The Effect of Cognitive Load on the Retrieval of Long-Term Memory: An fMRI Study

Abstract

One of the less well-understood aspects of memory function is the mechanism by which the brain responds to an increasing load of memory, either during encoding or retrieval. Identifying the brain structures which manage this increasing cognitive demand would enhance our knowledge of human memory. Despite numerous studies about the effect of cognitive loads on working memory processes, whether these can be applied to long-term memory processes is unclear. We asked 32 healthy young volunteers to memorize all possible details of 24 images over a 12-day period ending 2 days before the fMRI scan. The images were of 12 categories relevant to daily events, with each category including a high and a low load image. Behavioral assessments on a separate group of participants (#22) provided the average loads of the images. The participants had to retrieve these previously memorized images during the fMRI scan in 15 s, with their eyes closed. We observed seven brain structures showing the highest activation with increasing load of the retrieved images, viz. parahippocampus, cerebellum, superior lateral occipital, fusiform and lingual gyri, precuneus, and posterior cingulate gyrus. Some structures showed reduced activation when retrieving higher load images, such as the anterior cingulate, insula, and supramarginal and postcentral gyri. The findings of this study revealed that the mechanism by which a difficult-to-retrieve memory is handled is mainly by elevating the activation of the responsible brain areas and not by getting other brain regions involved, which is a help to better understand the LTM retrieval process in the human brain.

Keywords: cognitive load; functional MRI (fMRI); long-term memory; memory retrieval; visual memory.

FIGURE 1

The 24 imagesused in the study. The images are in 12 categories, with each category including a high load and a low load image. The names (used as the cue) devoted to the images (a two-word name in the Persian language) are also provided.

FIGURE 2

The schematic design of one trial of the fMRI task; white boxes are relevant to the auditory instructions, and the gray boxes show the periods the participant was taking rest or was performing an action. The same trial was repeated for the 24 images of the task.

FIGURE 3

The scores obtained for the 24 images, on their level of cognitive load (images are sorted here from 1 to 24 based on their average load). The blue circles show the scores obtained from the 22 participants of the behavioral group for each image (in most cases, a blue point accumulated the scorings of more than one participant), the red stars show the average scores, and the purple line shows the trend of the increasing load.

FIGURE 4

The graph shows the number of excellent, good, moderate, and weak responses obtained for any of the 24 images, during the retrieval phase in the MR scanner. There are 24 vertical lines corresponding to the 24 images sorted based on their average cognitive load, and on each vertical line, the number of excellent, good, moderate, and weak responses to each image are illustrated. The number of excellent responses declined with the increasing load of the images, and therefore, the number of good and moderate response increased; the number of weak responses seems unchanged.

FIGURE 5

The distribution of the retrieval duration for each of the 24 images, during the fMRI scan. The vertical axis shows the amount of time for retrieving an image by each participants (the numbers in each cell represent the number of participants). When the load increased, the number of people who took a longer time for retrieving an image increased. For example, 17 of the participants used the whole duration (15 s) for retrieving image 1, whereas all participants (#31) used the whole duration for retrieving image 24.

FIGURE 6

The brain maps obtained for (A) the high load greater than low load contrast; and (B) the high load smaller than low load contrast, using the categorical data analysis; (C) the brain areas with an association with cognitive load, in the parametric analysis; (D) the overlap of the high minus low (red) map and the parametric (blue) contrast map, which illustrates similarity of the brain areas between these two contrasts (illustrated in purple color); (E) the overlap of the high minus low (red) map and low minus high (blue) contrast map, which shows there is no similar areas between these two (no purple color voxel).

FIGURE 7

The functional connectivity of the 35 ROIs; (a) average connectivity in the “high load” condition; (b) average connectivity in the “low load” condition; (c) connectivities in the “high minus rest” condition; (d) connectivities in the “low minus rest” condition; (e) connectivities in the “high minus low” condition.