Semantic processing in older adults is associated with distributed neural activation which varies by association and abstractness of words

Abstract

The extent to which the neural systems underlying semantic processes degrade with advanced age remains unresolved, which motivated the current study of neural activation on functional magnetic resonance imaging (fMRI) during semantic judgments of associated vs. unassociated, semantic vs. rhyme, and abstract vs. rhyme word pairs. Thirty-eight older adults, 55-85 years of age, performed semantic association decision tasks in a mixed event-related block fMRI paradigm involving binary judgments as to whether word pairs were related (i.e., semantically associated). As hypothesized, significantly greater activation was evident during processing of associated (vs. unassociated) word pairs in cortical areas implicated in semantic processing, including the angular gyrus, temporal cortex, and inferior frontal cortex. Cortical areas showed greater activation to unassociated (vs. associated) word pairs, primarily within a large occipital cluster. Greater activation was evident in cortical areas when response to semantic vs. phonemic word pairs. Contrasting activation during abstract vs. concrete semantic processing revealed areas of co-activation to both semantic classes, and areas that had greater response to either abstract or concrete word pairs. Neural activation across conditions did not vary as a function of greater age, indicating only minimal age-associated perturbation in neural activation during semantic processing. Therefore, the response of the semantic hubs, semantic control, and secondary association areas appear to be largely preserved with advanced age among older adults exhibiting successful cognitive aging. These findings may provide a useful clinical contrast if compared to activation among adults experiencing cognitive decline due Alzheimer's, frontal-temporal dementia, and other neurodegenerative diseases.

Keywords: Aging; Neural activation; Semantics; fMRI.

Fig. 1

fMRI paradigm. A Semantic stimuli consisted of pairs of words with high or low associative relationship or that were unrelated. Each of the three conditions contained an equal number of abstract and concrete word pairs (i.e., semantic class). B Trials consisted of a sequence of visual stimuli; task instructions (e.g., “Associated?”) were presented for 3000 ms followed by a word pair for 4000 ms. Participants responded with a binary button press depending on whether the word pair was judged to be related (associated) or unrelated. A duration of between 2000 and 5000 ms occurred randomly between each trial, providing a variable lag to enable event-related analysis. C The word conditions (concrete, abstract, rhyme) occurred in a repeating block sequence alternating between the concrete, abstract, and rhyme word pair conditions, each 216 s in duration with a 15-s rest period between each block

Fig. 2

Activation of the semantic network. Areas in warm tones indicate greater activation during semantic versus rhyme processing (p < 0.05, FWE-corrected). T values are represented by the color bar. Robust activation is shown, predominantly in the left hemisphere, in lateral and medial frontal cortex including inferior frontal gyrus, as well as around the temporal-parietal junction (angular/supramarginal gyri)

Fig. 3

Activation of the abstract and concrete semantic systems. Voxels displaying significant activation for concrete > rhyme are shown in warm tones. Voxels displaying significant activation for abstract > rhyme are shown in cool tones

Fig. 4

Comparison of activation in 4 semantic regions. Comparison of concrete versus abstract word pairs for four cortical areas that are considered to serve as semantic hubs or control systems. Including the AG bilaterally, the left a-MTG, and left IFG. An effect reaching statistical significance (p < 0.05, FDR corrected) was found only in the IFG, with greater activation to abstract vs. concrete word pairs. Notably, the left AG exhibited the strongest intensity of activation, though without a significant difference based on concreteness

Fig. 5

Activation of low > high association word processing. Areas in warm tones indicate greater activation during low association word pairs versus high association word pairs (p < 0.05, FWE-corrected)