![]() |
||||||||||||||||||||
|
Spring 2008 Colloquia |
|||||||||||||||||||
| Thursdays |
4:00-5:30 |
N119 Elliot Hall - campus map | ||||||||||||||||||
February 14 |
![]() |
|
||||||||||||||||||
| The neocortex stores different items of long-term knowledge using distributed, partially overlapping representations in neural networks. This coding scheme affords benefits such as efficient storage of many items and enhanced generalization abilities. However, potential costs of this scheme have not been well examined. In particular, we hypothesized that strengthening a set of visual shape representations can detrimentally affect other representations with which they overlap. I will report behavioral and neuropsychological evidence for this impairment effect. Then, I will report functional magnetic resonance imaging (fMRI) evidence that cortical representations undergo strengthening or relearning with each use. These changes not only enhance the subsequent ability to use that information, but they also impair the ability to use other information, with the effect that increased neural activity is required to reestablish the impaired representations. Neurocomputational models of these effects provide a mechanistic hypothesis for their origins, and electrophysiological recordings corroborate the relearning interpretation of the fMRI results. These findings indicate that the ubiquitously observed neural repetition effect (lower neural activity elicited by repeated than by non- repeated items) may be reevaluated. It likely does not reflect changes solely to the neural representations of repeated stimuli (e.g., adaptation or suppression), but rather, the necessity to reestablish representations of non-repeated stimuli that previously have been weakened. |
||||||||||||||||||||