research highlights
Below are short highlights for some of the research taking place in the Perceptual Expertise Network, in some cases in collaboration with other members of the Temporal Dynamics Learning Center. Please contact investigators for more information.
A visual short-term memory advantage for faces
NSF-funded researchers from Vanderbilt University found that we can hold more faces than other objects in short term memory. The reason seems to be our expertise with faces, which allows us to encode them in a more efficient manner than other objects. But apparently, this expert skill requires time. Isabel Gauthier and Kim Curby found that when participants studied displays of faces or objects for only a brief amount of time (half a second), they could store fewer faces than objects such as watches and cars in visual short term memory. They believe this is because faces are more complex than the other objects and require more time to be encoded. But when participants were given additional time to encode the images (up to four seconds), an advantage for faces over objects emerged. Interestingly, only upright faces, with which we are most familiar, and not upside-down faces, show this advantage. The work conducted as part of the Temporal Dynamics Learning Center challenges previous models that assume the capacity of visual short-term memory is set in stone. Understanding the time constraints under which experts show the largest memory capacity may be useful in designing training protocols and software used in the workplace.
Curby, K.M. & Gauthier, I., (2007). A visual short-term memory advantage for faces. Psychonomic Bulletin and Review. 14(4): 620-8.
Investigators: Kim Curby (Temple University) and Isabel Gauthier (Vanderbilt University)
NSF Face Camp
Last summer over 100 children had the unique opportunity to learn about the science of human face recognition at the University of Victoria's NSF Face Camp. At camp, children, ages 7 through 12, were introduced to the psychological and biological principles of face processing through a series of fast-paced, interactive exercises. Face Camp kids participated in a range of activities from morphing their face with the face of their favorite celebrity to creating a computer sketch of the Face Camp burglar to practicing their face making abilities with Amigo, the clown. Dr. Jim Tanaka and his colleagues developed Face Camp as a special project for NSF's Temporal Dynamics Learning Center. The primary goal of Face Camp was to design an engaging science program on face recognition for elementary school aged children. However, the camp also gave researchers an opportunity to conduct basic research related to the development of face processes. Children at the camp participated in a cognitive experiment that compared their face recognition strategies to the strategies that they use to recognize non-face objects. In short, Face Camp is an innovative model that blends science education with scientific research where children learn about science of face recognition in a fun and engaging environment and at the same time, have a chance to contribute to a ongoing research project.
Investigators: Jim Tanaka (University of Victoria)
The Neural Basis of Audio/Visual Event Perception
Researchers from the Temporal Dynamics Learning Center are working as a team to explore how the brain puts together the sounds and sights of unfolding events (such as ripping paper or a bouncing ball). To better understand how such audio-visual stimuli come to be perceived as a single integrated event, Michael Tarr from Brown and his colleagues also at Brown, University of Colorado Boulder and UCSD created a functional magnetic resonance imaging (fMRI) paradigm in which participants view movies of events presented unimodally (just the movie or just the sound) or multimodally (congruent movie and sound, or incongruent movie and sound). The results to date reveal brain areas of strong activation within both visual and auditory processing areas, as well as robust multimodal integration areas in which activity for multimodal stimuli is greater than unimodal stimuli. They will also be looking to see if any brain area shows an effect of congruency (both semantic and temporal) within these specific regions. The team wishes to elucidate the computational mechanisms that "figure out" how a time-unfolding sound and visual action go together (integration) and hypothesized that these same brain mechanisms are the ones most likely to be sensitive to the fact that a particular sound does not match a particular event. Because fMRI is much better at telling us which brain area is active than when it is engaged, the team is planning to use the same basic research paradigm using a different method, event related potentials. Finally, other members of the team are working towards computational models capable of accounting for how disparate sensory information is bound into coherent percepts.
Investigators: Michael J. Tarr, David Sheinberg, Tim Curran, Ginny De Sa; Laurie Heller (NSF-funded, but not a TDLC participant) Students: Jean Vettel, Julia Green
The dynamics of manipulable objects: semantic, perceptual and motor interactions revealed by the hand in flight
Motor representations form part of our knowledge of everyday objects like calculators and spray-cans. There is increasing evidence that such representations, determining our ability to hold and use manipulable objects, are invoked dynamically during language and perceptual tasks. For example, spatial attention is modulated in unexpected ways when observers view graspable objects. The computational role played by higher level motor systems in cognition is therefore of fundamental interest. I have developed a novel experimental method to measure the evocation of hand actions to words, objects or sentences in real time. This approach requires subjects to carry out speeded reach and grasp actions on an eight-element response apparatus. Each element affords a unique action and subjects learn to produce a given manual action in response to a visual cue like color. We then measure the influence of an object, word or sentence (a priming event) on reach and grasp performance, time-locked to the visual cue, yielding evidence on the nature and time course of the motor representations evoked by the priming event. Until now, these measurements have been restricted to lift-off and transport time and so we lack crucial information on the actual dynamics of hand movement in response to perceptual or linguistic priming events. Recently, I have begun a collaboration with Howard Poizner to obtain detailed information on the parameters of manual actions dynamically altered by higher level contextual influences. The first phase of this project requires the construction of detailed spatiotemporal maps of grasping movements for each of the eight elements of the response apparatus. Multiple points on the hand will be tracked in 3D space at 120 Hz. as subjects are cued to reach for and grasp these elements. We will examine how grasping parameters, such as hand aperture and finger abduction, unfold over time depending on the object to be grasped. Of primary interest in this collaboration is the nature of control processes that determine a particular hand representation in a given task context.
Evocation of functional and volumetric gestural knowledge by objects and words
Cognition, Volume 106, Issue 1, January 2008, Pages 27-58
Investigators: Daniel N. Bub, Michael E.J. Masson and George S. Cree
Learning to become an expert
While all people make mistakes, experts learn best from the error in their ways. Funded by NSF, Dr. Jim Tanaka of the Temporal Dynamics Learning Center, his colleague Dr. Clay Holroyd and their students measured electrophysiological brain activity while participants learned to categorize perceptually similar "blobs" into blob families. When participants committed a categorization error, they were given a negative feedback message, which, in turn, elicited a brain potential called the Error-Related Negativity (ERN). For participants who were able to learn the blob families, the onset of the ERN brain potential shifted from the time of feedback to the time of response. This finding suggests that the expert categorizers were aware of their mistakes even before the feedback information was given. Interestingly, participants who were unable to learn the categorization task displayed no shift in their brain activity and remained reliant on the external feedback throughout the experiment.. This research has real world implications in the classroom as teachers could evaluate the type of feedback that might be most effective for student learning.
Investigators: Jim Tanaka (University of Victoria)
Musical brain sees faster
It may come as no surprise that the amount of practice that goes into learning to play a musical instrument changed the brain, but how much of the brain becomes engaged in the simple act of perceiving a single note did surprise two members of the NSF-funded Temporal Dynamics of Learning Center, Dr. Isabel Gauthier and her graduate student Yetta Wong, from Vanderbilt University. They used brain imaging to measure brain activity while people with varying degrees of musical experience (from none to extensive) made simple judgments about notes, letters and other shapes, shown one at a time, on a computer screen. Despite the fact that the task was far removed from the complexity of musical performance, several brain areas were found to be more engaged for notes than other images, in individuals who could read musical notation more than in novices. Several of these brain areas were not visual, but rather motor or auditory, suggesting that seeing a single note is sufficient to recruit a wide network of brain areas that have been specialized by the multifaceted experience of learning to translates notes into music. Finally, the researchers were able to demonstrate that activity in several of these non-visual regions of the brain could be used to predict performance on a test outside of the scanner that measured how fast participants could perceive note sequences. This suggests that visual judgments in experts may be performed by a distributed network of areas that have been fine-tuned by a rich learning experience. The work may lead to the development of multimodal training protocols to improve expert performance.
Investigators: Yetta Wong (Vanderbilt University), Isabel Gauthier (Vanderbilt University)
Knowing An Object Is There Does Not Necessarily Mean You Know What It Is
How do we recognize objects in our complex world? According to one prevailing view, when our visual system parses a visual scene into individual objects, at the same time it categorizes those objects in familiar categories such as "dog", "car", or "chair". In other words, as soon as you know an object is there you know what it is. New work by NSF-funded investigators from the Temporal Dynamics Learning Center overturns this view. One challenge for the scientific study of object recognition is that it happens so very fast: We can recognize objects in less than 1/10 of a second. Michael Mack, Isabel Gauthier, Javid Sadr, and Thomas Palmeri used a variety of experimental techniques to examine in detail how object recognition unfolds over time, from as little as 1/100 of a second of exposure. While under some circumstances, the time-course of detecting an object and categorizing an object are the same, but when the objects were upside-down or blurred, people could now detect the presence of objects before they could categorize them. Detecting an object and recognizing an object as a member of a category are separate decisions that rely on different visual information. This basic result has general implications for understanding the mechanisms of object recognition and how they might change over learning. Future work could investigate the time-course of categorization experts in different domains (e.g., radiologists and baggage screeners) and suggest the best way to present information for different types of decisions.
Mack, M., Gauthier, I., Sadr, J., & Palmeri, T.J. (in press). Object detection and basic-level categorization: Sometimes you know it is there before you know what it is. Psychonomic Bulletin & Review.
Investigators: Michael Mack, Isable Gauthier, Tom Palmeri (Vanderbilt University)
Making Collaborations Work
How do you create successful scientific collaborations that bring together methods, tools, concepts, and data from a wide range of disciplines and how do you know whether these collaborations have worked? When is the whole truly greater than the sum of its parts? Members of the TDLC presented a symposium on scientific collaboration at the Cognitive Science Society this summer in Nashville that tackled these tough questions. Collaborative interdisciplinary research is becoming increasingly attractive (and perhaps unavoidable) as the questions being tackled become more complex. Yet scientists vary greatly in their experience with collaborations and opportunities are rare to discuss how to make the most of collaborations, what factors make collaborations work or not work, and how to measure the success or failure of collaborations. This symposium examined successful models in which a number of scientists have bridged the extradisciplinary gap across disciplines and the intradisciplinary gap across methods and concepts within a discipline. It turned an analytic lens on the value of collaborative networks and teams of researchers and reported on the logistic, social, and scientific processes that drive intellectual growth and research and reflected on the social factors that can make or break a successful collaboration. It also discussed how to quantitative measure collaboration and interdisciplinarity at both small and large scales using sophisticated scientometric and bibiometric techniques. In a unique combination, the history of the Perceptual Expertise Network was related from both the perspectives of the scientists and the funding agency that originally supported its creation several years ago. The TDLC was highlighted for its efforts to expand the collaborative research network model to a network of networks. This symposium examined collaborations as entities worthy of scientific study by cognitive scientists in and of themselves, collaborations as engines for novel interdisciplinary research in cognitive science, and measures of collaborative activities in cognitive science as data to be modeled and analyzed.
