A mechanistic insight into sources of error of visual working memory in multiple sclerosis
Figures

Schematic design of visual working memory (WM) paradigms.
(A) In the memory-guided localization (MGL) paradigm, participants were asked to memorize and then localize the position of the target circle following a random delay interval of 0.5, 1, 2, 4, or 8 s. Following their response, visual feedback was presented. (B) In the sequential paradigm with 3 bar (high memory load condition), a sequence of three colored bars was presented consecutively. Participants were asked to match the orientation of the probe bar to the previously presented bar with the same color. Visual feedback was displayed following their response. (C) The 1 bar paradigm (low memory load condition) has the same structure as the 3 bar paradigm except for presenting one bar instead of three.

Recall error and precision of healthy control and multiple sclerosis (MS) subtypes (relapsing-remitting [RRMS] and secondary progressive [SPMS]) in visual working memory (WM) paradigms.
(A) Recall error, (B) recall precision, and (C) reaction time as a function of distance for the memory-guided localization (MGL) paradigm. (D–F) The same as a function of delay interval. (G) Recall error, (H) recall precision, and (I) reaction time as a function of bar order in the sequential paradigms with 3 bar (left of each subplot) and 1 bar (right of each subplot). Data are represented as mean ± SEM.
-
Figure 2—source data 1
Statistical reports corresponding to Figure 2.
- https://cdn.elifesciences.org/articles/87442/elife-87442-fig2-data1-v1.xlsx

Sources of recall error in high and low memory load conditions (3 bar and 1 bar, respectively).
(A) von Mises SD (circular standard deviation of von Mises distribution), (B) Target response (probability of response around the target value), (C) swap error (probability of response around the non-target values), and (D) uniform response (probability of random response) for healthy control and multiple sclerosis (MS) subtypes in the sequential paradigms with 3 bar (left of each subplot) and 1 bar (right of each subplot). Data are represented as mean ± SEM.
-
Figure 3—source data 1
Statistical reports corresponding to Figure 3 and Figure 3—figure supplement 1.
- https://cdn.elifesciences.org/articles/87442/elife-87442-fig3-data1-v1.xlsx

Isolated effect of orientation in the high and low memory load conditions.
The nearest-neighbor analysis determined the isolated effect of orientation for healthy control, relapsing-remitting multiple sclerosis (RRMS), and secondary progressive multiple sclerosis (SPMS) in the high memory load condition (left of each subplot). The effect of orientation for the same groups in the low memory load condition (right of each subplot).

Classifying performance of visual working memory (WM) paradigms in differentiating healthy control from multiple sclerosis (MS) and MS subtypes, and MS subtypes from each other.
Receiver operating characteristic (ROC) curve demonstrated the accuracy of (A) memory-guided localization (MGL) and sequential paradigms with (B) 3 bar and (C) 1 bar in distinguishing healthy control from MS patients. The precision of these paradigms in dissociating healthy control from MS subtypes (relapsing-remitting MS [RRMS] and secondary progressive MS [SPMS]) and MS subtypes from each other is represented as the area under the curve (AUC) for (D) MGL and sequential paradigms with (E) 3 bar and (F) 1 bar.
-
Figure 4—source data 1
Statistical reports corresponding to Figure 4.
- https://cdn.elifesciences.org/articles/87442/elife-87442-fig4-data1-v1.xlsx
Tables
Demographic and clinical profiles of participants in the MGL paradigm.
HC (n = 23) | RRMS (n = 16) | SPMS (n = 25) | p | |
---|---|---|---|---|
Gender (F:M) | 13:10 | 14:2 | 17:8 | 0.12 |
Age (year) | 35.9 ± 8.34 | 37.25 ± 6.63 | 39.28 ± 5.56 | 0.25 |
Education (years) | 13.30 ± 2.74 | 13.69 ± 3.34 | 13.56 ± 3.22 | 0.86 |
Cognitive ability† (NL:MCI) | 23:0 | 14:2 | 19:6 | <0.05* |
Disease duration (years) | N/A | 8.562 ± 3.20 | 11.56 ± 3.28 | <0.02* |
EDSS | N/A | 1.28 ± 0.79 | 2.740 ± 1.23 | <0.0002* |
DMT (platform: non-platform) | N/A | 2:14 | 0:25 | 0.07 |
-
All data, except for the categorical information, are presented as mean ± standard deviation.
-
Gender, cognitive ability, and DMT were compared using the chi-square test. Age and education were compared using one-way ANOVA and Kruskal–Wallis H test, respectively. Disease duration and EDSS score were compared using Mann–Whitney U test and independent-sample t-test, respectively.
-
HC = healthy control, RRMS = relapsing-remitting multiple sclerosis, SPMS = secondary progressive multiple sclerosis, NL = normal (MoCA score ≥26), MCI = mild cognitive impairment (MoCA score = 18–25), EDSS = expanded disability status scale, DMT = disease-modifying therapy, platform treatment = interferon β-1a and glatiramer acetate, non-platform treatment = rituximab, ocrelizumab, fingolimod, dimethyl fumarate, and natalizumab, N/A = not applicable.
-
*
p<0.05.
-
†
Assessed based on the Montreal Cognitive Assessment (MoCA) test classification.
-
Table 1—source data 1
Clinical and demographic data of participants in the memory-guided localization (MGL) paradigm.
- https://cdn.elifesciences.org/articles/87442/elife-87442-table1-data1-v1.xlsx
Demographic and clinical profiles of participants in the sequential paradigms.
HC (n = 46) | RRMS (n = 39) | SPMS (n = 32) | p | |
---|---|---|---|---|
Gender (F:M) | 16:30 | 23:16 | 22:10 | <0.008* |
Age (year) | 30.5 ± 10.37 | 32.03 ± 6.72 | 39.00 ± 6.43 | <10–6* |
Education (years) | 16.95 ± 2.23 | 13.87 ± 3.41 | 13.67 ± 2.73 | <10–7* |
Cognitive ability† (NL: MCI) | 42:4 | 28:11 | 18:13 | <0.003* |
Disease duration (years) | N/A | 6.60 ± 3.84 | 9.37 ± 4.43 | <0.007* |
EDSS | N/A | 1.49 ± 1.01 | 3.86 ± 1.74 | <10–7* |
DMT (platform: non-platform) | N/A | 5:34 | 1:31 | 0.14 |
-
All data, except for the categorical information, are presented as mean ± standard deviation.
-
One MoCA value in the SPMS group is missing. Gender, cognitive ability, and DMT were compared using the chi-square test. Age and education were compared using the Kruskal–Wallis H test. Disease duration and EDSS score were compared using an independent-sample t-test and Mann–Whitney U test, respectively. Dunn’s post hoc test was performed following significant results of age and education, and the adjusted p-value following Bonferroni correction for multiple tests are reported: Age: healthy vs. RRMS: p=0.27, healthy vs. SPMS: p<10–6*, and RRMS vs. SPMS: p<0.005*. Education: healthy vs. RRMS: p<10–5*, healthy vs. SPMS: p<10–5*, and RRMS vs. SPMS: p=1.
-
HC = healthy control, RRMS = relapsing-remitting multiple sclerosis, SPMS = secondary progressive multiple sclerosis, NL = normal (MoCA score ≥26), MCI = mild cognitive impairment (MoCA score = 18–25), EDSS = expanded disability status scale, DMT = disease-modifying therapy, platform treatment = interferon β-1a and glatiramer acetate, non-platform treatment = rituximab, ocrelizumab, fingolimod, dimethyl fumarate, and natalizumab, N/A = ot applicable.
-
*
p<0.05.
-
†
Assessed based on the Montreal Cognitive Assessment (MoCA) test classification.
-
Table 2—source data 1
Clinical and demographic data of participants in the sequential paradigms.
- https://cdn.elifesciences.org/articles/87442/elife-87442-table2-data1-v1.xlsx
Additional files
-
Supplementary file 1
Hierarchical regression analysis for the MGL paradigm.
- https://cdn.elifesciences.org/articles/87442/elife-87442-supp1-v1.docx
-
Supplementary file 2
Statistical results of reaction time for the MGL paradigm.
- https://cdn.elifesciences.org/articles/87442/elife-87442-supp2-v1.docx
-
Supplementary file 3
Hierarchical regression analysis for the sequential paradigm with 3 bar, high memory load condition.
- https://cdn.elifesciences.org/articles/87442/elife-87442-supp3-v1.docx
-
Supplementary file 4
Statistical results of reaction time for the sequential paradigms (3 bar and 1 bar).
- https://cdn.elifesciences.org/articles/87442/elife-87442-supp4-v1.docx
-
Supplementary file 5
Hierarchical regression analysis for the sequential paradigm with 1 bar, low memory load condition.
- https://cdn.elifesciences.org/articles/87442/elife-87442-supp5-v1.docx
-
MDAR checklist
- https://cdn.elifesciences.org/articles/87442/elife-87442-mdarchecklist1-v1.docx
-
Source data 1
Psychophysics data of participants in the MGL paradigm.
- https://cdn.elifesciences.org/articles/87442/elife-87442-data1-v1.xlsx
-
Source data 2
Psychophysics data of participants in the sequential paradigm with 3 bar.
- https://cdn.elifesciences.org/articles/87442/elife-87442-data2-v1.xlsx
-
Source data 3
Psychophysics data of participants in the sequential paradigm with 1 bar.
- https://cdn.elifesciences.org/articles/87442/elife-87442-data3-v1.xlsx