Visual Recognition Memory in Alzheimer's Disease: Repetition-Lag Effects

Abstract

There is considerable evidence that visual recognition memory is largely affected by Alzheimer's disease (AD). Deficits might concern the forming, maintaining, and matching of the memory representation of the visual stimulus, especially when long interitem lags occur. The aim of the present study was to assess the effect of repetition lag on picture identification in mild- and moderate-AD patients, as well as in elderly controls. Participants underwent an old/new paradigm. To manipulate the temporal gradient, short and long lags were introduced between the first and second presentations. Pictures were presented at different levels of spatial filtering, following a coarse-to-fine order. This allowed for the measurement of the amount of physical information required for the identification of stimuli as a function of prior exposure and repetition lag. In the elderly, the magnitude of repetition priming did not differ as a function of interitem lag. Instead, repetition-lag effects interacted with dementia severity, and the capacity to retain memory traces for longer intervals worsened as the disease progresses. Current findings suggest that severe cortical degeneration may render AD patients unable to maintain their perceptual memories, and that dementia severity is a critical variable in the visual recognition memory assessment.

Keywords:

• Alzheimer's disease
• memory
• object identification
• priming
• repetition lag

Notes

¹The stimuli were blurred, removing ranges of spatial frequencies from the spectrum of the image. This filtering process created a multiresolution representation of the scanned images. The different resolutions were obtained by means of a digital filter applied to the bidimensional array representing the original image scanned at the resolution of 300 dpi. The multiresolution filter that was selected was the gaussian mask that performs a low-pass filtering. Given an image resolution of 300 dpi, the maximum reproducible image frequency, f _max, is equal to 150 cycles/inch. Hence, in order to get the multiresolution representation, we filtered the original image using eight different filters characterized by different bandwidths in the range 0 to f _max.

²In order to balance the number of intervening stimuli across subjects, in the identification phase, all the 39 stimuli from the study phase were presented. However, in the identification phase, incorrectly named or non-recognized pictures were not included in the analyses.

³Note that each stimulus was displayed at all of the nine resolution levels regardless of the level at which it was actually identified (Viggiano & Kutas, 1998, 2000). For this reason, each interval between first and second stimulus presentations includes (i) the delay between the first presentation of the stimulus and the end of the study phase, and (ii) the delay between the beginning of the identification phase until the second stimulus presentation, multiplied by 9 (levels of filtering).