments demonstrate that individuals are capable of learning categories constructed from the integral dimensions of brightness and saturation, but this ability is generally limited to category structures requiring selective attention to brightness. As might be expected with integral dimensions, participants were often able to integrate brightness and saturation information in the absence of feedback--an ability not observed in previous studies with separable dimensions. Even so, there was a bias to weight brightness more heavily than saturation in the categorization process, suggesting a weak form of selective attention to brightness. These data present an important challenge for the development of models of unsupervised category learning.

When bad stress goes good: increased threat reactivity predicts improved category learning performance.

The way in which we respond to everyday stressors can have a profound impact on cognitive functioning. Maladaptive stress responses in particular are generally associated with impaired cognitive performance. We argue, however, that the cognitive system mediating task performance is also a critical determinant of the stress-cognition relationship. Consistent with this prediction, we observed that stress reactivity consistent with a maladaptive, threat response differentially predicted performance on two categorization tasks. Increased threat reactivity predicted enhanced performance on an information-integration task (i.e., learning is thought to depend upon a procedural-based memory system), and a (nonsignificant) trend for impaired performance on a rule-based task (i.e., learning is thought to depend upon a hypothesis-testing system). These data suggest that it is critical to consider both variability in the stress response and variability in the cognitive system mediating task performance in order to fully understand the stress-cognition relationship.

Rule-based categorization deficits in focal basal ganglia lesion and Parkinson's disease patients.

Patients with basal ganglia (BG) pathology are consistently found to be impaired on rule-based category learning tasks in which learning is thought to depend upon the use of an explicit, hypothesis-guided strategy. The factors that influence this impairment remain unclear. Moreover, it remains unknown if the impairments observed in patients with degenerative disorders such as Parkinson's disease (PD) are also observed in those with focal BG lesions. In the present study, we tested patients with either focal BG lesions or PD on two categorization tasks that varied in terms of their demands on selective attention and working memory. Individuals with focal BG lesions were impaired on the task in which working memory demand was high and performed similarly to healthy controls on the task in which selective-attention demand was high. In contrast, individuals with PD were impaired on both tasks, and accuracy rates did not differ between on and off medication states for a subset of patients who were also tested after abstaining from dopaminergic medication. Quantitative, model-based analyses attributed the performance deficit for both groups in the task with high working memory demand to the utilization of suboptimal strategies, whereas the PD-specific impairment on the task with high selective-attention demand was driven by the inconsistent use of an optimal strategy. These data suggest that the demands on selective attention and working memory affect the presence of impairment in patients with focal BG lesions and the nature of the impairment in patients with PD.

Criterial noise effects on rule-based category learning: the impact of delayed feedback.

Variability in the representation of the decision criterion is assumed in many category-learning models, yet few studies have directly examined its impact. On each trial, criterial noise should result in drift in the criterion and will negatively impact categorization accuracy, particularly in rule-based categorization tasks, where learning depends on the maintenance and manipulation of decision criteria. In three experiments, we tested this hypothesis and examined the impact of working memory on slowing the drift rate. In Experiment 1, we examined the effect of drift by inserting a 5-sec delay between the categorization response and the delivery of corrective feedback, and working memory demand was manipulated by varying the number of decision criteria to be learned. Delayed feedback adversely affected performance, but only when working memory demand was high. In Experiment 2, we built on a classic finding in the absolute identification literature and demonstrated that distributing the criteria across multiple dimensions decreases the impact of drift during the delay. In Experiment 3, we confirmed that the effect of drift during the delay is moderated by working memory. These results provide important insights into the interplay between criterial noise and working memory, as well as providing important constraints for models of rule-based category learning.