Published online by Cambridge University Press: 20 May 2010
Humans can effortlessly recognize objects in a fraction of a second despite large variability in the appearance of objects (Thorpe et al. 1996). What are the underlying representations and computations that enable this remarkable human ability? One way to answer these questions is to investigate the neural mechanisms of object recognition in the human brain. With the advent of functional magnetic resonance imaging (fMRI) about 15 years ago, neuroscientists and psychologists began to examine the neural bases of object recognition in humans. Functional magnetic resonance imaging (fMRI) is an attractive method because it is a noninvasive technique that allows multiple measurements of brain activation in the same awake behaving human. Among noninvasive techniques, it provides the best spatial resolution currently available, enabling us to localize cortical activations in the spatial resolution of millimeters (as fine as 1 mm) and at a reasonable time scale (in the order of seconds).
Before the advent of fMRI, knowledge about the function of the ventral stream was based on single-unit electrophysiology measurements in monkeys and on lesion studies. These studies showed that neurons in the monkey inferotemporal (IT) cortex respond to shapes (Fujita et al. 1992) and complex objects such as faces (Desimone et al. 1984), and that lesions to the ventral stream can produce specific deficits in object recognition, such as agnosia (inability to recognize objects) and prosopagnosia (inability to recognize faces, Farah 1995). However, interpreting lesion data is complicated because lesions are typically diffuse (usually more than one region is damaged), typically disrupt both a cortical region and its connectivity, and are not replicable across patients.