My Research Archive

Jeremiah Still, PhD

2017

Still, J. D. (2017). Web page attentional priority modelCognition, Technology, & Work, X, 1-12.

Designing an interface that is both information rich and easy to search is challenging. Successfully finding a solution depends on understanding an interface’s explicit and implicit influences. A cognitively inspired computational approach is taken to make the implicit influences apparent to designers. A saliency model has already been shown to predict the deployment of attention within web page interfaces. It predicts regions likely to be salient, based on local contrast stemming from the bottom-up channels (e.g., color, orientation). This research replicates these previous findings and extends the work by proposing a web page-specific attentional priority (AP) model. This AP model includes previous interaction experience history, manifested as conventions, within the already valuable saliency model. These sources of influence automatically nudge our attention to regions that usually contain useful visual information. This research shows that, by integrating spatial conventions with a saliency model, designers can better predict the deployment of attention within web page interfaces.

Still, J. D., Cain, A. A., & Schuster, D. (2017). Human-Centered authentication guidelines. Journal of Information and Computer Security 

Cybersecurity research and development has mainly focused on technical solutions to increase security. However, the greatest weakness of many systems is the user. We argue that authentication schemes with poor usability are inherently insecure, as users will inadvertently weaken the security in their efforts to use the system. We propose that designers need to consider the human factors that impact end-user behavior. Development from this perspective will address the greatest weakness in most security systems by increasing end-user compliance.

Hicks, J. M., Cain, A. A., & Still, J. D. (2017). Visual saliency predicts fixations in low clutter web pages. HFES Conference Proceedings, 

Previous research has shown a computational model of visual saliency can predict where people fixate in cluttered web pages (Masciocchi & Still, 2013). Over time, web site designers are moving towards simpler, less cluttered webpages to improve aesthetics and to make searches more efficient. Even with simpler interfaces, determining a saliency ranking among interface elements is a difficult task. Also, it is unclear whether the traditionally employed saliency model (Itti, Koch, & Niebur, 1998) can be applied to simpler interfaces. To examine the model’s ability to predict fixations in simple web pages we compared a distribution of observed fixations to a conservative measure of chance performance (a shuffled distribution). Simplicity was determined by using two visual clutter models (Rosenholz, Li, & Nakano, 2007). We found under free-viewing conditions that the saliency model was able to predict fixations within less cluttered web pages.

Still, M. L. & Still, J. D. (2017). Subliminal techniques: Considerations and recommendations for analyzing feasibilityInternational Journal of Human-Computer Interaction, 

In an attempt to provide users with more information while maintaining a calm ubiquitous environment, researchers have investigated the possibility of presenting information “subliminally”. We explore the historical issues associated with examining perception without awareness with special emphasis on the difficulty associated with ensuring stimuli have been presented below the participant’s subjective threshold of awareness. It may be possible to circumvent this issue by taking a results-oriented approach. However, the advancement of subliminal techniques requires stronger experimental evidence that information is actually being presented below the subject threshold of awareness and gaining the desired effect on user performance. We offer three considerations that help designers weigh the costs and benefits of employing a subliminal technique. We also offer three recommendations that help designers’ present information below the subjective threshold of awareness and measure their users’ awareness of the information.

Cain, A. A., Werner, S., & Still, J. D. (2017). Graphical authentication resistance to over-the-shoulder-attacksCHI Extended Astracts Proceedings,1-8 

Graphical passwords offer advantages for memorability over conventional alphanumeric passwords, but in some cases they have been vulnerable to over-the-shoulder-attacks (OSA). Thus, many second-generation graphic based schemes are specifically designed to be resistant to OSA. This is often achieved by not having users select targets directly, but by adding cognitive operations to create seemingly random response patterns. This study takes the first step to directly compare three prototypical graphical password schemes to determine their relative resistance to OSAs employing a within-subjects design. We found that schemes requiring cognitive operations in response to target patterns were superior to direct selection of targets. Convex Hull Click was most secure, followed by What You See is What You Enter, while Use Your Illusion showed high vulnerability to OSA. In addition, we discuss a diversity of previous measurements, which are meant to examine security strength of new approaches. We highlight the need for standard OSA resistance measures depending on threat model needs.

Still, J. D., Hicks, J. M., Cain, A. A., & Billman, D. (2017). Predicting stimulus-driven attentional selection within mobile interfacesProceedings of the 8th International Conference on Cognitive and Neuroergonomics,255-261 

Masciocchi and Still [1] suggested that biologically inspired computational saliency models could predict attentional deployment within webpages.Their stimuli were presented on a large desktop monitor. We explored whether a saliency model’s predictive performance can be applied to small mobile interface displays. We asked participants to free-view screenshots of NASA’s mobile application Playbook. The Itti et al. [2] saliency model was employed to produce the predictive stimulus-driven maps. The first six fixations were used to select values to form the saliency maps’ bins, which formed the observed distribution. This was compared to the shuffled distribution, which offers a very conservative chance comparison as it includes predictable spatial biases by using a within-subjects bootstrapping technique. The observed distribution values were higher than the shuffled distribution. This suggests that a saliency model was able to predict the deployment of attention within small mobile application interfaces.

2016

Grgic, J., Still, M. L., & Still, J. D. (2016). Effects of cognitive load on affordance-based interactionsJournal of Applied Cognitive Psychology, 30, 1042-1051.

Affordances are design clues that allow users to effortlessly determine how objects should be used. Since their introduction to the human–computer interaction design community, their use within interfaces has become a staple. But, despite work to understand how affordances make an interface's available actions transparent, few empirical studies have examined the cognitive origins of this interaction type. Therefore, we explored the impact of cognitive load on affordance-based interactions. Participants were asked to perform a series of affordance-based interactions while under different types of working memory load (no load, verbal, spatial, and central executive). Affordance-based interactions were consistently slowed under central executive load, but never when under verbal load. We conclude that affordance-based interactions do require cognitive resources, but resource costs may only manifest under heavy load. With this knowledge, designers may be able to better predict when an affordance will (or will not) act as if it is resource free.

Still, J. D., Hicks, J. M., & Gall, J. (2016, November). Using Computational Models to Quantify Stimulus-Driven Influences in Web Pages. Poster presented at the 57th Annual Meeting of the Psychonomic Society, Boston, MA.

One critical factor in whether a search will be fast and effortless is visual saliency. If we are searching for an object, it is easier to find one that is visually salient than one that is not. Although many researchers have examined the performance of stimulus-driven computational models within natural scenes, relatively few have explored interfaces. Still and Masciocchi (2010) suggested that the classic Itti, Koch, and Niebur (1998) Saliency model is capable of predicting the initial fixations in webpages. The present study expands on that research by comparing the performance of popular saliency models across a variety of webpages (i.e., mostly images, half text and images, mostly text). Results of this study reveal the strengths and weaknesses of each model within these atypical stimuli. These findings ought to help interface designers create easier to search interfaces by making implicit stimulus-driven information that guides attention visible.

Cain, A. A., Chiu, L., Santiago, Felicia, S., & Still, J. D. (2016). Swipe authentication: Exploring over-the-shoulder attack performance. In D. Nicholson (Ed.), Advances in Human Factors Cybersecurity (pp. 327-336). Walt Disney World, Florida, USA: Springer.

Swipe passwords are a popular method for authenticating on mobile phones. In public, these passwords may become visible to attackers who engage in shoulder surfing. There is a need for strategies that protect swipe passwords from over-the-shoulder attacks (OSAs). We empirically explored the impact of providing gesture visual feedback on OSA performance during successful and unsuccessful swipe login attempts on mobile phones. We found evidence that entry visual feedback facilitates OSAs. As users are biased towards symmetrical swipe patterns, we investigated their impact on attack performance. We found that symmetrical swipe patterns were less vulnerable than asymmetrical patterns; possibly due to the speed of entry. As users tend toward simple patterns, we investigated the impact that nonadjacent, diagonal knight moves have on OSAs. We found that knight moves significantly decreased OSA performance. We recommend users turn off gesture entry visual feedback and use knight moves for greater password security.

 

Cain, A. A., & Still, J. D. (2016). A rapid serial visual presentation method for graphical authentication. In D. Nicholson (Ed.), Advances in Human Factors Cybersecurity (pp. 3-12). Walt Disney World, Florida, USA: Springer.

We propose a Rapid Serial Visual Presentation (RSVP) graphical authentication method that is suited for multi-touch mobile devices. This method presents degraded pictures of everyday objects in a temporal stream. Considering all the other authentication methods employ a spatial visual search, our method is unique (i.e., searching across time versus space). A temporal method of presentation is used to decreases login times down to 14 seconds and to allow login with a simple touch of the screen. By degrading the images, over-the-shoulder attackers are prevented from easily capturing the passcode. This study shows that all participants could successfully login at least once when allowed up to three attempts. After becoming familiar with the RSVP authentication method, participants took on the role of an attacker. Notably, no one was able to identify the passcode. The RSVP method offers a memorable, usable, quick, and secure alternative for authentication on multi-touch mobile devices.

 

Long, S. K., Karpinsky, N. D., Doner, H., & Still, J. D. (2016). Using a mobile application to help visually impaired individuals explore the outdoors. In G. Di Bucchianico and P. Kercher (Eds.), Advances in Design for Inclusion (pp. 213-226). Walt Disney World, Florida, USA: Springer.

Visually impaired individuals face a variety of challenges when navigating outdoors, including uneven terrain, unexpected obstacles, safety concerns, and reliance on others for information. The goal of this study was to understand further the navigational needs of visually impaired individuals and to develop a mid-fidelity prototype to address these needs. Through interviews with visually impaired users and accessibility professionals, researchers found that present technology leads to an incomplete understanding of the trail and harmful situations. Currently, there is no known technology available that integrates real-time updates with static trail information for individuals navigating outdoors. In response, a mobile prototype was proposed, integrating user-provided updates with static trail information in a format that caters to all users. Our usability testing showed visually impaired users made few errors using the prototype and were satisfied with their experience.

 

McEvoy, P., & Still, J. D. (2016). Contextualizing mnemonic phrase passwords. In D. Nicholson (Ed.), Advances in Human Factors Cybersecurity (pp. 295-304). Walt Disney World, Florida, USA: Springer.

Our society depends on password-based authentication methods for accessing valuable information. However, the use of weak passwords is placing us at risk. Cyber security systems encourage users to employ strong passwords often by increasing requirements. Unfortunately, using a strong password requires more cognitive effort. This increase in effort pushes users to find workarounds that directly harm security. The paradox between security and usability has often resulted in simply blaming users rather than seeking a Human-Centered Design perspective. We introduce a strategy for developing strong passwords that embed contextual cues within mnemonic phrase passwords.  Using this strategy participants were able to create strong passwords and better remember them compared with a traditional mnemonic strategy.

 

Still, J. D. (2016). Cybersecurity needs you! ACM Interactions (May + June: Feature), 23(3), 54-58.

 

2015

Still, J. D., Still, M. L., & Grgic, J. (2015). Designing intuitive interactions: Exploring performance and reflection measuresInteracting with Computers, 27, 271-286. Impact Factor: 1.4

Intuitive interactions are supported by users’ implicit and explicit learning experiences. But, determining user knowledge can be difficult. With many options available for eliciting that knowledge, we tested the effectiveness of two methods—performance and reflection. Users were presented with simple interactions that had varying levels of intuitiveness (affordance, convention, bias). They were asked to perform the interaction or to describe how the interaction should be designed. These methods of knowledge elicitation produced inconsistent results; sometimes they produced the same result (affordance-based interactions), sometimes the opposite (convention-based interactions). Furthermore, when both methods were used, results obtained from the second measure were often contaminated by completion of the first measure. Carryover effects were present regardless of which measure was completed first. These results indicate that the method used to elicit knowledge should be selected based on the type of interaction that is being investigated and multiple measures should be used with caution.

Still, M. L. & Still, J. D. (2015). Contrasting traditional in-class exams with frequent online testingJournal of Teaching and Learning with Technology, 4, 30-40.

Although there are clear practical benefits to using online exams compared to in-class exams (e.g., reduced cost, increased scalability, flexible scheduling), the results of previous studies provide mixed evidence for the effectiveness of online testing. This uncertainty may discourage instructors from using online testing. To further investigate the effectiveness of online exams in a
naturalistic situation, we compared student learning outcomes associated with traditional in-class exams compared to frequent online exams. Online exams were administered more frequently in an attempt to mitigate potential negative effects associated with open-book testing. All students completed in-class and online exams with order of testing condition counterbalanced (in-class first, or online first) between students. We found no difference in long-term retention for material
that had originally been tested using frequent online or traditional in-class exams and no difference in self-reported study time. Overall, our results suggest that frequent online assessments do not harm student learning in comparison to traditional in-class exams and may impart positive subjective outcomes for students.

Lee, L. & Still, J. D. (2015). Re-designing warning messages to encourage BYOD policy adherenceProceedings of the 3rd Human Aspects of Information Security, Privacy, and Trust held at the 17th International Conference on Human-Computer Interaction, 369-378.

Many corporations and organizations support a Bring Your Own Device (BYOD) policy, which allows employees to use their personal smartphones for work-related purposes. Access to proprietary company data and information from an employee’s smartphone raises serious privacy and security concerns. Companies are vulnerable to data breaches if employees are unable to discern which applications are safe to install. Situating privacy requirements ought to encourage safer application install decisions and decrease risker ones. This study examines the use of context-relevant warning messages, which alert employees to be cautious when the company’s BYOD policy may be violated. We also explore the impact of presenting permission requirements before and after making the install decision. We provide evidence that the presence of warnings, despite the timing of when they were presented, facilitated a lower number of risky installations. In situations when it was safe to install an application, warning messages presented before the install decision drastically encouraged installations compared to when there were no warnings. Interestingly, the opposite pattern was found when warning messages were presented after the decision. Overall, better privacy and security decisions will be made if permission requirements are displayed with relevant warning messages. In addition, safe installations will be encouraged through the placement of these meaningful warnings on the description page of a mobile application before a user has decided to install it.

Schuster, D., Still, M. L., Still, J. D., Lim, J. J., Feria, C., & Rohrer, C. (2015). Opinions or algorithms: An investigation of trust in people versus automation in app store securityProceedings of the 3rd Human Aspects of Information Security, Privacy, and Trust held at the 17th International Conference on Human-Computer Interaction, 415-425.

Mobile application (app) stores are a critical source of information about risk in an uncertain environment. App stores ought to assess and communicate the risk associated with an installation so that users are discouraged from installing risky or harmful apps in app stores. However, only a limited number of studies offer designers information about how to communicate risk effectively. We focused on the user’s trust associated with security information stemming from crowd-sourced evaluations compared to those generated from an automated system. Both of these sources of security information are pervasively used to indicate possible risk associated with an app. We investigated whether biases exist for a particular source of information given similar amount of security information being available. We found that participants preferred to install apps rated by automation to those rated by humans despite equivalence in stated risk. Further, we found evidence of a gender difference in trust in automation.

 

2014

Still, J. D., Masciocchi, C. & Still, M. L. (2014, November). Stimulus- and Goal-Driven Influences on Visual Searches in Webpages. Poster to be presented at the 55th Annual Meeting of the Psychonomic Society, Long Beach, CA.

Although many studies have examined the processes that guide visual search, relatively few have used realistic tasks. This is problematic because, although it is assumed both stimulus-driven features and search goals contribute to "real world" visual search, the iteraction between these two influences is ill defined. This makes predicting where users will look during naturalistic visual searches difficult for interface designers. Masciocchi and Still (2013) began addressing this issue by tracking participants' eye movements while they free viewed webpages. They found that visual salience predicted users' initial fixations, demonstrating that stimulus-driven features pull attention even in complex, naturalistic displays. The present study expands on that research by having participants perform a realistic task - searching for specific items on webpages - while
manipulating the saliency of the target's location (high versus low). Results of this study thus address the interaction between stimulus-driven and goal-driven processes within realistic tasks and stimuli.

Still, J. D., Still, M. L., & Bell, J. (2014, August). Closing the knowledge gap with intuitive flat designs. UX Magazine: Feature, Article 1297, ISSN: 2168-5681.

Shokrpour, A., Palma, A., Gomez, M., Santiago, F., & Still, J. D. (2014, May). Swipe authentication: Exploring over-the-shoulder-attack performance. Poster presented at the 57th Annual Spartan Psychological Association Research Conference, San Jose, CA.

Swipe gestures are a popular method for authentication, because they are easier to enter than text based passwords on touch screen devices.Unfortunately, gesture-based authentication may fall prey to over-the-shoulder-attacks. Research based guidelines surround the creation of strong character-based passwords, but not for gesture-based passwords. In this study, we start to empirically explore the impact of providing gesture visual feedback and symmetry on over-the-shoulder-attacks performance. Participants viewed 64 swipe authentication videos. We captured their memory of the gesture seconds later using pencil and paper. This study showed that gesture authentication with visual feedback improves over-the-shoulder-attack performance.

 

2013

Still, J. D. & Dark, V. J. (2013). Cognitively describing and designing affordances. Journal of Design Studies, 13, 285-301.

The term affordance carries different meanings within design communities. Traditionally, affordances were discussed within a Gibsonian framework in which affordances arise from direct perception. Some authors now describe affordances as being mostly perceptual while others describe them as being culturally bound. We suggest that both of these descriptions are correct and that they can be explained from a cognitive conceptualization of perceived affordances. We suggest that perceived affordances are supported by automatic perceptual processes in the user developed over time through consistent interactions with the environment. Design consistency is critical for producing effortless usage, because interaction consistency facilitates the formation of long-term memory structures. We explore the underlying mechanisms that could explain how affordances arise and affect the cognitive system.

 

Steinhoff, C. & Still, J. D. (2013). Priming categorization in a card sort. Proceedings of the 15th
International Conference on Human-Computer Interaction
, 8004, 265-272.

When using the card sorting technique, the goal of a user experience researcher is to determine the user’s expected information architecture. Card sorting is a knowledge elicitation method where users are given labeled cards and are asked to place them into groups. This method is commonly used to determine a natural navigation structure for a group of users. We examine the impact of priming, an implicit memory effect in which exposure to a stimulus influences response to a later stimulus, on this popular user-centered design method. A control group did the card sort only, while the experimental group watched a short presentation before performing their card sorts. The dependent measure was the percentage of agreement of each card sort against the typical sort. The primed group sort was significantly more similar to the typical response than the control group. This study provides evidence that card sorting can be modulated by priming.

 

Masciocchi, C. M. & Still, J. D. (2013). Alternatives to eye tracking for predicting stimulus-driven attentional selection within interfacesJournal of Human-Computer Interaction, 34, 285-301.

The visual properties of a design contribute to the formation of regions with differing amounts of uniqueness, or salience, producing an initial stimulus-driven attentional bias. The colocation of salient regions and critical information should be maximized as this increases the interface's usability by decreasing search times. The determination of salient locations, however, is often difficult. In web page design, eye tracking has traditionally been used to measure where users attend, therefore indicating the salient regions. But, eye tracking as a descriptive technique has many known costs. We propose two alternative methods to eye tracking that can be used to predict which regions of a web page will draw users' stimulus-driven attention: interest point recording and saliency model predictions. Through an empirical investigation we show that the predictions of both methods correlate with the locations fixated by a separate group of participants, and thus these methods are effective alternatives to eye tracking during formative design testing.

 

2012

Still, M. L. & Still, J. D. (2012). Practical Recommendations for Examining Subliminal Techniques in Ubiquitous Computing [abstract]. Proceedings of the Effects of Subliminal Perception in Ubiquitous Computing Environments Workshop at the 14th International Conference on Ubiquitous Computing, Pittsburgh, PA.

In an attempt to provide users with more information while maintaining a calm ubiquitous environment, designers have investigated the possibility of presenting information “subliminally”. We explore the historical issues associated with examining perception without awareness; one of those issues is the difficulty associated with ensuring stimuli have been presented below the participant’s subjective threshold of awareness. It may be possible to circumvent this issue by taking a results-oriented approach. We make five recommendations for designers interested in using subliminal techniques. In these recommendations we provide methods to gauge participant awareness and encourage designers to examine the importance of perception without awareness within their specific ubiquitous information sources. Before we can continue the advance of insightful uses of subliminal techniques, we need to explore whether or not information presented below the subjective threshold of awareness will have a practical effect on user performance.

 

Still, J. D., & Masciocchi, C. M. (2012). Considering the influence of visual saliency during interface searches [abstract]. Book Chapter in Cognitively Informed Interfaces

In this chapter we highlight the influence of visual saliency, or local contrast, on users’ searches of interfaces. Designers have traditionally focused on the importance of goals and expectations (top-down processes) for the navigation of interfaces with little consideration for the influence of saliency (bottom-up processes). The Human-Computer Interaction literature does not discuss the influence of bottom-up processing, potentially neglecting an important aspect of interface-based searches. We review studies that demonstrate how a user’s attention is rapidly drawn to visually salient locations in a variety of tasks and scenes, including web pages.

 

2011

Still, J. D. & Still, M. L. (2011, May). No cost or benefit from frequent online quizzes compared to traditional exams. Poster presented at the Annual Meeting of the Society for the Teaching of Psychology, Chicago, IL.

Online examinations were explored as a viable replacement for in-class examinations. The goal was to save limited resources without jeopardizing academic integrity in an Introductory Psychology course. This study uniquely employed a within-subjects design to explore possible differences between frequent online examinations and traditional, less frequent in-class examinations. We hoped the use of multiple online examinations would promote student learning by encouraging spaced studying. One hundred and thirty nine students participated in this study; they were from two sections of the same course taught by the same instructor. Assignment to online and in-class examinations was counterbalanced across the two sections to minimize nuisance variables such as content difficulty. We found no difference in student study time for online or in-class examinations. Further, there was no evidence that spaced studying increased long-term retention. Although no study habit or comprehension differences were found, students reported a preference for online exams. In addition, the use of online assessments freed class time and minimized printing and grading costs. Our results suggest that online assessments do not harm student learning and may provide some positive, practical outcomes.

Wynn, J. & Still, J. D. (2011). Motivating change and reducing cost with the discount video data
analysis technique
. Proceedings of the 14th International Conference on Human-Computer
Interaction
, 6770, 321-328.

Testing the usability of an interface is a critical phase of product development. However, it is often reported that analyzing the data from such testing consumes too many limited resources. We attempted to reduce this consumption by proposing a new technique, Discount Video Data Analysis (DVDA). We compared it with another popular accelerated analysis technique, Instant Data Analysis (IDA). Using IDA, evaluators analyze data after a series of usability tests, whereas DVDA calls for analyzing the data after every test in the series. Immediate analysis decreases the chance that subsequent test data will negatively interfere with evaluators’ recall. Additionally, DVDA produces a video of the testing allowing the users’ emotional responses (e.g., frustration) to be shared with developers who may be resistant to interface modifications. We found evaluators using DVDA identified more usability issues and provided more supportive evidence for each issue than evaluators using IDA.

Still, J. D. (2011). Experimental design: Does external validity trump internal validity? ACM
Interactions (May + June: Feature), 18(13), 66-68.

 

2010

Still, J. D., & Dark, V. J. (2010). Examining working memory load and congruency effects on
affordances and conventions
. International Journal of Human-Computer Studies, 68(9),
561-571.

Although there is a debate about whether designers should draw a distinction between perceptual affordances and cultural conventions, there are few behavioral studies. We examined the impact of working memory load and expected button-to-action mapping congruency on affordances and conventions. The findings suggest both sides of the debate are correct. Learned conventions were found to structure responses towards expected actions, just like affordances, but affordance-based interactions were not affected by memory load while convention-based actions were. Therefore, designers ought to employ perceptual affordances when possible and when not feasible they ought to reuse established conventions. Additionally, evidence is presented that violating expected affordance-based and convention-based button-to-action mappings caused a similar performance cost. We believe that after the initial learning period, conventions play a critical role in the perception of a design’s available actions just as perceptual affordances do.

Grgic, J. & Still, J. D. (2010, May). Affordance interactions are affected by spatial working memory load. Poster presented at the meeting of the Midwestern Psychological Association, Chicago, IL.

According to Gibson (1979), affordances are available through direct perception of the environment. Interaction designers interpret this as meaning affordances are executable without cognitive cost. We provide evidence that this commonly accepted belief is not true. Affordance interaction performance was decreased by loading working memory with spatial information.

Still, J. D. & Masciocchi, C. M. (2010, April). A saliency model predicts fixations in web interfaces. Paper presented at the Computer-Human Interaction Conference, 5th International Workshop on Model-Driven Development of Advanced User Interfaces, Atlanta, GA.

User interfaces are visually rich and complex. Consequently, it is difficult for designers to predict which locations will be attended to first within a display. Designers currently depend on eye tracking data to determine fixated locations, which are naturally associated with the allocation of attention. A computational saliency model can make predictions about where individuals are
likely to fixate. Thus, we propose that the saliency model may facilitate successful interface development during the iterative design process by providing information about an interface’s stimulus-driven properties. To test its predictive power, the saliency model was used to render 50 web page screenshots; eye tracking data were gathered from participants on the same images. We found that the saliency model predicted fixated locations within web page interfaces. Thus, using computational models to determine regions high in visual saliency during web page development may be a cost effective alternative to eye tracking.

Wellman, D., Vey, M. & Still, J. D. (2010, March). Visual saliency and website design. Poster presented at the Great Plains Conference, St. Joseph, MO.

Wynn, J. S. & Still, J. D. (2010, March). Comparison of two accelerated usability test data analysis techniques. Poster presented at the Great Plains Conference, St. Joseph, MO.

Chowdhury, S. K. & Still, J. D. (2010, March). Effect of opacity of stimulus in deployment of interest in an interface. Poster presented at the Great Plains Conference, St. Joseph, MO. {*Awarded 1st Place*}

Grgic, J. E., Masciocchi, C. & Still, J. D. (2010, March). Designing human-centered notifications. Poster presented at the Great Plains Conference, St. Joseph, MO. {*Awarded 3rd Place*}

 

2009

Still, J. D., & Dark, V. J. (2009, April). The cost of violating design affordances and conventions. Poster presented at the Emerging Technologies Conference, Ames, IA.

Three Takeaways for Designers:
1. Perceptual affordances ought to be used when possible.
2. Violating a learned convention is just as costly as violating a perceptual affordance.
3. If implementing a perceptual affordance is not feasible, established conventions should be reused.

 

2008

Still, J. D ., & Dark, V. J. (2008). An empirical investigation of affordances and conventionsProceedings of the Third International Conference on Design Computing and Cognition, 457-472. {Awarded Best Paper}

Still, J. D., Dark, V. J. & Parkhurst, D. J. (2008). Dense Regions of View-Invariant Features Promote Object Recognition [abstract]. Poster presented at the 16th annual meeting of the Object Perception, Attention & Memory Conference, Chicago, IL.

Still, J. D., Dark, V. J., & Parkhurst, D. J. (2008). What determines where you look in photographs of complex objects? [abstract]. Paper presented at the meeting of the Midwestern Psychological Association, Chicago, IL.

Using SIFT (Scale Invariant Feature Transform), a computer vision algorithm, Still, Dark, and Parkhurst (2007) showed that eye fixations are linked to regions containing high proportions of SIFT view-invariant features. We examined whether regions of invariance affect object naming accuracy. Participants named fragmented images displaying 20%, 40%, 60%, or 80% of the pixels of photographs of complex objects. Displayed pixels were chosen based either on random or view-invariant features. Object naming accuracy was better when regions of invariance were shown, especially with 40% or fewer pixels. Regions of invariance appear to play a role in human object recognition.

Dohse, K. C., Dohse, T., Still, J. D., & Parkhurst, D. J. (2008). Enhancing multi-use interaction with multi-touch tabletop displays using hand tracking [abstract]. Paper presented at the meeting of the First International Conference on Advances in Computer-Human Interaction, Sainte Luce, Martinique.

A rear-projection multi-touch tabletop display was augmented with hand tracking utilizing computer vision techniques. Touch detection by frustrated total internal reflection is useful for achieving interaction with tabletop displays, but the technique is not always reliable when multiple users in close proximity simultaneously interact with the display. To solve this problem, we combine touch detection and hand tracking techniques in order to allow multiple users to simultaneously interact with the display without interference. Our hope is that by considering activities occurring on and above a tabletop display, multiuser interaction will become more natural and useful, which should ultimately support collaborative work.

Watch our Demo Video (over 18,000 views):

http://youtu.be/NNJS5zPhHTo

 

2007

Still, J. D., Dark, V. J., & Parkhurst, D. J. (2007). Viewpoint Invariant Object Features Attract Overt Visual Attention [abstract]. Poster presented at the meeting of the Vision Sciences Society, Sarasota, FL.

     Previous research suggests that salient image features attract overt shifts of attention when participants freely view complex artificial and natural scenes (Parkhurst, Law & Niebur, Vision Research, 2002). This evidence is consistent with the influence of a bottom-up stimulus-driven mechanism of attentional guidance. Attention to salient image features is a plausible default information selection strategy, especially in the absence of a well-defined task.

     Given that object recognition is necessary for most natural visual tasks, a plausible alternative default strategy is to preferentially select information likely to be important for object recognition. Object recognition depends in part on the presence of visual features that remain invariant across viewpoints (Biederman, Psychological Review, 1987). Thus, it is possible that viewpoint invariant features of objects will attract visual attention.

     To examine this possibility, the eye movements made by 12 participants freely viewing images of objects were recorded. Within each image, the Scale Invariant Feature Transform (SIFT) algorithm was used to identify highly invariant features (Lowe, I nternational Conference on Computer Vision, 1999) and the Saliency model was used to identifiy salient features (Itti, Koch & Niebur, IEEE Transactions on Pattern Analysis and Machine Intelligence, 1998). Because the predictions of the two models can be highly correlated, we selected images of 100 different objects from the Amsterdam Library of Object Images (Geusebroek, Burghouts & Smeulders, International Journal of Computer Vision, 2005) that maximally decorrelated the predictions.

     Both models performed significantly better than chance. The SIFT model performed signifcantly better than the Saliency model. These results suggest viewpoint invariant features of objects attract attention as reflected in eye movements. They also support the hypothesis that the default attentional selection strategy is biased to select visual features likely to be important for object recognition.

 

2006

Attentional Blink Posters:

Early eye fixations recorded during free viewing of photographs of natural objects were better described as regions with a high proportion of invariant features than as regions with highly salient features. This suggests that low level visual processing is biased towards visual features likely to be important for object recognition.

Tan, W. P., Still, J. D., & Dark, V. J. (2006). Measuring the AB Effect in an Irregular RVSP Stream [abstract].
Poster presented at the meeting of the Midwestern Psychological Association, Chicago, IL.

When the intervals between distractors in RSVP streams were irregular, attentional blink (AB) magnitude varied with lag (time between T1 and T2) but not number of distractors. Target identification was lower with irregular compared to regular intervals. Distractor regularity modulated AB; current AB models cannot account for this finding.

Tan, W. P., Still, J. D., & Dark, V. J. (2006). Attenuating the Attentional Blink by Including Salient Temporal Events in the RSVP Stream [abstract]. Poster presented at the meeting of the Psychonomic Society, Houston, TX.

We examined the effect on the attentional blink (AB) of changes in temporal dynamics among items presented in a rapid serial visual presentation (RSVP) stream. Manipulation of interstimulus intervals produced faster item presentation rates at different points in the stream. AB did not attenuate when the change occurred just prior to the onset of either the first target (T1) or the second (T2), or just after T2 offset. AB was attenuated when the change occurred just after T1 offset. AB attenuation did not vary with changes in T1 identification rate. These findings are latter varied with changes in item presentation rate. These findings are inconsistent with two-stage processing models of AB; however, they are consistent with attentional control models, in which AB occurs when transference of attentional control from T1 to T2 is inefficient. AB attenuation produced by rate change after T1 offset may indicate that the change facilitated disengagement of attention from T1.

 

2004

Undergraduate Independent Project Paper

Hands-Free Cellular Phones: Greatest Hidden Public Hazard on the Roadway

 

 

 

 

Return to JeremiahStill.info