Average Estimates in Line Graphs Are Biased Toward Areas of Higher Variability.

We investigate variability overweighting, a previously undocumented bias in line graphs, where estimates of average value are biased toward areas of higher variability in that line. We found this effect across two preregistered experiments with 140 and 420 participants. These experiments also show that the bias is reduced when using a dot encoding of the same series. We can model the bias with the average of the data series and the average of the points drawn along the line. This bias might arise because higher variability leads to stronger weighting in the average calculation, either due to the longer line segments (even though those segments contain the same number of data values) or line segments with higher variability being otherwise more visually salient. Understanding and predicting this bias is important for visualization design guidelines, recommendation systems, and tool builders, as the bias can adversely affect estimates of averages and trends.

Difficulty limits of visual mental imagery.

While past work has focused on the representational format of mental imagery, and the similarities of its operation and neural substrate to online perception, surprisingly little has tested the boundaries of the level of detail that mental imagery can generate. To answer this question, we take inspiration from the visual short-term memory literature, a related field which has found that memory capacity is affected by the number of items, whether they are unique, and whether and how they move. We test these factors of set size, color heterogeneity, and transformation in mental imagery through both subjective (Exp 1; Exp 2) and objective (Exp 2) measures - difficulty ratings and a change detection task, respectively - to determine the capacity limits of our mental imagery, and find that limits on mental imagery are similar to those for visual short-term memory. In Experiment 1, participants rated the difficulty of imagining 1-4 colored items as subjectively more difficult when there were more items, when the items had unique colors instead of an identical color, and when they scaled or rotated instead of merely linearly translating. Experiment 2 isolated these subjective difficulty ratings of rotation for uniquely colored items, and added a rotation distance manipulation (10° to 110°), again finding higher subjective difficulty for more items, and for when those items rotated farther; the objective measure showed a decrease in performance for more items, but not for rotational degree. Congruities between the subjective and objective results suggest similar costs, but some incongruities suggest that subjective reports can be overly optimistic, likely because they are biased by an illusion of detail.

Symmetry and spatial ability enhance change detection in visuospatial structures.

Science, Technology, Engineering, and Mathematics (STEM) domains require people to recognize and transform complex visuospatial displays that appear to vastly exceed the limits of visuospatial working memory. Here, we consider possible domain-general mechanisms that may explain this advantage: capitalizing on symmetry, a structural regularity that can produce more efficient representations. Participants briefly viewed a structure made up of three-dimensional connected cubes of different colors, which was either asymmetrical or symmetrical. After a short delay, they were asked to detect a change (colors swapping positions) within a rotated second view. In change trials, the second display always had an asymmetrical structure. The presence of symmetry in the initial view improved change detection, and performance also declined with angular disparity of the encoding and test displays. People with higher spatial ability performed better on the change-detection task, but there was no evidence that they were better at leveraging symmetry than low-spatial individuals. The results suggest that leveraging symmetrical structures can help people of all ability levels exceed typical working memory limits by constructing more efficient representations and substituting resource-demanding mental rotation operations with alternative orientation-independent strategies.

Visual working memory for connected 3D objects: effects of stimulus complexity, dimensionality and connectivity.

Visual working memory (VWM) is typically measured using arrays of two-dimensional isolated stimuli with simple visual identities (e.g., color or shape), and these studies typically find strong capacity limits. Science, technology, engineering and mathematics (STEM) experts are tasked with reasoning with representations of three-dimensional (3D) connected objects, raising questions about whether those stimuli would be subject to the same limits. Here, we use a color change detection task to examine working memory capacity for 3D objects made up of differently colored cubes. Experiment 1a shows that increasing the number of parts of an object leads to less sensitivity to color changes, while change-irrelevant structural dimensionality (the number of dimensions into which parts of the structure extend) does not. Experiment 1b shows that sensitivity to color changes decreases similarly with increased complexity for multipart 3D connected objects and disconnected 2D squares, while sensitivity is slightly higher with 3D objects. Experiments 2a and 2b find that when other stimulus characteristics, such as size and visual angle, are controlled, change-irrelevant dimensionality and connectivity have no effect on performance. These results suggest that detecting color changes on 3D connected objects and on displays of isolated 2D stimuli are subject to similar set size effects and are not affected by dimensionality and connectivity when these properties are change-irrelevant, ruling out one possible explanation for scientists' advantages in storing and manipulating representations of complex 3D objects.

The Science of Visual Data Communication: What Works.

Effectively designed data visualizations allow viewers to use their powerful visual systems to understand patterns in data across science, education, health, and public policy. But ineffectively designed visualizations can cause confusion, misunderstanding, or even distrust-especially among viewers with low graphical literacy. We review research-backed guidelines for creating effective and intuitive visualizations oriented toward communicating data to students, coworkers, and the general public. We describe how the visual system can quickly extract broad statistics from a display, whereas poorly designed displays can lead to misperceptions and illusions. Extracting global statistics is fast, but comparing between subsets of values is slow. Effective graphics avoid taxing working memory, guide attention, and respect familiar conventions. Data visualizations can play a critical role in teaching and communication, provided that designers tailor those visualizations to their audience.

Truth or Square: Aspect Ratio Biases Recall of Position Encodings.

Bar charts are among the most frequently used visualizations, in part because their position encoding leads them to convey data values precisely. Yet reproductions of single bars or groups of bars within a graph can be biased. Curiously, some previous work found that this bias resulted in an overestimation of reproduced data values, while other work found an underestimation. Across three empirical studies, we offer an explanation for these conflicting findings: this discrepancy is a consequence of the differing aspect ratios of the tested bar marks. Viewers are biased to remember a bar mark as being more similar to a prototypical square, leading to an overestimation of bars with a wide aspect ratio, and an underestimation of bars with a tall aspect ratio. Experiments 1 and 2 showed that the aspect ratio of the bar marks indeed influenced the direction of this bias. Experiment 3 confirmed that this pattern of misestimation bias was present for reproductions from memory, suggesting that this bias may arise when comparing values across sequential displays or views. We describe additional visualization designs that might be prone to this bias beyond bar charts (e.g., Mekko charts and treemaps), and speculate that other visual channels might hold similar biases toward prototypical values.

Spatial alignment facilitates visual comparison.

Humans have a uniquely sophisticated ability to see past superficial features and to understand the relational structure of the world around us. This ability often requires that we compare structures, finding commonalities and differences across visual depictions that are arranged in space, such as maps, graphs, or diagrams. Although such visual comparison of relational structures is ubiquitous in classrooms, textbooks, and news media, surprisingly little is known about how to facilitate this process. Here we suggest a new principle of spatial alignment, whereby visual comparison is substantially more efficient when visuals are placed perpendicular to their structural axes, such that the matching components of the visuals are in direct alignment. In four experiments, this direct alignment led to faster and more accurate comparison than other placements of the same patterns. We discuss the spatial alignment principle in connection to broader work on relational comparison and describe its implications for design and instruction. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Foveal gravity: A robust illusion of color-location misbinding.

Some types of object features, such as color, shape, or location, can be processed separately within the visual system, requiring that they be correctly "bound" to a single object via attentional selection of a subset of visual information. Forcing selection to spread too widely can cause an illusion where these features misbind to objects, creating illusory objects that were never present. Here, we present a novel display that produces a robust color-location misbinding illusion that we call foveal gravity (viewable at https://osf.io/2bndg/). When observers selected only a set of colored objects, colors were largely perceived in their correct locations. When observers additionally selected objects in the far periphery, colors in the near periphery migrated closer to the fovea on over 35% of trials. We speculate that foveal gravity occurs because locations closer to the fovea are more likely to defeat more peripheral locations in competitive interactions to "win" the task-relevant color.

Similarity Grouping as Feature-Based Selection.

Across the natural world as well as the artificial worlds of maps, diagrams, and data visualizations, feature similarity (e.g., color and shape) links spatially separate areas into sets. Despite a century of study, it is yet unclear what mechanism underlies this gestalt similarity grouping. One recent proposal is that similarity grouping-for example, seeing a red, vertical, or square group-is just global selection of those features. Although parsimonious, this account makes the counterintuitive prediction that similarity grouping is strictly serial: A green group cannot be constructed at the same time as a red group. We tested this prediction with a novel measure-a grouping illusion within number-estimation tasks that should work only if participants simultaneously construct groups-and found the strongest evidence yet in favor of serial feature-based attention ( Ns = 14, 12, and 12 for Experiment 1, Experiment 2, and Experiment 3, respectively).

Correlation Judgment and Visualization Features: A Comparative Study.

Recent visualization research efforts have incorporated experimental techniques and perceptual models from the vision science community. Perceptual laws such as Weber's law, for example, have been used to model the perception of correlation in scatterplots. While this thread of research has progressively refined the modeling of the perception of correlation in scatterplots, it remains unclear as to why such perception can be modeled using relatively simple functions, e.g., linear and log-linear. In this paper, we investigate a longstanding hypothesis that people use visual features in a chart as a proxy for statistical measures like correlation. For a given scatterplot, we extract 49 candidate visual features and evaluate which best align with existing models and participant judgments. The results support the hypothesis that people attend to a small number of visual features when discriminating correlation in scatterplots. We discuss how this result may account for prior conflicting findings, and how visual features provide a baseline for future model-based approaches in visualization evaluation and design.

Gestalt similarity groupings are not constructed in parallel.

Our visual system organizes spatially distinct areas with similar features into perceptual groups. To better understand the underlying mechanism of grouping, one route is to study its capacity and temporal progression. Intuitively, that capacity seems unlimited, and the temporal progression feels immediate. In contrast, here we show that in a visual search task that requires similarity grouping, search performance is consistent with serial processing of those groups. This was true across several experiments, for seeking a single ungrouped pair among grouped pairs, vice versa, and for displays with tiny spacings between the grouped items. In a control condition that ruled out display complexity confounds, when the small inter-object spacing was removed so that that pairs touched, removing the need to group by similarity, search became parallel. Why is similarity grouping so slow to develop? We argue that similarity grouping is 'just' feature selection - seeing a red, bright, or square group is global selection of those features. This account predicts serial processing of one feature group at a time, and makes new falsifiable predictions about how properties of feature-based selection should be reflected in similarity grouping.

Very young infants learn abstract rules in the visual modality.

Abstracting the structure or 'rules' underlying observed patterns is central to mature cognition, yet research with infants suggests this far-reaching capacity is initially restricted to certain stimuli. Infants successfully abstract rules from auditory sequences (e.g., language), but fail when the same rules are presented as visual sequences (e.g., shapes). We propose that this apparent gap between rule learning in the auditory and visual modalities reflects the distinct requirements of the perceptual systems that interface with cognition: The auditory system efficiently extracts patterns from sequences structured in time, but the visual system best extracts patterns from sequences structured in space. Here, we provide the first evidence for this proposal with adults in an abstract rule learning task. We then reveal strong developmental continuity: infants as young as 3 months of age also successfully learn abstract rules in the visual modality when sequences are structured in space. This provides the earliest evidence to date of abstract rule learning in any modality.

Topological Relations Between Objects Are Categorically Coded.

How do individuals compare images-for example, two graphs or diagrams-to identify differences between them? We argue that categorical relations between objects play a critical role. These relations divide continuous space into discrete categories, such as "above" and "below," or "containing" and "overlapping," which are remembered and compared more easily than precise metric values. These relations should lead to categorical perception, such that viewers find it easier to notice a change that crosses a category boundary (one object is now above, rather than below, another, or now contains, rather than overlaps with, another) than a change of equal magnitude that does not cross a boundary. We tested the influence of a set of topological categorical relations from the cognitive-modeling literature. In a visual same/different comparison task, viewers more accurately noticed changes that crossed relational category boundaries, compared with changes that did not cross these boundaries. The results highlight the potential of systematic exploration of the boundaries of between-object relational categories.

Visual routines are associated with specific graph interpretations.

We argue that people compare values in graphs with a visual routine - attending to data values in an ordered pattern over time. Do these visual routines exist to manage capacity limitations in how many values can be encoded at once, or do they actually affect the relations that are extracted? We measured eye movements while people judged configurations of a two-bar graph based on size only ("[short tall] or [tall short]?") and contrast only ("[light dark] or [dark light]?"). Participants exhibited visual routines in which they systematically attended to a specific feature (or "anchor point") in the graph; in the size task, most participants inspected the taller bar first, and in the contrast task, most participants attended to the darker bar first. Participants then judged configurations that varied in both size and contrast (e.g., [short-light tall-dark]); however, only one dimension was task-relevant (varied between subjects). During this orthogonal task, participants overwhelmingly relied on the same anchor point used in the single-dimension version, but only for the task-relevant dimension (e.g., taller bar for the size-relevant task). These results suggest that visual routines are associated with specific graph interpretations. Responses were also faster when task-relevant and task-irrelevant anchor points appeared on the same object (congruent) than on different objects (incongruent). This interference from the task-irrelevant dimension suggests that top-down control may be necessary to extract relevant relations from graphs. The effect of visual routines on graph comprehension has implications for both science, technology, engineering, and mathematics pedagogy and graph design.

Visual routines for extracting magnitude relations.

Linking relations described in text with relations in visualizations is often difficult. We used eye tracking to measure the optimal way to extract such relations in graphs, college students, and young children (6- and 8-year-olds). Participants compared relational statements ("Are there more blueberries than oranges?") with simple graphs, and two systematic patterns emerged: eye movements that followed the verbal order of the question (inspecting the "blueberry" value first) versus those that followed a left-first bias (regardless of the left value's identity). Question-order patterns led substantially to faster responses and increased in prevalence with age, whereas the left-first pattern led to far slower responses and was the dominant strategy for younger children. We argue that the optimal way to verify a verbally expressed relation's consistency with visualization is for the eyes to mimic the verbal ordering but that this strategy requires executive control and coordination with language.

The Connected Scatterplot for Presenting Paired Time Series.

The connected scatterplot visualizes two related time series in a scatterplot and connects the points with a line in temporal sequence. News media are increasingly using this technique to present data under the intuition that it is understandable and engaging. To explore these intuitions, we (1) describe how paired time series relationships appear in a connected scatterplot, (2) qualitatively evaluate how well people understand trends depicted in this format, (3) quantitatively measure the types and frequency of misinter pretations, and (4) empirically evaluate whether viewers will preferentially view graphs in this format over the more traditional format. The results suggest that low-complexity connected scatterplots can be understood with little explanation, and that viewers are biased towards inspecting connected scatterplots over the more traditional format. We also describe misinterpretations of connected scatterplots and propose further research into mitigating these mistakes for viewers unfamiliar with the technique.

Sudden insight is associated with shutting out visual inputs.

Creative ideas seem often to appear when we close our eyes, stare at a blank wall, or gaze out of a window--all signs of shutting out distractions and turning attention inward. Prior research has demonstrated that attention-related brain areas are differently active when people solve problems with sudden insight (the Aha! phenomenon), relative to deliberate, analytic solving. We directly investigated the relationship between attention deployment and problem solving by recording eye movements and blinks, which are overt indicators of attention, as people solved short, visually presented problems. In the preparation period, before problems eventually solved by insight, participants blinked more frequently and longer, and made fewer fixations, than before problems eventually solved by analysis. Immediately prior to solutions, participants blinked longer and looked away from the problem more often when solving by insight than when solving analytically. These phenomena extend prior research with a direct demonstration of dynamic differences in attention as people solve problems with sudden insight versus analytically.

Selecting and tracking multiple objects.

When interacting with the world, people can dynamically split attention across multiple objects in the environment, both when the objects are stationary and when the objects are moving. This type of visual processing is commonly studied in lab settings using either static selection tasks or moving tracking tasks. We describe performance limits that are common to both tasks, including limits on capacity, crowding, visual hemifield arrangement, and speed. Because these shared limits on performance suggest common underlying mechanisms, we examine a set of models that might account for limits across both. We also review cognitive neuroscience data relevant to these limits, which can provide constraints on the set of models. Finally, we examine performance limits that are unique to tracking tasks, such as trajectory encoding, and identity encoding. We argue that a complete model of multiple object tracking must account for both those limits shared between static selection and dynamic tracking, as well as limits unique to tracking. It must also provide neurally plausible mechanisms for the underlying processing resources.

Capacity for Visual Features in Mental Rotation.

Although mental rotation is a core component of scientific reasoning, little is known about its underlying mechanisms. For instance, how much visual information can someone rotate at once? We asked participants to rotate a simple multipart shape, requiring them to maintain attachments between features and moving parts. The capacity of this aspect of mental rotation was strikingly low: Only one feature could remain attached to one part. Behavioral and eye-tracking data showed that this single feature remained "glued" via a singular focus of attention, typically on the object's top. We argue that the architecture of the human visual system is not suited for keeping multiple features attached to multiple parts during mental rotation. Such measurement of capacity limits may prove to be a critical step in dissecting the suite of visuospatial tools involved in mental rotation, leading to insights for improvement of pedagogy in science-education contexts.

Attentional deployment is not necessary for successful emotion regulation via cognitive reappraisal or expressive suppression.

According to appraisal theories of emotion, cognitive reappraisal is a successful emotion regulation strategy because it involves cognitively changing our thoughts, which, in turn, change our emotions. However, recent evidence has challenged the importance of cognitive change and, instead, has suggested that attentional deployment may at least partly explain the emotion regulation success of cognitive reappraisal. The purpose of the current study was to examine the causal relationship between attentional deployment and emotion regulation success. We examined 2 commonly used emotion regulation strategies--cognitive reappraisal and expressive suppression-because both depend on attention but have divergent behavioral, experiential, and physiological outcomes. Participants were either instructed to regulate emotions during free-viewing (unrestricted image viewing) or gaze-controlled (restricted image viewing) conditions and to self-report negative emotional experience. For both emotion regulation strategies, emotion regulation success was not altered by changes in participant control over the (a) direction of attention (free-viewing vs. gaze-controlled) during image viewing and (b) valence (negative vs. neutral) of visual stimuli viewed when gaze was controlled. Taken together, these findings provide convergent evidence that attentional deployment does not alter subjective negative emotional experience during either cognitive reappraisal or expressive suppression, suggesting that strategy-specific processes, such as cognitive appraisal and response modulation, respectively, may have a greater impact on emotional regulation success than processes common to both strategies, such as attention.