Visual category learning

Individual differences in working memory impact the trajectory of non-native speech category learning

What is the role of working memory over the course of non-native speech category learning? Prior work has predominantly focused on how working memory might influence learning assessed at a single timepoint. Here, we substantially extend this prior work by examining the role of working memory on speech learning performance over time (i.e., over several months) and leverage a multifaceted approach that provides key insights into how working memory influences learning accuracy, maintenance of knowledge over time, generalization ability, and decision processes. We found that the role of working memory in non-native speech learning depends on the timepoint of learning and whether individuals learned the categories at all. Among learners, across all stages of learning, working memory was associated with higher accuracy as well as faster and slightly more cautious decision making. Further, while learners and non-learners did not have substantially different working memory performance, learners had faster evidence accumulation and more cautious decision thresholds throughout all sessions. Working memory may enhance learning by facilitating rapid category acquisition in initial stages and enabling faster and slightly more careful decision-making strategies that may reduce the overall effort needed to learn. Our results have important implications for developing interventions to improve learning in naturalistic language contexts.

Object recognition ability predicts category learning with medical images

We investigated the relationship between category learning and domain-general object recognition ability (o). We assessed this relationship in a radiological context, using a category learning test in which participants judged whether white blood cells were cancerous. In study 1, Bayesian evidence negated a relationship between o and category learning. This lack of correlation occurred despite high reliability in all measurements. However, participants only received feedback on the first 10 of 60 trials. In study 2, we assigned participants to one of two conditions: feedback on only the first 10 trials, or on all 60 trials of the category learning test. We found strong Bayesian evidence for a correlation between o and categorisation accuracy in the full-feedback condition, but not when feedback was limited to early trials. Moderate Bayesian evidence supported a difference between these correlations. Without feedback, participants may stick to simple rules they formulate at the start of category learning, when trials are easier. Feedback may encourage participants to abandon less effective rules and switch to exemplar learning. This work provides the first evidence relating o to a specific learning mechanism, suggesting this ability is more dependent upon exemplar learning mechanisms than rule abstraction. Object-recognition ability could complement other sources of individual differences when predicting accuracy of medical image interpretation.

Multiple components of statistical word learning are resource dependent: Evidence from a dual-task learning paradigm

It is increasingly understood that people may learn new word/object mappings in part via a form of statistical learning in which they track co-occurrences between words and objects across situations (cross-situational learning). Multiple learning processes contribute to this, thought to reflect the simultaneous influence of real-time hypothesis testing and graduate learning. It is unclear how these processes interact, and if any require explicit cognitive resources. To manipulate the availability of working memory resources for explicit processing, participants completed a dual-task paradigm in which a cross-situational word-learning task was interleaved with a short-term memory task. We then used trial-by-trial analyses to estimate how different learning processes that play out simultaneously are impacted by resource availability. Critically, we found that the effect of hypothesis testing and gradual learning effects showed a small reduction under limited resources, and that the effect of memory load was not fully mediated by these processes. This suggests that neither is purely explicit, and there may be additional resource-dependent processes at play. Consistent with a hybrid account, these findings suggest that these two aspects of learning may reflect different aspects of a single system gated by attention, rather than competing learning systems.

Visual statistical learning is modulated by arbitrary and natural categories

Visual statistical learning (VSL) describes the unintentional extraction of statistical regularities from visual environments across time or space, and is typically studied using novel stimuli (e.g., symbols unfamiliar to participants) and using familiarization procedures that are passive or require only basic vigilance. The natural visual world, however, is rich with a variety of complex visual stimuli, and we experience that world in the presence of goal-driven behavior including overt learning of other kinds. To examine how VSL responds to such contexts, we exposed subjects to statistical contingencies as they learned arbitrary categorical mappings of unfamiliar stimuli (fractals, Experiment 1) or familiar stimuli with preexisting categorical boundaries (faces and scenes, Experiment 2). In a familiarization stage, subjects learned by trial and error the arbitrary mappings between stimuli and one of two responses. Unbeknownst to participants, items were paired such that they always appeared together in the stream. Pairs were equally likely to be of the same or different category. In a pair recognition stage to assess VSL, subjects chose between a target pair and a foil pair. In both experiments, subjects' VSL was shaped by arbitrary categories: same-category pairs were learned better than different-category pairs. Natural categories (Experiment 2) also played a role, with subjects learning same-natural-category pairs at higher rates than different-category pairs, an effect that did not interact with arbitrary mappings. We conclude that learning goals of the observer and preexisting knowledge about the structure of the world play powerful roles in the incidental learning of novel statistical information.

A computational cognitive model of judgments of relative direction

In the past several decades, considerable theoretical progress has been made in understanding the role of reference frames in the encoding and retrieval of spatial information about the environment. Many of these insights have come from participants making judgments of relative direction using their memories of spatial layouts. In this task, participants are asked to imagine standing at a given location and facing a certain direction, and to point to a target location. Although this task has been widely and productively used, a computational cognitive model of judgments of relative direction has yet to be introduced. Computational modeling of judgments of relative direction is a critical next step to formulating and testing hypotheses about the cognitive processes involved in establishing and using spatial reference frames. We present an initial attempt to model judgments of relative direction and fit the model to two datasets exhibiting behavioral patterns commonly observed in the spatial memory literature. The model was able to predict many important features of these data, most notably alignment effects. We discuss directions for future modeling efforts.

Development of Attention and Accuracy in Learning a Categorization Task

Being able to categorize objects as similar or different is an essential skill. An important aspect of learning to categorize is learning to attend to relevant features (i.e., features that determine category membership) and ignore irrelevant features of the to-be-categorized objects. Feature variability across objects of different categories is informative, because it allows inferring the rules underlying category membership. In this study, participants learned to categorize fictitious creatures (i.e., aliens). We measured attention to the aliens during learning using eye-tracking and calculated the attentional focus as the ratio of attention to relevant versus irrelevant features. As expected, participants’ categorization accuracy improved with practice; however, in contrast to our expectations, their attentional focus did not improve with practice. When computing the attentional focus, attention to the aliens’ eyes was disregarded, because while eyes attract a lot of attention, they did not vary across aliens (non-informative feature). Yet, an explorative analysis of attention to eyes suggested that participants’ attentional focus did become somewhat more efficient in that over time they learned to ignore the eyes. Results are discussed in the context of the need for instructional methods to improve attentional focus in learning to categorize.

Mouse visual cortex areas represent perceptual and semantic features of learned visual categories

Associative memories are stored in distributed networks extending across multiple brain regions. However, it is unclear to what extent sensory cortical areas are part of these networks. Using a paradigm for visual category learning in mice, we investigated whether perceptual and semantic features of learned category associations are already represented at the first stages of visual information processing in the neocortex. Mice learned categorizing visual stimuli, discriminating between categories and generalizing within categories. Inactivation experiments showed that categorization performance was contingent on neuronal activity in the visual cortex. Long-term calcium imaging in nine areas of the visual cortex identified changes in feature tuning and category tuning that occurred during this learning process, most prominently in the postrhinal area (POR). These results provide evidence for the view that associative memories form a brain-wide distributed network, with learning in early stages shaping perceptual representations and supporting semantic content downstream.

Perceptual category learning in autism spectrum disorder: Truth and consequences

The ability to categorize is fundamental to cognitive development. Some categories emerge effortlessly and rapidly while others can take years of experience to acquire. Children with autism spectrum disorder (ASD) are often able to name and sort objects, suggesting that their categorization abilities are largely intact. However, recent experimental work shows that the categories formed by individuals with ASD may diverge substantially from those that most people learn. This review considers how atypical perceptual category learning can affect cognitive development in children with ASD and how atypical categorization may contribute to many of the socially problematic symptoms associated with this disorder. Theoretical approaches to understanding perceptual processing and category learning at both the behavioral and neural levels are assessed in relation to known alterations in perceptual category learning associated with ASD. Mismatches between the ways in which children learn to organize perceived events relative to their peers and adults can accumulate over time, leading to difficulties in communication, social interactions, academic performance, and behavioral flexibility.

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.

Which Matters More in Incidental Category Learning: Edge-Based Versus Surface-Based Features

Although many researches have shown that edge-based information is more important than surface-based information in object recognition, it remains unclear whether edge-based features play a more crucial role than surface-based features in category learning. To address this issue, a modified prototype distortion task was adopted in the present study, in which each category was defined by a rule or a similarity about either the edge-based features (i.e., contours or shapes) or the corresponding surface-based features (i.e., color and textures). The results of Experiments 1 and 2 showed that when the category was defined by a rule, the performance was significantly better in the edge-based condition than in the surface-based condition in the testing phase, and increasing the defined dimensions enhanced rather than reduced performance in the edge-based condition but not in the surface-based condition. The results of Experiment 3 showed that when each category was defined by a similarity, there was also a larger learning effect when the category was defined by edge-based dimensions than by surface-based dimensions in the testing phase. The current study is the first to provide convergent evidence that the edge-based information matters more than surface-based information in incidental category learning.

On the Role of Cortex-Basal Ganglia Interactions for Category Learning: A Neurocomputational Approach

In addition to the prefrontal cortex (PFC), the basal ganglia (BG) have been increasingly often reported to play a fundamental role in category learning, but the circuit mechanisms mediating their interaction remain to be explored. We developed a novel neurocomputational model of category learning that particularly addresses the BG-PFC interplay. We propose that the BG bias PFC activity by removing the inhibition of cortico-thalamo-cortical loop and thereby provide a teaching signal to guide the acquisition of category representations in the corticocortical associations to the PFC. Our model replicates key behavioral and physiological data of macaque monkey learning a prototype distortion task from Antzoulatos and Miller (2011) Our simulations allowed us to gain a deeper insight into the observed drop of category selectivity in striatal neurons seen in the experimental data and in the model. The simulation results and a new analysis of the experimental data based on the model's predictions show that the drop in category selectivity of the striatum emerges as the variability of responses in the striatum rises when confronting the BG with an increasingly larger number of stimuli to be classified. The neurocomputational model therefore provides new testable insights of systems-level brain circuits involved in category learning that may also be generalized to better understand other cortico-BG-cortical loops.SIGNIFICANCE STATEMENT Inspired by the idea that basal ganglia (BG) teach the prefrontal cortex (PFC) to acquire category representations, we developed a novel neurocomputational model and tested it on a task that was recently applied in monkey experiments. As an advantage over previous models of category learning, our model allows to compare simulation data with single-cell recordings in PFC and BG. We not only derived model predictions, but already verified a prediction to explain the observed drop in striatal category selectivity. When testing our model with a simple, real-world face categorization task, we observed that the fast striatal learning with a performance of 85% correct responses can teach the slower PFC learning to push the model performance up to almost 100%.

Bayesian statistical approaches to evaluating cognitive models

Cognitive models aim to explain complex human behavior in terms of hypothesized mechanisms of the mind. These mechanisms can be formalized in terms of mathematical structures containing parameters that are theoretically meaningful. For example, in the case of perceptual decision making, model parameters might correspond to theoretical constructs like response bias, evidence quality, response caution, and the like. Formal cognitive models go beyond verbal models in that cognitive mechanisms are instantiated in terms of mathematics and they go beyond statistical models in that cognitive model parameters are psychologically interpretable. We explore three key elements used to formally evaluate cognitive models: parameter estimation, model prediction, and model selection. We compare and contrast traditional approaches with Bayesian statistical approaches to performing each of these three elements. Traditional approaches rely on an array of seemingly ad hoc techniques, whereas Bayesian statistical approaches rely on a single, principled, internally consistent system. We illustrate the Bayesian statistical approach to evaluating cognitive models using a running example of the Linear Ballistic Accumulator model of decision making (Brown SD, Heathcote A. The simplest complete model of choice response time: linear ballistic accumulation. Cogn Psychol 2008, 57:153-178). WIREs Cogn Sci 2018, 9:e1458. doi: 10.1002/wcs.1458 This article is categorized under: Neuroscience > Computation Psychology > Reasoning and Decision Making Psychology > Theory and Methods.

Getting to know you: The development of mechanisms underlying face learning

Nearly every study investigating the development of face recognition has focused on the ability to tell people apart using one or two tightly controlled images to represent each identity. Such research ignores the challenge of recognizing the same person despite variability in appearance. Whereas natural variation in appearance makes unfamiliar faces difficult to recognize, by 6 years of age people easily recognize multiple images of familiar faces. Two mechanisms are proposed to underlie the process by which adults become familiar with newly encountered faces. We provide the first examination of the development of these mechanisms during childhood (6-11 years). In Experiment 1, we examined children's (6- to 10-year-olds') and adults' ability to engage in ensemble coding-the ability to rapidly extract an average representation of an identity from several instances. In Experiment 2, we examined children's ability to use within-person variability in appearance to recognize novel instances of a newly encountered identity. We created a child-friendly perceptual matching task, and the number of images to which participants were exposed varied across targets. Although children were less accurate than adults overall in Experiment 2, we found no age-related improvement in either ensemble coding or the ability to benefit from exposure to within-person variability in appearance when learning a new face, suggesting that both abilities are developed by 6 years of age. We discuss the implications of these findings for understanding the nature of mechanisms underlying face learning and other developmental processes such as language and music.

Differential Impact of Visuospatial Working Memory on Rule-based and Information-integration Category Learning

Previous studies have indicated that the category learning system is a mechanism with multiple processing systems, and that working memory has different effects on category learning. But how does visuospatial working memory affect perceptual category learning? As there is no definite answer to this question, we conducted three experiments. In Experiment 1, the dual-task paradigm with sequential presentation was adopted to investigate the influence of visuospatial working memory on rule-based and information-integration category learning. The results showed that visuospatial working memory interferes with rule-based but not information-integration category learning. In Experiment 2, the dual-task paradigm with simultaneous presentation was used, in which the categorization task was integrated into the visuospatial working memory task. The results indicated that visuospatial working memory affects information-integration category learning but not rule-based category learning. In Experiment 3, the dual-task paradigm with simultaneous presentation was employed, in which visuospatial working memory was integrated into the category learning task. The results revealed that visuospatial working memory interferes with both rule-based and information-integration category learning. Through these three experiments, we found that, regarding the rule-based category learning, working memory load is the main mechanism by which visuospatial working memory influences the discovery of the category rules. In addition, regarding the information-integration category learning, visual resources mainly operates on the category representation.

Tracking the emergence of memories: A category-learning paradigm to explore schema-driven recognition

Previous research has shown that prior knowledge structures or schemas affect recognition memory. However, since the acquisition of schemas occurs over prolonged periods of time, few paradigms allow the direct manipulation of schema acquisition to study their effect on memory performance. Recently, a number of parallelisms in recognition memory between studies involving schemas and studies involving category learning have been identified. The current paper capitalizes on these findings and offers a novel experimental paradigm that allows manipulation of category learning between individuals to study the effects of schema acquisition on recognition. First, participants learn to categorize computer-generated items whose category-inclusion criteria differ between participants. Next, participants study items that belong to either the learned category, the non-learned category, both, or neither. Finally, participants receive a recognition test that includes old and new items, either from the learned, the non-learned, or neither category. Using variations on this paradigm, four experiments were conducted. The results from the first three studies suggest that learning a category increases hit rates for old category-consistent items and false alarm rates for new category-consistent lures. Absent the category learning, no such effects are evident, even when participants are exposed to the same learning trials as those who learned the categories. The results from the fourth experiment suggest that, at least for false alarm rates, the effects of category learning are not solely attributable to frequency of occurrence of category-consistent items during learning. Implications for recognition memory as well as advantages of the proposed paradigm are discussed.

The style of a stranger: Identification expertise generalizes to coarser level categories

Experience identifying visual objects and categories improves generalization within the same class (e.g., discriminating bird species improves transfer to new bird species), but does such perceptual expertise transfer to coarser category judgments? We tested whether fingerprint experts, who spend their days comparing pairs of prints and judging whether they were left by the same finger or two different fingers, can generalize their finger discrimination expertise to people more broadly. That is, can these experts identify prints from Jones's right thumb and prints from Jones's right index finger as instances of the same "Jones" category? Novices and experts were both sensitive to the style of a stranger's prints; despite lower levels of confidence, experts were significantly more sensitive to this style than novices. This expert advantage persisted even when we reduced the number of exemplars provided. Our results demonstrate that perceptual expertise can be flexible to upwards shifts in the level of specificity, suggesting a dynamic memory retrieval process.

A Taxonomic Analysis of Abstraction

An important characteristic of knowledge is that it exists at multiple levels of abstraction. This article illustrates how different levels of abstraction influence perception, comprehension, categorization, memory, and thought. Theories exist for how abstraction influences each of these cognitive processes, but there are few unifying principles for discussing these theories within a common conceptual framework. My proposed taxonomy examines three senses of abstraction: (a) an abstract entity is a concept that has no material referent, (b) abstraction focuses on only some attributes of multicomponent stimuli, and (c) an abstract idea applies to many particular instances of a category. The first refers to instances, the second to attributes of instances, and the third to classes of instances. Concrete mental representations consist of modal images for instances, equivalent attributes, and exemplars or episodes for categories. Abstract mental representations consist of amodal propositions for instances, distinctive attributes, and rules or prototypes or schema for categories. I first apply the taxonomy to words, pictures, and problems. The next section shows how categorization strategies combine with abstraction at the attribute, instance, and category levels. The subsequent section applies the taxonomy to hierarchical (subordinate, basic, superordinate) levels. A concluding section proposes directions for further development.

Modelling individual difference in visual categorization

Recent years has seen growing interest in understanding, characterizing, and explaining individual differences in visual cognition. We focus here on individual differences in visual categorization. Categorization is the fundamental visual ability to group different objects together as the same kind of thing. Research on visual categorization and category learning has been significantly informed by computational modeling, so our review will focus both on how formal models of visual categorization have captured individual differences and how individual difference have informed the development of formal models. We first examine the potential sources of individual differences in leading models of visual categorization, providing a brief review of a range of different models. We then describe several examples of how computational models have captured individual differences in visual categorization. This review also provides a bit of an historical perspective, starting with models that predicted no individual differences, to those that captured group differences, to those that predict true individual differences, and to more recent hierarchical approaches that can simultaneously capture both group and individual differences in visual categorization. Via this selective review, we see how considerations of individual differences can lead to important theoretical insights into how people visually categorize objects in the world around them. We also consider new directions for work examining individual differences in visual categorization.

Explaining the Timing of Natural Scene Understanding with a Computational Model of Perceptual Categorization

Observers can rapidly perform a variety of visual tasks such as categorizing a scene as open, as outdoor, or as a beach. Although we know that different tasks are typically associated with systematic differences in behavioral responses, to date, little is known about the underlying mechanisms. Here, we implemented a single integrated paradigm that links perceptual processes with categorization processes. Using a large image database of natural scenes, we trained machine-learning classifiers to derive quantitative measures of task-specific perceptual discriminability based on the distance between individual images and different categorization boundaries. We showed that the resulting discriminability measure accurately predicts variations in behavioral responses across categorization tasks and stimulus sets. We further used the model to design an experiment, which challenged previous interpretations of the so-called "superordinate advantage." Overall, our study suggests that observed differences in behavioral responses across rapid categorization tasks reflect natural variations in perceptual discriminability.

The dynamics of categorization: Unraveling rapid categorization

We explore a puzzle of visual object categorization: Under normal viewing conditions, you spot something as a dog fastest, but at a glance, you spot it faster as an animal. During speeded category verification, a classic basic-level advantage is commonly observed (Rosch, Mervis, Gray, Johnson, & Boyes-Braem, 1976), with categorization as a dog faster than as an animal (superordinate) or Golden Retriever (subordinate). A different story emerges during ultra-rapid categorization with limited exposure duration (<30 ms), with superordinate categorization faster than basic or subordinate categorization (Thorpe, Fize, & Marlot, 1996). These two widely cited findings paint contrary theoretical pictures about the time course of categorization, yet no previous study has investigated them together. We systematically examined two experimental factors that could explain the qualitative difference in categorization across the two paradigms: exposure duration and category trial context. Mapping out the time course of object categorization by manipulating exposure duration and the timing of a post-stimulus mask revealed that brief exposure durations favor superordinate-level categorization, but with more time a basic-level advantage emerges. However, these advantages were modulated by category trial context. With randomized target categories, the superordinate advantage was eliminated; and with only four repetitions of superordinate categorization within an otherwise randomized context, the basic-level advantage was eliminated. Contrary to theoretical accounts that dictate a fixed priority for certain levels of abstraction in visual processing and access to semantic knowledge, the dynamics of object categorization are flexible, depending jointly on the level of abstraction, time for perceptual encoding, and category context.

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.

Category Learning Stretches Neural Representations in Visual Cortex

We review recent work that shows how learning to categorize objects changes how those objects are represented in the mind and the brain. After category learning, visual perception of objects is enhanced along perceptual dimensions that were relevant to the learned categories, an effect we call dimensional modulation (DM). DM stretches object representations along category-relevant dimensions and shrinks them along category-irrelevant dimensions. The perceptual advantage for category-relevant dimensions extends beyond categorization and can be observed during visual discrimination and other tasks that do not depend on the learned categories. fMRI shows that category learning causes ventral stream neural populations in visual cortex representing objects along a category-relevant dimension to become more distinct. These results are consistent with a view that specific aspects of cognitive tasks associated with objects can account for how our visual system responds to objects.

Studying real-world perceptual expertise

Significant insights into visual cognition have come from studying real-world perceptual expertise. Many have previously reviewed empirical findings and theoretical developments from this work. Here we instead provide a brief perspective on approaches, considerations, and challenges to studying real-world perceptual expertise. We discuss factors like choosing to use real-world versus artificial object domains of expertise, selecting a target domain of real-world perceptual expertise, recruiting experts, evaluating their level of expertise, and experimentally testing experts in the lab and online. Throughout our perspective, we highlight expert birding (also called birdwatching) as an example, as it has been used as a target domain for over two decades in the perceptual expertise literature.

An exemplar of model-based cognitive neuroscience

Are categories learned by forming abstract prototypes or by remembering specific exemplars? Mack, Preston, and Love observed that patterns of functional MRI (fMRI) brain activity were more consistent with patterns of representations predicted by exemplar models than by prototype models. Their work represents the theoretical power of emerging approaches to model-based cognitive neuroscience.