The Perceptual Expertise Network: Innovation on Collaboration
By Isabel Gauthier and Danielle D. Brown
A Discipline that Requires a New Collaborative Model
A cognitive neuroscientist explaining to her grandmother the nature of her field could describe how it relies on different techniques, such as measuring response times and accuracy from undergraduate students, recording brain activity from children, human adults and from animals or building computer models of how humans solve complex problems like recognizing objects or faces. It is hoped that evidence from these different techniques will converge and help us understand the nature of the relationship between the mind and the brain. In reality, very few researchers or laboratories can master all of these techniques, and progress in cognitive neuroscience has relied heavily on collaborations. Most scientists today utilize long distance collaborations that typically depend on phone calls and email. A few years ago, a group of cognitive neuroscientists took this collaborative approach to a new level when they established the Perceptual Expertise Network (PEN).
In 2000, Isabel Gauthier learned about a new program from the James S. McDonnell Foundation called "Bridging Brain, Mind and Behavior." The Foundation targeted inter-disciplinary efforts and recognized the difficulties of conducting highly collaborative and innovative research using traditional funding sources. Gauthier and Michael Tarr, her PhD mentor, had been looking for a way to bring together a diverse group of colleagues working on issues relevant to those explored in Gauthier's dissertation, completed in 1998. Her thesis focused on a simple question: why does face recognition appear to be special? There is good evidence that faces are processed in a more "holistic" manner than other objects and that there are specialized neural substrates in the brain for their processing. When those are damaged, individuals can become selectively impaired in face recognition. Gauthier and Tarr tested the hypothesis that it is in fact our expertise with faces that accounts for these effects, and they found evidence that expert observers for non-face objects exhibited similar behavior and neural patterns to those found for faces (Gauthier & Tarr, 1997). This was not a completely new idea. But their work was unique in demonstrating that expertise can be achieved within a few days of training, which makes it possible to study its development experimentally in the laboratory. At the time PEN was created, Gauthier and Tarr had already teamed up with many of the current PEN members (see Table 1) to work on specific projects: however, within PEN several of these individuals who had not met before have now moved on to develop their own collaborations (see Figure 1).
PEN was founded on two principles: First, to do science in a question-motivated framework rather than one motivated by techniques. PEN members do not ask themselves what should be the next fMRI or ERP experiment, just because those happen to be the techniques with which they are most familiar or in which they have invested their lab resources. Instead, they brainstorm, free from such technical constraints, and later capitalize on the diversity of their group and choose the tool that best suits the question. Second, PEN members believe that students can distinguish themselves by grounding their education in a highly collaborative experience. Because good cognitive neuroscience requires collaboration, individuals who master its challenges early in their training should be better equipped to succeed.
Developing a New Collaborative Model
PEN consists of approximately 30 people, including principle investigators, post-doctoral fellows, graduate students and undergraduate research assistants at 7 institutions around the US and Canada. An administrative assistant based at Vanderbilt coordinates the activities of the meeting. The entire group meets in person twice a year, generally with two outside researchers invited to present their work to the group. PEN workshops are unlike most scientific meetings, in fact they all begin with a warning to invited guests: expect to be interrupted--a lot. New results are presented, studies are proposed and concepts are debated - and much of it occurs over dinners and extended lunchtimes where members are encouraged to form smaller groups to discuss specific projects. In the first workshops, PEN members discussed definitions of perceptual expertise, reviewed empirical markers of expertise (i.e. those effects, behavioral or neural, that reveal expert performance), identified some empirical holes in the expertise literature, and discussed the importance of computational modeling in the group's common efforts. As the group matured, the discussion has progressed to encompass new research questions. What is the role of semantic knowledge in the acquisition of visual expertise? How can expertise with letters be compared to expertise with faces or other objects like cars and birds? Does the issue of the overlap (or non-overlap) of the neural substrates or face and object processing really address the question of whether faces are processed by a modular system? The success of the enterprise depends on the participation of students and post docs, because they are best positioned to turn a new question into an entire research program.
The group also holds regular joint lab meetings using speakerphones and shared PowerPoint files, to keep everybody in touch between face-to-face meetings. These teleconferences allow a more informal discussion and spark follow-up discussions in individual labs and on the PEN website's Bulletin Board. Other PEN activities have included a workshop on modeling organized by graduate students from different labs and a course on fMRI methods prepared by a PEN post doc. Recently, PEN members introduced this new collaborative model at the 2003 Cognitive Neuroscience Meeting. Four PEN faculty presented a symposium on the research conducted in the network. The symposium was unique in many ways: first, each speaker presented not only research from their own lab but discussed many findings from other PEN members; second, the four talks were conceived as parts of a whole. They were practiced together at a PEN workshop to get feedback from the entire group, and again in a conference call before the meeting. It is rare for a scientific audience to hear speakers who have made a concerted effort to deliver a coherent message that covers many aspects of the same field, but feedback at the conference indicated that it was very well received.
Examples of Collaborative Projects
Dozens of new research projects are currently underway within PEN, several of which would have not been possible without the network, and many spearheaded by our students. For example, Kim Curby, a graduate student in the Gauthier lab, had been working on a new paradigm to test interference between car and face expertise in subjects who are experts in both domains. The idea was to gather evidence that face and expert object recognition may not function independently. The paradigm is a dual-task in which holistic processing (HP) for both cars and faces is measured, while observers are required to process both categories at once. Results showed that the level of interference between HP for each category was a function of a subject's car expertise. As HP for cars increased, HP for faces was reduced in this dual task, demonstrating that they are not functionally independent. When presented to the group, a new question arose: when during processing does this interference occur, was it perceptual or did it occur later, maybe in the working memory stages necessary to perform the task? Gauthier and Curby had access to fMRI but this technique does not have the temporal resolution to resolve this question. A collaboration with Tim Curran (U.C. Boulder) developed as it became clear that ERPs would be better suited for this question. The ERPs results provided clear evidence for interference between car and face processing at the perceptual level: there was a significant correlation between the interference produced by car expertise and the amplitude of the earliest face-sensitive potential called the N170 (Gauthier, Curran, Curby, & Collins, 2003). Along similar lines, Bruno Rossion and colleagues in the Tarr lab at Brown University designed a different interference paradigm using face flankers shown around novel objects called Greebles. Following expertise training with Greebles, the N170 potential in response to faces was substantially decreased when it was preceded by a central Greeble, but not when preceded by a control object. This further supported the conclusion that face perception relies on processes common to expertise with other categories. In the spirit of PEN, these first studies of interference are spawning other projects: Curran and Rossion recently replicated this flanker interference effect with car experts, and Curby and Gauthier are working on an fMRI study to investigate where in the visual system such interference occurs.
Another example of PEN collaboration is one that has some clinical applications. Years before PEN, Bob Schultz recruited Gauthier to help out on a neuroimaging study of face processing in people with Autism. Autism and a closely related condition, Asperger Syndrome, are characterized by impairments in social functioning and interactions. People with Autism also have some difficulties with face recognition. The three-year study led to the first report of abnormal specialization for faces in the brain of individuals with this developmental disorder (Schultz et al., 2000). Subsequently, Schultz developed an interest in the study of expertise, because of the possibility that a lack of face expertise could explain the abnormal specialization for faces in the autistic population. Upon joining PEN, Schultz partnered with Rossion to study the N170 potential in people with autism, simultaneously continuing his earlier collaboration with Gauthier by investigating the behavioral aspect of face perception deficits in autism using a computerized face parts-whole faces assessment battery. Schultz and Jim Tanaka (University of Victoria) have since teamed up to create a computer-based training program, "Let's Face It!", meant to develop and sharpen perceptual skills involved in the recognition of facial identity and expression. In effect, they seek to artificially create expertise for faces in children with Autism who may not have developed this skill naturally.
Looking to the Future
The success of PEN, as measured by increased funding of the labs involved, their publications, and the volume of ongoing collaborative studies, can be mainly attributed to the involvement of students and post docs in all of the network's activities. Thus it is likely that the impact of PEN will be exponential as a new generation of cognitive neuroscientists is trained in a novel tradition of interdisciplinary collaboration. The network looks forward to graduating the first crop of PEN trainees and beginning anew with the next generation of students interested in approaching questions about the mind and brain in a different and innovative manner.
Gauthier, I., Curran, T., Curby, K. M., & Collins, D. (2003). Perceptual interference supports a non-modular account of face processing. Nat Neurosci, 6(4), 428-432.
Gauthier, I., & Tarr, M. J. (1997). Becoming a "Greeble" expert: exploring mechanisms for face recognition. Vision Research, 37(12), 1673-1682.
Schultz, R. T., Gauthier, I., Klin, A., Fulbright, R. K., Anderson, A. W., Volkmar, F., et al. (2000). Abnormal ventral temporal cortical activity during face discrimination among individuals with autism and Asperger syndrome. Archives of General Psychiatry, 37, 331-340.
|Carnegie Mellon University||Neuropsychology|
|Daniel Bub (PI)||University of Victoria||Neuropsychology|
|Garison Cottrell (PI)||University of California
at San Diego
|Neural Nets, fMRI|
|Tim Curran||University of Colorado
Chairs the Network
|Vanderbilt University||Psychophysics, fMRI|
|Thomas Palmeri||Vanderbilt University||Psychophysics,
|Robert Schultz||Yale Child Study Center||Autism, fMRI, ERPs|
|David Sheinberg||Brown University||Neurophysiology|
|James Tanaka||University of Victoria||Psychophysics, ERPs|
|Michael J. Tarr||Brown University||Psychophysics, EPRs, fMRI, Computer Graphics|
Schematic of ongoing collaborations within the Perceptual Expertise Network.
Each set of colored lines represents a different collaborative project between PEN labs. A number of other collaborations are taking place with PEN post docs who have now left the lab, or PEN guests. PEN projects that involve a single PEN lab are not represented.
About the Authors
Isabel Gauthier received her B.S. in Psychology from Université du Québec à Montréal and her PhD in cognitive psychology from Yale University in 1998. She is currently an Assistant Professor in the Department of Psychology and the Vanderbilt Vision Research Center at Vanderbilt University. She heads the Object Perception Laboratory at Vanderbilt University and chairs the Perceptual Expertise Network.
Danielle D. Brown received her B.S. in Biology from Cornell University. She is currently a research assistant in the Object Perception Laboratory at Vanderbilt University and is moving on to start a graduate degree in Behavioral Ecology in the fall of 2004.