Perceptual Learning and Consolidation Studied with Perfusion fMRI
Geoffrey Aguirre, M.D., Ph.D.
University of Pennsylvania School of Medicine, Philadelphia, PA
David Mahoney Neuroimaging Program
December 2004, for 4 years
Imaging Imagery While We Sleep, with Perfusion fMRI
University of Pennsylvania researchers will study what occurs in the awake and sleeping brain while people learn to recognize faces or buildings. The findings should help to explain the role of sleep in this essential learning function, called visual perception, and also may provide a basis for restoring visual perception capacity in people who have sustained a brain injury.
Prior research suggests that people have an initial period of fast improvement in learning to recognize a face or a building, followed by slow, step-wise, improvement seen following sleep. Improvement is associated with the fine-tuning of an internal (mental) representation of facial appearance. Using fMRI to provide a quantitative measure of cerebral blood flow over hours to days, the researchers will examine how the brain processes visual perception learning, and describe the role of REM (rapid eye movement) sleep in this process. (REM is the deepest stage of sleep.) They will image the neural activity of patients as they practice identifying faces and while they sleep (in the scanner). Then, investigators will undertake REM sleep scanning in participants who learn to discriminate pictures of either faces or buildings. Depending on which type of visual learning is taking place, the investigators hypothesize, specific parts of the brain will show increased neural activity, and REM sleep will play a role in the consolidation of this perceptual experience.
Significance: The findings of how REM sleep contributes to visual perception learning should provide new information on this vital brain function, and may provide a basis for devising rehabilitation strategies for people with brain injuries that resulted in diminished visual perception capacity.
Perceptual Learning and Consolidation Studied with Perfusion fMRI
Training on a visual discrimination task improves subject performance over several days. Behavioral studies of this perceptual learning suggest that there is an initial period of fast improvement in performance that occurs during the first training session, followed by step-wise enhancement of performance after additional training. There is evidence that rapid-eye-movement (REM) sleep is critical to this slow consolidation of learning over training periods. Behavioral studies have also characterized the systems-level changes in information processing that accompany improvement in performance. In subjects trained to discriminate faces, for example, improved performance appears driven by enhancement of the stimulus representation, as opposed to reduction of external or internal perceptual noise. Characterizing the neural basis of perceptual learning should help us understand the properties of rehabilitative therapy for recovery of visual function following brain injury.
The proposed experiments make use of a recently developed neuroimaging technique, perfusion functional magnetic resonance imaging (fMRI), that is ideally suited to the study of the long time-scale changes in neural activity presumed to accompany perceptual learning. By observing cerebral blood flow in subjects while they perform a perceptual learning task and while they sleep after training, I will test the following hypotheses: 1) different neural mechanisms underlie initial, fast improvement in task performance compared to subsequent slow improvement; 2) improved performance is manifest on a neural level by augmented responses to trained stimuli within category-specific cortical areas; and 3) step-wise improvements in performance are associated with reactivation of category-specific cortical areas during REM sleep.
Training on an object discrimination task improves subject performance over several days. Perfusion fMRI will be used to test the hypothesis that slow changes in regional neural activity accompany perceptual learning.
1. To characterize the neural correlates of the fast and slow phases of perceptual learning.
2. To characterize the neural basis of signal enhancement during perceptual learning.
3. To determine if perceptual learning induces neural activity within category-specific cortical areas during sleep.
Subjects will perform a perceptual learning task in which they learn to identify particular faces within visual noise. Their performance will be measured during four training sessions over 4-8 days. Perfusion fMRI scanning, which obtains whole-brain images of CBF every 6 seconds, will be performed during baseline, training, and transfer periods. The initial scanning session will also include high-resolution anatomical images and "localizer" tasks to identity the FFA and PPA in each subject. Data analysis will compare the time-course of signal change during initial and subsequent training sessions within the FFA and PPA as well as earlier visual cortical areas. The categorical and exemplar specificity of any training effects will be demonstrated by evaluation of the degree of evoked neural activity during the baseline and transfer sessions.
For the sleep studies, subjects will perform the training sessions in the evening, outside of the scanner, followed by a sleep period within the bore of the MRI scanner. Continuous electroencephalogram (EEG) and perfusion MRI data will be acquired, allowing the segregation of CBF images to drowsiness, successive stages of slow-wave sleep, and REM. Analysis will compare average CBF values for different stages of sleep within the FFA and PPA regions of interest between the populations trained in the face or house tasks. It will not be necessary to scan subjects during an entire night, as even brief (60-90 minute) naps induce step-wise improvements in perceptual performance.
Harris A. and Aguirre G.K. The representation of parts and wholes in face-selective cortex. J Cogn Neurosci. 2008 May;20(5):863-78 .
Wang Z., Aguirre G.K., Rao H., Wang J., Fernández-Seara M.A., Childress A.R., and Detre J.A. Empirical optimization of ASL data analysis using an ASL data processing toolbox: Asltbx. Magn Reson Imaging. 2008 Feb;26(2):261-9 .
Aguirre G.K., Detre J.A., Zarahn E., and Alsop D.C. Experimental design and the relative sensitivity of BOLD and perfusion fMRI. Neuroimage. 2002 Mar;15(3):488-500.
Sledge K., Olson I.R., Rao H., Wang J., Detre J.A., and Aguirre G.K. (2005). fMRI Responses of Face and Object Cortical Areas to Parametric Manipulations of Face Processing and Inversion. Paper to be presented at the Cognitive Neuroscience Society meeting, New York, New York.
Rao H., Dinges D.F., Censits D., DuRousseau D., Roalf D., Wang Z., Aguirre G.K., Detre J.A., and Wang J. Simultaneous EEG and ASL Perfusion fMRI during Resting and Mental Calculation: A Preliminary Study. ISMRM Annual Meeting. (2006)
Harris A. and Aguirre G.K. Effects of Parts, Wholes, and Familiarity on Face-Selective Responses in MEG. Journal of Vision (In Press)
Aguirre G.K. and Thomas A. Distributed representation of facial identity studied with fMRI. Paper presented at Cognitive Neuroscience Society Meeting, San Diego, CA (2006)