Background: Previous investigations in young adults have demonstrated the remarkable speed with which individuals can accurately process and discriminate different classes of objects within visual scenes, and suggest that such rapid categorization is mediated primarily by feed-forward mechanisms within the ventral visual processing stream. The loss of ef- fective connectivity within the visual system in AD due to disruption of corticocortical projections should produce marked deficits on tasks re- quiring fast, efficient communication across processing stages within the ventral stream. Given the high demands that ultra rapid categorization tasks place on feed-forward connections within visual cortex, these tasks may prove to be particularly sensitive markers of AD pathology in even the earliest stages of the disease. Methods: Groups of healthy young, healthy elderly, and patients diagnosed with amnestic mild cognitive im- pairment (MCI) performed a rapid classification task in which partici- pants viewed briefly presented scenes containing natural and artificial objects, and were asked to press a button as quickly as possible if they detected an animal within the scene. Stimulus duration was individually thresholded to produce about 80% accuracy under unmasked conditions. Task difficulty was manipulated by presenting a masking stimulus after the scene at varying temporal intervals (ISI: 14ms, 35ms, no mask), and by varying the focal distance of the target object in the scene (i.e., head, close body, mid-body). Results: Young and elderly controls dis- played similar patterns of performance, displaying good discrimination under no mask conditions; decreased but still robust discrimination at 35ms mask (although elderly performed significantly worse than young); and near-chance performance at 14ms mask). In contrast, MCI patients displayed robust classification performance under the no mask conditions but performed at near-chance levels under both masked conditions. Con- clusions: Results suggest that, despite neural changes associated with normal aging, feed-forward mechanisms within the ventral visual stream remain relatively intact in healthy elderly. In contrast, impaired performance of the MCI group may reflect subtle disruptions of cortico- cortical projections within the ventral stream even at the preclinical stage of the disease. These findings serve as a critical first step in the de- velopment of a sensitive tool for early detection of neurocognitive changes in AD. P2-256 PERCEPTUAL BINDING OF AUDIOVISUAL SPEECH INFORMATION IS DISRUPTED AT THE PRECLINICAL STAGE OFALZHEIMER’S DISEASE Elena Festa, Andrew Katz, Brian Ott, William Heindel, Brown University, Providence, Rhode Island, United States. Contact e-mail: elena_festa@ brown.edu Background: The McGurk effect is a classic phenomenon in which au- ditory speech perception is altered by the presence of conflicting visual speech information (lip movements), and reflects audiovisual binding at early stages of speech processing: The effect appears to depend on syn- chronized activity between auditory and visual cortices mediated by feed- back from superior temporal sulcus. The disruption of corticortical connections within these regions in early-stage AD should produce a marked disruption of the McGurk effect. Indeed, a recent study using phonemic stimuli found a reduced McGurk effect in AD patients relative to healthy elderly. This study investigates whether disrupted audiovisual binding is detectable even in preclinical AD patients, and whether perfor- mance measures could serve as effective diagnostic markers of the dis- ease. Methods: Groups of healthy young, healthy elderly, and patients diagnosed with amnestic mild cognitive impairment (MCI) watched movie clips of an individual mouthing either words or non-words with the voice providing either consistent or inconsistent audio information, and were asked to report the initial consonant sound of each stimulus they heard. The strength of the perceptual binding was manipulated by using different consonant combinations and lexical relations between the audio and visual information that elicit different degrees of audiovi- sual binding in healthy young adults. Perception of the isolated auditory and visual components of stimuli was also assessed following the audio- visual binding task. Results: Both healthy groups displayed robust McGurk effects at levels consistent with the variation in stimulus strength (although elderly showed overall significantly more perceptual binding than young). Despite showing normal perception of the unimodal stimuli, the MCI patient group displayed significantly reduced McGurk effects for the perceptually weaker (but not stronger) audiovisual stimuli, relative to healthy elderly. Conclusions: Results suggest that neural changes associ- ated with normal aging may increase the perceptual binding of audiovi- sual speech information. The demonstration of reduced audiovisual binding in the MCI group may reflect subtle disruptions of corticocortical projections within unimodal and heteromodal cortexes even at the pre- clinical stage of the disease. These findings serve as a critical first step in the development of a sensitive tool for early detection of neurocogni- tive changes in AD. P2-257 VALIDATIONOF THE MONTREAL COGNITIVE ASSESSMENT – SPANISH VERSION TEST (MOCA- S) AS A SCREENING TOOL FOR MILD COGNITIVE IMPAIRMENTAND MILD DEMENTIA IN BOGOT  A, COLOMBIA Laura Gil, Juan G omez Vega, Carolina Ruiz de Sanchez, Felipe Pretelt Burgos, Pontificia Universidad Javeriana, B ogota D.C., Colombia. Contact e-mail: gilcita@hotmail.com Background: MoCA test was developed as a brief instrument to screen co- gnoscitive impairment. This is the first validation in Latin America of MoCA in Spanish (MoCA-S) developed in Colombia. Methods: Study of concordance by conformity to evaluated MoCA test in comparison with gold standard (Diagnostic in consensus of interdisciplinary assessment by Clinic of Memory). Assessment of psychometric properties of MoCA-S with interrater reliability, internal consistency and convergent validity. Results: The internal consistency measure with a Cronbach index was 0,863 value similar to that obtained by the original designers of the MoCA test (Nasreddine et al 2005). The interrater reliability was almost perfect with a concordance coefficient of Lin¼0.903 confidence interval (CI) ¼ 95% (0,78-1.00) and the convergent validity was evaluated by means of the high correlation with the MMSE and the concordance with the gold standard of 85% patients perfectly classified by the test with a nomial kappa index of 0,7 (IC: 0,60-0.79).256 subjects were evaluated (64.4% women), of which 138 were patients and 118 healthy individuals. Patients were divided into two subgroups: MCI¼ 44 and MD ¼ 94. Subjects with cognitive im- pairment averaged MoCA-S ¼ 17.5 and MMSE ¼ 25.3. The healthy group averaged MoCA-S ¼ 25.39 and MMSE ¼ 28.7. MCI subgroup averaged MoCA-S ¼ 20.9 and MMSE ¼ 27.5 (normal range). MD subgroup showed an average of MoCA-S ¼ 15.4 (abnormal range) and MMSE ¼ 24.4 (normal range). The difference of scores in the subgroups is statistically significant Poster Presentations: P2 P452