Abstract :
[en] Comparison of representational similarity of episodic memory traces at encoding and retrieval in younger and healthy older adults
Relevant questions of advanced functional Magnetic Resonance Imaging (fMRI) aging studies are lately rather focusing on the investigation of neuronal activity patterns across voxels between younger and older adults. Multivariate Pattern Analyses (MVPA) approaches are gaining ground for the investigation of neural activity by employing voxel-by-voxel variability assessment to explain cognition. One of the MVPA approaches namely Representational Similarity Analysis (RSA) is a relatively novel technique to assess neural pattern similarity across distinct experimental conditions.
For our episodic memory/recognition study aiming to disentangle memory performance differences, if any, between older and younger adults, we have used RSA among age groups and conditions. Fifty-five young (28 men; M age = 23.47 years old; SD = 3.12; range = 18-30) and 45 older (19 men; M age = 67.37 years old; SD = 4.9; range = 60-75) participants were selected from a larger cohort which underwent an fMRI recognition memory task / episodic memory task.
The procedure began with an incidental learning paradigm. During encoding phase participants were successively presented once (hard condition), or twice (easy condition) pictures representing objects, and were asked to determine whether the presented object would fit in a shoebox. During the retrieval phase, participants were presented randomly shuffled previously seen and never seen/new pictures and were tasked to determine if the object was previously presented [remember (with some contextual detail) / know (without contextual information) /new (unstudied picture) paradigm].
Employing MVPA searchlight technique, encoding and recognition neural pattern maps were created for each subject/each picture individually, for both phases. Further, we have computed pattern similarity during encoding and retrieval phases for a single image (item level). Then, we have computed pattern similarity of one image at encoding phase and the average activity for the remaining images during recognition phase (set level). Our design matrix consisted of 100x100 elements (similarity values) at item and set levels. Further, neural similarity patterns were analyzed at the level of remember/know responses, item/set level, respectively groups (older vs younger) level. Statistical analyses were conducted by using SPM12 ANOVA 2(Group older/younger) X 2(Level=Set/Item). Additional ANOVAs were carried out for remember/know responses.
At the brain level, larger encoding-recognition similarity patterns are observed in younger by comparison to older in brain regions processing visual characteristics (occipital pole)
and embodied cognition (postcentral area).
At remember/know level, results denoted reduced neural similarity patterns among older adults.
At item/set level, we observed greater neural similarity patterns among younger adults.
Globally, between groups we observed higher similarity patterns among younger adults when compared to older adults.
Our findings suggests that older adults’ neural similarity patterns were reduced when compared to younger adults.
Lower encoding/recognition similarity values constantly observed among older adults indicate less specific reactivation of individual memory traces for pictures between encoding and recognition.
One possible explanation of the observed aging effect on recognition/performance is that poor visual and sensorimotor encoding processes results in less distinctive recognition memory traces.
