Here I am looking smug (and a little sweaty) after just
scampering up the 288 steps of the Batu Caves Hindu
Shrine outside Kuala Lumpur. The grey limb hanging down
in the upper right corner of the photo is a monkey's leg.
1. Shape and Object Recognition
2. Scene Perception
3. Face Recognition
4. Perceptual and Cognitive Pleasure
Our lab is currently engaged in research in four domains of cognitive neuroscience: a) Shape and Object Recognition, b) Scene Perception, c) Face Recognition, and d) Perceptual and Cognitive Pleasure. We attack these problems with a broad range of methodologies: behavioral testing both in LA and undeveloped regions of our planet, fMRI, EEG, eye tracking in humans and, in collaboration, TMS, single-neuron recordings in the macaque, behavioral testing with pigeons, and, in the coming months, human patients undergoing surgery.
Shape and Object Recognition In a fraction of a second -- from a single visual fixation -- humans are able to comprehend novel images of objects, often under highly degraded and novel viewing conditions. To account for this extraordinary capacity, we have proposed that objects are represented as an arrangement of simple, convex, viewpoint-invariant 3D shape primitives, termed geons, such as bricks, cylinders, wedges, and cones and their 2D projections, that serve to distinguish visual entities at a basic (or entry) level, so that a given image can be determined to be that of a chair, fork, or penguin (Biederman, 1987). The geons can be distinguished by properties of edges that are invariant with orientation in depth (such as straight vs. curved contours) so representations distinguished by geons possess the same invariance. As long as two or three geons in their specified relations can be extracted from the image, entry-level classification will almost always be successful despite drastic variations in the object's silhouette, specific contours, and occlusion of large regions of the object.
Scene Perception Objects tend to occur not by themselves but in concert with others, in scenes. We not only can identify the objects that make up a scene in a glance, but how they are related to each other. The neural basis of the coding of invariant relations, so that if X is above Y that relation would remain true no matter how X and Y were translated. This problem of the representation of relations is one of the great challenges of cognition, being central not only to scene perception but to language as well. We can understand, for example, the relation changes in “John gave the book to Lisa.” We have an active program of research in which we are investigating where in the brain relations are coded and how changes in relations are represented.
Face Recognition We have been studying prosopagnosia, the inability to individuate faces while retaining a normal capacity for judging the age, sex, and attractiveness of a face and individuating highly similar objects. Whereas an arrangement of geons appears to suffice for object recognition, a representation that preserves the metrics implicit in the original cortical activation may suffice for face individuation, which his why we can be highly accurate in distinguishing Tom Cruise from John Travolta but not be able to articulate how (Yue, Tjan, & Biederman, 2006).
Perceptual and Cognitive Pleasure
With every visual fixation and every decision as to what to read or what movie to see or conversation to keep going or wanting a room with a view, we are driven to seek out new information. What is the neural basis for such behavior? We have been exploring the extent to which the answer lies in a gradient of opioid receptors in the cortex (Biederman & Vessel, 2006). The receptors are sparse in early sensory areas but grow in density with each succeeding stage and reach their maximum in association cortex, where perceptual inputs are interpreted and understood. The simple idea is that we try to maximize the rate of opioid release and we can achieve that by having rich, highly interpretable experiences--experiences that generate lots of associations. However, once we have an experience, adaptation (based on competitive learning or “neural Darwinism,”) sets in, caused by the few strongly activated neurons inhibiting the vast number of weakly activated neurons leading to less opioid release. So its “been there, done that.” We are thus infovores, always seeking out novel but richly interpretable experiences. Stay tuned.