On the Generalization of Simple Alternating Category Structures.

A fundamental question in the study of human cognition is how people learn to predict the category membership of an example from its properties. Leading approaches account for a wide range of data in terms of comparison to stored examples, abstractions capturing statistical regularities, or logical rules. Across three experiments, participants learned a category structure in a low-dimension, continuous-valued space consisting of regularly alternating regions of class membership (A B A B). The dependent measure was generalization performance for novel items outside the range of the training space. Human learners often extended the alternation pattern--a finding of critical interest given that leading theories of categorization based on similarity or dimensional rules fail to predict this behavior. In addition, we provide novel theoretical interpretations of the observed phenomenon.

Object understanding: Investigating the path from percept to meaning.

Researchers tend to follow two paths when investigating categorization: 1) artificial classification learning tasks and 2) studies of natural conceptual organization involving reasoning from prior category knowledge. Largely separate, another body of research addresses the process of object recognition, i.e., how people identify what they are looking at strictly in terms of visual as opposed to semantic properties. The present work brings together elements from each of these approaches in order to address object understanding: the ubiquitous natural process of accessing meaning based on a realistic image of an everyday object. According to a widely held features-first framework, a stimulus is initially encoded as a set of features that is compared to stored category representations to find the best match. This approach has been successful for explaining artificial classification learning, but it bypasses how items are encoded and fails to include a role for top-down processing in constructing item representations. We used a speeded verification task to evaluate the features-first account using realistic stimuli. Participants saw photographic images of everyday objects and judged as quickly as possible whether a provided verbal description matched the picture. Category descriptions (basic-level labels) were verified significantly faster than descriptions of physical or functional properties. This suggests that people access the category of the stimulus prior to accessing its parsed features. We outline a construal account whereby the category is accessed first to construct a featural item interpretation rather than features being the basis for determining the category.

Preventing inert knowledge: Category status promotes spontaneous structure-based retrieval of prior knowledge.

Analogical comparison of 2 provided cases promotes spontaneous analogical transfer by encouraging a more abstract representation of a target principle. This is widely understood as a process of schema abstraction that aids retrieval from memory in the absence of superficial similarity. The category status hypothesis states that if knowledge about a target principle is represented as a relational category, it is easier to activate as a result of categorizing (as opposed to cue-based reminding). To investigate these 2 pathways to spontaneous transfer (schema retrieval vs. categorization), participants were assigned to different study conditions: (a) standard comparison of 2 analogs; (b) standard comparison followed by a second comparison of 2 new analogs; or (c) a guided Category-building task based on a foundation of sequential summarization with additional categorization supports. Experiment 1 showed improved spontaneous transfer in the Category-building task relative to Single-comparison-the gold standard for encouraging spontaneous transfer. Experiment 2 demonstrated that the Category-building task led to more frequent spontaneous structure-based retrieval than either comparison-based task. This suggests that the transfer improvements observed in Experiment 1 are attributable to the Category-building task making superficially dissimilar prior knowledge more accessible. Experiments 3a and 3b tested alternative explanations of this finding and provided evidence that the effectiveness of the Category-building task arises from encouraging the construal of a target principle as a relational category. These findings support categorization as a critical factor in explaining successful spontaneous analogical retrieval and toward instructional approaches that promote portable rather than inert knowledge. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Similarity judgments predict N400 amplitude differences between taxonomic category members and thematic associates.

Human similarity judgments do not reliably conform to the predictions of leading theories of psychological similarity. Evidence from the triad similarity judgment task shows that people often identify thematic associates like dog and bone as more similar than taxonomic category members like dog and cat, even though thematic associates lack the type of featural or relational similarity that is foundational to theories of psychological similarity. This specific failure to predict human behavior has been addressed as a consequence of education and other individual differences, an artifact of the triad similarity judgment paradigm, or a shortcoming in psychological accounts of similarity. We investigated the judged similarity of semantically-related concepts (taxonomic category members and thematic associates) as it relates to other task-independent measures of semantic knowledge and access. Participants were assessed on reading and language ability, then event-related potentials (ERPs) were collected during a passive, sequential word reading task that presented pseudowords and taxonomically-related, thematically-related, and unrelated word sequences, and, finally, similarity judgments were collected with the classic two-alternative forced-choice triad task. The results uncovered a correspondence between ERP amplitude and triad-based similarity judgments-similarity judgment behavior reliably predicts ERP amplitude during passive word reading, absent of any instruction to consider similarity. It was also found that individual differences in reading and language ability independently predicted ERP amplitude. This evidence suggests that similarity judgments are driven by reliable patterns of thought that are not solely rooted in the interpretation of task goals or reading and language ability.

Comparison-based learning of relational categories (you'll never guess).

In accord with structural alignment theory, same-category comparison opportunities within a classification learning task should promote relational category acquisition. However, a straightforward merging of the classification paradigm with copresentation of same-category item pairs does not yield an advantage relative to an equal number of single-item exposures. In 3 experiments, we explore the hypothesis that the traditional classification learning mode (guess-and-correct) and comparison have a previously unforeseen incompatibility. In Experiment 1, we test this hypothesis by contrasting classification with supervised observational learning (passive study of labeled examples) under 3 presentation formats: same-category pairs, mixed pairs, and single-item. We find an observational advantage with same-category pairs and produce the elusive advantage over single-item exposures. In Experiment 2, we assess the generality of the learning mode effect by testing both same- and different-category comparison. The observational advantage replicates and extends to different-category comparison-although, we do not find a significant difference between the 2 types of comparison. In Experiment 3, relative to the classification mode, we find enhanced performance in an intermediate learning mode between classification and observation in which participants are instructed to make a covert category guess (without making an actual response) before seeing the correct category label. Implications and interpretations-including our interpretation that the performance emphasis inherent in classification learning undermines the benefits that arise from comparison opportunities-are discussed. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Revisiting the linear separability constraint: New implications for theories of human category learning.

While the ability to acquire non-linearly separable (NLS) classifications is well documented in the study of human category learning, the relative ease of learning compared to a linear separable structure is difficult to evaluate without potential confounds. Medin and Schwanenflugel (Journal of Experimental Psychology: Human Learning and Memory, 7, 355-368, 1981) were the first to demonstrate that NLS classifications are not more difficult to acquire than linearly separable ones when structures are equated in terms of within- and between-category similarities. However, their evidence is less sturdy than might be expected due to non-standard methodology and low sample size. We conducted a conceptual replication to clarify the behavioral picture and perform qualitative testing of formal models. The behavioral results not only showed a lack of advantage for the linearly separable (LS) structure, but revealed a stronger finding: the NLS structure was reliably easier to acquire. Differences in the relative ease of NLS learners to master certain items yielded evidence for the existence of distinct learner subgroups, one marked by significantly easier (not harder) learning of exception items. Comparing the qualitative fits of leading computational models to the human learning performance confirmed that a pure prototype account, even with contemporary updates, remains incompatible with the data. However, exemplar models and similarity-based models grounded in sophisticated forms of abstraction-based learning successfully account for the NLS advantage. In sum, evidence against a linear separability constraint is redoubled, and the observed NLS advantage along with behavioral patterns seen at the subgroup and item level provide a valuable basis for comprehensive evaluation of competing theoretical accounts and models.

Sorting out the problem of inert knowledge: Category construction to promote spontaneous transfer.

A fundamental goal in the study of human cognition is to understand the transfer of knowledge. This goes hand-in-hand with the translational goal of promoting such transfer via instructional techniques. Despite a rich history of research using the analogical problem-solving paradigm, no study activity has been found to produce a robust rate of successful spontaneous transfer-even when the test is immediate. We propose the category status hypothesis as an explanation of the difficulty of transfer and as motivation for a novel approach to promoting transfer. We report a set of experiments evaluating a category construction technique based on a sorting task. In Experiment 1a, we found category construction to be significantly more effective than the "gold standard" of schema abstraction through comparison of 2 analogous cases. In Experiment 1b, we explored a variation of the category construction technique that did not reliably differ in effectiveness from comparison-based schema abstraction-we also verified that both study tasks were superior to a baseline task of separate summarization of 2 cases. In Experiment 2, we conducted a replication of the initial design with higher power and confirmed the significant advantage for category construction over schema abstraction via comparison. In Experiment 3, we compared category construction to an information-consistent reading comprehension control to evaluate competing interpretations of the category construction advantage found in Experiments 1 and 2. We discuss theoretical and applied implications of these findings. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Evidence of analogical re-representation from a change detection task.

The flexibility to map similar, but non-identical relations, is a key characteristic of human analogical reasoning. Understanding how this flexibility is implemented is necessary for a complete accounting of analogical processes. The structure mapping theory of analogy addresses this issue by invoking re-representation-an online transformation of conceptually similar relational content that reveals potential partial identity matches between predicates. Despite the critical importance of re-representation to structure mapping, very little empirical work has validated the psychological reality of this mechanism, with the existing evidence being no more than suggestive. The present work investigates the likelihood of re-representation across two experiments using a novel change detection task. The resultant findings demonstrate precise evidence of representational change in relational content of analogs. Experiment 2 further explores the relationship between lower-order relational similarity and the likelihood of re-representation.

Similarity is as similarity does? A critical inquiry into the effect of thematic association on similarity.

Leading theories of psychological similarity are based on the degree of match in semantic content between compared cases (i.e., shared features, low dimensional distance, alignable relations). Broader forms of semantic relatedness such as the degree of association between cases (e.g., egg and spatula) are generally not considered to contribute to similarity judgments. However, empirical work has demonstrated a behavioral tendency to choose associated pairs over proximal pairs (i.e., high semantic content overlap) in similarity judgement tasks. As a result, models have been proposed that combine thematic integration and content match as component processes of similarity. The present experiments investigate the thematic association effect in similarity in order to more clearly determine whether such a theoretical redirection is warranted. An alternative viewpoint is that confusion between similarity and association is the cause of the reported thematic bias. Experiment 1 introduces a modified similarity judgement task and addresses the impact of task instructions as a potential causal factor underlying the thematic association effect on similarity. Experiment 2 specifically compares the novel similarity task to a traditional two-alternative, forced choice triad task. Experiment 3 addresses the possibility of bias in the stimulus sets used in Experiments 1 and 2. Across the experiments we find association-based responding to be much less prevalent than in previous demonstrations: the traditional finding of a thematic preference only occurred when participants were specifically asked to select based on associativity ("goes with"). Modifications to conventional methodology that minimize biasing factors clearly attenuate the effect of association on similarity. We interpret these findings as evidence that that the thematic association effect derives from intrusions on psychological similarity, not from an additional component intrinsic to psychological similarity.

Analogies Without Commonalities? Evidence of Re-representation via Relational Category Activation.

Analogies between cases with matching sets of connected relational structure is well-explained by existing theory. Re-representation is posited as an important mechanism to increase the flexibility of analogical processing by allowing the alignment of non-identical predicates across compared cases. It has been proposed that certain kind of categories can be characterized in terms of the relational structure that its exemplars tend to satisfy. Such relational categories have the property that all members of the category are analogous to one another. We ask whether a process of re-representation can alter the construal of a case and bring two evidently non-analogous cases into analogical alignment if they are both seen as members of the same relational category. We examine analogies between pairs of cases where the base is a canonical example of a relational category and the target would not be considered a member of the category on its own - critically, the cases themselves share no evident relational identities or similarities. In Experiment 1, we ask whether presenting a target case as part of an analogical pairing alters its construal. In Experiment 2, the pairs are presented for judgment as potential analogies. In both studies, participants interpret the target cases differently (consistent with the relational category) as a result of processing the analogy. There are two main implications: (1) a form of re-representation is at work in which the activation of a relational category triggers an alternate construal of the target case; and (2) this suggests a path to analogical status for cases that lack relational identities or similarities if the cases can both be fit to the same relational category.

Learning relational concepts from within- versus between-category comparisons.

This article examines relational category learning in light of 2 influential theories of concept acquisition: the structure-mapping theory of analogy and theories of feature-based category learning. According to current theories of analogy, comparing 2 instances of a relational concept enables alignment of their elements and reveals their shared relational structure. Therefore, learning relationally defined categories should be faster when comparing items of the same category than when comparing items of different categories. By contrast, feature-based theories predict a benefit of between-category comparisons, because such comparisons direct attention to the features that discriminate the categories. The present experiments tested these predictions using a 2-category classification-learning task in which 2 items are presented on every trial: either in the same category (match condition) or in different categories (contrast condition). Subjects in the contrast condition outperformed those in the match condition for feature-based categories (Experiment 1) and across 4 different types of relational categories (Experiments 1-4). Although theorists have posited that structure-mapping theory is directly applicable to relational category learning, the present findings pose a distinct challenge to this assertion. We propose that many relational categories are learnable based solely on which relations are present in the stimulus rather than requiring explicitly compositional representations based on role-filler binding. This process would be akin to feature processing and would not require structural alignment. This theoretical proposal, together with the empirical results, may lead to a better understanding of when people do and do not engage in the cognitively demanding process of structural alignment. (PsycINFO Database Record (c) 2018 APA, all rights reserved).

Solving Nonlinearly Separable Classifications in a Single-Layer Neural Network.

Since the work of Minsky and Papert ( 1969 ), it has been understood that single-layer neural networks cannot solve nonlinearly separable classifications (i.e., XOR). We describe and test a novel divergent autoassociative architecture capable of solving nonlinearly separable classifications with a single layer of weights. The proposed network consists of class-specific linear autoassociators. The power of the model comes from treating classification problems as within-class feature prediction rather than directly optimizing a discriminant function. We show unprecedented learning capabilities for a simple, single-layer network (i.e., solving XOR) and demonstrate that the famous limitation in acquiring nonlinearly separable problems is not just about the need for a hidden layer; it is about the choice between directly predicting classes or learning to classify indirectly by predicting features.

Similar to the category, but not the exemplars: A study of generalization.

Reference point approaches have dominated the study of categorization for decades by explaining classification learning in terms of similarity to stored exemplars or averages of exemplars. The most successful reference point models are firmly grounded in the associative learning tradition-treating categorization as a stimulus generalization process based on inverse exponential distance in psychological space augmented by a dimensional selective attention mechanism. We present experiments that pose a significant challenge to popular reference point accounts which explain categorization in terms of stimulus generalization from exemplars, prototypes, or adaptive clusters. DIVA, a similarity-based alternative to the reference point framework, provides a successful account of the human data. These findings suggest that a successful psychology of categorization may need to look beyond stimulus generalization and toward a view of category learning as the induction of a richer model of the data.

Observation versus classification in supervised category learning.

The traditional supervised classification paradigm encourages learners to acquire only the knowledge needed to predict category membership (a discriminative approach). An alternative that aligns with important aspects of real-world concept formation is learning with a broader focus to acquire knowledge of the internal structure of each category (a generative approach). Our work addresses the impact of a particular component of the traditional classification task: the guess-and-correct cycle. We compare classification learning to a supervised observational learning task in which learners are shown labeled examples but make no classification response. The goals of this work sit at two levels: (1) testing for differences in the nature of the category representations that arise from two basic learning modes; and (2) evaluating the generative/discriminative continuum as a theoretical tool for understand learning modes and their outcomes. Specifically, we view the guess-and-correct cycle as consistent with a more discriminative approach and therefore expected it to lead to narrower category knowledge. Across two experiments, the observational mode led to greater sensitivity to distributional properties of features and correlations between features. We conclude that a relatively subtle procedural difference in supervised category learning substantially impacts what learners come to know about the categories. The results demonstrate the value of the generative/discriminative continuum as a tool for advancing the psychology of category learning and also provide a valuable constraint for formal models and associated theories.

Detecting anomalous features in complex stimuli: the role of structured comparison.

The ability to detect anomalies in perceived stimuli is critical to a broad range of cognitive tasks, yet acquiring this ability often requires lengthy practice. In this research, we asked whether findings from research on analogical comparison can be used to aid in the acquisition of perceptual expertise. Building on findings that comparison can facilitate the detection of differences, the present research addressed two questions: (1) Does having an alignable comparison standard improve performance on a difficult detection task? (2) Can such comparison experience improve subsequent detection performance on single anomalous targets? Across 3 experiments, university undergraduates were asked to find an anomalous bone in drawings of animal skeletons. Target items including an anomaly were presented either alone or with a correct standard. Furthermore, to evaluate the impact of ease of alignment, the correct standard was presented either mirror-reversed (low alignable) or regular (high alignable). Results showed increased accuracy when a comparison standard was present and further gains when the standard was more easily alignable. In Experiment 3, we used a between-subjects design to reveal that advance comparison (as opposed to single-item training) led to improved detection of anomalies in subsequent novel examples presented as isolated targets. We conclude that the availability of a standard and ease of alignment promote encoding and processing. Furthermore, comparison-based learning confers an ongoing advantage even without standards for comparison. Therefore, task performance in application areas requiring detection of nonobvious anomalies can be improved by providing alignable standards next to targets or in advance training.

Comparison promotes learning and transfer of relational categories.

We investigated the effect of co-presenting training items during supervised classification learning of novel relational categories. Strong evidence exists that comparison induces a structural alignment process that renders common relational structure more salient. We hypothesized that comparisons between exemplars would facilitate learning and transfer of categories that cohere around a common relational property. The effect of comparison was investigated using learning trials that elicited a separate classification response for each item in presentation pairs that could be drawn from the same or different categories. This methodology ensures consideration of both items and invites comparison through an implicit same-different judgment inherent in making the two responses. In a test phase measuring learning and transfer, the comparison group significantly outperformed a control group receiving an equivalent training session of single-item classification learning. Comparison-based learners also outperformed the control group on a test of far transfer, that is, the ability to accurately classify items from a novel domain that was relationally alike, but surface-dissimilar, to the training materials. Theoretical and applied implications of this comparison advantage are discussed.

Human learning of elemental category structures: revising the classic result of Shepard, Hovland, and Jenkins (1961).

The findings of Shepard, Hovland, and Jenkins (1961) on the relative ease of learning 6 elemental types of 2-way classifications have been deeply influential 2 times over: 1st, as a rebuke to pure stimulus generalization accounts, and again as the leading benchmark for evaluating formal models of human category learning. The litmus test for models is the ability to simulate an observed advantage in learning a category structure based on an exclusive-or (XOR) rule over 2 relevant dimensions (Type II) relative to category structures that have no perfectly predictive cue or cue combination (including the linearly-separable Type IV). However, a review of the literature reveals that a Type II advantage over Type IV is found only under highly specific experimental conditions. We investigate when and why a Type II advantage exists to determine the appropriate benchmark for models and the psychological theories they represent. A series of 8 experiments link particular conditions of learning to outcomes ranging from a traditional Type II advantage to compelling non-differences and reversals (i.e., Type IV advantage). Common interpretations of the Type II advantage as either a broad-based phenomenon of human learning or as strong evidence for an attention-mediated similarity-based account are called into question by our findings. Finally, a role for verbalization in the category learning process is supported.

Category learning in the context of co-presented items.

A series of four studies explore how the presentation of multiple items on each trial of a categorization task affects the course of category learning. In a three-category supervised classification task involving multi-dimensionally varying artificial organism-like stimuli, learners are shown a target plus two context items on every trial, with the context items' category membership explicitly identified. These triads vary in whether one, two, or all three categories are represented. This presentation context can support within-category comparison and/or between-category contrast. The most successful learning occurs when all categories are represented in each trial. This pattern occurs across two different underlying category structures and across variations in learners' prior knowledge of the relationship between the target and context items. These results appear to contrast with some other recent findings and make clear the potential importance of context-based inter-item evaluation in human category learning, which has implications for psychological theory and for real-world learning environments.

Relations, objects, and the composition of analogies.

This research addresses the kinds of matching elements that determine analogical relatedness and literal similarity. Despite theoretical agreement on the importance of relational match, the empirical evidence is neither systematic nor definitive. In 3 studies, participants performed online evaluations of relatedness of sentence pairs that varied in either the object or relational match. Results show a consistent focus on relational matches as the main determinant of analogical acceptance. In addition, analogy does not require strict overall identity of relational concepts. Semantically overlapping but nonsynonymous relations were commonly accepted, but required more processing time. Finally, performance in a similarity rating task partly paralleled analogical acceptance; however, relatively more weight was given to object matches. Implications for psychological theories of analogy and similarity are addressed.