VSS, May 13-18

Search and Attention: Capture, real-world, lifespan

Talk Session: Monday, May 16, 2022, 10:45 am – 12:15 pm EDT, Talk Room 1
Moderator: Trafton Drew, University of Utah

Times are being displayed in EDT timezone (Florida time): Wednesday, July 6, 4:08 am EDT America/New_York.
To see the V-VSS schedule in your timezone, Log In and set your timezone.

Search Abstracts | VSS Talk Sessions | VSS Poster Sessions | V-VSS Talk Sessions | V-VSS Poster Sessions

Talk 1, 10:45 am, 42.11

Contextual learning determines early attentional orienting in visual selection

Chris Jungerius1,2, Dirk Van Moorselaar2, Heleen A. Slagter2; 1University of Amsterdam, 2Vrije Universiteit

Background: While previous studies have shown that it is possible to proactively suppress attentional capture by a singleton distractor in a visual search task, this suppression disappears in larger set size conditions. One possible explanation for these findings is that the singleton has higher local feature contrast in the larger set size conditions, which increases its physical saliency and makes suppression impossible. However, increasing the number of items also affects target and singleton distractor location uncertainty and the number of possible display configurations, which might provide an alternative explanation for these findings. Methods: We performed a visual search study with 8-item search displays, where occasional probe trials tested the perception of letters at each location in the search display. In a series of online experiments, we manipulated singleton and target location uncertainty by limiting the locations they appeared at, and manipulated the number of possible display configurations by fixing part of the display. We compared probe letter recall in these conditions to performance in normal 4-item and 8-item display conditions. In a follow-up experiment, participants performed 2 sessions of the task, where only a subset of all possible search display configurations were used. Results: Recall of letters at singleton locations remains higher than at non-singleton distractor locations under reduced location uncertainty, but when configuration uncertainty was reduced, either by fixing the display or after two sessions of the restricted configuration condition, this difference disappeared. Currently, we are examining the interaction between longer learning and context uncertainty, by training participants with a limited set of configurations in 3 sessions over 3 days, and testing whether the reduced early attentional capture we observed in our previous experiment only applies to learned configurations. Conclusions: Our findings suggest that configuration uncertainty, but not location uncertainty, influences early attentional capture by singletons in visual search.

Talk 2, 11:00 am, 42.12

Visual search asymmetries are explained by visual homogeneity

Georgin Jacob1,2 (), SP Arun1,2; 1Centre for Neuroscience, Indian Institute of Science, Bengaluru, INDIA, 2Department of Electrical Comminication Engineering, Indian Institute of Science, Bengaluru, INDIA

Visual search is often asymmetric. Search for a Q among Os is often easier than search for O among Qs. But what image property causes asymmetry in visual search? The most popular explanation is that searching for the presence of a feature is usually easier than searching for its absence – but even this does not explain many asymmetries where no obvious feature can be named. Here, we report a novel and empirically measurable image property – visual homogeneity – that explains visual search asymmetries. For each image, its visual homogeneity is the time it takes to confirm the absence of an oddball target in a search array containing this image. We hypothesized that a search target is easy to find when the distractors are visually homogeneous. More generally, the visual homogeneity of the target and distractor would determine whether a given visual search is asymmetric. To investigate this hypothesis, we asked human participants to search for an oddball target that could be present or absent. We selected 40 popular search asymmetries reported in the literature. Search times were highly consistent, as evidenced by a high split-half correlation across participants (r = 0.92, p < 0.00005). Search asymmetries, measured as the difference between the search times of the hard and easy targets, were also likewise consistent across participants (split-half correlation: r = 0.64, p < 0.0005). Importantly, search asymmetry was accurately predicted by a weighted sum of the visual homogeneity of the target and distractor (r = 0.56, p < 0.0005). These predictions were driven primarily by the distractor, suggesting that search is asymmetric when the distractor is visually homogeneous. Taken together, our findings demonstrate for the first time, an empirically measurable image property that can explain visual search asymmetry.

Acknowledgements: This research was funded through a Senior Fellowship from the DBT-Wellcome India Alliance (Grant # IA/S/17/1/503081) and the DBT-IISc partnership programme (both to SPA).

Talk 3, 11:15 am, 42.13

The development of attention to social interactions in naturalistic scenes

Ioana Mihai1 (), Simona Skripkauskaite2, Kami Koldewyn1; 1Bangor University, 2University of Oxford

Human attention is preferentially captured by social information in scenes. Recent research suggests that people also preferentially attend to interacting dyads compared to non-interactors. Thus, interactions might capture visual attention above other social information, a "preference" that could contribute, across development, to social learning. However, very little work has directly examined these processes in complex scenes, nor across development. We investigated the development of attention to social interactions in naturalistic scenarios across three free-viewing eye-tracking experiments with 54 children (age: 6-12) and 99 adults (18-35). Scenes depicted dyads who were either interacting or not (exp1), or dyads were presented together with one or two non-interactors (exp2). In exp3, scenes were ambiguous and participants indicated whether they perceived the scenes as interactive or not. Across experiments, we compared attentional engagement and capture by social areas of interest (humans) with non-social information (background), contrasted by whether scenes were interactive or not. Results revealed that children and adults are strikingly similar, showing a strong social attentional bias in the first two experiments, but a weaker bias in ambiguous social scenes. For both children and adults, interactions increase this bias (exp1), and capture and engage attention more strongly compared to another agent in the same scene (exp2). However, when interactive dyads compete with two other agents, the interactive bias is smaller for children than for adults. In the third experiment, while adults were more likely to view ambiguous scenarios as interactive compared to children, this difference was not reflected in the way attention was oriented to social information, suggesting no top-down modulation. Together these data suggest interactions take attentional priority, but this “interaction bias” increases across development. Crucially, our data also suggest this ‘bias’ to attend to social interactions is present as early as six, with important implications for social attention and social development.

Talk 4, 11:30 am, 42.14

Incorrect Computer Aided Detection (CAD) marks lead to early quitting: A potential mechanism for poor CAD performance in clinical practice

Trafton Drew1 (), Anna Carissa Delos Reyes1, Jeff Moher2; 1University of Utah, 2Connecticut College

Prior work has shown that irrelevant distractors with an abrupt onset result in early quitting in a visual search task (Moher, 2020). Thus, while salient distractors often increase reaction time when a target is present, the same distractors elicit early quitting when no target is present. This may have important implications for applied tasks like medical screening, particularly in the context of CAD. In cancer screening, radiologists face an extremely difficult task where small differences could cost lives. There is a great deal of enthusiasm to help improve radiologist performance through AI-assisted cueing like CAD. Unfortunately, there are number of instances, such as breast cancer-screening, where CAD does not yield improved performance when deployed in the field. Prior work suggests that disappointing performance may be driven by attentional capture by CAD marks. The current examination sought to determine if the previously identified early quitting effect replicates in a situation more similar to CAD in screening radiology. Here, unlike Moher (2020), CAD marks sometimes highlighted targets (Ts embedded in 1/f noise) and appeared contemporaneously with the search array. Observers were divided into CAD and no CAD groups. We found that RTs greatly decreased (~500ms faster) on trials where there was an incorrect CAD mark on a distractor item (Ls). In addition, when a target was present and the CAD mark was on a distractor, accuracy was decreased relative to performance on the same targets in the No CAD condition. These results demonstrate that the previously identified phenomenon of early quitting in response to irrelevant distractor cues generalizes to a paradigm designed to emulate important aspects of breast cancer screening. More generally, these results suggest that early quitting may be an under-appreciated factor in understanding why CAD cues are less effective than expected in medical practice.

Talk 5, 11:45 am, 42.15

The reduced fidelity of selective sensory information processing in the elderly with mild cognitive impairment

Kanyarat Benjasupawan1,2 (), Panchalee Sookprao1,2, Thiparat Chotibut4, Itti Chatnuntawech5, Sirawaj Itthipuripat1,6, Chaipat Chunharas2,3; 1Neuroscience Center for Research and Innovation, Learning Institute, King Mongkut’s University of Technology Thonburi, Bangkok, 10140, Thailand, 2Cognitive Clinical and Computational Neuroscience lab, Faculty of Medicine, Chulalongkorn University, Bangkok, 10330, Thailand, 3Chula Neuroscience Center, King Chulalongkorn Memorial Hospital, Bangkok, 10330, Thailand, 4Department of Physics, Faculty of Science, Chulalongkorn University, Bangkok 10330, Thailand, 5National Nanotechnology Center, National Science and Technology Development Agency, Pathum Thani, 12120, Thailand, 6Big Data Experience Center, King Mongkut’s University of Technology Thonburi,10140, Bangkok, Thailand

Mild cognitive impairment (MCI) is a neurocognitive disorder found in ~30% of the elderly population. Declines in visuospatial and executive functions are common in MCI populations. However, it is unclear if these deficits are due to a general decline in sensory processing or dysfunction in selective attention. Here, we compared neural processes that indicate both sensory and selective information processing across the elderly subjects with and without MCI (60-72 years old, age-matched). We measured behavioral and EEG responses from the volunteers while they performed the attention-cueing Eriksen Flanker task. In this task, they discriminated shapes of a cued target surrounded by the distractors (i.e. flankers) whose shapes could be congruent or incongruent to the target shape. When task difficulty was matched, the elderly with MCI had slower response times than the healthy control, and this was observed in both congruent and incongruent conditions. The amplitudes of the target-evoked event-related potentials (ERPs), particularly the P1 and P3 components, were reduced in these MCI individuals. This suggested declines in early sensory and post-sensory decision-related processes, reflecting their behavioral responses. Furthermore, to quantify the spatial precision and timing of attentional focus, we employed inverted encoding models to reconstruct the spatially selective representations of the attentional focus centered at the cue onset and the target locations based on their alpha-band activity (~8-12Hz) in the EEG data. The healthy control exhibited more precise alpha-based spatial representations with an earlier onset. On the other hand, the MCI patients showed significantly broader spatial representations with a slower onset. Together, our findings provide strong neural evidence suggesting that declines in visuospatial and executive functions in MCI is contributed by the diminishing fidelity of selective sensory information processing.

Acknowledgements: The National Research Council of Thailand, the Thailand Science Research and Innovation, the Asahi Glass Foundation, the KMUTT Partnering Initiative, and the KMUTT’s Frontier Research Unit Grant for Neuroscience Center for Research and Innovation.

Talk 6, 12:00 pm, 42.16

Neural correlates associated with a virtual reality based dynamic visual search in Cerebral Visual Impairment

Zahide Pamir1 (), Corinna M. Bauer1, Claire E. Manley1, Daniel D. Dilks2, Lotfi B. Merabet1; 1The Laboratory for Visual Neuroplasticity. Department of Ophthalmology, Massachusetts Eye and Ear, Harvard Medical School. Boston, MA USA, 2Department of Psychology, Emory University, Atlanta, GA 30322, USA

Cerebral visual impairment (CVI) is a brain based visual impairment characterized by deficits in visual function and higher order visual perception. Individuals with CVI report feeling overwhelmed in cluttered environments and have difficulties recognizing familiar individuals in a crowd and following moving traffic. We investigated how visual performance in CVI is affected by perceptual load in a dynamic visual environment using eye tracking and fMRI methodologies. We developed a virtual reality based naturalistic visual search task (a school corridor) in participants were required to find and track the principal and three load conditions (low, medium, high) were tested by manipulating the number of people walking through the corridor. In the eye-tracking experiment, success rate, reaction time, and gaze error were compared across CVI (N=7) and controls (N=9). In the fMRI experiment, activations in visual areas within early visual cortex, dorsal and ventral visual streams were compared (CVI: N=4; controls: N=4). Eye tracking revealed that individuals with CVI were more likely to miss the target, took longer to find the target, and had higher gaze errors. Qualitatively, CVI appeared to be more sensitive to increasing task load. In both groups, fMRI results showed comparable response profiles with respect to increasing load within early visual areas (including V1, V2, V3). In contrast, visual areas associated with the dorsal stream (including hMT+, V3a&V3b and IPS subdivisions) showed overall lower responses in CVI and less sensitivity to task load. Ventral visual areas (including hV4, VO1&VO2) showed comparable responses in both groups with respect to load. Interestingly, certain areas including PHC1&PHC2 and V6 showed the opposite profile, with CVI having higher neural responses than controls, but again more sensitivity to increasing task demands. Overall, these results show a complex response profile in higher level visual areas in CVI with respect to a dynamic search task.

Acknowledgements: This work is supported by a grant from NEI (R01 EY03097 to LBM).