Sex, Age, and Handedness Modulate the Neural Correlates of Active Learning

Abstract

Background: Self-generation of material compared to passive learning results in mproved memory performance; this may be related to recruitment of a fronto-temporal encoding network. Using a verbal paired-associate learning fMRI task, we examined the effects of sex, age, and handedness on the neural correlates of self-generation.

Methods: Data from 174 healthy English-speaking participants (78M, 56 atypically handed; ages 19-76) were preprocessed using AFNI and FSL. Independent component analysis was conducted using GIFT (Group ICA fMRI Toolbox). Forty-one independent components were temporally sorted by task time series. Retaining correlations (r > 0.25) resulted in three task-positive ("generate") and three task-negative ("read") components. Using participants' back-projected components, we evaluated the effects of sex, handedness, and aging on activation lateralization and localization in task-relevant networks with two-sample t-tests. Further, we examined the linear relationship between sex and neuroimaging data with multiple regression, covarying for scanner, age, and handedness.

Results: Task-positive components identified using ICA revealed a fronto-parietal network involved with self-generation, while task-negative components reflecting passive reading showed temporo-occipital involvement. Compared to older adults, younger adults exhibited greater task-positive involvement of the left inferior frontal gyrus and insula, whereas older adults exhibited reduced prefrontal lateralization. Greater involvement of the left angular gyrus in task-positive encoding networks among right-handed individuals suggests the reliance on left dominant semantic processing areas may be modulated by handedness. Sex effects on task-related encoding networks while controlling for age and handedness suggest increased right hemisphere recruitment among males compared to females, specifically in the paracentral lobe during self-generation and the suparmarginal gyrus during passive reading.

Implications: Identified neuroimaging differences suggest that sex, age, and handedness are factors in the differential recruitment of encoding network regions for both passive and active learning.

Keywords: age; associate learning; fMRI; handedness; sex; verbal memory.

FIGURE 1

A schematic of the two parts of the verbal paired-associate learning task consisting of (A) an in-scanner encoding task, and a (B) post-fMRI recognition test. During the encoding task, word pairs are presented on the screen for 6 s, and participants are instructed to read (or generate) the second word of the word pair aloud. The STOP screen instructs participants to stop talking while three image volumes are acquired for a total of 6 s. This continues for all 60 word-pair stimuli, for a total of 12 min. After scanning, participants conduct a recognition post-test in a nearby facility testing room. Participants are presented with three words: one target word they had been exposed to during the in-scanner encoding task, and two foils, and participants are instructed to indicate which of the three words they recognize from the task.

FIGURE 2

Hemodynamics Unrelated to Sounds of Hardware (HUSH) Image Acquisition. A schematic of one stimulus presentation and image acquisition period (total of 12 s) using a clustered-sparse temporal acquisition, HUSH. Stimulus is presented for the first 6 s, followed by acquisition of three image volumes with a TR of 2 s. This is repeated for all 60 word-pair stimuli for 12 min and a total of 180 image volumes (HRF, Hemodynamic response function).

FIGURE 3

A schematic of the analysis pipeline. HUSH volumes were split into the 3 volumes acquired, and subsequent preprocessing and analysis in GIFT (PCA/ICA) was applied to volumes separately. After sorting components using the binary task time series, regionally similar components were identified across the three analyses and the volume with the highest correlated to the task time series was retained for further analyses. (Superscripts correspond to sections in the methods that further describe that stage in the analysis. HUSH, Hemodynamics Unrelated to Sounds of Hardware; FLIRT, FMRIB’s linear image registration tool; FWHM, full width at half maximum; PCA, principle component analysis; ICA, independent component analysis; SMs, spatial maps).

FIGURE 4

Task-positive and task-negative components (correlated with the generate and read conditions |r| > 0.25). Images are presented in neurological orientation (R = R). Group average independent components positively correlated (r > 0.25), (A–C), and negatively correlated (r < –0.25), (E–G) with the task time course, along with composite images for task-positive (D) and task-negative (H) components presented below. Components span the following brain regions: (A) bilateral occipital; (B) bilateral (left-lateralized) inferior frontal gyrus and precentral gyrus; (C) anterior insula and superior temporal gyrus; (D) composite image of task-positive components (A–C); (E) bilateral superior temporal gyrus; (F) bilateral precuneus and posterior cingulate cortex; (G) posterior cingulate cortex and culmen; (H) composite image of task-negative components (E–G).

FIGURE 5

Relationships between sex and task-related components: Two-sample t-tests between males and females in task-positive components (A,B) and task-negative components (C,D) reveal males recruit additional brain areas during active encoding compared to females: (A) right postcentral gyrus, (B) left insula; and during passive encoding: (C) left supramarginal gyrus, (D) right superior temporal gyrus (F, females; M, males; p < 0.05 corrected).

FIGURE 6

Relationships between handedness and task-related components. Two-sample t-tests reveal differences between atypically handed and right-handed individuals in task-positive components (A–C) and task-negative components (D). In terms of spatial extent of task-related activation, atypical-handers and right-handed individuals recruited additional regions during active encoding: (A) left cuneus, (B) posterior cingulate cortex, (C) left angular gyrus; and during passive encoding: (D) left insula. Orange depicts areas where atypical-handers task-related extent was greater than right-handed individuals, and blue depicts areas where right-handed individuals task-related extent was greater than atypical-handers (A, atypical-handed; R, right-handed; p < 0.05 corrected).

FIGURE 7

Relationships between age and task-related components. Two-sample t-tests reveal differences between younger (<50 years old) and older (≥50 years old) adults in task-positive components (A–D) and task-negative components (E,F). Recruitment among young adults seemed more widespread compared to older adults during self-generation, with greater recruitment across (A) bilateral insula, anterior and middle cingulate cortices, and left inferior frontal gyrus, (C) bilateral middle occipital gyrus, (D) left inferior frontal gyrus and left insula. During self-generation, older adults showed greater recruitment compared to younger adults in panel (B) left middle frontal gyrus. During reading, younger adults showed greater activity compared to older adults across: (E) right precuneus, (F) bilateral precuneus and PCC, (G) right inferior parietal lobe, and (H) bilateral PCC [Y, younger adults (<50 years old), O, older adults (≥50 years old); p < 0.05 corrected].

FIGURE 8

Relationship between sex and task-related components, controlling for scanner, age, and handedness. Males show greater task-positive network extent when controlling for scanner, age, and handedness in panel (A) right paracentral lobe and (C) left insula during self-generation and greater task-negative network extent in panel (D) right supramarginal gyrus/STG during passive reading. Females show greater task-positive recruitment during self-generation of panel (B) right middle temporal gyrus (F, females; M, males; p < 0.01 corrected).

FIGURE 9

Linear mixed-effects model revealed (A) an interaction between sex and age while controlling for scanner and handedness in the left supramarginal gyrus in a task-positive component. Correlations between subjects’ mean z-score for the left supramarginal gyrus ROI and age suggests that (B) as age increases, males show increased recruitment of left SMG during passive reading while females show a decline in BOLD signal intensity as age increases (p < 0.01 corrected).