Training visual pattern recognition in ophthalmology using a perceptual and adaptive learning module.

OBJECTIVE: To evaluate the efficacy of a perceptual and adaptive learning module (PALM) for teaching the identification of 5 optic nerve findings. METHODS: Second- through fourth-year medical students were randomized to the PALM or a video didactic lecture. The PALM presented the learner with short classification tasks consisting of optic nerve images. Learner accuracy and response time guided the sequencing of successive tasks until mastery was achieved. The lecture was a narrated video designed to simulate a portion of a traditional medical school lecture. Accuracy and fluency on a pretest, post-test, and 1-month delayed test were compared within and between groups. RESULTS: Eighty-three students participated. Accuracy and fluency improved significantly (p < 0.001) from pretest to post-test for both the PALM (accuracy, Cohen's d = 2.94; fluency, d = 3.39) and the lecture (accuracy, d = 2.32; fluency, d = 1.06). For the delayed test, PALM performance remained significantly greater (p < 0.001) than the pretest in both accuracy (d = 0.89) and fluency (d = 1.16), whereas lecture performance remained greater in accuracy only (d = 0.44; p = 0.02). CONCLUSIONS: The PALM facilitated visual pattern recognition for optic nerve diseases among novice learners using a single brief self-guided session. The PALM may be applied alongside traditional didactic lectures to expedite visual pattern recognition in ophthalmology.

For deep networks, the whole equals the sum of the parts.

Deep convolutional networks exceed humans in sensitivity to local image properties, but unlike biological vision systems, do not discover and encode abstract relations that capture important properties of objects and events in the world. Coupling network architectures with additional machinery for encoding abstract relations will make deep networks better models of human abilities and more versatile and capable artificial devices.

Independent mechanisms for processing local contour features and global shape.

The visual system can extract the global shape of an object from highly variable local contour features. We propose that there are separate systems for processing local and global shape. These systems are independent and process information differently. Global shape encoding accurately represents the form of low-frequency contour variations, whereas the local system encodes only summary statistics that describe typical features of high-frequency elements. In Experiments 1-4, we tested this hypothesis by obtaining same/different judgments for shapes that differed in local features, global features, or both. We found low sensitivity to changed local features that shared the same summary statistics, and no advantage in sensitivity for shapes that differed in both local and global features compared to shapes that differed only in global features. This sensitivity difference persisted when physical contour differences were equated and when shape feature sizes and exposure durations were increased. In Experiment 5, we compared sensitivity to sets of local contour features with matched or unmatched statistical properties. Sensitivity was higher for unmatched statistical properties than for properties sampled from the same statistical distribution. Experiment 6 directly tested our hypothesis of independent local and global systems using visual search. Search based on either local or global shape differences produced pop-out effects, but search for a target based on a conjunction of local and global differences required focal attention. These findings support the notion that separate mechanisms process local and global contour information and that the kinds of information these mechanisms encode are fundamentally different. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Multiple expressions of "expert" abnormality gist in novices following perceptual learning.

With a brief half-second presentation, a medical expert can determine at above chance levels whether a medical scan she sees is abnormal based on a first impression arising from an initial global image process, termed "gist." The nature of gist processing is debated but this debate stems from results in medical experts who have years of perceptual experience. The aim of the present study was to determine if gist processing for medical images occurs in naïve (non-medically trained) participants who received a brief perceptual training and to tease apart the nature of that gist signal. We trained 20 naïve participants on a brief perceptual-adaptive training of histology images. After training, naïve observers were able to obtain abnormality detection and abnormality categorization above chance, from a brief 500 ms masked presentation of a histology image, hence showing "gist." The global signal demonstrated in perceptually trained naïve participants demonstrated multiple dissociable components, with some of these components relating to how rapidly naïve participants learned a normal template during perceptual learning. We suggest that multiple gist signals are present when experts view medical images derived from the tens of thousands of images that they are exposed to throughout their training and careers. We also suggest that a directed learning of a normal template may produce better abnormality detection and identification in radiologists and pathologists.

Connecting Adaptive Perceptual Learning and Signal Detection Theory in Skin Cancer Screening.

Combining perceptual learning techniques with adaptive learning algorithms has been shown to accelerate the development of expertise in medical and STEM learning domains (Kellman & Massey, 2013; Kellman, Jacoby, Massey & Krasne, 2022). Virtually all adaptive learning systems have relied on simple accuracy data that does not take into account response bias, a problem that may be especially consequential in multi-category perceptual classifications. We investigated whether adaptive perceptual learning in skin cancer screening can be enhanced by incorporating signal detection theory (SDT) methods that separate sensitivity from criterion. SDT-style concepts were used to alter sequencing, and separately to define mastery (category retirement). SDT retirement used a running d' estimate calculated from a recent window of trials based on hit and false alarm rates. Undergraduate participants used a Skin Cancer PALM (perceptual adaptive learning module) to learn classification of 10 cancerous and readily-confused non-cancerous skin lesion types. Four adaptive conditions varied either the type of adaptive sequencing (standard vs. SDT) or retirement criteria (standard vs. SDT). A non-adaptive control condition presented didactic instruction on dermatologic screening in video form, including images, classification schemes, and detailed explanations. All adaptive conditions robustly outperformed the non-adaptive control in both learning efficiency and fluency (large effect sizes). Between adaptive conditions, SDT retirement criteria produced greater learning efficiency than standard, accuracy-based mastery criteria at both immediate and delayed posttests (medium effect sizes). SDT sequencing and standard adaptive sequencing did not differ. SDT enhancements to adaptive perceptual learning procedures have potential to enhance learning efficiency.

Configural relations in humans and deep convolutional neural networks.

Deep convolutional neural networks (DCNNs) have attracted considerable interest as useful devices and as possible windows into understanding perception and cognition in biological systems. In earlier work, we showed that DCNNs differ dramatically from human perceivers in that they have no sensitivity to global object shape. Here, we investigated whether those findings are symptomatic of broader limitations of DCNNs regarding the use of relations. We tested learning and generalization of DCNNs (AlexNet and ResNet-50) for several relations involving objects. One involved classifying two shapes in an otherwise empty field as same or different. Another involved enclosure. Every display contained a closed figure among contour noise fragments and one dot; correct responding depended on whether the dot was inside or outside the figure. The third relation we tested involved a classification that depended on which of two polygons had more sides. One polygon always contained a dot, and correct classification of each display depended on whether the polygon with the dot had a greater number of sides. We used DCNNs that had been trained on the ImageNet database, and we used both restricted and unrestricted transfer learning (connection weights at all layers could change with training). For the same-different experiment, there was little restricted transfer learning (82.2%). Generalization tests showed near chance performance for new shapes. Results for enclosure were at chance for restricted transfer learning and somewhat better for unrestricted (74%). Generalization with two new kinds of shapes showed reduced but above-chance performance (≈66%). Follow-up studies indicated that the networks did not access the enclosure relation in their responses. For the relation of more or fewer sides of polygons, DCNNs showed successful learning with polygons having 3-5 sides under unrestricted transfer learning, but showed chance performance in generalization tests with polygons having 6-10 sides. Experiments with human observers showed learning from relatively few examples of all of the relations tested and complete generalization of relational learning to new stimuli. These results using several different relations suggest that DCNNs have crucial limitations that derive from their lack of computations involving abstraction and relational processing of the sort that are fundamental in human perception.

Comparisons in Adaptive Perceptual Category Learning.

Recent work suggests that learning perceptual classifications can be enhanced by combining single item classifications with adaptive comparisons triggered by each learner's confusions. Here, we asked whether learning might work equally well using all comparison trials. In a face identification paradigm, we tested single item classifications, paired comparisons, and dual instance classifications that resembled comparisons but required two identification responses. In initial results, the comparisons condition showed evidence of greater efficiency (learning gain divided by trials or time invested). We suspected that this effect may have been driven by easier attainment of mastery criteria in the comparisons condition, and a negatively accelerated learning curve. To test this idea, we fit learning curves and found data consistent with the same underlying learning rate in all conditions. These results suggest that paired comparison trials may be as effective in driving learning of multiple perceptual classifications as more demanding single item classifications.

Novel Education Modules Addressing the Underrepresentation of Skin of Color in Dermatology Training.

BACKGROUND: Representative images of pathology in patients with skin of color are lacking in most medical education resources. This particularly affects training in dermatology, which relies heavily on the use of images to teach pattern recognition. The presentation of skin pathology can vary greatly among different skin tones, and this lack of representation of dark skin phototypes challenges providers' abilities to provide quality care to patients of color.In Botswana and other countries in sub-Saharan Africa, this challenge is further compounded by limited resources and access to dermatologists. There is a need for improved and accessible educational resources to train medical students and local medical providers in basic skin lesion description and diagnosis. OBJECTIVES: We examined whether online Perceptual and Adaptive Learning Modules (PALMs) composed of representative dark skin images could efficiently train University of Botswana medical students to more accurately describe and diagnose common skin conditions in their community. METHODS: Year 4 and 5 medical students voluntarily completed PALMs that teach skin morphology, configuration, and distribution terminology and diagnosis of the most common dermatologic conditions in their community. Pre-tests, post-tests and delayed-tests assessed knowledge acquisition and retention. RESULTS: PALMs training produced statistically significant (P < .0001) improvements in accuracy and fluency with large effect sizes (1.5, 3.7) and good retention after a 12.5-21-week median delay. Limitations were a self-selected group of students, a single institution, slow internet connections, and high drop-out rates. CONCLUSIONS: Overall, population-specific PALMs are a useful tool for efficient development of pattern recognition in skin disease description and diagnosis.

Constant curvature modeling of abstract shape representation.

How abstract shape is perceived and represented poses crucial unsolved problems in human perception and cognition. Recent findings suggest that the visual system may encode contours as sets of connected constant curvature segments. Here we describe a model for how the visual system might recode a set of boundary points into a constant curvature representation. The model includes two free parameters that relate to the degree to which the visual system encodes shapes with high fidelity vs. the importance of simplicity in shape representations. We conducted two experiments to estimate these parameters empirically. Experiment 1 tested the limits of observers' ability to discriminate a contour made up of two constant curvature segments from one made up of a single constant curvature segment. Experiment 2 tested observers' ability to discriminate contours generated from cubic splines (which, mathematically, have no constant curvature segments) from constant curvature approximations of the contours, generated at various levels of precision. Results indicated a clear transition point at which discrimination becomes possible. The results were used to fix the two parameters in our model. In Experiment 3, we tested whether outputs from our parameterized model were predictive of perceptual performance in a shape recognition task. We generated shape pairs that had matched physical similarity but differed in representational similarity (i.e., the number of segments needed to describe the shapes) as assessed by our model. We found that pairs of shapes that were more representationally dissimilar were also easier to discriminate in a forced choice, same/different task. The results of these studies provide evidence for constant curvature shape representation in human visual perception and provide a testable model for how abstract shape descriptions might be encoded.

Mastering Electrocardiogram Interpretation Skills Through a Perceptual and Adaptive Learning Module.

Although accurate interpretation of the standard 12-lead electrocardiogram (ECG) is fundamental to diagnosing heart disease, several prior studies report low accuracy rates among medical students, residents, and practicing physicians. The objective of this study was to determine if an online ECG Perceptual and Adaptive Learning Module (ECG PALM) is an efficient instrument to teach ECG interpretation. The ECG PALM consists of 415 unique ECG tracings with associated pretest, posttest, and delayed tests, each using 30 additional ECGs to gauge the effectiveness and durability of training. Between 2013 and 2015, a total of 113 third-year and 156 fourth-year medical students and 34 first-year, 41 second-year, and 37 third-year emergency medicine residents completed the PALM and associated tests. We measured two mastery criteria: accuracy, the percentage of correct interpretations, and fluency, the percentage of images interpreted accurately within 15 seconds. The ECG PALM produced statistically significant improvements (0.0001 < p < 0.0045) in student and resident performance for both accuracy (effect size = 0.9 to 3.2) and fluency (effect size = 2.5 to 3.1) following training ranging from 46 ± 24 minutes (R3s) to 88 ± 32 minutes (third-year medical students). Medical students and residents performed significantly better on a test the year following training (delayed test) than those without prior ECG PALM training (pretest). The fluency of R3 residents in classifying the 15 diagnostic categories was less than 60% for nine of the 15 diagnoses and greater than 80% for only one. Following PALM training, fluency was higher than 80% for seven of the 15 categories and less than 60% for only two categories. Accuracy in recognizing ST-elevation myocardial infarctions (STEMIs) was high both before and after PALM training for R3s, but fluency was only 64% for anterior STEMIs on the pretest, increasing to 93% following PALM training. These observations suggest that the ECG PALM is an effective and durable supplemental tool for developing mastery in interpreting common ECG abnormalities.

Evaluating the Use of Supplemental Training Technologies in Dermatology Education.

Physician assistants (PAs) are licensed to evaluate, diagnose, and treat dermatologic skin conditions. Data show that medical students have less than optimal dermatology diagnostic abilities. Although no known data exists for PA students, similar medical school and PA school training methods highlight a need for improved dermatology education in medical and PA programs. This project explored the use of perceptual and adaptive learning modules (PALMs) that target pattern recognition skills with PA students to hopefully improve PA knowledge of dermatology concepts.

Constant curvature segments as building blocks of 2D shape representation.

How the visual system represents shape, and how shape representations might be computed by neural mechanisms, are fundamental and unanswered questions. Here, we investigated the hypothesis that 2-dimensional (2D) contour shapes are encoded structurally, as sets of connected constant curvature segments. We report 3 experiments investigating constant curvature segments as fundamental units of contour shape representations in human perception. Our results showed better performance in a path detection paradigm for constant curvature targets, as compared with locally matched targets that lacked this global regularity (Experiment 1), and that participants can learn to segment contours into constant curvature parts with different curvature values, but not into similarly different parts with linearly increasing curvatures (Experiment 2). We propose a neurally plausible model of contour shape representation based on constant curvature, built from oriented units known to exist in early cortical areas, and we confirmed the model's prediction that changes to the angular extent of a segment will be easier to detect than changes to relative curvature (Experiment 3). Together, these findings suggest the human visual system is specially adapted to detect and encode regions of constant curvature and support the notion that constant curvature segments are the building blocks from which abstract contour shape representations are composed. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Local features and global shape information in object classification by deep convolutional neural networks.

Deep convolutional neural networks (DCNNs) show impressive similarities to the human visual system. Recent research, however, suggests that DCNNs have limitations in recognizing objects by their shape. We tested the hypothesis that DCNNs are sensitive to an object's local contour features but have no access to global shape information that predominates human object recognition. We employed transfer learning to assess local and global shape processing in trained networks. In Experiment 1, we used restricted and unrestricted transfer learning to retrain AlexNet, VGG-19, and ResNet-50 to classify circles and squares. We then probed these networks with stimuli with conflicting global shape and local contour information. We presented networks with overall square shapes comprised of curved elements and circles comprised of corner elements. Networks classified the test stimuli by local contour features rather than global shapes. In Experiment 2, we changed the training data to include circles and squares comprised of different elements so that the local contour features of the object were uninformative. This considerably increased the network's tendency to produce global shape responses, but deeper analyses in Experiment 3 revealed the network still showed no sensitivity to the spatial configuration of local elements. These findings demonstrate that DCNNs' performance is an inversion of human performance with respect to global and local shape processing. Whereas abstract relations of elements predominate in human perception of shape, DCNNs appear to extract only local contour fragments, with no representation of how they spatially relate to each other to form global shapes.

Comparing Adaptive and Random Spacing Schedules during Learning to Mastery Criteria.

Adaptive generation of spacing intervals in learning using response times improves learning relative to both adaptive systems that do not use response times and fixed spacing schemes (Mettler, Massey & Kellman, 2016). Studies have often used limited presentations (e.g., 4) of each learning item. Does adaptive practice benefit learning if items are presented until attainment of objective mastery criteria? Does it matter if mastered items drop out of the active learning set? We compared adaptive and non-adaptive spacing under conditions of mastery and dropout. Experiment 1 compared random presentation order with no dropout to adaptive spacing and mastery using the ARTS (Adaptive Response-time-based Sequencing) system. Adaptive spacing produced better retention than random presentation. Experiment 2 showed clear learning advantages for adaptive spacing compared to random schedules that also included dropout. Adaptive spacing performs better than random schedules of practice, including when learning proceeds to mastery and items drop out when mastered.

Adaptive vs. Fixed Spacing of Learning Items: Evidence from Studies of Learning and Transfer in Chemistry Education.

Spacing presentations of learning items across time improves memory relative to massed schedules of practice - the well-known spacing effect. Spaced practice can be further enhanced by adaptively scheduling the presentation of learning items to deliver customized spacing intervals for individual items and learners. ARTS - Adaptive Response-time-based Sequencing (Mettler, Massey, & Kellman 2016) determines spacing dynamically in relation to each learner's ongoing speed and accuracy in interactive learning trials. We demonstrate the effectiveness of ARTS when applied to chemistry nomenclature in community college chemistry courses by comparing adaptive schedules to fixed schedules consisting of continuously expanding spacing intervals. Adaptive spacing enhanced the efficiency and durability of learning, with learning gains persisting after a two-week delay and generalizing to a standardized assessment of chemistry knowledge after 2-3 months. Two additional experiments confirmed and extended these results in both laboratory and community college settings.

The Synergy of Passive and Active Learning Modes in Adaptive Perceptual Learning.

Adaptive learning systems that generate spacing intervals based on learner performance enhance learning efficiency and retention (Mettler, Massey & Kellman, 2016). Recent research in factual learning suggests that initial blocks of passive trials, where learners observe correct answers without overtly responding, produce greater learning than passive or active trials alone (Mettler, Massey, Burke, Garrigan & Kellman, 2018). Here we tested whether this passive + active advantage generalizes beyond factual learning to perceptual learning. Participants studied and classified images of butterfly genera using either: 1) Passive Only presentations, 2) Passive Initial Blocks followed by active, adaptive scheduling, 3) Passive Initial Category Exemplar followed by active, adaptive scheduling, or 4) Active Only learning. We found an advantage for combinations of active and passive presentations over Passive Only or Active Only presentations. Passive trials presented in initial blocks showed the best performance, paralleling earlier findings in factual learning. Combining active and passive learning produces greater learning gains than either alone, and these effects occur for diverse forms of learning, including perceptual learning.

Deep convolutional networks do not classify based on global object shape.

Deep convolutional networks (DCNNs) are achieving previously unseen performance in object classification, raising questions about whether DCNNs operate similarly to human vision. In biological vision, shape is arguably the most important cue for recognition. We tested the role of shape information in DCNNs trained to recognize objects. In Experiment 1, we presented a trained DCNN with object silhouettes that preserved overall shape but were filled with surface texture taken from other objects. Shape cues appeared to play some role in the classification of artifacts, but little or none for animals. In Experiments 2-4, DCNNs showed no ability to classify glass figurines or outlines but correctly classified some silhouettes. Aspects of these results led us to hypothesize that DCNNs do not distinguish object's bounding contours from other edges, and that DCNNs access some local shape features, but not global shape. In Experiment 5, we tested this hypothesis with displays that preserved local features but disrupted global shape, and vice versa. With disrupted global shape, which reduced human accuracy to 28%, DCNNs gave the same classification labels as with ordinary shapes. Conversely, local contour changes eliminated accurate DCNN classification but caused no difficulty for human observers. These results provide evidence that DCNNs have access to some local shape information in the form of local edge relations, but they have no access to global object shapes.

Accelerating expertise: Perceptual and adaptive learning technology in medical learning.

RATIONALE: Recent advances in the learning sciences offer remarkable potential for improving medical learning and performance. Difficult to teach pattern recognition skills can be systematically accelerated using techniques of perceptual learning (PL). The effectiveness of PL interventions is amplified when they are combined with adaptive learning (AL) technology in perceptual-adaptive learning modules (PALMs). INNOVATION: Specifically, PALMs incorporate the Adaptive Response Time-based Sequencing (ARTS) system, which leverages learner performance (accuracy and speed) in interactive learning episodes to guide the course of factual, perceptual, or procedural learning, optimize spacing, and lead learners to comprehensive mastery. Here we describe elements and scientific foundations of PL and its embodiment in learning technology. We also consider evidence that AL systems utilizing both accuracy and speed enhance learning efficiency and provide a unified account and potential optimization of spacing effects in learning, as well as supporting accuracy, transfer, and fluency as goals of learning. RESULTS: To illustrate this process, we review some results of earlier PALMs and present new data from a PALM designed to accelerate and improve diagnosis in electrocardiography. CONCLUSIONS: Through relatively short training interventions, PALMs produce large and durable improvements in trainees' abilities to accurately and fluently interpret clinical signs and tests, helping to bridge the gap between novice and expert clinicians.

Abstract shape representation in human visual perception.

The ability to form shape representations from visual input is crucial to perception, thought, and action. Perceived shape is abstract, as evidenced when we can see a contour specified only by discrete dots, when a cloud appears to resemble a fish, or when we match shapes across transformations of scale and orientation. Surprisingly little is known about the formation of abstract shape representations in biological vision. We report experiments that demonstrate the existence of abstract shape representations in visual perception and identify the time course of their formation. In Experiment 1, we varied stimulus exposure time in a task that required abstract shape and found that it emerges about 100 ms after stimulus onset. The results also showed that abstract shape representations are invariant across certain transformations and that they can be recovered from spatially separated dots. Experiment 2 found that encoding of basic visual features, such as dot locations, occurs during the first 30 ms after stimulus onset, indicating that shape representations require processing time beyond that needed to extract spatial features. Experiment 3 used a convergent method to confirm the timing and importance of abstract shape representations. Given sufficient time, shape representations form automatically and obligatorily, affecting performance even in a task in which neither instructions nor accurate responding involved shape. These results provide evidence for the existence, emergence, and functional importance of abstract shape representations in visual perception. We contrast these results with "deep learning" systems and with proposals that deny the importance of abstract representations in human perception and cognition. (PsycINFO Database Record

Differentiating global and local contour completion using a dot localization paradigm.

Competing theories of partially occluded object perception (amodal completion) emphasize either relatively local contour relationships or global factors such as symmetry. These disparate theories may reflect 2 separate processes: a low-level contour interpolation process and a higher-order global recognition process. The 2 could be distinguished experimentally if only the former produces precise representations of occluded object boundaries. Using a dot localization paradigm, we measured the precision and accuracy of perceived object boundaries for participants instructed to complete occluded objects with divergent local and global interpretations. On each trial, a small red dot was flashed on top of an occluder. Participants reported whether the dot fell inside or outside the occluded object's boundaries. Interleaved, 2-up, 1-down staircases estimated points on the psychometric function where the probability was .707 that the dot would be seen as either outside or inside the occluded object's boundaries. The results reveal that local contour interpolation produces precise and accurate representations of occluded contours, and consistency across observers, but completion according to global symmetry does not. These results support a distinction between local, automatic contour interpolation processes and global processes based on recognition from partial information. (PsycINFO Database Record